diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,3933 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.75,
+  "eval_steps": 500,
+  "global_step": 300,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "completion_length": 135.25000762939453,
+      "epoch": 0.0025,
+      "grad_norm": 4.3187419389243376,
+      "kl": 0.0,
+      "learning_rate": 9.9875e-07,
+      "loss": -0.0,
+      "reward": 0.1666666716337204,
+      "reward_std": 0.23899273574352264,
+      "rewards/accuracy_reward": 0.125,
+      "rewards/format_reward": 0.0416666679084301,
+      "step": 1
+    },
+    {
+      "completion_length": 117.58333587646484,
+      "epoch": 0.005,
+      "grad_norm": 4.103500404066666,
+      "kl": 0.00040721893310546875,
+      "learning_rate": 9.975e-07,
+      "loss": 0.0,
+      "reward": 0.125,
+      "reward_std": 0.1369306445121765,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/format_reward": 0.125,
+      "step": 2
+    },
+    {
+      "completion_length": 118.87500381469727,
+      "epoch": 0.0075,
+      "grad_norm": 6.369368643333465,
+      "kl": 0.0008068084716796875,
+      "learning_rate": 9.9625e-07,
+      "loss": 0.0,
+      "reward": 0.2083333358168602,
+      "reward_std": 0.4352857172489166,
+      "rewards/accuracy_reward": 0.0833333358168602,
+      "rewards/format_reward": 0.1250000037252903,
+      "step": 3
+    },
+    {
+      "completion_length": 155.0833396911621,
+      "epoch": 0.01,
+      "grad_norm": 20.444652962817383,
+      "kl": 0.001117706298828125,
+      "learning_rate": 9.95e-07,
+      "loss": 0.0,
+      "reward": 0.2500000111758709,
+      "reward_std": 0.4403265342116356,
+      "rewards/accuracy_reward": 0.125,
+      "rewards/format_reward": 0.1250000037252903,
+      "step": 4
+    },
+    {
+      "completion_length": 106.08333587646484,
+      "epoch": 0.0125,
+      "grad_norm": 7.9727528386684146,
+      "kl": 0.0011138916015625,
+      "learning_rate": 9.9375e-07,
+      "loss": 0.0,
+      "reward": 0.3333333358168602,
+      "reward_std": 0.6224541813135147,
+      "rewards/accuracy_reward": 0.2083333358168602,
+      "rewards/format_reward": 0.125,
+      "step": 5
+    },
+    {
+      "completion_length": 137.20833587646484,
+      "epoch": 0.015,
+      "grad_norm": 37.719903416343634,
+      "kl": 0.00411224365234375,
+      "learning_rate": 9.925e-07,
+      "loss": 0.0002,
+      "reward": 0.8333333432674408,
+      "reward_std": 0.8705713450908661,
+      "rewards/accuracy_reward": 0.4166666865348816,
+      "rewards/format_reward": 0.4166666865348816,
+      "step": 6
+    },
+    {
+      "completion_length": 94.45833587646484,
+      "epoch": 0.0175,
+      "grad_norm": 15.811856429811904,
+      "kl": 0.005523681640625,
+      "learning_rate": 9.912499999999998e-07,
+      "loss": 0.0002,
+      "reward": 0.4583333432674408,
+      "reward_std": 0.5918936878442764,
+      "rewards/accuracy_reward": 0.1666666679084301,
+      "rewards/format_reward": 0.291666679084301,
+      "step": 7
+    },
+    {
+      "completion_length": 115.79166793823242,
+      "epoch": 0.02,
+      "grad_norm": 15.46124375226576,
+      "kl": 0.012359619140625,
+      "learning_rate": 9.9e-07,
+      "loss": 0.0005,
+      "reward": 0.4791666865348816,
+      "reward_std": 0.7572853863239288,
+      "rewards/accuracy_reward": 0.2708333432674408,
+      "rewards/format_reward": 0.2083333358168602,
+      "step": 8
+    },
+    {
+      "completion_length": 111.66667175292969,
+      "epoch": 0.0225,
+      "grad_norm": 26.742629502915644,
+      "kl": 0.008148193359375,
+      "learning_rate": 9.8875e-07,
+      "loss": 0.0003,
+      "reward": 0.791666716337204,
+      "reward_std": 0.7401774525642395,
+      "rewards/accuracy_reward": 0.3750000223517418,
+      "rewards/format_reward": 0.4166666865348816,
+      "step": 9
+    },
+    {
+      "completion_length": 86.37500381469727,
+      "epoch": 0.025,
+      "grad_norm": 8.114967888040798,
+      "kl": 0.0157470703125,
+      "learning_rate": 9.875e-07,
+      "loss": 0.0006,
+      "reward": 0.7083333432674408,
+      "reward_std": 0.6621108055114746,
+      "rewards/accuracy_reward": 0.2916666716337204,
+      "rewards/format_reward": 0.4166666865348816,
+      "step": 10
+    },
+    {
+      "completion_length": 102.875,
+      "epoch": 0.0275,
+      "grad_norm": 9.33610011715284,
+      "kl": 0.01385498046875,
+      "learning_rate": 9.862499999999999e-07,
+      "loss": 0.0006,
+      "reward": 1.0833333730697632,
+      "reward_std": 0.6234788298606873,
+      "rewards/accuracy_reward": 0.291666679084301,
+      "rewards/format_reward": 0.7916666865348816,
+      "step": 11
+    },
+    {
+      "completion_length": 72.04166793823242,
+      "epoch": 0.03,
+      "grad_norm": 28.44553169004122,
+      "kl": 0.056884765625,
+      "learning_rate": 9.849999999999999e-07,
+      "loss": 0.0023,
+      "reward": 1.3333333730697632,
+      "reward_std": 0.7672148048877716,
+      "rewards/accuracy_reward": 0.5416666865348816,
+      "rewards/format_reward": 0.7916666865348816,
+      "step": 12
+    },
+    {
+      "completion_length": 94.25000381469727,
+      "epoch": 0.0325,
+      "grad_norm": 7.7534755603476295,
+      "kl": 0.03045654296875,
+      "learning_rate": 9.8375e-07,
+      "loss": 0.0012,
+      "reward": 0.9166666865348816,
+      "reward_std": 0.8366837203502655,
+      "rewards/accuracy_reward": 0.3750000149011612,
+      "rewards/format_reward": 0.5416666865348816,
+      "step": 13
+    },
+    {
+      "completion_length": 85.45833587646484,
+      "epoch": 0.035,
+      "grad_norm": 8.954286118853258,
+      "kl": 0.0286865234375,
+      "learning_rate": 9.825e-07,
+      "loss": 0.0011,
+      "reward": 1.3333333730697632,
+      "reward_std": 0.49133746325969696,
+      "rewards/accuracy_reward": 0.583333358168602,
+      "rewards/format_reward": 0.75,
+      "step": 14
+    },
+    {
+      "completion_length": 84.75000381469727,
+      "epoch": 0.0375,
+      "grad_norm": 6.787317409547862,
+      "kl": 0.0479736328125,
+      "learning_rate": 9.8125e-07,
+      "loss": 0.0019,
+      "reward": 1.2916666865348816,
+      "reward_std": 0.5646616071462631,
+      "rewards/accuracy_reward": 0.4166666865348816,
+      "rewards/format_reward": 0.8750000298023224,
+      "step": 15
+    },
+    {
+      "completion_length": 79.00000381469727,
+      "epoch": 0.04,
+      "grad_norm": 12.000529715665422,
+      "kl": 0.0469970703125,
+      "learning_rate": 9.8e-07,
+      "loss": 0.0019,
+      "reward": 1.2083333730697632,
+      "reward_std": 0.5714030414819717,
+      "rewards/accuracy_reward": 0.4166666716337204,
+      "rewards/format_reward": 0.7916666865348816,
+      "step": 16
+    },
+    {
+      "completion_length": 72.58333587646484,
+      "epoch": 0.0425,
+      "grad_norm": 6.631858893642277,
+      "kl": 0.0296630859375,
+      "learning_rate": 9.7875e-07,
+      "loss": 0.0012,
+      "reward": 1.125,
+      "reward_std": 0.42718589305877686,
+      "rewards/accuracy_reward": 0.3333333432674408,
+      "rewards/format_reward": 0.7916666865348816,
+      "step": 17
+    },
+    {
+      "completion_length": 80.20833587646484,
+      "epoch": 0.045,
+      "grad_norm": 7.461874304434306,
+      "kl": 0.033447265625,
+      "learning_rate": 9.775e-07,
+      "loss": 0.0013,
+      "reward": 1.2500000298023224,
+      "reward_std": 0.3332235887646675,
+      "rewards/accuracy_reward": 0.3333333544433117,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 18
+    },
+    {
+      "completion_length": 77.45833587646484,
+      "epoch": 0.0475,
+      "grad_norm": 15.843571456536381,
+      "kl": 0.0758056640625,
+      "learning_rate": 9.7625e-07,
+      "loss": 0.003,
+      "reward": 1.2083333730697632,
+      "reward_std": 0.6795306503772736,
+      "rewards/accuracy_reward": 0.3750000149011612,
+      "rewards/format_reward": 0.8333333730697632,
+      "step": 19
+    },
+    {
+      "completion_length": 74.16666793823242,
+      "epoch": 0.05,
+      "grad_norm": 10.26314464339147,
+      "kl": 0.0343017578125,
+      "learning_rate": 9.75e-07,
+      "loss": 0.0014,
+      "reward": 1.3333333730697632,
+      "reward_std": 0.46232303977012634,
+      "rewards/accuracy_reward": 0.4166666865348816,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 20
+    },
+    {
+      "completion_length": 75.70833587646484,
+      "epoch": 0.0525,
+      "grad_norm": 74.55790491289811,
+      "kl": 0.0345458984375,
+      "learning_rate": 9.7375e-07,
+      "loss": 0.0014,
+      "reward": 1.3333333730697632,
+      "reward_std": 0.6360993385314941,
+      "rewards/accuracy_reward": 0.4166666716337204,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 21
+    },
+    {
+      "completion_length": 57.45833396911621,
+      "epoch": 0.055,
+      "grad_norm": 8.310598933918627,
+      "kl": 0.148681640625,
+      "learning_rate": 9.725e-07,
+      "loss": 0.006,
+      "reward": 1.4166667461395264,
+      "reward_std": 0.3680921941995621,
+      "rewards/accuracy_reward": 0.5000000149011612,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 22
+    },
+    {
+      "completion_length": 54.62500190734863,
+      "epoch": 0.0575,
+      "grad_norm": 17.70309158653018,
+      "kl": 0.072021484375,
+      "learning_rate": 9.712499999999998e-07,
+      "loss": 0.0029,
+      "reward": 1.5000000596046448,
+      "reward_std": 0.3680921494960785,
+      "rewards/accuracy_reward": 0.5000000298023224,
+      "rewards/format_reward": 1.0,
+      "step": 23
+    },
+    {
+      "completion_length": 60.416669845581055,
+      "epoch": 0.06,
+      "grad_norm": 8.426296910342877,
+      "kl": 0.052734375,
+      "learning_rate": 9.7e-07,
+      "loss": 0.0021,
+      "reward": 1.6875000596046448,
+      "reward_std": 0.3540605455636978,
+      "rewards/accuracy_reward": 0.6875000298023224,
+      "rewards/format_reward": 1.0,
+      "step": 24
+    },
+    {
+      "completion_length": 65.70833587646484,
+      "epoch": 0.0625,
+      "grad_norm": 11.31257030642023,
+      "kl": 0.05615234375,
+      "learning_rate": 9.6875e-07,
+      "loss": 0.0022,
+      "reward": 1.4166666865348816,
+      "reward_std": 0.5373477190732956,
+      "rewards/accuracy_reward": 0.4583333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 25
+    },
+    {
+      "completion_length": 59.625003814697266,
+      "epoch": 0.065,
+      "grad_norm": 52.11448679039176,
+      "kl": 0.0579833984375,
+      "learning_rate": 9.675e-07,
+      "loss": 0.0023,
+      "reward": 1.0833333432674408,
+      "reward_std": 0.23899271339178085,
+      "rewards/accuracy_reward": 0.125,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 26
+    },
+    {
+      "completion_length": 56.29166793823242,
+      "epoch": 0.0675,
+      "grad_norm": 6.130006627510818,
+      "kl": 0.099365234375,
+      "learning_rate": 9.6625e-07,
+      "loss": 0.004,
+      "reward": 1.4166666865348816,
+      "reward_std": 0.3251238167285919,
+      "rewards/accuracy_reward": 0.4583333358168602,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 27
+    },
+    {
+      "completion_length": 50.75000190734863,
+      "epoch": 0.07,
+      "grad_norm": 9.439568241641684,
+      "kl": 0.0650634765625,
+      "learning_rate": 9.649999999999999e-07,
+      "loss": 0.0026,
+      "reward": 1.5416667461395264,
+      "reward_std": 0.23116152733564377,
+      "rewards/accuracy_reward": 0.5416666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 28
+    },
+    {
+      "completion_length": 44.583335876464844,
+      "epoch": 0.0725,
+      "grad_norm": 4.611117210184683,
+      "kl": 0.0576171875,
+      "learning_rate": 9.637499999999999e-07,
+      "loss": 0.0023,
+      "reward": 1.5416666865348816,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.5416666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 29
+    },
+    {
+      "completion_length": 73.00000381469727,
+      "epoch": 0.075,
+      "grad_norm": 0.9935764144637615,
+      "kl": 0.0751953125,
+      "learning_rate": 9.624999999999999e-07,
+      "loss": 0.003,
+      "reward": 1.5,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 0.5416666865348816,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 30
+    },
+    {
+      "completion_length": 53.33333396911621,
+      "epoch": 0.0775,
+      "grad_norm": 10.787013479194396,
+      "kl": 0.07568359375,
+      "learning_rate": 9.6125e-07,
+      "loss": 0.003,
+      "reward": 1.5416666865348816,
+      "reward_std": 0.39512956142425537,
+      "rewards/accuracy_reward": 0.5416666716337204,
+      "rewards/format_reward": 1.0,
+      "step": 31
+    },
+    {
+      "completion_length": 54.79166793823242,
+      "epoch": 0.08,
+      "grad_norm": 15.902958221650678,
+      "kl": 0.055908203125,
+      "learning_rate": 9.6e-07,
+      "loss": 0.0022,
+      "reward": 1.4583333730697632,
+      "reward_std": 0.2601759508252144,
+      "rewards/accuracy_reward": 0.5000000298023224,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 32
+    },
+    {
+      "completion_length": 54.83333396911621,
+      "epoch": 0.0825,
+      "grad_norm": 9.68655515316935,
+      "kl": 0.05126953125,
+      "learning_rate": 9.5875e-07,
+      "loss": 0.002,
+      "reward": 1.5416667461395264,
+      "reward_std": 0.3061862215399742,
+      "rewards/accuracy_reward": 0.5833333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 33
+    },
+    {
+      "completion_length": 46.58333396911621,
+      "epoch": 0.085,
+      "grad_norm": 7.9623502770430195,
+      "kl": 0.05859375,
+      "learning_rate": 9.575e-07,
+      "loss": 0.0023,
+      "reward": 1.7083333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.7083333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 34
+    },
+    {
+      "completion_length": 53.16666793823242,
+      "epoch": 0.0875,
+      "grad_norm": 8.168996294540337,
+      "kl": 0.08251953125,
+      "learning_rate": 9.5625e-07,
+      "loss": 0.0033,
+      "reward": 1.5416666865348816,
+      "reward_std": 0.26603007316589355,
+      "rewards/accuracy_reward": 0.5416666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 35
+    },
+    {
+      "completion_length": 59.333335876464844,
+      "epoch": 0.09,
+      "grad_norm": 8.505271344801097,
+      "kl": 0.053466796875,
+      "learning_rate": 9.55e-07,
+      "loss": 0.0021,
+      "reward": 1.3333333730697632,
+      "reward_std": 0.3332235887646675,
+      "rewards/accuracy_reward": 0.3333333544433117,
+      "rewards/format_reward": 1.0,
+      "step": 36
+    },
+    {
+      "completion_length": 65.08333587646484,
+      "epoch": 0.0925,
+      "grad_norm": 7.953549431636307,
+      "kl": 0.0482177734375,
+      "learning_rate": 9.5375e-07,
+      "loss": 0.0019,
+      "reward": 1.7083333730697632,
+      "reward_std": 0.37592338025569916,
+      "rewards/accuracy_reward": 0.75,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 37
+    },
+    {
+      "completion_length": 57.91666793823242,
+      "epoch": 0.095,
+      "grad_norm": 6.172142062081599,
+      "kl": 0.059814453125,
+      "learning_rate": 9.525e-07,
+      "loss": 0.0024,
+      "reward": 1.5833333730697632,
+      "reward_std": 0.23899271339178085,
+      "rewards/accuracy_reward": 0.5833333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 38
+    },
+    {
+      "completion_length": 60.166669845581055,
+      "epoch": 0.0975,
+      "grad_norm": 63.318354629363114,
+      "kl": 0.054931640625,
+      "learning_rate": 9.5125e-07,
+      "loss": 0.0022,
+      "reward": 1.8750000596046448,
+      "reward_std": 0.23116153478622437,
+      "rewards/accuracy_reward": 0.8750000298023224,
+      "rewards/format_reward": 1.0,
+      "step": 39
+    },
+    {
+      "completion_length": 51.41666793823242,
+      "epoch": 0.1,
+      "grad_norm": 17.054147779969167,
+      "kl": 0.0771484375,
+      "learning_rate": 9.499999999999999e-07,
+      "loss": 0.0031,
+      "reward": 1.2500000596046448,
+      "reward_std": 0.4431168884038925,
+      "rewards/accuracy_reward": 0.3333333432674408,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 40
+    },
+    {
+      "completion_length": 57.625,
+      "epoch": 0.1025,
+      "grad_norm": 7.977114237328186,
+      "kl": 0.043212890625,
+      "learning_rate": 9.487499999999999e-07,
+      "loss": 0.0017,
+      "reward": 1.7083333730697632,
+      "reward_std": 0.3410547822713852,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 41
+    },
+    {
+      "completion_length": 58.375003814697266,
+      "epoch": 0.105,
+      "grad_norm": 9.771980963826733,
+      "kl": 0.056884765625,
+      "learning_rate": 9.474999999999999e-07,
+      "loss": 0.0023,
+      "reward": 1.7083333730697632,
+      "reward_std": 0.39512956142425537,
+      "rewards/accuracy_reward": 0.7083333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 42
+    },
+    {
+      "completion_length": 67.00000381469727,
+      "epoch": 0.1075,
+      "grad_norm": 6.345190754294424,
+      "kl": 0.056884765625,
+      "learning_rate": 9.462499999999999e-07,
+      "loss": 0.0023,
+      "reward": 1.6250000596046448,
+      "reward_std": 0.3892754018306732,
+      "rewards/accuracy_reward": 0.7083333730697632,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 43
+    },
+    {
+      "completion_length": 68.75000381469727,
+      "epoch": 0.11,
+      "grad_norm": 7.078370979919857,
+      "kl": 0.0556640625,
+      "learning_rate": 9.45e-07,
+      "loss": 0.0022,
+      "reward": 1.7500000596046448,
+      "reward_std": 0.3332235887646675,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 44
+    },
+    {
+      "completion_length": 55.125,
+      "epoch": 0.1125,
+      "grad_norm": 12.46450565501957,
+      "kl": 0.05859375,
+      "learning_rate": 9.4375e-07,
+      "loss": 0.0023,
+      "reward": 1.8750000596046448,
+      "reward_std": 0.23116152733564377,
+      "rewards/accuracy_reward": 0.8750000298023224,
+      "rewards/format_reward": 1.0,
+      "step": 45
+    },
+    {
+      "completion_length": 66.45833587646484,
+      "epoch": 0.115,
+      "grad_norm": 9.74330758017454,
+      "kl": 0.151123046875,
+      "learning_rate": 9.425e-07,
+      "loss": 0.006,
+      "reward": 1.4583333730697632,
+      "reward_std": 0.5133170187473297,
+      "rewards/accuracy_reward": 0.5833333432674408,
+      "rewards/format_reward": 0.8750000298023224,
+      "step": 46
+    },
+    {
+      "completion_length": 59.45833396911621,
+      "epoch": 0.1175,
+      "grad_norm": 3.2665063407104244,
+      "kl": 0.0550537109375,
+      "learning_rate": 9.4125e-07,
+      "loss": 0.0022,
+      "reward": 1.5416666865348816,
+      "reward_std": 0.10206207633018494,
+      "rewards/accuracy_reward": 0.5833333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 47
+    },
+    {
+      "completion_length": 66.08333396911621,
+      "epoch": 0.12,
+      "grad_norm": 7.049241940629922,
+      "kl": 0.078369140625,
+      "learning_rate": 9.399999999999999e-07,
+      "loss": 0.0031,
+      "reward": 1.5416666865348816,
+      "reward_std": 0.33826446533203125,
+      "rewards/accuracy_reward": 0.5833333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 48
+    },
+    {
+      "completion_length": 54.58333396911621,
+      "epoch": 0.1225,
+      "grad_norm": 19.39524628115419,
+      "kl": 0.0606689453125,
+      "learning_rate": 9.387499999999999e-07,
+      "loss": 0.0024,
+      "reward": 1.6666667461395264,
+      "reward_std": 0.4779854714870453,
+      "rewards/accuracy_reward": 0.7083333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 49
+    },
+    {
+      "completion_length": 49.79166793823242,
+      "epoch": 0.125,
+      "grad_norm": 6.231498983069941,
+      "kl": 0.05029296875,
+      "learning_rate": 9.374999999999999e-07,
+      "loss": 0.002,
+      "reward": 1.7916666865348816,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 50
+    },
+    {
+      "completion_length": 68.66666793823242,
+      "epoch": 0.1275,
+      "grad_norm": 22.55936033645877,
+      "kl": 0.0743408203125,
+      "learning_rate": 9.3625e-07,
+      "loss": 0.003,
+      "reward": 1.5625000596046448,
+      "reward_std": 0.5052542686462402,
+      "rewards/accuracy_reward": 0.6041666865348816,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 51
+    },
+    {
+      "completion_length": 66.20833587646484,
+      "epoch": 0.13,
+      "grad_norm": 8.735323797284012,
+      "kl": 0.091796875,
+      "learning_rate": 9.35e-07,
+      "loss": 0.0037,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206207633018494,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 52
+    },
+    {
+      "completion_length": 65.29166793823242,
+      "epoch": 0.1325,
+      "grad_norm": 27.60906731867457,
+      "kl": 0.0546875,
+      "learning_rate": 9.3375e-07,
+      "loss": 0.0022,
+      "reward": 1.4166667461395264,
+      "reward_std": 0.3872983753681183,
+      "rewards/accuracy_reward": 0.416666679084301,
+      "rewards/format_reward": 1.0,
+      "step": 53
+    },
+    {
+      "completion_length": 63.583335876464844,
+      "epoch": 0.135,
+      "grad_norm": 8.445620379510913,
+      "kl": 0.14404296875,
+      "learning_rate": 9.325e-07,
+      "loss": 0.0058,
+      "reward": 1.7916667461395264,
+      "reward_std": 0.341054804623127,
+      "rewards/accuracy_reward": 0.8750000298023224,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 54
+    },
+    {
+      "completion_length": 77.79166793823242,
+      "epoch": 0.1375,
+      "grad_norm": 4.846653160352893,
+      "kl": 0.0599365234375,
+      "learning_rate": 9.3125e-07,
+      "loss": 0.0024,
+      "reward": 1.7083333730697632,
+      "reward_std": 0.306186206638813,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 0.8750000298023224,
+      "step": 55
+    },
+    {
+      "completion_length": 66.95833587646484,
+      "epoch": 0.14,
+      "grad_norm": 8.841402167892682,
+      "kl": 0.0660400390625,
+      "learning_rate": 9.3e-07,
+      "loss": 0.0026,
+      "reward": 1.5416666865348816,
+      "reward_std": 0.48936043679714203,
+      "rewards/accuracy_reward": 0.583333358168602,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 56
+    },
+    {
+      "completion_length": 69.62500381469727,
+      "epoch": 0.1425,
+      "grad_norm": 7.967652723826169,
+      "kl": 0.046142578125,
+      "learning_rate": 9.287499999999999e-07,
+      "loss": 0.0018,
+      "reward": 1.4166666865348816,
+      "reward_std": 0.3971066102385521,
+      "rewards/accuracy_reward": 0.458333358168602,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 57
+    },
+    {
+      "completion_length": 63.375003814697266,
+      "epoch": 0.145,
+      "grad_norm": 7.358323587086431,
+      "kl": 0.0455322265625,
+      "learning_rate": 9.274999999999999e-07,
+      "loss": 0.0018,
+      "reward": 1.6250000596046448,
+      "reward_std": 0.42718592286109924,
+      "rewards/accuracy_reward": 0.6666666865348816,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 58
+    },
+    {
+      "completion_length": 60.875,
+      "epoch": 0.1475,
+      "grad_norm": 15.192207461658905,
+      "kl": 0.186279296875,
+      "learning_rate": 9.2625e-07,
+      "loss": 0.0075,
+      "reward": 1.6458333730697632,
+      "reward_std": 0.1801304817199707,
+      "rewards/accuracy_reward": 0.6875,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 59
+    },
+    {
+      "completion_length": 69.70833587646484,
+      "epoch": 0.15,
+      "grad_norm": 11.20470960960183,
+      "kl": 0.059814453125,
+      "learning_rate": 9.25e-07,
+      "loss": 0.0024,
+      "reward": 1.6250000596046448,
+      "reward_std": 0.43528565764427185,
+      "rewards/accuracy_reward": 0.7500000298023224,
+      "rewards/format_reward": 0.875,
+      "step": 60
+    },
+    {
+      "completion_length": 58.00000190734863,
+      "epoch": 0.1525,
+      "grad_norm": 6.997738487865672,
+      "kl": 0.0654296875,
+      "learning_rate": 9.237499999999999e-07,
+      "loss": 0.0026,
+      "reward": 1.75,
+      "reward_std": 0.3680921643972397,
+      "rewards/accuracy_reward": 0.75,
+      "rewards/format_reward": 1.0,
+      "step": 61
+    },
+    {
+      "completion_length": 61.41666793823242,
+      "epoch": 0.155,
+      "grad_norm": 16.323714568650235,
+      "kl": 0.057861328125,
+      "learning_rate": 9.225e-07,
+      "loss": 0.0023,
+      "reward": 1.166666716337204,
+      "reward_std": 0.6123371124267578,
+      "rewards/accuracy_reward": 0.291666679084301,
+      "rewards/format_reward": 0.875,
+      "step": 62
+    },
+    {
+      "completion_length": 72.16666793823242,
+      "epoch": 0.1575,
+      "grad_norm": 8.688472666286163,
+      "kl": 0.0504150390625,
+      "learning_rate": 9.2125e-07,
+      "loss": 0.002,
+      "reward": 1.7083333730697632,
+      "reward_std": 0.23116152733564377,
+      "rewards/accuracy_reward": 0.7083333730697632,
+      "rewards/format_reward": 1.0,
+      "step": 63
+    },
+    {
+      "completion_length": 67.58333396911621,
+      "epoch": 0.16,
+      "grad_norm": 0.3043968655675616,
+      "kl": 0.0498046875,
+      "learning_rate": 9.2e-07,
+      "loss": 0.002,
+      "reward": 1.5,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 0.5,
+      "rewards/format_reward": 1.0,
+      "step": 64
+    },
+    {
+      "completion_length": 63.12500190734863,
+      "epoch": 0.1625,
+      "grad_norm": 5.6487717317467725,
+      "kl": 0.0587158203125,
+      "learning_rate": 9.187499999999999e-07,
+      "loss": 0.0024,
+      "reward": 1.6666666865348816,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 0.6666666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 65
+    },
+    {
+      "completion_length": 59.875,
+      "epoch": 0.165,
+      "grad_norm": 11.66756846162878,
+      "kl": 0.069091796875,
+      "learning_rate": 9.174999999999999e-07,
+      "loss": 0.0028,
+      "reward": 1.5000000596046448,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 0.5000000298023224,
+      "rewards/format_reward": 1.0,
+      "step": 66
+    },
+    {
+      "completion_length": 73.125,
+      "epoch": 0.1675,
+      "grad_norm": 9.079921508300968,
+      "kl": 0.0567626953125,
+      "learning_rate": 9.1625e-07,
+      "loss": 0.0023,
+      "reward": 1.5625,
+      "reward_std": 0.2249610871076584,
+      "rewards/accuracy_reward": 0.5625,
+      "rewards/format_reward": 1.0,
+      "step": 67
+    },
+    {
+      "completion_length": 57.41666793823242,
+      "epoch": 0.17,
+      "grad_norm": 4.604079232678911,
+      "kl": 0.06005859375,
+      "learning_rate": 9.15e-07,
+      "loss": 0.0024,
+      "reward": 1.875,
+      "reward_std": 0.1369306445121765,
+      "rewards/accuracy_reward": 0.875,
+      "rewards/format_reward": 1.0,
+      "step": 68
+    },
+    {
+      "completion_length": 68.70833587646484,
+      "epoch": 0.1725,
+      "grad_norm": 24.43050848592979,
+      "kl": 0.0577392578125,
+      "learning_rate": 9.137499999999999e-07,
+      "loss": 0.0023,
+      "reward": 1.3750000596046448,
+      "reward_std": 0.48936043679714203,
+      "rewards/accuracy_reward": 0.416666679084301,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 69
+    },
+    {
+      "completion_length": 68.25000381469727,
+      "epoch": 0.175,
+      "grad_norm": 7.065846830514221,
+      "kl": 0.0543212890625,
+      "learning_rate": 9.124999999999999e-07,
+      "loss": 0.0022,
+      "reward": 1.7083333730697632,
+      "reward_std": 0.37592338025569916,
+      "rewards/accuracy_reward": 0.7083333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 70
+    },
+    {
+      "completion_length": 51.79166793823242,
+      "epoch": 0.1775,
+      "grad_norm": 7.912103229675263,
+      "kl": 0.072021484375,
+      "learning_rate": 9.1125e-07,
+      "loss": 0.0029,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.23899273574352264,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 71
+    },
+    {
+      "completion_length": 67.08333396911621,
+      "epoch": 0.18,
+      "grad_norm": 7.4195867306219725,
+      "kl": 0.07373046875,
+      "learning_rate": 9.1e-07,
+      "loss": 0.0029,
+      "reward": 1.4583333730697632,
+      "reward_std": 0.4481577202677727,
+      "rewards/accuracy_reward": 0.5833333432674408,
+      "rewards/format_reward": 0.8750000298023224,
+      "step": 72
+    },
+    {
+      "completion_length": 60.33333396911621,
+      "epoch": 0.1825,
+      "grad_norm": 4.732110104154176,
+      "kl": 0.0615234375,
+      "learning_rate": 9.087499999999999e-07,
+      "loss": 0.0025,
+      "reward": 1.6666666865348816,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 0.6666666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 73
+    },
+    {
+      "completion_length": 76.83333587646484,
+      "epoch": 0.185,
+      "grad_norm": 15.673564911876664,
+      "kl": 0.077392578125,
+      "learning_rate": 9.074999999999999e-07,
+      "loss": 0.0031,
+      "reward": 1.75,
+      "reward_std": 0.3680921643972397,
+      "rewards/accuracy_reward": 0.75,
+      "rewards/format_reward": 1.0,
+      "step": 74
+    },
+    {
+      "completion_length": 60.75000190734863,
+      "epoch": 0.1875,
+      "grad_norm": 11.536043510370401,
+      "kl": 0.05859375,
+      "learning_rate": 9.0625e-07,
+      "loss": 0.0023,
+      "reward": 1.9375000596046448,
+      "reward_std": 0.11558076739311218,
+      "rewards/accuracy_reward": 0.9375000298023224,
+      "rewards/format_reward": 1.0,
+      "step": 75
+    },
+    {
+      "completion_length": 68.25000190734863,
+      "epoch": 0.19,
+      "grad_norm": 0.5206115311671419,
+      "kl": 0.070556640625,
+      "learning_rate": 9.05e-07,
+      "loss": 0.0028,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 76
+    },
+    {
+      "completion_length": 70.95833587646484,
+      "epoch": 0.1925,
+      "grad_norm": 9.429791436668038,
+      "kl": 0.0654296875,
+      "learning_rate": 9.0375e-07,
+      "loss": 0.0026,
+      "reward": 1.7916667461395264,
+      "reward_std": 0.23116152733564377,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 77
+    },
+    {
+      "completion_length": 65.75000190734863,
+      "epoch": 0.195,
+      "grad_norm": 6.729037338982755,
+      "kl": 0.060546875,
+      "learning_rate": 9.024999999999999e-07,
+      "loss": 0.0024,
+      "reward": 1.7916666865348816,
+      "reward_std": 0.26603010296821594,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 78
+    },
+    {
+      "completion_length": 64.41666984558105,
+      "epoch": 0.1975,
+      "grad_norm": 9.457055927403056,
+      "kl": 0.056640625,
+      "learning_rate": 9.0125e-07,
+      "loss": 0.0023,
+      "reward": 1.7708333730697632,
+      "reward_std": 0.5128307789564133,
+      "rewards/accuracy_reward": 0.8958333432674408,
+      "rewards/format_reward": 0.8750000298023224,
+      "step": 79
+    },
+    {
+      "completion_length": 60.833335876464844,
+      "epoch": 0.2,
+      "grad_norm": 10.88321497488877,
+      "kl": 0.06396484375,
+      "learning_rate": 9e-07,
+      "loss": 0.0026,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.23899271339178085,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 80
+    },
+    {
+      "completion_length": 74.00000381469727,
+      "epoch": 0.2025,
+      "grad_norm": 9.14141506172264,
+      "kl": 0.052978515625,
+      "learning_rate": 8.9875e-07,
+      "loss": 0.0021,
+      "reward": 1.7500000596046448,
+      "reward_std": 0.3680921494960785,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 81
+    },
+    {
+      "completion_length": 63.541669845581055,
+      "epoch": 0.205,
+      "grad_norm": 8.58175119212839,
+      "kl": 0.0615234375,
+      "learning_rate": 8.974999999999999e-07,
+      "loss": 0.0025,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.23899272084236145,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 82
+    },
+    {
+      "completion_length": 56.62500190734863,
+      "epoch": 0.2075,
+      "grad_norm": 4.683616930700102,
+      "kl": 0.077392578125,
+      "learning_rate": 8.9625e-07,
+      "loss": 0.0031,
+      "reward": 1.7916666865348816,
+      "reward_std": 0.10206207633018494,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 83
+    },
+    {
+      "completion_length": 77.50000381469727,
+      "epoch": 0.21,
+      "grad_norm": 19.854891318758874,
+      "kl": 0.07177734375,
+      "learning_rate": 8.95e-07,
+      "loss": 0.0029,
+      "reward": 1.4583333730697632,
+      "reward_std": 0.5203921049833298,
+      "rewards/accuracy_reward": 0.5416666716337204,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 84
+    },
+    {
+      "completion_length": 65.04166984558105,
+      "epoch": 0.2125,
+      "grad_norm": 6.598335432224304,
+      "kl": 0.050048828125,
+      "learning_rate": 8.9375e-07,
+      "loss": 0.002,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206207633018494,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 85
+    },
+    {
+      "completion_length": 60.416669845581055,
+      "epoch": 0.215,
+      "grad_norm": 8.543498204974801,
+      "kl": 0.097412109375,
+      "learning_rate": 8.924999999999999e-07,
+      "loss": 0.0039,
+      "reward": 1.7500000596046448,
+      "reward_std": 0.20412414520978928,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 86
+    },
+    {
+      "completion_length": 78.16667175292969,
+      "epoch": 0.2175,
+      "grad_norm": 4.966080621186072,
+      "kl": 0.061279296875,
+      "learning_rate": 8.912499999999999e-07,
+      "loss": 0.0024,
+      "reward": 1.625,
+      "reward_std": 0.3061861991882324,
+      "rewards/accuracy_reward": 0.7083333432674408,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 87
+    },
+    {
+      "completion_length": 63.33333396911621,
+      "epoch": 0.22,
+      "grad_norm": 8.582726087717052,
+      "kl": 0.0543212890625,
+      "learning_rate": 8.9e-07,
+      "loss": 0.0022,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.23899271339178085,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 88
+    },
+    {
+      "completion_length": 78.75000381469727,
+      "epoch": 0.2225,
+      "grad_norm": 23.545799089019614,
+      "kl": 0.0528564453125,
+      "learning_rate": 8.8875e-07,
+      "loss": 0.0021,
+      "reward": 1.3750000596046448,
+      "reward_std": 0.42718585580587387,
+      "rewards/accuracy_reward": 0.4166666865348816,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 89
+    },
+    {
+      "completion_length": 68.29166793823242,
+      "epoch": 0.225,
+      "grad_norm": 0.47046993267117027,
+      "kl": 0.052490234375,
+      "learning_rate": 8.874999999999999e-07,
+      "loss": 0.0021,
+      "reward": 1.5,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 0.5,
+      "rewards/format_reward": 1.0,
+      "step": 90
+    },
+    {
+      "completion_length": 70.00000381469727,
+      "epoch": 0.2275,
+      "grad_norm": 6.2355460311295685,
+      "kl": 0.05224609375,
+      "learning_rate": 8.8625e-07,
+      "loss": 0.0021,
+      "reward": 1.7916667461395264,
+      "reward_std": 0.23116153478622437,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 91
+    },
+    {
+      "completion_length": 68.95833587646484,
+      "epoch": 0.23,
+      "grad_norm": 4.0370719278724545,
+      "kl": 0.0537109375,
+      "learning_rate": 8.85e-07,
+      "loss": 0.0021,
+      "reward": 1.875,
+      "reward_std": 0.1369306445121765,
+      "rewards/accuracy_reward": 0.875,
+      "rewards/format_reward": 1.0,
+      "step": 92
+    },
+    {
+      "completion_length": 73.66666793823242,
+      "epoch": 0.2325,
+      "grad_norm": 7.618350439185457,
+      "kl": 0.060302734375,
+      "learning_rate": 8.8375e-07,
+      "loss": 0.0024,
+      "reward": 1.7083333730697632,
+      "reward_std": 0.43528565764427185,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 93
+    },
+    {
+      "completion_length": 83.29166793823242,
+      "epoch": 0.235,
+      "grad_norm": 7.33749565606432,
+      "kl": 0.0596923828125,
+      "learning_rate": 8.824999999999999e-07,
+      "loss": 0.0024,
+      "reward": 1.5833333730697632,
+      "reward_std": 0.4431168735027313,
+      "rewards/accuracy_reward": 0.6250000149011612,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 94
+    },
+    {
+      "completion_length": 75.875,
+      "epoch": 0.2375,
+      "grad_norm": 4.8133899271670435,
+      "kl": 0.063720703125,
+      "learning_rate": 8.812499999999999e-07,
+      "loss": 0.0026,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 0.9166666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 95
+    },
+    {
+      "completion_length": 62.66666793823242,
+      "epoch": 0.24,
+      "grad_norm": 8.492203007126976,
+      "kl": 0.056640625,
+      "learning_rate": 8.799999999999999e-07,
+      "loss": 0.0023,
+      "reward": 1.7500000596046448,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 0.7500000298023224,
+      "rewards/format_reward": 1.0,
+      "step": 96
+    },
+    {
+      "completion_length": 72.41667175292969,
+      "epoch": 0.2425,
+      "grad_norm": 13.606138214943282,
+      "kl": 0.06298828125,
+      "learning_rate": 8.7875e-07,
+      "loss": 0.0025,
+      "reward": 1.7083333730697632,
+      "reward_std": 0.26603007316589355,
+      "rewards/accuracy_reward": 0.7083333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 97
+    },
+    {
+      "completion_length": 69.75,
+      "epoch": 0.245,
+      "grad_norm": 0.5453979042575031,
+      "kl": 0.060546875,
+      "learning_rate": 8.774999999999999e-07,
+      "loss": 0.0024,
+      "reward": 1.75,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 0.75,
+      "rewards/format_reward": 1.0,
+      "step": 98
+    },
+    {
+      "completion_length": 68.25000381469727,
+      "epoch": 0.2475,
+      "grad_norm": 9.180577120924026,
+      "kl": 0.10546875,
+      "learning_rate": 8.7625e-07,
+      "loss": 0.0042,
+      "reward": 1.6875000596046448,
+      "reward_std": 0.3087066560983658,
+      "rewards/accuracy_reward": 0.7291666865348816,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 99
+    },
+    {
+      "completion_length": 89.54167175292969,
+      "epoch": 0.25,
+      "grad_norm": 5.517684116055758,
+      "kl": 0.065673828125,
+      "learning_rate": 8.75e-07,
+      "loss": 0.0026,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.25819891691207886,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.8333333730697632,
+      "step": 100
+    },
+    {
+      "completion_length": 82.00000381469727,
+      "epoch": 0.2525,
+      "grad_norm": 12.707071758177568,
+      "kl": 0.06005859375,
+      "learning_rate": 8.7375e-07,
+      "loss": 0.0024,
+      "reward": 1.6666666865348816,
+      "reward_std": 0.25819891691207886,
+      "rewards/accuracy_reward": 0.6666666716337204,
+      "rewards/format_reward": 1.0,
+      "step": 101
+    },
+    {
+      "completion_length": 79.20833587646484,
+      "epoch": 0.255,
+      "grad_norm": 7.511693516580632,
+      "kl": 0.062744140625,
+      "learning_rate": 8.725e-07,
+      "loss": 0.0025,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.31122709065675735,
+      "rewards/accuracy_reward": 0.8750000298023224,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 102
+    },
+    {
+      "completion_length": 73.20833587646484,
+      "epoch": 0.2575,
+      "grad_norm": 8.840679094835005,
+      "kl": 0.0966796875,
+      "learning_rate": 8.712499999999999e-07,
+      "loss": 0.0039,
+      "reward": 1.4166667461395264,
+      "reward_std": 0.3332235887646675,
+      "rewards/accuracy_reward": 0.4583333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 103
+    },
+    {
+      "completion_length": 77.25000381469727,
+      "epoch": 0.26,
+      "grad_norm": 3.274184002670797,
+      "kl": 0.046630859375,
+      "learning_rate": 8.699999999999999e-07,
+      "loss": 0.0019,
+      "reward": 1.625,
+      "reward_std": 0.1369306445121765,
+      "rewards/accuracy_reward": 0.625,
+      "rewards/format_reward": 1.0,
+      "step": 104
+    },
+    {
+      "completion_length": 76.79166793823242,
+      "epoch": 0.2625,
+      "grad_norm": 3.9080005650231326,
+      "kl": 0.056884765625,
+      "learning_rate": 8.687499999999999e-07,
+      "loss": 0.0023,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 105
+    },
+    {
+      "completion_length": 75.75000381469727,
+      "epoch": 0.265,
+      "grad_norm": 4.484978903474679,
+      "kl": 0.0535888671875,
+      "learning_rate": 8.675000000000001e-07,
+      "loss": 0.0021,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 0.9166666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 106
+    },
+    {
+      "completion_length": 80.58333587646484,
+      "epoch": 0.2675,
+      "grad_norm": 4.072450072560431,
+      "kl": 0.0406494140625,
+      "learning_rate": 8.6625e-07,
+      "loss": 0.0016,
+      "reward": 1.875,
+      "reward_std": 0.3061861991882324,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 107
+    },
+    {
+      "completion_length": 69.58333587646484,
+      "epoch": 0.27,
+      "grad_norm": 3.9109965229285257,
+      "kl": 0.0396728515625,
+      "learning_rate": 8.65e-07,
+      "loss": 0.0016,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 108
+    },
+    {
+      "completion_length": 76.875,
+      "epoch": 0.2725,
+      "grad_norm": 15.308422489270624,
+      "kl": 0.05322265625,
+      "learning_rate": 8.6375e-07,
+      "loss": 0.0021,
+      "reward": 1.5833333730697632,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 0.5833333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 109
+    },
+    {
+      "completion_length": 62.12500190734863,
+      "epoch": 0.275,
+      "grad_norm": 5.051061764048627,
+      "kl": 0.06494140625,
+      "learning_rate": 8.625e-07,
+      "loss": 0.0026,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 110
+    },
+    {
+      "completion_length": 81.79167175292969,
+      "epoch": 0.2775,
+      "grad_norm": 3.9187181374700106,
+      "kl": 0.0545654296875,
+      "learning_rate": 8.612499999999999e-07,
+      "loss": 0.0022,
+      "reward": 1.5416666865348816,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.5416666679084301,
+      "rewards/format_reward": 1.0,
+      "step": 111
+    },
+    {
+      "completion_length": 61.333335876464844,
+      "epoch": 0.28,
+      "grad_norm": 6.496734022552879,
+      "kl": 0.06103515625,
+      "learning_rate": 8.599999999999999e-07,
+      "loss": 0.0024,
+      "reward": 1.7083333730697632,
+      "reward_std": 0.3061862587928772,
+      "rewards/accuracy_reward": 0.75,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 112
+    },
+    {
+      "completion_length": 67.41666984558105,
+      "epoch": 0.2825,
+      "grad_norm": 40.521332963582154,
+      "kl": 0.058349609375,
+      "learning_rate": 8.587499999999999e-07,
+      "loss": 0.0023,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 0.9166666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 113
+    },
+    {
+      "completion_length": 64.70833587646484,
+      "epoch": 0.285,
+      "grad_norm": 13.390181016483035,
+      "kl": 0.078125,
+      "learning_rate": 8.575e-07,
+      "loss": 0.0031,
+      "reward": 1.9166667461395264,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 0.9166666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 114
+    },
+    {
+      "completion_length": 61.708335876464844,
+      "epoch": 0.2875,
+      "grad_norm": 0.2579672215056597,
+      "kl": 0.05859375,
+      "learning_rate": 8.5625e-07,
+      "loss": 0.0023,
+      "reward": 1.75,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 0.75,
+      "rewards/format_reward": 1.0,
+      "step": 115
+    },
+    {
+      "completion_length": 65.54166984558105,
+      "epoch": 0.29,
+      "grad_norm": 24.693248847842778,
+      "kl": 0.064208984375,
+      "learning_rate": 8.55e-07,
+      "loss": 0.0026,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 116
+    },
+    {
+      "completion_length": 59.958335876464844,
+      "epoch": 0.2925,
+      "grad_norm": 4.32030389823459,
+      "kl": 0.076904296875,
+      "learning_rate": 8.5375e-07,
+      "loss": 0.0031,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 0.9166666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 117
+    },
+    {
+      "completion_length": 58.16666793823242,
+      "epoch": 0.295,
+      "grad_norm": 7.601593598404612,
+      "kl": 0.20361328125,
+      "learning_rate": 8.525e-07,
+      "loss": 0.0081,
+      "reward": 1.75,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 0.75,
+      "rewards/format_reward": 1.0,
+      "step": 118
+    },
+    {
+      "completion_length": 57.708335876464844,
+      "epoch": 0.2975,
+      "grad_norm": 5.946498248735497,
+      "kl": 0.066650390625,
+      "learning_rate": 8.512499999999999e-07,
+      "loss": 0.0027,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 119
+    },
+    {
+      "completion_length": 61.45833396911621,
+      "epoch": 0.3,
+      "grad_norm": 3.0579917552588554,
+      "kl": 0.131103515625,
+      "learning_rate": 8.499999999999999e-07,
+      "loss": 0.0053,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 120
+    },
+    {
+      "completion_length": 56.00000190734863,
+      "epoch": 0.3025,
+      "grad_norm": 1.5124769776550189,
+      "kl": 0.103515625,
+      "learning_rate": 8.487499999999999e-07,
+      "loss": 0.0041,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 121
+    },
+    {
+      "completion_length": 56.25000190734863,
+      "epoch": 0.305,
+      "grad_norm": 7.31028449777576,
+      "kl": 0.0908203125,
+      "learning_rate": 8.475e-07,
+      "loss": 0.0036,
+      "reward": 1.7083333730697632,
+      "reward_std": 0.3061862140893936,
+      "rewards/accuracy_reward": 0.7083333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 122
+    },
+    {
+      "completion_length": 69.20833396911621,
+      "epoch": 0.3075,
+      "grad_norm": 7.086257081155087,
+      "kl": 0.072998046875,
+      "learning_rate": 8.462499999999999e-07,
+      "loss": 0.0029,
+      "reward": 1.7916666865348816,
+      "reward_std": 0.26603010296821594,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 123
+    },
+    {
+      "completion_length": 66.58333587646484,
+      "epoch": 0.31,
+      "grad_norm": 4.867315965330029,
+      "kl": 0.048095703125,
+      "learning_rate": 8.45e-07,
+      "loss": 0.0019,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 124
+    },
+    {
+      "completion_length": 66.25000190734863,
+      "epoch": 0.3125,
+      "grad_norm": 16.64611686310609,
+      "kl": 0.0623779296875,
+      "learning_rate": 8.4375e-07,
+      "loss": 0.0025,
+      "reward": 1.7291666865348816,
+      "reward_std": 0.25515516847372055,
+      "rewards/accuracy_reward": 0.7291666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 125
+    },
+    {
+      "completion_length": 69.91666793823242,
+      "epoch": 0.315,
+      "grad_norm": 4.5944308739615245,
+      "kl": 0.05615234375,
+      "learning_rate": 8.425e-07,
+      "loss": 0.0023,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 0.9166666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 126
+    },
+    {
+      "completion_length": 83.25,
+      "epoch": 0.3175,
+      "grad_norm": 3.4492654552916115,
+      "kl": 0.056884765625,
+      "learning_rate": 8.4125e-07,
+      "loss": 0.0023,
+      "reward": 1.8958333730697632,
+      "reward_std": 0.2002602517604828,
+      "rewards/accuracy_reward": 0.9375,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 127
+    },
+    {
+      "completion_length": 84.29166793823242,
+      "epoch": 0.32,
+      "grad_norm": 3.9532152770578914,
+      "kl": 0.0654296875,
+      "learning_rate": 8.399999999999999e-07,
+      "loss": 0.0026,
+      "reward": 1.9166667461395264,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 0.9166666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 128
+    },
+    {
+      "completion_length": 75.58333587646484,
+      "epoch": 0.3225,
+      "grad_norm": 10.437554081102167,
+      "kl": 0.05078125,
+      "learning_rate": 8.387499999999999e-07,
+      "loss": 0.002,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 129
+    },
+    {
+      "completion_length": 72.25000381469727,
+      "epoch": 0.325,
+      "grad_norm": 3.6422939374830814,
+      "kl": 0.0484619140625,
+      "learning_rate": 8.375e-07,
+      "loss": 0.0019,
+      "reward": 1.7916666865348816,
+      "reward_std": 0.10206207633018494,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 130
+    },
+    {
+      "completion_length": 77.87500381469727,
+      "epoch": 0.3275,
+      "grad_norm": 0.44791979427142387,
+      "kl": 0.065673828125,
+      "learning_rate": 8.3625e-07,
+      "loss": 0.0026,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 131
+    },
+    {
+      "completion_length": 68.66666793823242,
+      "epoch": 0.33,
+      "grad_norm": 5.385790752053767,
+      "kl": 0.0596923828125,
+      "learning_rate": 8.349999999999999e-07,
+      "loss": 0.0024,
+      "reward": 1.9791666865348816,
+      "reward_std": 0.05103103443980217,
+      "rewards/accuracy_reward": 0.9791666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 132
+    },
+    {
+      "completion_length": 74.04166793823242,
+      "epoch": 0.3325,
+      "grad_norm": 3.5356298442034095,
+      "kl": 0.0518798828125,
+      "learning_rate": 8.3375e-07,
+      "loss": 0.0021,
+      "reward": 1.625,
+      "reward_std": 0.20916502177715302,
+      "rewards/accuracy_reward": 0.6666666716337204,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 133
+    },
+    {
+      "completion_length": 79.54166793823242,
+      "epoch": 0.335,
+      "grad_norm": 13.519892690256926,
+      "kl": 0.051513671875,
+      "learning_rate": 8.325e-07,
+      "loss": 0.0021,
+      "reward": 1.7083333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.7083333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 134
+    },
+    {
+      "completion_length": 90.95833587646484,
+      "epoch": 0.3375,
+      "grad_norm": 8.248631988755827,
+      "kl": 0.0509033203125,
+      "learning_rate": 8.3125e-07,
+      "loss": 0.002,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.25819891691207886,
+      "rewards/accuracy_reward": 0.8333333730697632,
+      "rewards/format_reward": 1.0,
+      "step": 135
+    },
+    {
+      "completion_length": 63.208335876464844,
+      "epoch": 0.34,
+      "grad_norm": 0.27092369317618625,
+      "kl": 0.0523681640625,
+      "learning_rate": 8.299999999999999e-07,
+      "loss": 0.0021,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 136
+    },
+    {
+      "completion_length": 78.25,
+      "epoch": 0.3425,
+      "grad_norm": 8.134027797348121,
+      "kl": 0.06689453125,
+      "learning_rate": 8.287499999999999e-07,
+      "loss": 0.0027,
+      "reward": 1.7291666865348816,
+      "reward_std": 0.5334838479757309,
+      "rewards/accuracy_reward": 0.8125000298023224,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 137
+    },
+    {
+      "completion_length": 69.75000190734863,
+      "epoch": 0.345,
+      "grad_norm": 11.145798458704071,
+      "kl": 0.0457763671875,
+      "learning_rate": 8.275e-07,
+      "loss": 0.0018,
+      "reward": 1.8750000596046448,
+      "reward_std": 0.23116153478622437,
+      "rewards/accuracy_reward": 0.8750000298023224,
+      "rewards/format_reward": 1.0,
+      "step": 138
+    },
+    {
+      "completion_length": 76.375,
+      "epoch": 0.3475,
+      "grad_norm": 4.648506226378103,
+      "kl": 0.068603515625,
+      "learning_rate": 8.2625e-07,
+      "loss": 0.0027,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.31122709810733795,
+      "rewards/accuracy_reward": 0.8750000298023224,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 139
+    },
+    {
+      "completion_length": 72.08333587646484,
+      "epoch": 0.35,
+      "grad_norm": 5.49537540817778,
+      "kl": 0.0550537109375,
+      "learning_rate": 8.249999999999999e-07,
+      "loss": 0.0022,
+      "reward": 1.5833333730697632,
+      "reward_std": 0.25819891691207886,
+      "rewards/accuracy_reward": 0.583333358168602,
+      "rewards/format_reward": 1.0,
+      "step": 140
+    },
+    {
+      "completion_length": 74.12500381469727,
+      "epoch": 0.3525,
+      "grad_norm": 3.23463816741491,
+      "kl": 0.0396728515625,
+      "learning_rate": 8.2375e-07,
+      "loss": 0.0016,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 141
+    },
+    {
+      "completion_length": 73.79167175292969,
+      "epoch": 0.355,
+      "grad_norm": 8.949390923947067,
+      "kl": 0.0540771484375,
+      "learning_rate": 8.225e-07,
+      "loss": 0.0022,
+      "reward": 1.8750000596046448,
+      "reward_std": 0.23116152733564377,
+      "rewards/accuracy_reward": 0.8750000298023224,
+      "rewards/format_reward": 1.0,
+      "step": 142
+    },
+    {
+      "completion_length": 81.33333587646484,
+      "epoch": 0.3575,
+      "grad_norm": 13.970481624687274,
+      "kl": 0.14892578125,
+      "learning_rate": 8.2125e-07,
+      "loss": 0.0059,
+      "reward": 1.8125000596046448,
+      "reward_std": 0.2682070881128311,
+      "rewards/accuracy_reward": 0.9791666865348816,
+      "rewards/format_reward": 0.8333333432674408,
+      "step": 143
+    },
+    {
+      "completion_length": 77.45833587646484,
+      "epoch": 0.36,
+      "grad_norm": 13.619028198353357,
+      "kl": 0.06884765625,
+      "learning_rate": 8.199999999999999e-07,
+      "loss": 0.0028,
+      "reward": 1.8750000596046448,
+      "reward_std": 0.3061862215399742,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 144
+    },
+    {
+      "completion_length": 70.91666793823242,
+      "epoch": 0.3625,
+      "grad_norm": 8.050368007065988,
+      "kl": 0.071533203125,
+      "learning_rate": 8.187499999999999e-07,
+      "loss": 0.0029,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 145
+    },
+    {
+      "completion_length": 83.79167175292969,
+      "epoch": 0.365,
+      "grad_norm": 8.301452022148478,
+      "kl": 0.0537109375,
+      "learning_rate": 8.175e-07,
+      "loss": 0.0021,
+      "reward": 1.8125000596046448,
+      "reward_std": 0.3357440382242203,
+      "rewards/accuracy_reward": 0.8541666865348816,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 146
+    },
+    {
+      "completion_length": 89.83333587646484,
+      "epoch": 0.3675,
+      "grad_norm": 5.225379290022837,
+      "kl": 0.0634765625,
+      "learning_rate": 8.1625e-07,
+      "loss": 0.0025,
+      "reward": 1.9166667461395264,
+      "reward_std": 0.20412414520978928,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 147
+    },
+    {
+      "completion_length": 73.91667175292969,
+      "epoch": 0.37,
+      "grad_norm": 0.3862241193439462,
+      "kl": 0.0545654296875,
+      "learning_rate": 8.149999999999999e-07,
+      "loss": 0.0022,
+      "reward": 1.75,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 0.75,
+      "rewards/format_reward": 1.0,
+      "step": 148
+    },
+    {
+      "completion_length": 75.83333587646484,
+      "epoch": 0.3725,
+      "grad_norm": 0.16866310269978918,
+      "kl": 0.0380859375,
+      "learning_rate": 8.137499999999999e-07,
+      "loss": 0.0015,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 149
+    },
+    {
+      "completion_length": 71.54166793823242,
+      "epoch": 0.375,
+      "grad_norm": 0.336328483289417,
+      "kl": 0.04541015625,
+      "learning_rate": 8.125e-07,
+      "loss": 0.0018,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 150
+    },
+    {
+      "completion_length": 83.41667175292969,
+      "epoch": 0.3775,
+      "grad_norm": 0.18233740920926422,
+      "kl": 0.043701171875,
+      "learning_rate": 8.1125e-07,
+      "loss": 0.0017,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 151
+    },
+    {
+      "completion_length": 74.04166793823242,
+      "epoch": 0.38,
+      "grad_norm": 2.4615804123735074,
+      "kl": 0.0394287109375,
+      "learning_rate": 8.1e-07,
+      "loss": 0.0016,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 152
+    },
+    {
+      "completion_length": 67.83333587646484,
+      "epoch": 0.3825,
+      "grad_norm": 4.506961936813221,
+      "kl": 0.05419921875,
+      "learning_rate": 8.087499999999999e-07,
+      "loss": 0.0022,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206207633018494,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 153
+    },
+    {
+      "completion_length": 75.04166793823242,
+      "epoch": 0.385,
+      "grad_norm": 13.894215119687294,
+      "kl": 0.05029296875,
+      "learning_rate": 8.075e-07,
+      "loss": 0.002,
+      "reward": 1.7916666865348816,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 154
+    },
+    {
+      "completion_length": 77.91667175292969,
+      "epoch": 0.3875,
+      "grad_norm": 5.786917651350198,
+      "kl": 0.0419921875,
+      "learning_rate": 8.0625e-07,
+      "loss": 0.0017,
+      "reward": 1.7916666865348816,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 155
+    },
+    {
+      "completion_length": 75.54166793823242,
+      "epoch": 0.39,
+      "grad_norm": 10.916507056002851,
+      "kl": 0.044921875,
+      "learning_rate": 8.05e-07,
+      "loss": 0.0018,
+      "reward": 1.7500000596046448,
+      "reward_std": 0.4431168735027313,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 156
+    },
+    {
+      "completion_length": 72.45833587646484,
+      "epoch": 0.3925,
+      "grad_norm": 0.6036085775451172,
+      "kl": 0.052978515625,
+      "learning_rate": 8.037499999999999e-07,
+      "loss": 0.0021,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 157
+    },
+    {
+      "completion_length": 69.20833587646484,
+      "epoch": 0.395,
+      "grad_norm": 0.20269296867808964,
+      "kl": 0.0343017578125,
+      "learning_rate": 8.024999999999999e-07,
+      "loss": 0.0014,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 158
+    },
+    {
+      "completion_length": 70.91666793823242,
+      "epoch": 0.3975,
+      "grad_norm": 5.002628339877501,
+      "kl": 0.05078125,
+      "learning_rate": 8.0125e-07,
+      "loss": 0.002,
+      "reward": 1.5833333730697632,
+      "reward_std": 0.23899273574352264,
+      "rewards/accuracy_reward": 0.625,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 159
+    },
+    {
+      "completion_length": 83.79166984558105,
+      "epoch": 0.4,
+      "grad_norm": 14.264997208415672,
+      "kl": 0.1083984375,
+      "learning_rate": 8e-07,
+      "loss": 0.0043,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.25819891691207886,
+      "rewards/accuracy_reward": 0.8333333730697632,
+      "rewards/format_reward": 1.0,
+      "step": 160
+    },
+    {
+      "completion_length": 80.16667175292969,
+      "epoch": 0.4025,
+      "grad_norm": 3.4840572426344765,
+      "kl": 0.0389404296875,
+      "learning_rate": 7.9875e-07,
+      "loss": 0.0016,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 0.9166666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 161
+    },
+    {
+      "completion_length": 80.37500381469727,
+      "epoch": 0.405,
+      "grad_norm": 4.2899724011799245,
+      "kl": 0.0386962890625,
+      "learning_rate": 7.975e-07,
+      "loss": 0.0015,
+      "reward": 1.7083333730697632,
+      "reward_std": 0.10206207633018494,
+      "rewards/accuracy_reward": 0.7083333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 162
+    },
+    {
+      "completion_length": 72.58333587646484,
+      "epoch": 0.4075,
+      "grad_norm": 6.888283826236383,
+      "kl": 0.05029296875,
+      "learning_rate": 7.9625e-07,
+      "loss": 0.002,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.23899271339178085,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 163
+    },
+    {
+      "completion_length": 83.87500381469727,
+      "epoch": 0.41,
+      "grad_norm": 10.672136555947715,
+      "kl": 0.053466796875,
+      "learning_rate": 7.95e-07,
+      "loss": 0.0021,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.3332235962152481,
+      "rewards/accuracy_reward": 0.9166666865348816,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 164
+    },
+    {
+      "completion_length": 63.833335876464844,
+      "epoch": 0.4125,
+      "grad_norm": 0.5939326240650639,
+      "kl": 0.0518798828125,
+      "learning_rate": 7.937499999999999e-07,
+      "loss": 0.0021,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 165
+    },
+    {
+      "completion_length": 84.45833587646484,
+      "epoch": 0.415,
+      "grad_norm": 6.343742530614501,
+      "kl": 0.0601806640625,
+      "learning_rate": 7.924999999999999e-07,
+      "loss": 0.0024,
+      "reward": 1.7500000596046448,
+      "reward_std": 0.3872983753681183,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 166
+    },
+    {
+      "completion_length": 74.16666984558105,
+      "epoch": 0.4175,
+      "grad_norm": 0.5061062272176906,
+      "kl": 0.06103515625,
+      "learning_rate": 7.912499999999999e-07,
+      "loss": 0.0025,
+      "reward": 1.75,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 0.75,
+      "rewards/format_reward": 1.0,
+      "step": 167
+    },
+    {
+      "completion_length": 93.41666793823242,
+      "epoch": 0.42,
+      "grad_norm": 3.3456995801068596,
+      "kl": 0.0562744140625,
+      "learning_rate": 7.9e-07,
+      "loss": 0.0023,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 0.9166666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 168
+    },
+    {
+      "completion_length": 68.87500381469727,
+      "epoch": 0.4225,
+      "grad_norm": 3.248130898572834,
+      "kl": 0.0615234375,
+      "learning_rate": 7.8875e-07,
+      "loss": 0.0025,
+      "reward": 1.7083333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.7083333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 169
+    },
+    {
+      "completion_length": 73.04167175292969,
+      "epoch": 0.425,
+      "grad_norm": 21.57626593100111,
+      "kl": 0.0567626953125,
+      "learning_rate": 7.875e-07,
+      "loss": 0.0023,
+      "reward": 1.7083333730697632,
+      "reward_std": 0.3061862215399742,
+      "rewards/accuracy_reward": 0.7500000298023224,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 170
+    },
+    {
+      "completion_length": 74.0,
+      "epoch": 0.4275,
+      "grad_norm": 0.8393141771605732,
+      "kl": 0.04345703125,
+      "learning_rate": 7.8625e-07,
+      "loss": 0.0017,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 171
+    },
+    {
+      "completion_length": 74.75000381469727,
+      "epoch": 0.43,
+      "grad_norm": 8.933723968345006,
+      "kl": 0.0443115234375,
+      "learning_rate": 7.85e-07,
+      "loss": 0.0018,
+      "reward": 1.7083333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.75,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 172
+    },
+    {
+      "completion_length": 78.70833587646484,
+      "epoch": 0.4325,
+      "grad_norm": 17.572461684097632,
+      "kl": 0.05419921875,
+      "learning_rate": 7.837499999999999e-07,
+      "loss": 0.0022,
+      "reward": 1.875,
+      "reward_std": 0.23116151988506317,
+      "rewards/accuracy_reward": 0.9166666865348816,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 173
+    },
+    {
+      "completion_length": 82.41666793823242,
+      "epoch": 0.435,
+      "grad_norm": 2.572194060503535,
+      "kl": 0.0504150390625,
+      "learning_rate": 7.824999999999999e-07,
+      "loss": 0.002,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 174
+    },
+    {
+      "completion_length": 80.75000381469727,
+      "epoch": 0.4375,
+      "grad_norm": 5.220117972645122,
+      "kl": 0.08984375,
+      "learning_rate": 7.812499999999999e-07,
+      "loss": 0.0036,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.23899271339178085,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 175
+    },
+    {
+      "completion_length": 80.33333587646484,
+      "epoch": 0.44,
+      "grad_norm": 7.129240956071849,
+      "kl": 0.04315185546875,
+      "learning_rate": 7.799999999999999e-07,
+      "loss": 0.0017,
+      "reward": 1.7083333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.7083333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 176
+    },
+    {
+      "completion_length": 83.75000381469727,
+      "epoch": 0.4425,
+      "grad_norm": 2.7325871171100524,
+      "kl": 0.076904296875,
+      "learning_rate": 7.787500000000001e-07,
+      "loss": 0.0031,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 177
+    },
+    {
+      "completion_length": 78.625,
+      "epoch": 0.445,
+      "grad_norm": 7.552601468940335,
+      "kl": 0.0445556640625,
+      "learning_rate": 7.775e-07,
+      "loss": 0.0018,
+      "reward": 1.5833333730697632,
+      "reward_std": 0.3332235962152481,
+      "rewards/accuracy_reward": 0.625,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 178
+    },
+    {
+      "completion_length": 75.75000381469727,
+      "epoch": 0.4475,
+      "grad_norm": 0.19732564393416355,
+      "kl": 0.044677734375,
+      "learning_rate": 7.7625e-07,
+      "loss": 0.0018,
+      "reward": 1.75,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 0.75,
+      "rewards/format_reward": 1.0,
+      "step": 179
+    },
+    {
+      "completion_length": 75.83333587646484,
+      "epoch": 0.45,
+      "grad_norm": 5.413407541750669,
+      "kl": 0.0576171875,
+      "learning_rate": 7.75e-07,
+      "loss": 0.0023,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 0.9166666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 180
+    },
+    {
+      "completion_length": 78.00000381469727,
+      "epoch": 0.4525,
+      "grad_norm": 5.254043089084881,
+      "kl": 0.04833984375,
+      "learning_rate": 7.7375e-07,
+      "loss": 0.0019,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 181
+    },
+    {
+      "completion_length": 85.29166793823242,
+      "epoch": 0.455,
+      "grad_norm": 12.874216139003746,
+      "kl": 0.054931640625,
+      "learning_rate": 7.724999999999999e-07,
+      "loss": 0.0022,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 182
+    },
+    {
+      "completion_length": 61.83333396911621,
+      "epoch": 0.4575,
+      "grad_norm": 22.282393167218782,
+      "kl": 0.05322265625,
+      "learning_rate": 7.712499999999999e-07,
+      "loss": 0.0021,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206207633018494,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 183
+    },
+    {
+      "completion_length": 75.58333587646484,
+      "epoch": 0.46,
+      "grad_norm": 72.84277045431853,
+      "kl": 0.1082763671875,
+      "learning_rate": 7.699999999999999e-07,
+      "loss": 0.0043,
+      "reward": 1.7083333730697632,
+      "reward_std": 0.43528565764427185,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 184
+    },
+    {
+      "completion_length": 83.50000381469727,
+      "epoch": 0.4625,
+      "grad_norm": 4.987506561476635,
+      "kl": 0.0479736328125,
+      "learning_rate": 7.6875e-07,
+      "loss": 0.0019,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.31122709065675735,
+      "rewards/accuracy_reward": 0.9166666865348816,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 185
+    },
+    {
+      "completion_length": 69.91666793823242,
+      "epoch": 0.465,
+      "grad_norm": 0.3185214597818445,
+      "kl": 0.05712890625,
+      "learning_rate": 7.675e-07,
+      "loss": 0.0023,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 186
+    },
+    {
+      "completion_length": 79.625,
+      "epoch": 0.4675,
+      "grad_norm": 3.2685263168504877,
+      "kl": 0.0455322265625,
+      "learning_rate": 7.6625e-07,
+      "loss": 0.0018,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206207633018494,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 187
+    },
+    {
+      "completion_length": 75.00000381469727,
+      "epoch": 0.47,
+      "grad_norm": 8.325018779424052,
+      "kl": 0.051513671875,
+      "learning_rate": 7.65e-07,
+      "loss": 0.0021,
+      "reward": 1.5833333730697632,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 0.5833333358168602,
+      "rewards/format_reward": 1.0,
+      "step": 188
+    },
+    {
+      "completion_length": 76.54166793823242,
+      "epoch": 0.4725,
+      "grad_norm": 0.3920675734655507,
+      "kl": 0.0504150390625,
+      "learning_rate": 7.6375e-07,
+      "loss": 0.002,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 189
+    },
+    {
+      "completion_length": 84.91666793823242,
+      "epoch": 0.475,
+      "grad_norm": 8.99147320074875,
+      "kl": 0.0390625,
+      "learning_rate": 7.624999999999999e-07,
+      "loss": 0.0016,
+      "reward": 1.8958333730697632,
+      "reward_std": 0.20664457976818085,
+      "rewards/accuracy_reward": 0.8958333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 190
+    },
+    {
+      "completion_length": 95.83333587646484,
+      "epoch": 0.4775,
+      "grad_norm": 69.97651402716114,
+      "kl": 0.072998046875,
+      "learning_rate": 7.612499999999999e-07,
+      "loss": 0.0029,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 191
+    },
+    {
+      "completion_length": 76.79166793823242,
+      "epoch": 0.48,
+      "grad_norm": 3.2673606951083007,
+      "kl": 0.0509033203125,
+      "learning_rate": 7.599999999999999e-07,
+      "loss": 0.002,
+      "reward": 1.9791666865348816,
+      "reward_std": 0.05103103816509247,
+      "rewards/accuracy_reward": 0.9791666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 192
+    },
+    {
+      "completion_length": 71.54166793823242,
+      "epoch": 0.4825,
+      "grad_norm": 6.555166981922478,
+      "kl": 0.043701171875,
+      "learning_rate": 7.5875e-07,
+      "loss": 0.0018,
+      "reward": 1.75,
+      "reward_std": 0.3332235813140869,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 193
+    },
+    {
+      "completion_length": 74.79166793823242,
+      "epoch": 0.485,
+      "grad_norm": 88.79934942718205,
+      "kl": 0.0692138671875,
+      "learning_rate": 7.575e-07,
+      "loss": 0.0028,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.23899271339178085,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 194
+    },
+    {
+      "completion_length": 80.37500381469727,
+      "epoch": 0.4875,
+      "grad_norm": 3.537706923150665,
+      "kl": 0.052734375,
+      "learning_rate": 7.5625e-07,
+      "loss": 0.0021,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206207633018494,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 195
+    },
+    {
+      "completion_length": 103.33333587646484,
+      "epoch": 0.49,
+      "grad_norm": 5.717780246230898,
+      "kl": 0.088134765625,
+      "learning_rate": 7.55e-07,
+      "loss": 0.0035,
+      "reward": 1.6666666865348816,
+      "reward_std": 0.3622380346059799,
+      "rewards/accuracy_reward": 0.7500000298023224,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 196
+    },
+    {
+      "completion_length": 78.25000381469727,
+      "epoch": 0.4925,
+      "grad_norm": 0.26404344551615316,
+      "kl": 0.0494384765625,
+      "learning_rate": 7.5375e-07,
+      "loss": 0.002,
+      "reward": 1.75,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 0.75,
+      "rewards/format_reward": 1.0,
+      "step": 197
+    },
+    {
+      "completion_length": 58.70833396911621,
+      "epoch": 0.495,
+      "grad_norm": 8.884717272881257,
+      "kl": 0.0595703125,
+      "learning_rate": 7.524999999999999e-07,
+      "loss": 0.0024,
+      "reward": 1.5833333730697632,
+      "reward_std": 0.29025524109601974,
+      "rewards/accuracy_reward": 0.6250000298023224,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 198
+    },
+    {
+      "completion_length": 78.91666793823242,
+      "epoch": 0.4975,
+      "grad_norm": 0.1968209030109873,
+      "kl": 0.0447998046875,
+      "learning_rate": 7.512499999999999e-07,
+      "loss": 0.0018,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 199
+    },
+    {
+      "completion_length": 75.41666793823242,
+      "epoch": 0.5,
+      "grad_norm": 0.14338764769370782,
+      "kl": 0.0352783203125,
+      "learning_rate": 7.5e-07,
+      "loss": 0.0014,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 200
+    },
+    {
+      "completion_length": 71.41666793823242,
+      "epoch": 0.5025,
+      "grad_norm": 2.678875577316952,
+      "kl": 0.066650390625,
+      "learning_rate": 7.4875e-07,
+      "loss": 0.0027,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 201
+    },
+    {
+      "completion_length": 63.16666793823242,
+      "epoch": 0.505,
+      "grad_norm": 2.9167153600838405,
+      "kl": 0.04144287109375,
+      "learning_rate": 7.475e-07,
+      "loss": 0.0017,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.20412415266036987,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 202
+    },
+    {
+      "completion_length": 70.75,
+      "epoch": 0.5075,
+      "grad_norm": 8.191465178041433,
+      "kl": 0.0460205078125,
+      "learning_rate": 7.4625e-07,
+      "loss": 0.0018,
+      "reward": 1.7916666865348816,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 203
+    },
+    {
+      "completion_length": 77.75000381469727,
+      "epoch": 0.51,
+      "grad_norm": 0.34559278800136956,
+      "kl": 0.0435791015625,
+      "learning_rate": 7.45e-07,
+      "loss": 0.0017,
+      "reward": 1.75,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 0.75,
+      "rewards/format_reward": 1.0,
+      "step": 204
+    },
+    {
+      "completion_length": 69.87500190734863,
+      "epoch": 0.5125,
+      "grad_norm": 4.670473448875982,
+      "kl": 0.0528564453125,
+      "learning_rate": 7.4375e-07,
+      "loss": 0.0021,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206207633018494,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 205
+    },
+    {
+      "completion_length": 74.95833587646484,
+      "epoch": 0.515,
+      "grad_norm": 8.764562343746835,
+      "kl": 0.06884765625,
+      "learning_rate": 7.425e-07,
+      "loss": 0.0028,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 206
+    },
+    {
+      "completion_length": 71.875,
+      "epoch": 0.5175,
+      "grad_norm": 6.055581273594887,
+      "kl": 0.076904296875,
+      "learning_rate": 7.412499999999999e-07,
+      "loss": 0.0031,
+      "reward": 1.5833333730697632,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 0.6666666716337204,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 207
+    },
+    {
+      "completion_length": 63.125003814697266,
+      "epoch": 0.52,
+      "grad_norm": 8.948612653935108,
+      "kl": 0.07373046875,
+      "learning_rate": 7.4e-07,
+      "loss": 0.0029,
+      "reward": 1.8750000596046448,
+      "reward_std": 0.306186206638813,
+      "rewards/accuracy_reward": 0.9166666865348816,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 208
+    },
+    {
+      "completion_length": 62.45833396911621,
+      "epoch": 0.5225,
+      "grad_norm": 5.496964254978537,
+      "kl": 0.043701171875,
+      "learning_rate": 7.3875e-07,
+      "loss": 0.0017,
+      "reward": 1.7916667461395264,
+      "reward_std": 0.23116153478622437,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 209
+    },
+    {
+      "completion_length": 60.16666793823242,
+      "epoch": 0.525,
+      "grad_norm": 2.802340467430377,
+      "kl": 0.0654296875,
+      "learning_rate": 7.375e-07,
+      "loss": 0.0026,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 210
+    },
+    {
+      "completion_length": 68.08333587646484,
+      "epoch": 0.5275,
+      "grad_norm": 0.3889704730731261,
+      "kl": 0.0616455078125,
+      "learning_rate": 7.362499999999999e-07,
+      "loss": 0.0025,
+      "reward": 1.75,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 0.75,
+      "rewards/format_reward": 1.0,
+      "step": 211
+    },
+    {
+      "completion_length": 78.04167175292969,
+      "epoch": 0.53,
+      "grad_norm": 5.437383846615231,
+      "kl": 0.0552978515625,
+      "learning_rate": 7.35e-07,
+      "loss": 0.0022,
+      "reward": 1.75,
+      "reward_std": 0.22360679507255554,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 212
+    },
+    {
+      "completion_length": 62.041669845581055,
+      "epoch": 0.5325,
+      "grad_norm": 0.29430882644776396,
+      "kl": 0.053466796875,
+      "learning_rate": 7.3375e-07,
+      "loss": 0.0021,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 213
+    },
+    {
+      "completion_length": 72.95833396911621,
+      "epoch": 0.535,
+      "grad_norm": 0.2340500925907288,
+      "kl": 0.053466796875,
+      "learning_rate": 7.325e-07,
+      "loss": 0.0021,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 214
+    },
+    {
+      "completion_length": 76.125,
+      "epoch": 0.5375,
+      "grad_norm": 5.190113249057551,
+      "kl": 0.046875,
+      "learning_rate": 7.312499999999999e-07,
+      "loss": 0.0019,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206207633018494,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 215
+    },
+    {
+      "completion_length": 91.58333587646484,
+      "epoch": 0.54,
+      "grad_norm": 3.904416501586602,
+      "kl": 0.036376953125,
+      "learning_rate": 7.3e-07,
+      "loss": 0.0015,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 0.875,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 216
+    },
+    {
+      "completion_length": 79.25000381469727,
+      "epoch": 0.5425,
+      "grad_norm": 0.20631593608442583,
+      "kl": 0.0389404296875,
+      "learning_rate": 7.2875e-07,
+      "loss": 0.0016,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 217
+    },
+    {
+      "completion_length": 72.25,
+      "epoch": 0.545,
+      "grad_norm": 4.8107379341458465,
+      "kl": 0.042724609375,
+      "learning_rate": 7.275e-07,
+      "loss": 0.0017,
+      "reward": 1.7916667461395264,
+      "reward_std": 0.23116152733564377,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 218
+    },
+    {
+      "completion_length": 84.95833587646484,
+      "epoch": 0.5475,
+      "grad_norm": 3.1614660219253885,
+      "kl": 0.041015625,
+      "learning_rate": 7.262499999999999e-07,
+      "loss": 0.0016,
+      "reward": 1.9791666865348816,
+      "reward_std": 0.05103103443980217,
+      "rewards/accuracy_reward": 0.9791666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 219
+    },
+    {
+      "completion_length": 77.41666793823242,
+      "epoch": 0.55,
+      "grad_norm": 15.034209953389713,
+      "kl": 0.045166015625,
+      "learning_rate": 7.249999999999999e-07,
+      "loss": 0.0018,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 220
+    },
+    {
+      "completion_length": 76.87500381469727,
+      "epoch": 0.5525,
+      "grad_norm": 4.1373132955961545,
+      "kl": 0.1297607421875,
+      "learning_rate": 7.2375e-07,
+      "loss": 0.0052,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.20412415266036987,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 221
+    },
+    {
+      "completion_length": 83.33333587646484,
+      "epoch": 0.555,
+      "grad_norm": 6.0890022474681,
+      "kl": 0.0511474609375,
+      "learning_rate": 7.225e-07,
+      "loss": 0.002,
+      "reward": 1.8958333730697632,
+      "reward_std": 0.20664458721876144,
+      "rewards/accuracy_reward": 0.9375000298023224,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 222
+    },
+    {
+      "completion_length": 96.33333587646484,
+      "epoch": 0.5575,
+      "grad_norm": 6.051302437198327,
+      "kl": 0.0531005859375,
+      "learning_rate": 7.212499999999999e-07,
+      "loss": 0.0021,
+      "reward": 1.875,
+      "reward_std": 0.1369306445121765,
+      "rewards/accuracy_reward": 0.875,
+      "rewards/format_reward": 1.0,
+      "step": 223
+    },
+    {
+      "completion_length": 82.66667175292969,
+      "epoch": 0.56,
+      "grad_norm": 2.83249582786607,
+      "kl": 0.0518798828125,
+      "learning_rate": 7.2e-07,
+      "loss": 0.0021,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206207633018494,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 224
+    },
+    {
+      "completion_length": 78.29166793823242,
+      "epoch": 0.5625,
+      "grad_norm": 4.662406270036471,
+      "kl": 0.051025390625,
+      "learning_rate": 7.1875e-07,
+      "loss": 0.002,
+      "reward": 1.75,
+      "reward_std": 0.15811388194561005,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 225
+    },
+    {
+      "completion_length": 100.0,
+      "epoch": 0.565,
+      "grad_norm": 3.1936222808478574,
+      "kl": 0.1239013671875,
+      "learning_rate": 7.175e-07,
+      "loss": 0.005,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.20412415266036987,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 226
+    },
+    {
+      "completion_length": 95.83333587646484,
+      "epoch": 0.5675,
+      "grad_norm": 3.9506167101538026,
+      "kl": 0.03521728515625,
+      "learning_rate": 7.1625e-07,
+      "loss": 0.0014,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 227
+    },
+    {
+      "completion_length": 73.33333587646484,
+      "epoch": 0.57,
+      "grad_norm": 3.123680615411211,
+      "kl": 0.0498046875,
+      "learning_rate": 7.149999999999999e-07,
+      "loss": 0.002,
+      "reward": 1.6666666865348816,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 0.6666666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 228
+    },
+    {
+      "completion_length": 94.54167175292969,
+      "epoch": 0.5725,
+      "grad_norm": 6.3149431746393345,
+      "kl": 0.049560546875,
+      "learning_rate": 7.137499999999999e-07,
+      "loss": 0.002,
+      "reward": 1.7083333730697632,
+      "reward_std": 0.2601759433746338,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 229
+    },
+    {
+      "completion_length": 78.95833587646484,
+      "epoch": 0.575,
+      "grad_norm": 5.342916500056161,
+      "kl": 0.0555419921875,
+      "learning_rate": 7.125e-07,
+      "loss": 0.0022,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 230
+    },
+    {
+      "completion_length": 77.04166793823242,
+      "epoch": 0.5775,
+      "grad_norm": 5.455421780647997,
+      "kl": 0.044677734375,
+      "learning_rate": 7.1125e-07,
+      "loss": 0.0018,
+      "reward": 1.6666667461395264,
+      "reward_std": 0.25819891691207886,
+      "rewards/accuracy_reward": 0.6666666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 231
+    },
+    {
+      "completion_length": 97.16666793823242,
+      "epoch": 0.58,
+      "grad_norm": 5.258370348056576,
+      "kl": 0.057861328125,
+      "learning_rate": 7.1e-07,
+      "loss": 0.0023,
+      "reward": 1.9166667461395264,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 232
+    },
+    {
+      "completion_length": 66.75000381469727,
+      "epoch": 0.5825,
+      "grad_norm": 7.272895732537956,
+      "kl": 0.061767578125,
+      "learning_rate": 7.0875e-07,
+      "loss": 0.0025,
+      "reward": 1.8750000596046448,
+      "reward_std": 0.3061862215399742,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 233
+    },
+    {
+      "completion_length": 89.83333587646484,
+      "epoch": 0.585,
+      "grad_norm": 9.942526724808292,
+      "kl": 0.05224609375,
+      "learning_rate": 7.075e-07,
+      "loss": 0.0021,
+      "reward": 1.7291667461395264,
+      "reward_std": 0.5663822889328003,
+      "rewards/accuracy_reward": 0.8958333730697632,
+      "rewards/format_reward": 0.8333333730697632,
+      "step": 234
+    },
+    {
+      "completion_length": 82.04166793823242,
+      "epoch": 0.5875,
+      "grad_norm": 7.132091789592793,
+      "kl": 0.0484619140625,
+      "learning_rate": 7.0625e-07,
+      "loss": 0.0019,
+      "reward": 1.7083333730697632,
+      "reward_std": 0.10206207633018494,
+      "rewards/accuracy_reward": 0.75,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 235
+    },
+    {
+      "completion_length": 84.66666793823242,
+      "epoch": 0.59,
+      "grad_norm": 4.743784716908782,
+      "kl": 0.043212890625,
+      "learning_rate": 7.049999999999999e-07,
+      "loss": 0.0017,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.3332235962152481,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.8750000298023224,
+      "step": 236
+    },
+    {
+      "completion_length": 91.54167175292969,
+      "epoch": 0.5925,
+      "grad_norm": 4.935952755278898,
+      "kl": 0.049072265625,
+      "learning_rate": 7.037499999999999e-07,
+      "loss": 0.002,
+      "reward": 1.875,
+      "reward_std": 0.23116151988506317,
+      "rewards/accuracy_reward": 0.9166666865348816,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 237
+    },
+    {
+      "completion_length": 73.50000381469727,
+      "epoch": 0.595,
+      "grad_norm": 0.3195324306878552,
+      "kl": 0.0552978515625,
+      "learning_rate": 7.024999999999999e-07,
+      "loss": 0.0022,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 238
+    },
+    {
+      "completion_length": 75.54166793823242,
+      "epoch": 0.5975,
+      "grad_norm": 0.2519214939758411,
+      "kl": 0.0498046875,
+      "learning_rate": 7.0125e-07,
+      "loss": 0.002,
+      "reward": 1.75,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 0.75,
+      "rewards/format_reward": 1.0,
+      "step": 239
+    },
+    {
+      "completion_length": 71.70833587646484,
+      "epoch": 0.6,
+      "grad_norm": 8.86356803041963,
+      "kl": 0.054931640625,
+      "learning_rate": 7e-07,
+      "loss": 0.0022,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 240
+    },
+    {
+      "completion_length": 65.0,
+      "epoch": 0.6025,
+      "grad_norm": 3.813454727671433,
+      "kl": 0.0618896484375,
+      "learning_rate": 6.9875e-07,
+      "loss": 0.0025,
+      "reward": 1.875,
+      "reward_std": 0.1369306445121765,
+      "rewards/accuracy_reward": 0.875,
+      "rewards/format_reward": 1.0,
+      "step": 241
+    },
+    {
+      "completion_length": 63.45833396911621,
+      "epoch": 0.605,
+      "grad_norm": 9.339337313546885,
+      "kl": 0.054931640625,
+      "learning_rate": 6.975e-07,
+      "loss": 0.0022,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 242
+    },
+    {
+      "completion_length": 84.20833587646484,
+      "epoch": 0.6075,
+      "grad_norm": 6.81096039877308,
+      "kl": 0.06591796875,
+      "learning_rate": 6.9625e-07,
+      "loss": 0.0026,
+      "reward": 1.7916667461395264,
+      "reward_std": 0.43528565019369125,
+      "rewards/accuracy_reward": 0.9166666865348816,
+      "rewards/format_reward": 0.8750000298023224,
+      "step": 243
+    },
+    {
+      "completion_length": 65.20833587646484,
+      "epoch": 0.61,
+      "grad_norm": 6.783355140115774,
+      "kl": 0.060791015625,
+      "learning_rate": 6.949999999999999e-07,
+      "loss": 0.0024,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 244
+    },
+    {
+      "completion_length": 73.04166793823242,
+      "epoch": 0.6125,
+      "grad_norm": 5.741815475332929,
+      "kl": 0.1298828125,
+      "learning_rate": 6.937499999999999e-07,
+      "loss": 0.0052,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.20412415266036987,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 245
+    },
+    {
+      "completion_length": 89.54166793823242,
+      "epoch": 0.615,
+      "grad_norm": 4.665578526736793,
+      "kl": 0.06298828125,
+      "learning_rate": 6.924999999999999e-07,
+      "loss": 0.0025,
+      "reward": 1.9375,
+      "reward_std": 0.1530930995941162,
+      "rewards/accuracy_reward": 0.9375,
+      "rewards/format_reward": 1.0,
+      "step": 246
+    },
+    {
+      "completion_length": 78.08333587646484,
+      "epoch": 0.6175,
+      "grad_norm": 4.511986133714191,
+      "kl": 0.13525390625,
+      "learning_rate": 6.9125e-07,
+      "loss": 0.0054,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 247
+    },
+    {
+      "completion_length": 86.00000381469727,
+      "epoch": 0.62,
+      "grad_norm": 20.96488879364711,
+      "kl": 0.087890625,
+      "learning_rate": 6.9e-07,
+      "loss": 0.0035,
+      "reward": 1.6666666865348816,
+      "reward_std": 0.4943794012069702,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 0.8333333730697632,
+      "step": 248
+    },
+    {
+      "completion_length": 90.20833587646484,
+      "epoch": 0.6225,
+      "grad_norm": 2.77922410156699,
+      "kl": 0.048095703125,
+      "learning_rate": 6.8875e-07,
+      "loss": 0.0019,
+      "reward": 1.75,
+      "reward_std": 0.15811388194561005,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 249
+    },
+    {
+      "completion_length": 100.58333587646484,
+      "epoch": 0.625,
+      "grad_norm": 2.817284456850442,
+      "kl": 0.072021484375,
+      "learning_rate": 6.875e-07,
+      "loss": 0.0029,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 250
+    },
+    {
+      "completion_length": 62.791669845581055,
+      "epoch": 0.6275,
+      "grad_norm": 17.030079043631407,
+      "kl": 0.40380859375,
+      "learning_rate": 6.8625e-07,
+      "loss": 0.0161,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.40824826806783676,
+      "rewards/accuracy_reward": 0.8750000298023224,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 251
+    },
+    {
+      "completion_length": 82.37500381469727,
+      "epoch": 0.63,
+      "grad_norm": 0.21518633979859408,
+      "kl": 0.053955078125,
+      "learning_rate": 6.85e-07,
+      "loss": 0.0022,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 252
+    },
+    {
+      "completion_length": 70.20833396911621,
+      "epoch": 0.6325,
+      "grad_norm": 5.134130102315088,
+      "kl": 0.0743408203125,
+      "learning_rate": 6.837499999999999e-07,
+      "loss": 0.003,
+      "reward": 1.9166667461395264,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 253
+    },
+    {
+      "completion_length": 65.83333587646484,
+      "epoch": 0.635,
+      "grad_norm": 0.20873842927370562,
+      "kl": 0.0518798828125,
+      "learning_rate": 6.824999999999999e-07,
+      "loss": 0.0021,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 254
+    },
+    {
+      "completion_length": 93.41666793823242,
+      "epoch": 0.6375,
+      "grad_norm": 6.749135582171699,
+      "kl": 0.04541015625,
+      "learning_rate": 6.8125e-07,
+      "loss": 0.0018,
+      "reward": 1.7916666865348816,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 255
+    },
+    {
+      "completion_length": 85.16666793823242,
+      "epoch": 0.64,
+      "grad_norm": 2.1300871235777383,
+      "kl": 0.0584716796875,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 0.0023,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 256
+    },
+    {
+      "completion_length": 79.62500381469727,
+      "epoch": 0.6425,
+      "grad_norm": 10.324275974811723,
+      "kl": 0.060546875,
+      "learning_rate": 6.7875e-07,
+      "loss": 0.0024,
+      "reward": 1.9166667461395264,
+      "reward_std": 0.20412414520978928,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 257
+    },
+    {
+      "completion_length": 92.54166793823242,
+      "epoch": 0.645,
+      "grad_norm": 1.8564813514423362,
+      "kl": 0.0521240234375,
+      "learning_rate": 6.775e-07,
+      "loss": 0.0021,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 258
+    },
+    {
+      "completion_length": 79.5,
+      "epoch": 0.6475,
+      "grad_norm": 5.487610324008745,
+      "kl": 0.06298828125,
+      "learning_rate": 6.7625e-07,
+      "loss": 0.0025,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 0.9166666865348816,
+      "rewards/format_reward": 1.0,
+      "step": 259
+    },
+    {
+      "completion_length": 90.79166793823242,
+      "epoch": 0.65,
+      "grad_norm": 4.315055069139165,
+      "kl": 0.061767578125,
+      "learning_rate": 6.75e-07,
+      "loss": 0.0025,
+      "reward": 1.7083333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.75,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 260
+    },
+    {
+      "completion_length": 99.12500381469727,
+      "epoch": 0.6525,
+      "grad_norm": 3.429932230965528,
+      "kl": 0.0589599609375,
+      "learning_rate": 6.737499999999999e-07,
+      "loss": 0.0024,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.23899273574352264,
+      "rewards/accuracy_reward": 0.875,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 261
+    },
+    {
+      "completion_length": 103.58333587646484,
+      "epoch": 0.655,
+      "grad_norm": 6.967938391967251,
+      "kl": 0.0594482421875,
+      "learning_rate": 6.724999999999999e-07,
+      "loss": 0.0024,
+      "reward": 1.7500000596046448,
+      "reward_std": 0.515351228415966,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 262
+    },
+    {
+      "completion_length": 71.70833587646484,
+      "epoch": 0.6575,
+      "grad_norm": 1.8714191151976278,
+      "kl": 0.0615234375,
+      "learning_rate": 6.7125e-07,
+      "loss": 0.0025,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 263
+    },
+    {
+      "completion_length": 70.20833587646484,
+      "epoch": 0.66,
+      "grad_norm": 5.8396222255495775,
+      "kl": 0.0709228515625,
+      "learning_rate": 6.7e-07,
+      "loss": 0.0028,
+      "reward": 1.8750000596046448,
+      "reward_std": 0.3061862140893936,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 264
+    },
+    {
+      "completion_length": 102.41666793823242,
+      "epoch": 0.6625,
+      "grad_norm": 5.33364534247676,
+      "kl": 0.060791015625,
+      "learning_rate": 6.6875e-07,
+      "loss": 0.0024,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.3332235887646675,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.8750000298023224,
+      "step": 265
+    },
+    {
+      "completion_length": 70.16666793823242,
+      "epoch": 0.665,
+      "grad_norm": 3.28533974788896,
+      "kl": 0.07861328125,
+      "learning_rate": 6.675e-07,
+      "loss": 0.0031,
+      "reward": 1.9166667461395264,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 266
+    },
+    {
+      "completion_length": 77.08333587646484,
+      "epoch": 0.6675,
+      "grad_norm": 3.939626781384261,
+      "kl": 0.06298828125,
+      "learning_rate": 6.6625e-07,
+      "loss": 0.0025,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 267
+    },
+    {
+      "completion_length": 77.79167175292969,
+      "epoch": 0.67,
+      "grad_norm": 4.81275939334038,
+      "kl": 0.08203125,
+      "learning_rate": 6.65e-07,
+      "loss": 0.0033,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206207633018494,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 268
+    },
+    {
+      "completion_length": 84.91666793823242,
+      "epoch": 0.6725,
+      "grad_norm": 2.1249954195251233,
+      "kl": 0.0745849609375,
+      "learning_rate": 6.637499999999999e-07,
+      "loss": 0.003,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 269
+    },
+    {
+      "completion_length": 84.08333587646484,
+      "epoch": 0.675,
+      "grad_norm": 5.3477759440562105,
+      "kl": 0.05859375,
+      "learning_rate": 6.624999999999999e-07,
+      "loss": 0.0023,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 270
+    },
+    {
+      "completion_length": 68.50000381469727,
+      "epoch": 0.6775,
+      "grad_norm": 0.2047033038704335,
+      "kl": 0.058837890625,
+      "learning_rate": 6.6125e-07,
+      "loss": 0.0023,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 271
+    },
+    {
+      "completion_length": 65.41666793823242,
+      "epoch": 0.68,
+      "grad_norm": 5.6259746624836104,
+      "kl": 0.10302734375,
+      "learning_rate": 6.6e-07,
+      "loss": 0.0041,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.3332236111164093,
+      "rewards/accuracy_reward": 0.8750000298023224,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 272
+    },
+    {
+      "completion_length": 62.875003814697266,
+      "epoch": 0.6825,
+      "grad_norm": 0.223192104904964,
+      "kl": 0.054443359375,
+      "learning_rate": 6.587499999999999e-07,
+      "loss": 0.0022,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 273
+    },
+    {
+      "completion_length": 72.625,
+      "epoch": 0.685,
+      "grad_norm": 5.697380639686242,
+      "kl": 0.118408203125,
+      "learning_rate": 6.575e-07,
+      "loss": 0.0047,
+      "reward": 1.9166667461395264,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 274
+    },
+    {
+      "completion_length": 65.75000381469727,
+      "epoch": 0.6875,
+      "grad_norm": 4.3740083420763645,
+      "kl": 0.063720703125,
+      "learning_rate": 6.5625e-07,
+      "loss": 0.0025,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 275
+    },
+    {
+      "completion_length": 67.5,
+      "epoch": 0.69,
+      "grad_norm": 6.47041338291971,
+      "kl": 0.052490234375,
+      "learning_rate": 6.55e-07,
+      "loss": 0.0021,
+      "reward": 1.8333333730697632,
+      "reward_std": 0.23899273574352264,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 276
+    },
+    {
+      "completion_length": 68.62500381469727,
+      "epoch": 0.6925,
+      "grad_norm": 3.326676452375192,
+      "kl": 0.08056640625,
+      "learning_rate": 6.5375e-07,
+      "loss": 0.0032,
+      "reward": 1.8750000596046448,
+      "reward_std": 0.306186206638813,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 277
+    },
+    {
+      "completion_length": 69.66666793823242,
+      "epoch": 0.695,
+      "grad_norm": 9.572473982368715,
+      "kl": 0.097412109375,
+      "learning_rate": 6.524999999999999e-07,
+      "loss": 0.0039,
+      "reward": 1.7708333730697632,
+      "reward_std": 0.1530931033194065,
+      "rewards/accuracy_reward": 0.7708333730697632,
+      "rewards/format_reward": 1.0,
+      "step": 278
+    },
+    {
+      "completion_length": 62.208335876464844,
+      "epoch": 0.6975,
+      "grad_norm": 0.2513839533013271,
+      "kl": 0.0548095703125,
+      "learning_rate": 6.5125e-07,
+      "loss": 0.0022,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 279
+    },
+    {
+      "completion_length": 59.79166793823242,
+      "epoch": 0.7,
+      "grad_norm": 0.3018927564996281,
+      "kl": 0.064453125,
+      "learning_rate": 6.5e-07,
+      "loss": 0.0026,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 280
+    },
+    {
+      "completion_length": 67.58333587646484,
+      "epoch": 0.7025,
+      "grad_norm": 3.022076491480647,
+      "kl": 0.1083984375,
+      "learning_rate": 6.4875e-07,
+      "loss": 0.0043,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 281
+    },
+    {
+      "completion_length": 70.58333587646484,
+      "epoch": 0.705,
+      "grad_norm": 0.4005563509564593,
+      "kl": 0.060546875,
+      "learning_rate": 6.474999999999999e-07,
+      "loss": 0.0024,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 282
+    },
+    {
+      "completion_length": 67.83333587646484,
+      "epoch": 0.7075,
+      "grad_norm": 3.3165615677681215,
+      "kl": 0.0682373046875,
+      "learning_rate": 6.4625e-07,
+      "loss": 0.0027,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 283
+    },
+    {
+      "completion_length": 76.5,
+      "epoch": 0.71,
+      "grad_norm": 7.157915025138244,
+      "kl": 0.0810546875,
+      "learning_rate": 6.45e-07,
+      "loss": 0.0032,
+      "reward": 1.6666667461395264,
+      "reward_std": 0.3680921941995621,
+      "rewards/accuracy_reward": 0.7083333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 284
+    },
+    {
+      "completion_length": 78.95833587646484,
+      "epoch": 0.7125,
+      "grad_norm": 2.4773582650919574,
+      "kl": 0.073974609375,
+      "learning_rate": 6.4375e-07,
+      "loss": 0.003,
+      "reward": 1.7083333730697632,
+      "reward_std": 0.10206207633018494,
+      "rewards/accuracy_reward": 0.75,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 285
+    },
+    {
+      "completion_length": 73.29166793823242,
+      "epoch": 0.715,
+      "grad_norm": 3.5419075074377937,
+      "kl": 0.0643310546875,
+      "learning_rate": 6.424999999999999e-07,
+      "loss": 0.0026,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206207633018494,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 286
+    },
+    {
+      "completion_length": 66.29166793823242,
+      "epoch": 0.7175,
+      "grad_norm": 4.248074629554753,
+      "kl": 0.072998046875,
+      "learning_rate": 6.4125e-07,
+      "loss": 0.0029,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.20412415266036987,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 287
+    },
+    {
+      "completion_length": 65.91666984558105,
+      "epoch": 0.72,
+      "grad_norm": 4.09755233598868,
+      "kl": 0.1015625,
+      "learning_rate": 6.4e-07,
+      "loss": 0.004,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.20412415266036987,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 288
+    },
+    {
+      "completion_length": 59.16666793823242,
+      "epoch": 0.7225,
+      "grad_norm": 5.392706724219783,
+      "kl": 0.061279296875,
+      "learning_rate": 6.3875e-07,
+      "loss": 0.0024,
+      "reward": 1.875,
+      "reward_std": 0.1369306445121765,
+      "rewards/accuracy_reward": 0.875,
+      "rewards/format_reward": 1.0,
+      "step": 289
+    },
+    {
+      "completion_length": 65.87500190734863,
+      "epoch": 0.725,
+      "grad_norm": 0.6612757501656221,
+      "kl": 0.08935546875,
+      "learning_rate": 6.374999999999999e-07,
+      "loss": 0.0036,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 290
+    },
+    {
+      "completion_length": 69.75000381469727,
+      "epoch": 0.7275,
+      "grad_norm": 2.017457988128028,
+      "kl": 0.078369140625,
+      "learning_rate": 6.362499999999999e-07,
+      "loss": 0.0031,
+      "reward": 1.75,
+      "reward_std": 0.15811388194561005,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 291
+    },
+    {
+      "completion_length": 87.20833587646484,
+      "epoch": 0.73,
+      "grad_norm": 3.7133430453190055,
+      "kl": 0.063720703125,
+      "learning_rate": 6.35e-07,
+      "loss": 0.0026,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 1.0,
+      "step": 292
+    },
+    {
+      "completion_length": 71.875,
+      "epoch": 0.7325,
+      "grad_norm": 0.2544317598109972,
+      "kl": 0.066650390625,
+      "learning_rate": 6.3375e-07,
+      "loss": 0.0027,
+      "reward": 2.0,
+      "reward_std": 0.0,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 1.0,
+      "step": 293
+    },
+    {
+      "completion_length": 81.95833587646484,
+      "epoch": 0.735,
+      "grad_norm": 8.180192030414677,
+      "kl": 0.262451171875,
+      "learning_rate": 6.324999999999999e-07,
+      "loss": 0.0105,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 294
+    },
+    {
+      "completion_length": 81.41666793823242,
+      "epoch": 0.7375,
+      "grad_norm": 2.144538696812927,
+      "kl": 0.06494140625,
+      "learning_rate": 6.3125e-07,
+      "loss": 0.0026,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 295
+    },
+    {
+      "completion_length": 93.875,
+      "epoch": 0.74,
+      "grad_norm": 2.7966760839221565,
+      "kl": 0.062255859375,
+      "learning_rate": 6.3e-07,
+      "loss": 0.0025,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206207633018494,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 296
+    },
+    {
+      "completion_length": 101.16666793823242,
+      "epoch": 0.7425,
+      "grad_norm": 2.40302610010562,
+      "kl": 0.0517578125,
+      "learning_rate": 6.2875e-07,
+      "loss": 0.0021,
+      "reward": 1.9166666865348816,
+      "reward_std": 0.20412413775920868,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 297
+    },
+    {
+      "completion_length": 90.25000381469727,
+      "epoch": 0.745,
+      "grad_norm": 2.4971803044396963,
+      "kl": 0.0684814453125,
+      "learning_rate": 6.274999999999999e-07,
+      "loss": 0.0027,
+      "reward": 1.875,
+      "reward_std": 0.20916502177715302,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 298
+    },
+    {
+      "completion_length": 108.16666793823242,
+      "epoch": 0.7475,
+      "grad_norm": 5.629330551809588,
+      "kl": 0.0693359375,
+      "learning_rate": 6.262499999999999e-07,
+      "loss": 0.0028,
+      "reward": 1.9583333730697632,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 1.0,
+      "rewards/format_reward": 0.9583333432674408,
+      "step": 299
+    },
+    {
+      "completion_length": 66.75,
+      "epoch": 0.75,
+      "grad_norm": 7.215837542303994,
+      "kl": 0.0615234375,
+      "learning_rate": 6.249999999999999e-07,
+      "loss": 0.0025,
+      "reward": 1.875,
+      "reward_std": 0.3061861991882324,
+      "rewards/accuracy_reward": 0.9583333432674408,
+      "rewards/format_reward": 0.9166666865348816,
+      "step": 300
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 800,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 50,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}