{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 7.0,
  "eval_steps": 500,
  "global_step": 280,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.05,
      "learning_rate": 1e-05,
      "loss": 0.3723,
      "step": 2
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.964028776978418e-06,
      "loss": 0.3063,
      "step": 4
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.892086330935252e-06,
      "loss": 0.1585,
      "step": 6
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.820143884892086e-06,
      "loss": 0.1134,
      "step": 8
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.748201438848922e-06,
      "loss": 0.0801,
      "step": 10
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.676258992805757e-06,
      "loss": 0.0699,
      "step": 12
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.60431654676259e-06,
      "loss": 0.0568,
      "step": 14
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.532374100719425e-06,
      "loss": 0.0654,
      "step": 16
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.46043165467626e-06,
      "loss": 0.0608,
      "step": 18
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.388489208633095e-06,
      "loss": 0.0469,
      "step": 20
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.31654676258993e-06,
      "loss": 0.0453,
      "step": 22
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.244604316546764e-06,
      "loss": 0.0412,
      "step": 24
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.172661870503598e-06,
      "loss": 0.0335,
      "step": 26
    },
    {
      "epoch": 0.7,
      "learning_rate": 9.100719424460432e-06,
      "loss": 0.0382,
      "step": 28
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.028776978417268e-06,
      "loss": 0.0356,
      "step": 30
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.956834532374102e-06,
      "loss": 0.0363,
      "step": 32
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.884892086330936e-06,
      "loss": 0.034,
      "step": 34
    },
    {
      "epoch": 0.9,
      "learning_rate": 8.81294964028777e-06,
      "loss": 0.0371,
      "step": 36
    },
    {
      "epoch": 0.95,
      "learning_rate": 8.741007194244605e-06,
      "loss": 0.0316,
      "step": 38
    },
    {
      "epoch": 1.0,
      "learning_rate": 8.66906474820144e-06,
      "loss": 0.0295,
      "step": 40
    },
    {
      "epoch": 1.05,
      "learning_rate": 8.597122302158273e-06,
      "loss": 0.0196,
      "step": 42
    },
    {
      "epoch": 1.1,
      "learning_rate": 8.525179856115109e-06,
      "loss": 0.0191,
      "step": 44
    },
    {
      "epoch": 1.15,
      "learning_rate": 8.453237410071943e-06,
      "loss": 0.0198,
      "step": 46
    },
    {
      "epoch": 1.2,
      "learning_rate": 8.381294964028778e-06,
      "loss": 0.0169,
      "step": 48
    },
    {
      "epoch": 1.25,
      "learning_rate": 8.309352517985614e-06,
      "loss": 0.0151,
      "step": 50
    },
    {
      "epoch": 1.3,
      "learning_rate": 8.237410071942446e-06,
      "loss": 0.0151,
      "step": 52
    },
    {
      "epoch": 1.35,
      "learning_rate": 8.165467625899282e-06,
      "loss": 0.0191,
      "step": 54
    },
    {
      "epoch": 1.4,
      "learning_rate": 8.093525179856116e-06,
      "loss": 0.0189,
      "step": 56
    },
    {
      "epoch": 1.45,
      "learning_rate": 8.02158273381295e-06,
      "loss": 0.0184,
      "step": 58
    },
    {
      "epoch": 1.5,
      "learning_rate": 7.949640287769785e-06,
      "loss": 0.019,
      "step": 60
    },
    {
      "epoch": 1.55,
      "learning_rate": 7.877697841726619e-06,
      "loss": 0.0349,
      "step": 62
    },
    {
      "epoch": 1.6,
      "learning_rate": 7.805755395683455e-06,
      "loss": 0.0206,
      "step": 64
    },
    {
      "epoch": 1.65,
      "learning_rate": 7.733812949640287e-06,
      "loss": 0.0179,
      "step": 66
    },
    {
      "epoch": 1.7,
      "learning_rate": 7.661870503597123e-06,
      "loss": 0.0182,
      "step": 68
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.589928057553958e-06,
      "loss": 0.0171,
      "step": 70
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.517985611510792e-06,
      "loss": 0.0194,
      "step": 72
    },
    {
      "epoch": 1.85,
      "learning_rate": 7.446043165467627e-06,
      "loss": 0.0192,
      "step": 74
    },
    {
      "epoch": 1.9,
      "learning_rate": 7.374100719424461e-06,
      "loss": 0.018,
      "step": 76
    },
    {
      "epoch": 1.95,
      "learning_rate": 7.302158273381296e-06,
      "loss": 0.019,
      "step": 78
    },
    {
      "epoch": 2.0,
      "learning_rate": 7.230215827338129e-06,
      "loss": 0.0196,
      "step": 80
    },
    {
      "epoch": 2.05,
      "learning_rate": 7.1582733812949644e-06,
      "loss": 0.0081,
      "step": 82
    },
    {
      "epoch": 2.1,
      "learning_rate": 7.0863309352517995e-06,
      "loss": 0.0085,
      "step": 84
    },
    {
      "epoch": 2.15,
      "learning_rate": 7.014388489208634e-06,
      "loss": 0.0072,
      "step": 86
    },
    {
      "epoch": 2.2,
      "learning_rate": 6.942446043165468e-06,
      "loss": 0.0071,
      "step": 88
    },
    {
      "epoch": 2.25,
      "learning_rate": 6.870503597122302e-06,
      "loss": 0.0075,
      "step": 90
    },
    {
      "epoch": 2.3,
      "learning_rate": 6.798561151079137e-06,
      "loss": 0.0065,
      "step": 92
    },
    {
      "epoch": 2.35,
      "learning_rate": 6.726618705035972e-06,
      "loss": 0.0084,
      "step": 94
    },
    {
      "epoch": 2.4,
      "learning_rate": 6.654676258992806e-06,
      "loss": 0.0063,
      "step": 96
    },
    {
      "epoch": 2.45,
      "learning_rate": 6.582733812949641e-06,
      "loss": 0.0079,
      "step": 98
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.510791366906475e-06,
      "loss": 0.0076,
      "step": 100
    },
    {
      "epoch": 2.55,
      "learning_rate": 6.43884892086331e-06,
      "loss": 0.0069,
      "step": 102
    },
    {
      "epoch": 2.6,
      "learning_rate": 6.366906474820145e-06,
      "loss": 0.0086,
      "step": 104
    },
    {
      "epoch": 2.65,
      "learning_rate": 6.2949640287769785e-06,
      "loss": 0.0063,
      "step": 106
    },
    {
      "epoch": 2.7,
      "learning_rate": 6.2230215827338136e-06,
      "loss": 0.0082,
      "step": 108
    },
    {
      "epoch": 2.75,
      "learning_rate": 6.151079136690648e-06,
      "loss": 0.0063,
      "step": 110
    },
    {
      "epoch": 2.8,
      "learning_rate": 6.079136690647483e-06,
      "loss": 0.0059,
      "step": 112
    },
    {
      "epoch": 2.85,
      "learning_rate": 6.007194244604317e-06,
      "loss": 0.0078,
      "step": 114
    },
    {
      "epoch": 2.9,
      "learning_rate": 5.935251798561151e-06,
      "loss": 0.0062,
      "step": 116
    },
    {
      "epoch": 2.95,
      "learning_rate": 5.863309352517986e-06,
      "loss": 0.0072,
      "step": 118
    },
    {
      "epoch": 3.0,
      "learning_rate": 5.79136690647482e-06,
      "loss": 0.0053,
      "step": 120
    },
    {
      "epoch": 3.05,
      "learning_rate": 5.719424460431655e-06,
      "loss": 0.0026,
      "step": 122
    },
    {
      "epoch": 3.1,
      "learning_rate": 5.64748201438849e-06,
      "loss": 0.0026,
      "step": 124
    },
    {
      "epoch": 3.15,
      "learning_rate": 5.575539568345324e-06,
      "loss": 0.003,
      "step": 126
    },
    {
      "epoch": 3.2,
      "learning_rate": 5.503597122302159e-06,
      "loss": 0.0026,
      "step": 128
    },
    {
      "epoch": 3.25,
      "learning_rate": 5.4316546762589925e-06,
      "loss": 0.0023,
      "step": 130
    },
    {
      "epoch": 3.3,
      "learning_rate": 5.359712230215828e-06,
      "loss": 0.0021,
      "step": 132
    },
    {
      "epoch": 3.35,
      "learning_rate": 5.287769784172663e-06,
      "loss": 0.0025,
      "step": 134
    },
    {
      "epoch": 3.4,
      "learning_rate": 5.215827338129497e-06,
      "loss": 0.0027,
      "step": 136
    },
    {
      "epoch": 3.45,
      "learning_rate": 5.143884892086332e-06,
      "loss": 0.0028,
      "step": 138
    },
    {
      "epoch": 3.5,
      "learning_rate": 5.071942446043165e-06,
      "loss": 0.0028,
      "step": 140
    },
    {
      "epoch": 3.55,
      "learning_rate": 5e-06,
      "loss": 0.0034,
      "step": 142
    },
    {
      "epoch": 3.6,
      "learning_rate": 4.928057553956835e-06,
      "loss": 0.0033,
      "step": 144
    },
    {
      "epoch": 3.65,
      "learning_rate": 4.856115107913669e-06,
      "loss": 0.0027,
      "step": 146
    },
    {
      "epoch": 3.7,
      "learning_rate": 4.784172661870504e-06,
      "loss": 0.0021,
      "step": 148
    },
    {
      "epoch": 3.75,
      "learning_rate": 4.712230215827339e-06,
      "loss": 0.0027,
      "step": 150
    },
    {
      "epoch": 3.8,
      "learning_rate": 4.640287769784173e-06,
      "loss": 0.003,
      "step": 152
    },
    {
      "epoch": 3.85,
      "learning_rate": 4.5683453237410074e-06,
      "loss": 0.0028,
      "step": 154
    },
    {
      "epoch": 3.9,
      "learning_rate": 4.496402877697842e-06,
      "loss": 0.0039,
      "step": 156
    },
    {
      "epoch": 3.95,
      "learning_rate": 4.424460431654677e-06,
      "loss": 0.0034,
      "step": 158
    },
    {
      "epoch": 4.0,
      "learning_rate": 4.352517985611511e-06,
      "loss": 0.0033,
      "step": 160
    },
    {
      "epoch": 4.05,
      "learning_rate": 4.280575539568346e-06,
      "loss": 0.0014,
      "step": 162
    },
    {
      "epoch": 4.1,
      "learning_rate": 4.20863309352518e-06,
      "loss": 0.0012,
      "step": 164
    },
    {
      "epoch": 4.15,
      "learning_rate": 4.1366906474820145e-06,
      "loss": 0.0014,
      "step": 166
    },
    {
      "epoch": 4.2,
      "learning_rate": 4.0647482014388495e-06,
      "loss": 0.0013,
      "step": 168
    },
    {
      "epoch": 4.25,
      "learning_rate": 3.992805755395684e-06,
      "loss": 0.0015,
      "step": 170
    },
    {
      "epoch": 4.3,
      "learning_rate": 3.920863309352518e-06,
      "loss": 0.001,
      "step": 172
    },
    {
      "epoch": 4.35,
      "learning_rate": 3.848920863309353e-06,
      "loss": 0.0011,
      "step": 174
    },
    {
      "epoch": 4.4,
      "learning_rate": 3.7769784172661873e-06,
      "loss": 0.001,
      "step": 176
    },
    {
      "epoch": 4.45,
      "learning_rate": 3.7050359712230215e-06,
      "loss": 0.0018,
      "step": 178
    },
    {
      "epoch": 4.5,
      "learning_rate": 3.6330935251798566e-06,
      "loss": 0.0009,
      "step": 180
    },
    {
      "epoch": 4.55,
      "learning_rate": 3.561151079136691e-06,
      "loss": 0.0009,
      "step": 182
    },
    {
      "epoch": 4.6,
      "learning_rate": 3.4892086330935254e-06,
      "loss": 0.0013,
      "step": 184
    },
    {
      "epoch": 4.65,
      "learning_rate": 3.4172661870503596e-06,
      "loss": 0.001,
      "step": 186
    },
    {
      "epoch": 4.7,
      "learning_rate": 3.3453237410071943e-06,
      "loss": 0.0011,
      "step": 188
    },
    {
      "epoch": 4.75,
      "learning_rate": 3.2733812949640294e-06,
      "loss": 0.0011,
      "step": 190
    },
    {
      "epoch": 4.8,
      "learning_rate": 3.2014388489208636e-06,
      "loss": 0.0012,
      "step": 192
    },
    {
      "epoch": 4.85,
      "learning_rate": 3.1294964028776982e-06,
      "loss": 0.0014,
      "step": 194
    },
    {
      "epoch": 4.9,
      "learning_rate": 3.0575539568345324e-06,
      "loss": 0.0011,
      "step": 196
    },
    {
      "epoch": 4.95,
      "learning_rate": 2.985611510791367e-06,
      "loss": 0.001,
      "step": 198
    },
    {
      "epoch": 5.0,
      "learning_rate": 2.9136690647482017e-06,
      "loss": 0.0011,
      "step": 200
    },
    {
      "epoch": 5.05,
      "learning_rate": 2.8417266187050364e-06,
      "loss": 0.0004,
      "step": 202
    },
    {
      "epoch": 5.1,
      "learning_rate": 2.7697841726618706e-06,
      "loss": 0.0005,
      "step": 204
    },
    {
      "epoch": 5.15,
      "learning_rate": 2.6978417266187052e-06,
      "loss": 0.0003,
      "step": 206
    },
    {
      "epoch": 5.2,
      "learning_rate": 2.6258992805755395e-06,
      "loss": 0.0003,
      "step": 208
    },
    {
      "epoch": 5.25,
      "learning_rate": 2.5539568345323745e-06,
      "loss": 0.0005,
      "step": 210
    },
    {
      "epoch": 5.3,
      "learning_rate": 2.4820143884892088e-06,
      "loss": 0.0004,
      "step": 212
    },
    {
      "epoch": 5.35,
      "learning_rate": 2.4100719424460434e-06,
      "loss": 0.0007,
      "step": 214
    },
    {
      "epoch": 5.4,
      "learning_rate": 2.3381294964028776e-06,
      "loss": 0.0003,
      "step": 216
    },
    {
      "epoch": 5.45,
      "learning_rate": 2.2661870503597123e-06,
      "loss": 0.0005,
      "step": 218
    },
    {
      "epoch": 5.5,
      "learning_rate": 2.194244604316547e-06,
      "loss": 0.0004,
      "step": 220
    },
    {
      "epoch": 5.55,
      "learning_rate": 2.1223021582733816e-06,
      "loss": 0.0005,
      "step": 222
    },
    {
      "epoch": 5.6,
      "learning_rate": 2.050359712230216e-06,
      "loss": 0.0003,
      "step": 224
    },
    {
      "epoch": 5.65,
      "learning_rate": 1.9784172661870504e-06,
      "loss": 0.0005,
      "step": 226
    },
    {
      "epoch": 5.7,
      "learning_rate": 1.906474820143885e-06,
      "loss": 0.0005,
      "step": 228
    },
    {
      "epoch": 5.75,
      "learning_rate": 1.8345323741007195e-06,
      "loss": 0.0004,
      "step": 230
    },
    {
      "epoch": 5.8,
      "learning_rate": 1.762589928057554e-06,
      "loss": 0.0003,
      "step": 232
    },
    {
      "epoch": 5.85,
      "learning_rate": 1.6906474820143886e-06,
      "loss": 0.0006,
      "step": 234
    },
    {
      "epoch": 5.9,
      "learning_rate": 1.618705035971223e-06,
      "loss": 0.0004,
      "step": 236
    },
    {
      "epoch": 5.95,
      "learning_rate": 1.5467625899280579e-06,
      "loss": 0.0005,
      "step": 238
    },
    {
      "epoch": 6.0,
      "learning_rate": 1.474820143884892e-06,
      "loss": 0.0003,
      "step": 240
    },
    {
      "epoch": 6.05,
      "learning_rate": 1.4028776978417265e-06,
      "loss": 0.0002,
      "step": 242
    },
    {
      "epoch": 6.1,
      "learning_rate": 1.3309352517985614e-06,
      "loss": 0.0002,
      "step": 244
    },
    {
      "epoch": 6.15,
      "learning_rate": 1.2589928057553958e-06,
      "loss": 0.0001,
      "step": 246
    },
    {
      "epoch": 6.2,
      "learning_rate": 1.1870503597122303e-06,
      "loss": 0.0001,
      "step": 248
    },
    {
      "epoch": 6.25,
      "learning_rate": 1.115107913669065e-06,
      "loss": 0.0001,
      "step": 250
    },
    {
      "epoch": 6.3,
      "learning_rate": 1.0431654676258993e-06,
      "loss": 0.0001,
      "step": 252
    },
    {
      "epoch": 6.35,
      "learning_rate": 9.71223021582734e-07,
      "loss": 0.0002,
      "step": 254
    },
    {
      "epoch": 6.4,
      "learning_rate": 8.992805755395684e-07,
      "loss": 0.0001,
      "step": 256
    },
    {
      "epoch": 6.45,
      "learning_rate": 8.27338129496403e-07,
      "loss": 0.0001,
      "step": 258
    },
    {
      "epoch": 6.5,
      "learning_rate": 7.553956834532375e-07,
      "loss": 0.0002,
      "step": 260
    },
    {
      "epoch": 6.55,
      "learning_rate": 6.83453237410072e-07,
      "loss": 0.0002,
      "step": 262
    },
    {
      "epoch": 6.6,
      "learning_rate": 6.115107913669066e-07,
      "loss": 0.0001,
      "step": 264
    },
    {
      "epoch": 6.65,
      "learning_rate": 5.39568345323741e-07,
      "loss": 0.0002,
      "step": 266
    },
    {
      "epoch": 6.7,
      "learning_rate": 4.676258992805756e-07,
      "loss": 0.0001,
      "step": 268
    },
    {
      "epoch": 6.75,
      "learning_rate": 3.956834532374101e-07,
      "loss": 0.0002,
      "step": 270
    },
    {
      "epoch": 6.8,
      "learning_rate": 3.237410071942446e-07,
      "loss": 0.0001,
      "step": 272
    },
    {
      "epoch": 6.85,
      "learning_rate": 2.5179856115107916e-07,
      "loss": 0.0001,
      "step": 274
    },
    {
      "epoch": 6.9,
      "learning_rate": 1.7985611510791368e-07,
      "loss": 0.0001,
      "step": 276
    },
    {
      "epoch": 6.95,
      "learning_rate": 1.0791366906474822e-07,
      "loss": 0.0001,
      "step": 278
    },
    {
      "epoch": 7.0,
      "learning_rate": 3.597122302158274e-08,
      "loss": 0.0001,
      "step": 280
    }
  ],
  "logging_steps": 2,
  "max_steps": 280,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 7,
  "save_steps": 40,
  "total_flos": 58523658485760.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}