{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 3120,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0016025641025641025,
      "grad_norm": 213.62364196777344,
      "learning_rate": 4.998397435897436e-05,
      "loss": 9.3378,
      "step": 1
    },
    {
      "epoch": 0.049679487179487176,
      "grad_norm": 43.086360931396484,
      "learning_rate": 4.950320512820513e-05,
      "loss": 2.5252,
      "step": 31
    },
    {
      "epoch": 0.09935897435897435,
      "grad_norm": 48.219451904296875,
      "learning_rate": 4.9006410256410256e-05,
      "loss": 1.5213,
      "step": 62
    },
    {
      "epoch": 0.14903846153846154,
      "grad_norm": 30.23198127746582,
      "learning_rate": 4.850961538461539e-05,
      "loss": 1.9636,
      "step": 93
    },
    {
      "epoch": 0.1987179487179487,
      "grad_norm": 52.96772766113281,
      "learning_rate": 4.8012820512820516e-05,
      "loss": 1.3545,
      "step": 124
    },
    {
      "epoch": 0.2483974358974359,
      "grad_norm": 48.79069900512695,
      "learning_rate": 4.751602564102564e-05,
      "loss": 1.4998,
      "step": 155
    },
    {
      "epoch": 0.2980769230769231,
      "grad_norm": 38.8032341003418,
      "learning_rate": 4.701923076923077e-05,
      "loss": 1.4654,
      "step": 186
    },
    {
      "epoch": 0.34775641025641024,
      "grad_norm": 106.62429809570312,
      "learning_rate": 4.65224358974359e-05,
      "loss": 1.8027,
      "step": 217
    },
    {
      "epoch": 0.3974358974358974,
      "grad_norm": 25.370744705200195,
      "learning_rate": 4.602564102564102e-05,
      "loss": 1.4394,
      "step": 248
    },
    {
      "epoch": 0.44711538461538464,
      "grad_norm": 82.52824401855469,
      "learning_rate": 4.5528846153846157e-05,
      "loss": 1.4573,
      "step": 279
    },
    {
      "epoch": 0.4967948717948718,
      "grad_norm": 19.148210525512695,
      "learning_rate": 4.503205128205128e-05,
      "loss": 1.2547,
      "step": 310
    },
    {
      "epoch": 0.5464743589743589,
      "grad_norm": 46.984779357910156,
      "learning_rate": 4.453525641025642e-05,
      "loss": 1.3297,
      "step": 341
    },
    {
      "epoch": 0.5961538461538461,
      "grad_norm": 45.50709915161133,
      "learning_rate": 4.403846153846154e-05,
      "loss": 1.3359,
      "step": 372
    },
    {
      "epoch": 0.6458333333333334,
      "grad_norm": 36.45267105102539,
      "learning_rate": 4.354166666666667e-05,
      "loss": 1.2466,
      "step": 403
    },
    {
      "epoch": 0.6955128205128205,
      "grad_norm": 20.8362979888916,
      "learning_rate": 4.30448717948718e-05,
      "loss": 1.1206,
      "step": 434
    },
    {
      "epoch": 0.7451923076923077,
      "grad_norm": 64.35557556152344,
      "learning_rate": 4.2548076923076924e-05,
      "loss": 1.3676,
      "step": 465
    },
    {
      "epoch": 0.7948717948717948,
      "grad_norm": 12.930615425109863,
      "learning_rate": 4.205128205128206e-05,
      "loss": 1.3871,
      "step": 496
    },
    {
      "epoch": 0.844551282051282,
      "grad_norm": 51.6870002746582,
      "learning_rate": 4.1554487179487184e-05,
      "loss": 1.0038,
      "step": 527
    },
    {
      "epoch": 0.8942307692307693,
      "grad_norm": 14.677131652832031,
      "learning_rate": 4.105769230769231e-05,
      "loss": 1.2283,
      "step": 558
    },
    {
      "epoch": 0.9439102564102564,
      "grad_norm": 43.24152755737305,
      "learning_rate": 4.056089743589744e-05,
      "loss": 1.0872,
      "step": 589
    },
    {
      "epoch": 0.9935897435897436,
      "grad_norm": 25.641794204711914,
      "learning_rate": 4.006410256410257e-05,
      "loss": 0.9991,
      "step": 620
    },
    {
      "epoch": 1.0432692307692308,
      "grad_norm": 36.622703552246094,
      "learning_rate": 3.956730769230769e-05,
      "loss": 0.9941,
      "step": 651
    },
    {
      "epoch": 1.092948717948718,
      "grad_norm": 17.249757766723633,
      "learning_rate": 3.9070512820512824e-05,
      "loss": 0.9877,
      "step": 682
    },
    {
      "epoch": 1.142628205128205,
      "grad_norm": 31.591552734375,
      "learning_rate": 3.857371794871795e-05,
      "loss": 0.8917,
      "step": 713
    },
    {
      "epoch": 1.1923076923076923,
      "grad_norm": 28.23312759399414,
      "learning_rate": 3.807692307692308e-05,
      "loss": 0.9264,
      "step": 744
    },
    {
      "epoch": 1.2419871794871795,
      "grad_norm": 37.707183837890625,
      "learning_rate": 3.7580128205128204e-05,
      "loss": 0.849,
      "step": 775
    },
    {
      "epoch": 1.2916666666666667,
      "grad_norm": 12.67361831665039,
      "learning_rate": 3.708333333333334e-05,
      "loss": 0.8255,
      "step": 806
    },
    {
      "epoch": 1.3413461538461537,
      "grad_norm": 39.1461181640625,
      "learning_rate": 3.658653846153846e-05,
      "loss": 0.7372,
      "step": 837
    },
    {
      "epoch": 1.391025641025641,
      "grad_norm": 33.27256393432617,
      "learning_rate": 3.608974358974359e-05,
      "loss": 0.8227,
      "step": 868
    },
    {
      "epoch": 1.4407051282051282,
      "grad_norm": 545.9265747070312,
      "learning_rate": 3.559294871794872e-05,
      "loss": 0.7941,
      "step": 899
    },
    {
      "epoch": 1.4903846153846154,
      "grad_norm": 23.55372428894043,
      "learning_rate": 3.5096153846153845e-05,
      "loss": 0.8686,
      "step": 930
    },
    {
      "epoch": 1.5400641025641026,
      "grad_norm": 22.64493179321289,
      "learning_rate": 3.459935897435898e-05,
      "loss": 0.7059,
      "step": 961
    },
    {
      "epoch": 1.5897435897435899,
      "grad_norm": 79.4326171875,
      "learning_rate": 3.4102564102564105e-05,
      "loss": 0.7567,
      "step": 992
    },
    {
      "epoch": 1.6394230769230769,
      "grad_norm": 13.156254768371582,
      "learning_rate": 3.360576923076923e-05,
      "loss": 0.7138,
      "step": 1023
    },
    {
      "epoch": 1.689102564102564,
      "grad_norm": 39.97475051879883,
      "learning_rate": 3.310897435897436e-05,
      "loss": 0.676,
      "step": 1054
    },
    {
      "epoch": 1.7387820512820513,
      "grad_norm": 24.62538719177246,
      "learning_rate": 3.261217948717949e-05,
      "loss": 0.683,
      "step": 1085
    },
    {
      "epoch": 1.7884615384615383,
      "grad_norm": 18.064680099487305,
      "learning_rate": 3.211538461538462e-05,
      "loss": 0.5746,
      "step": 1116
    },
    {
      "epoch": 1.8381410256410255,
      "grad_norm": 10.56497573852539,
      "learning_rate": 3.1618589743589746e-05,
      "loss": 0.5881,
      "step": 1147
    },
    {
      "epoch": 1.8878205128205128,
      "grad_norm": 43.007911682128906,
      "learning_rate": 3.112179487179487e-05,
      "loss": 0.78,
      "step": 1178
    },
    {
      "epoch": 1.9375,
      "grad_norm": 25.748680114746094,
      "learning_rate": 3.0625000000000006e-05,
      "loss": 0.592,
      "step": 1209
    },
    {
      "epoch": 1.9871794871794872,
      "grad_norm": 20.511089324951172,
      "learning_rate": 3.012820512820513e-05,
      "loss": 0.6807,
      "step": 1240
    },
    {
      "epoch": 2.0368589743589745,
      "grad_norm": 26.275846481323242,
      "learning_rate": 2.963141025641026e-05,
      "loss": 0.6935,
      "step": 1271
    },
    {
      "epoch": 2.0865384615384617,
      "grad_norm": 69.18566131591797,
      "learning_rate": 2.913461538461539e-05,
      "loss": 0.511,
      "step": 1302
    },
    {
      "epoch": 2.136217948717949,
      "grad_norm": 4.620207786560059,
      "learning_rate": 2.8637820512820513e-05,
      "loss": 0.5338,
      "step": 1333
    },
    {
      "epoch": 2.185897435897436,
      "grad_norm": 44.329566955566406,
      "learning_rate": 2.8141025641025643e-05,
      "loss": 0.4471,
      "step": 1364
    },
    {
      "epoch": 2.235576923076923,
      "grad_norm": 24.95821762084961,
      "learning_rate": 2.7644230769230773e-05,
      "loss": 0.5295,
      "step": 1395
    },
    {
      "epoch": 2.28525641025641,
      "grad_norm": 32.0108757019043,
      "learning_rate": 2.7147435897435896e-05,
      "loss": 0.5075,
      "step": 1426
    },
    {
      "epoch": 2.3349358974358974,
      "grad_norm": 15.610992431640625,
      "learning_rate": 2.6650641025641026e-05,
      "loss": 0.4898,
      "step": 1457
    },
    {
      "epoch": 2.3846153846153846,
      "grad_norm": 6.399534225463867,
      "learning_rate": 2.6153846153846157e-05,
      "loss": 0.4273,
      "step": 1488
    },
    {
      "epoch": 2.434294871794872,
      "grad_norm": 22.35305404663086,
      "learning_rate": 2.565705128205128e-05,
      "loss": 0.4826,
      "step": 1519
    },
    {
      "epoch": 2.483974358974359,
      "grad_norm": 26.82170867919922,
      "learning_rate": 2.516025641025641e-05,
      "loss": 0.4472,
      "step": 1550
    },
    {
      "epoch": 2.5336538461538463,
      "grad_norm": 31.885419845581055,
      "learning_rate": 2.466346153846154e-05,
      "loss": 0.3986,
      "step": 1581
    },
    {
      "epoch": 2.5833333333333335,
      "grad_norm": 21.020639419555664,
      "learning_rate": 2.4166666666666667e-05,
      "loss": 0.4856,
      "step": 1612
    },
    {
      "epoch": 2.6330128205128203,
      "grad_norm": 10.991323471069336,
      "learning_rate": 2.3669871794871794e-05,
      "loss": 0.5008,
      "step": 1643
    },
    {
      "epoch": 2.6826923076923075,
      "grad_norm": 17.077383041381836,
      "learning_rate": 2.3173076923076924e-05,
      "loss": 0.3585,
      "step": 1674
    },
    {
      "epoch": 2.7323717948717947,
      "grad_norm": 6.274226188659668,
      "learning_rate": 2.2676282051282054e-05,
      "loss": 0.4101,
      "step": 1705
    },
    {
      "epoch": 2.782051282051282,
      "grad_norm": 7.574007511138916,
      "learning_rate": 2.217948717948718e-05,
      "loss": 0.3604,
      "step": 1736
    },
    {
      "epoch": 2.831730769230769,
      "grad_norm": 15.08122730255127,
      "learning_rate": 2.168269230769231e-05,
      "loss": 0.407,
      "step": 1767
    },
    {
      "epoch": 2.8814102564102564,
      "grad_norm": 32.56257247924805,
      "learning_rate": 2.1185897435897437e-05,
      "loss": 0.4002,
      "step": 1798
    },
    {
      "epoch": 2.9310897435897436,
      "grad_norm": 34.24089431762695,
      "learning_rate": 2.0689102564102564e-05,
      "loss": 0.429,
      "step": 1829
    },
    {
      "epoch": 2.980769230769231,
      "grad_norm": 5.287622928619385,
      "learning_rate": 2.0192307692307694e-05,
      "loss": 0.352,
      "step": 1860
    },
    {
      "epoch": 3.030448717948718,
      "grad_norm": 33.49539566040039,
      "learning_rate": 1.969551282051282e-05,
      "loss": 0.3348,
      "step": 1891
    },
    {
      "epoch": 3.0801282051282053,
      "grad_norm": 3.116619825363159,
      "learning_rate": 1.919871794871795e-05,
      "loss": 0.2613,
      "step": 1922
    },
    {
      "epoch": 3.1298076923076925,
      "grad_norm": 5.825009346008301,
      "learning_rate": 1.8701923076923078e-05,
      "loss": 0.279,
      "step": 1953
    },
    {
      "epoch": 3.1794871794871793,
      "grad_norm": 2.2447965145111084,
      "learning_rate": 1.8205128205128204e-05,
      "loss": 0.3405,
      "step": 1984
    },
    {
      "epoch": 3.2291666666666665,
      "grad_norm": 6.297438144683838,
      "learning_rate": 1.7708333333333335e-05,
      "loss": 0.3294,
      "step": 2015
    },
    {
      "epoch": 3.2788461538461537,
      "grad_norm": 17.42154884338379,
      "learning_rate": 1.721153846153846e-05,
      "loss": 0.296,
      "step": 2046
    },
    {
      "epoch": 3.328525641025641,
      "grad_norm": 11.834112167358398,
      "learning_rate": 1.6714743589743588e-05,
      "loss": 0.3767,
      "step": 2077
    },
    {
      "epoch": 3.378205128205128,
      "grad_norm": 2.2308290004730225,
      "learning_rate": 1.6217948717948718e-05,
      "loss": 0.2352,
      "step": 2108
    },
    {
      "epoch": 3.4278846153846154,
      "grad_norm": 13.97610855102539,
      "learning_rate": 1.5721153846153845e-05,
      "loss": 0.2937,
      "step": 2139
    },
    {
      "epoch": 3.4775641025641026,
      "grad_norm": 4.005247116088867,
      "learning_rate": 1.5224358974358973e-05,
      "loss": 0.283,
      "step": 2170
    },
    {
      "epoch": 3.52724358974359,
      "grad_norm": 5.180654525756836,
      "learning_rate": 1.4727564102564103e-05,
      "loss": 0.2773,
      "step": 2201
    },
    {
      "epoch": 3.5769230769230766,
      "grad_norm": 6.487146854400635,
      "learning_rate": 1.423076923076923e-05,
      "loss": 0.2665,
      "step": 2232
    },
    {
      "epoch": 3.626602564102564,
      "grad_norm": 2.202457904815674,
      "learning_rate": 1.373397435897436e-05,
      "loss": 0.2541,
      "step": 2263
    },
    {
      "epoch": 3.676282051282051,
      "grad_norm": 47.195499420166016,
      "learning_rate": 1.3237179487179489e-05,
      "loss": 0.2362,
      "step": 2294
    },
    {
      "epoch": 3.7259615384615383,
      "grad_norm": 21.26088523864746,
      "learning_rate": 1.2740384615384615e-05,
      "loss": 0.2331,
      "step": 2325
    },
    {
      "epoch": 3.7756410256410255,
      "grad_norm": 16.77073860168457,
      "learning_rate": 1.2243589743589744e-05,
      "loss": 0.2134,
      "step": 2356
    },
    {
      "epoch": 3.8253205128205128,
      "grad_norm": 2.246936559677124,
      "learning_rate": 1.1746794871794872e-05,
      "loss": 0.2437,
      "step": 2387
    },
    {
      "epoch": 3.875,
      "grad_norm": 1.4183766841888428,
      "learning_rate": 1.125e-05,
      "loss": 0.2012,
      "step": 2418
    },
    {
      "epoch": 3.9246794871794872,
      "grad_norm": 3.9523444175720215,
      "learning_rate": 1.0753205128205129e-05,
      "loss": 0.2204,
      "step": 2449
    },
    {
      "epoch": 3.9743589743589745,
      "grad_norm": 3.564764976501465,
      "learning_rate": 1.0256410256410256e-05,
      "loss": 0.2034,
      "step": 2480
    },
    {
      "epoch": 4.024038461538462,
      "grad_norm": 1.7849295139312744,
      "learning_rate": 9.759615384615384e-06,
      "loss": 0.1819,
      "step": 2511
    },
    {
      "epoch": 4.073717948717949,
      "grad_norm": 1.7893842458724976,
      "learning_rate": 9.262820512820514e-06,
      "loss": 0.1585,
      "step": 2542
    },
    {
      "epoch": 4.123397435897436,
      "grad_norm": 3.4071810245513916,
      "learning_rate": 8.766025641025641e-06,
      "loss": 0.1596,
      "step": 2573
    },
    {
      "epoch": 4.173076923076923,
      "grad_norm": 28.766258239746094,
      "learning_rate": 8.26923076923077e-06,
      "loss": 0.1563,
      "step": 2604
    },
    {
      "epoch": 4.222756410256411,
      "grad_norm": 3.066229820251465,
      "learning_rate": 7.772435897435898e-06,
      "loss": 0.1612,
      "step": 2635
    },
    {
      "epoch": 4.272435897435898,
      "grad_norm": 1.4539713859558105,
      "learning_rate": 7.275641025641026e-06,
      "loss": 0.151,
      "step": 2666
    },
    {
      "epoch": 4.322115384615385,
      "grad_norm": 3.9048240184783936,
      "learning_rate": 6.778846153846154e-06,
      "loss": 0.1514,
      "step": 2697
    },
    {
      "epoch": 4.371794871794872,
      "grad_norm": 1.2496715784072876,
      "learning_rate": 6.282051282051282e-06,
      "loss": 0.1632,
      "step": 2728
    },
    {
      "epoch": 4.421474358974359,
      "grad_norm": 1.516411542892456,
      "learning_rate": 5.785256410256411e-06,
      "loss": 0.1497,
      "step": 2759
    },
    {
      "epoch": 4.471153846153846,
      "grad_norm": 2.217137575149536,
      "learning_rate": 5.288461538461538e-06,
      "loss": 0.1514,
      "step": 2790
    },
    {
      "epoch": 4.520833333333333,
      "grad_norm": 1.7278540134429932,
      "learning_rate": 4.791666666666667e-06,
      "loss": 0.1409,
      "step": 2821
    },
    {
      "epoch": 4.57051282051282,
      "grad_norm": 2.3950905799865723,
      "learning_rate": 4.294871794871795e-06,
      "loss": 0.1524,
      "step": 2852
    },
    {
      "epoch": 4.6201923076923075,
      "grad_norm": 2.2806148529052734,
      "learning_rate": 3.798076923076923e-06,
      "loss": 0.1498,
      "step": 2883
    },
    {
      "epoch": 4.669871794871795,
      "grad_norm": 2.3029630184173584,
      "learning_rate": 3.3012820512820517e-06,
      "loss": 0.1471,
      "step": 2914
    },
    {
      "epoch": 4.719551282051282,
      "grad_norm": 1.592606782913208,
      "learning_rate": 2.8044871794871797e-06,
      "loss": 0.14,
      "step": 2945
    },
    {
      "epoch": 4.769230769230769,
      "grad_norm": 3.534292221069336,
      "learning_rate": 2.307692307692308e-06,
      "loss": 0.1426,
      "step": 2976
    },
    {
      "epoch": 4.818910256410256,
      "grad_norm": 2.344146966934204,
      "learning_rate": 1.810897435897436e-06,
      "loss": 0.1399,
      "step": 3007
    },
    {
      "epoch": 4.868589743589744,
      "grad_norm": 11.425804138183594,
      "learning_rate": 1.3141025641025643e-06,
      "loss": 0.1374,
      "step": 3038
    },
    {
      "epoch": 4.918269230769231,
      "grad_norm": 2.704406499862671,
      "learning_rate": 8.173076923076924e-07,
      "loss": 0.1389,
      "step": 3069
    },
    {
      "epoch": 4.967948717948718,
      "grad_norm": 6.8166728019714355,
      "learning_rate": 3.205128205128205e-07,
      "loss": 0.1721,
      "step": 3100
    },
    {
      "epoch": 5.0,
      "step": 3120,
      "total_flos": 1.845867535870722e+19,
      "train_loss": 0.6135548925552613,
      "train_runtime": 6226.8155,
      "train_samples_per_second": 2.003,
      "train_steps_per_second": 0.501
    }
  ],
  "logging_steps": 31,
  "max_steps": 3120,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.845867535870722e+19,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}