config.json

{
  "architectures": [
    "GromaModel"
  ],
  "nms_thres": 0.6,
  "box_score_thres": 0.15,
  "max_region_num": 100,
  "llm_cfg": {
    "_name_or_path": "vicuna-7b-v1.5",
    "architectures": [
      "LlamaForCausalLM"
    ],
    "bos_token_id": 1,
    "eos_token_id": 2,
    "hidden_act": "silu",
    "hidden_size": 4096,
    "initializer_range": 0.02,
    "intermediate_size": 11008,
    "max_position_embeddings": 4096,
    "model_type": "llama",
    "num_attention_heads": 32,
    "num_hidden_layers": 32,
    "num_key_value_heads": 32,
    "pad_token_id": 0,
    "pretraining_tp": 1,
    "rms_norm_eps": 1e-05,
    "rope_scaling": null,
    "tie_word_embeddings": false,
    "torch_dtype": "float16",
    "transformers_version": "4.32.0",
    "use_cache": true,
    "vocab_size": 32000
  },
  "model_type": "groma",
  "num_new_token": 114,
  "perceiver_cfg": {
    "architectures": [
      "CustomDDETRModel"
    ],
    "ddetr_cfg": {
      "auxiliary_loss": true,
      "class_cost": 2,
      "cls_loss_coefficient": 2,
      "id2label": {
        "0": "LABEL_0"
      },
      "label2id": {
        "LABEL_0": 0
      },
      "model_type": "deformable_detr",
      "num_feature_levels": 1,
      "two_stage": true,
      "with_box_refine": true
    },
    "model_type": "ddetr",
    "torch_dtype": "float32",
    "transformers_version": "4.32.0",
    "vis_encoder_cfg": {
      "architectures": [
        "Dinov2Model"
      ],
      "hidden_size": 1024,
      "image_size": 518,
      "model_type": "dinov2",
      "num_attention_heads": 16,
      "num_hidden_layers": 24,
      "patch_size": 14,
      "torch_dtype": "float32"
    },
    "vis_output_layer": -2,
    "zs_weight_path": null
  },
  "torch_dtype": "float32",
  "transformers_version": "4.32.0",
  "vocab_size": 32114
}