{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.1066027296200664,
  "global_step": 3000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 9.975408828230666e-05,
      "loss": 0.9349,
      "step": 20
    },
    {
      "epoch": 0.01,
      "learning_rate": 9.950817656461331e-05,
      "loss": 0.9288,
      "step": 40
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.926226484691996e-05,
      "loss": 0.929,
      "step": 60
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.901635312922661e-05,
      "loss": 0.9254,
      "step": 80
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.877044141153327e-05,
      "loss": 0.9303,
      "step": 100
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.852452969383992e-05,
      "loss": 0.9255,
      "step": 120
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.827861797614657e-05,
      "loss": 0.9197,
      "step": 140
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.803270625845322e-05,
      "loss": 0.9275,
      "step": 160
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.778679454075987e-05,
      "loss": 0.919,
      "step": 180
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.754088282306653e-05,
      "loss": 0.9246,
      "step": 200
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.729497110537318e-05,
      "loss": 0.917,
      "step": 220
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.704905938767983e-05,
      "loss": 0.9211,
      "step": 240
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.680314766998648e-05,
      "loss": 0.9204,
      "step": 260
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.655723595229314e-05,
      "loss": 0.9127,
      "step": 280
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.631132423459979e-05,
      "loss": 0.9077,
      "step": 300
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.606541251690644e-05,
      "loss": 0.9176,
      "step": 320
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.581950079921309e-05,
      "loss": 0.9119,
      "step": 340
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.557358908151974e-05,
      "loss": 0.9135,
      "step": 360
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.53276773638264e-05,
      "loss": 0.9098,
      "step": 380
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.508176564613305e-05,
      "loss": 0.9043,
      "step": 400
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.48358539284397e-05,
      "loss": 0.9167,
      "step": 420
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.458994221074635e-05,
      "loss": 0.9076,
      "step": 440
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.4344030493053e-05,
      "loss": 0.9039,
      "step": 460
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.409811877535964e-05,
      "loss": 0.9147,
      "step": 480
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.38522070576663e-05,
      "loss": 0.9108,
      "step": 500
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.360629533997295e-05,
      "loss": 0.9068,
      "step": 520
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.33603836222796e-05,
      "loss": 0.9123,
      "step": 540
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.311447190458625e-05,
      "loss": 0.9084,
      "step": 560
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.28685601868929e-05,
      "loss": 0.9049,
      "step": 580
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.262264846919956e-05,
      "loss": 0.9044,
      "step": 600
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.237673675150621e-05,
      "loss": 0.9074,
      "step": 620
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.213082503381286e-05,
      "loss": 0.9024,
      "step": 640
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.188491331611951e-05,
      "loss": 0.9017,
      "step": 660
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.163900159842617e-05,
      "loss": 0.895,
      "step": 680
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.139308988073282e-05,
      "loss": 0.8943,
      "step": 700
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.114717816303947e-05,
      "loss": 0.9036,
      "step": 720
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.090126644534612e-05,
      "loss": 0.8968,
      "step": 740
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.065535472765277e-05,
      "loss": 0.8997,
      "step": 760
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.040944300995943e-05,
      "loss": 0.8935,
      "step": 780
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.016353129226608e-05,
      "loss": 0.891,
      "step": 800
    },
    {
      "epoch": 0.3,
      "learning_rate": 8.991761957457273e-05,
      "loss": 0.8986,
      "step": 820
    },
    {
      "epoch": 0.31,
      "learning_rate": 8.967170785687938e-05,
      "loss": 0.8976,
      "step": 840
    },
    {
      "epoch": 0.32,
      "learning_rate": 8.942579613918604e-05,
      "loss": 0.9008,
      "step": 860
    },
    {
      "epoch": 0.32,
      "learning_rate": 8.917988442149269e-05,
      "loss": 0.9023,
      "step": 880
    },
    {
      "epoch": 0.33,
      "learning_rate": 8.893397270379934e-05,
      "loss": 0.8904,
      "step": 900
    },
    {
      "epoch": 0.34,
      "learning_rate": 8.868806098610599e-05,
      "loss": 0.8892,
      "step": 920
    },
    {
      "epoch": 0.35,
      "learning_rate": 8.844214926841264e-05,
      "loss": 0.8949,
      "step": 940
    },
    {
      "epoch": 0.35,
      "learning_rate": 8.81962375507193e-05,
      "loss": 0.8982,
      "step": 960
    },
    {
      "epoch": 0.36,
      "learning_rate": 8.795032583302595e-05,
      "loss": 0.888,
      "step": 980
    },
    {
      "epoch": 0.37,
      "learning_rate": 8.77044141153326e-05,
      "loss": 0.8866,
      "step": 1000
    },
    {
      "epoch": 0.38,
      "learning_rate": 8.745850239763925e-05,
      "loss": 0.8859,
      "step": 1020
    },
    {
      "epoch": 0.38,
      "learning_rate": 8.72125906799459e-05,
      "loss": 0.8952,
      "step": 1040
    },
    {
      "epoch": 0.39,
      "learning_rate": 8.696667896225256e-05,
      "loss": 0.8855,
      "step": 1060
    },
    {
      "epoch": 0.4,
      "learning_rate": 8.672076724455921e-05,
      "loss": 0.8951,
      "step": 1080
    },
    {
      "epoch": 0.41,
      "learning_rate": 8.647485552686586e-05,
      "loss": 0.8875,
      "step": 1100
    },
    {
      "epoch": 0.41,
      "learning_rate": 8.622894380917251e-05,
      "loss": 0.8919,
      "step": 1120
    },
    {
      "epoch": 0.42,
      "learning_rate": 8.598303209147917e-05,
      "loss": 0.8889,
      "step": 1140
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.573712037378582e-05,
      "loss": 0.8857,
      "step": 1160
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.549120865609247e-05,
      "loss": 0.8876,
      "step": 1180
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.524529693839912e-05,
      "loss": 0.8905,
      "step": 1200
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.499938522070577e-05,
      "loss": 0.886,
      "step": 1220
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.475347350301243e-05,
      "loss": 0.8839,
      "step": 1240
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.450756178531908e-05,
      "loss": 0.8851,
      "step": 1260
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.426165006762573e-05,
      "loss": 0.8784,
      "step": 1280
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.401573834993238e-05,
      "loss": 0.8897,
      "step": 1300
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.376982663223904e-05,
      "loss": 0.883,
      "step": 1320
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.352391491454569e-05,
      "loss": 0.8819,
      "step": 1340
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.327800319685234e-05,
      "loss": 0.883,
      "step": 1360
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.303209147915899e-05,
      "loss": 0.8864,
      "step": 1380
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.278617976146564e-05,
      "loss": 0.8875,
      "step": 1400
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.254026804377228e-05,
      "loss": 0.8749,
      "step": 1420
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.229435632607894e-05,
      "loss": 0.878,
      "step": 1440
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.204844460838559e-05,
      "loss": 0.8786,
      "step": 1460
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.180253289069224e-05,
      "loss": 0.8844,
      "step": 1480
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.155662117299889e-05,
      "loss": 0.8768,
      "step": 1500
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.131070945530554e-05,
      "loss": 0.8751,
      "step": 1520
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.10647977376122e-05,
      "loss": 0.8795,
      "step": 1540
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.081888601991885e-05,
      "loss": 0.8782,
      "step": 1560
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.05729743022255e-05,
      "loss": 0.8789,
      "step": 1580
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.032706258453215e-05,
      "loss": 0.8801,
      "step": 1600
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.00811508668388e-05,
      "loss": 0.8772,
      "step": 1620
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.983523914914546e-05,
      "loss": 0.8722,
      "step": 1640
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.958932743145211e-05,
      "loss": 0.8749,
      "step": 1660
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.934341571375876e-05,
      "loss": 0.8728,
      "step": 1680
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.909750399606541e-05,
      "loss": 0.8811,
      "step": 1700
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.885159227837207e-05,
      "loss": 0.8741,
      "step": 1720
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.860568056067872e-05,
      "loss": 0.8697,
      "step": 1740
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.835976884298537e-05,
      "loss": 0.8746,
      "step": 1760
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.811385712529202e-05,
      "loss": 0.878,
      "step": 1780
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.786794540759867e-05,
      "loss": 0.8766,
      "step": 1800
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.762203368990533e-05,
      "loss": 0.8694,
      "step": 1820
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.737612197221198e-05,
      "loss": 0.8821,
      "step": 1840
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.713021025451863e-05,
      "loss": 0.8776,
      "step": 1860
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.688429853682528e-05,
      "loss": 0.8678,
      "step": 1880
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.663838681913194e-05,
      "loss": 0.8709,
      "step": 1900
    },
    {
      "epoch": 0.71,
      "learning_rate": 7.639247510143859e-05,
      "loss": 0.8683,
      "step": 1920
    },
    {
      "epoch": 0.72,
      "learning_rate": 7.614656338374524e-05,
      "loss": 0.8754,
      "step": 1940
    },
    {
      "epoch": 0.72,
      "learning_rate": 7.590065166605189e-05,
      "loss": 0.8724,
      "step": 1960
    },
    {
      "epoch": 0.73,
      "learning_rate": 7.565473994835854e-05,
      "loss": 0.8679,
      "step": 1980
    },
    {
      "epoch": 0.74,
      "learning_rate": 7.54088282306652e-05,
      "loss": 0.8747,
      "step": 2000
    },
    {
      "epoch": 0.75,
      "learning_rate": 7.516291651297185e-05,
      "loss": 0.8669,
      "step": 2020
    },
    {
      "epoch": 0.75,
      "learning_rate": 7.49170047952785e-05,
      "loss": 0.8712,
      "step": 2040
    },
    {
      "epoch": 0.76,
      "learning_rate": 7.467109307758515e-05,
      "loss": 0.8656,
      "step": 2060
    },
    {
      "epoch": 0.77,
      "learning_rate": 7.44251813598918e-05,
      "loss": 0.8736,
      "step": 2080
    },
    {
      "epoch": 0.77,
      "learning_rate": 7.417926964219846e-05,
      "loss": 0.868,
      "step": 2100
    },
    {
      "epoch": 0.78,
      "learning_rate": 7.393335792450511e-05,
      "loss": 0.8747,
      "step": 2120
    },
    {
      "epoch": 0.79,
      "learning_rate": 7.368744620681176e-05,
      "loss": 0.8693,
      "step": 2140
    },
    {
      "epoch": 0.8,
      "learning_rate": 7.344153448911841e-05,
      "loss": 0.8704,
      "step": 2160
    },
    {
      "epoch": 0.8,
      "learning_rate": 7.319562277142507e-05,
      "loss": 0.8616,
      "step": 2180
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.294971105373172e-05,
      "loss": 0.8692,
      "step": 2200
    },
    {
      "epoch": 0.82,
      "learning_rate": 7.270379933603837e-05,
      "loss": 0.8721,
      "step": 2220
    },
    {
      "epoch": 0.83,
      "learning_rate": 7.245788761834502e-05,
      "loss": 0.8632,
      "step": 2240
    },
    {
      "epoch": 0.83,
      "learning_rate": 7.221197590065168e-05,
      "loss": 0.8698,
      "step": 2260
    },
    {
      "epoch": 0.84,
      "learning_rate": 7.196606418295833e-05,
      "loss": 0.8606,
      "step": 2280
    },
    {
      "epoch": 0.85,
      "learning_rate": 7.172015246526498e-05,
      "loss": 0.8626,
      "step": 2300
    },
    {
      "epoch": 0.86,
      "learning_rate": 7.147424074757163e-05,
      "loss": 0.865,
      "step": 2320
    },
    {
      "epoch": 0.86,
      "learning_rate": 7.122832902987828e-05,
      "loss": 0.8651,
      "step": 2340
    },
    {
      "epoch": 0.87,
      "learning_rate": 7.098241731218494e-05,
      "loss": 0.8676,
      "step": 2360
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.073650559449157e-05,
      "loss": 0.8629,
      "step": 2380
    },
    {
      "epoch": 0.89,
      "learning_rate": 7.049059387679823e-05,
      "loss": 0.8636,
      "step": 2400
    },
    {
      "epoch": 0.89,
      "learning_rate": 7.024468215910488e-05,
      "loss": 0.8632,
      "step": 2420
    },
    {
      "epoch": 0.9,
      "learning_rate": 6.999877044141153e-05,
      "loss": 0.8632,
      "step": 2440
    },
    {
      "epoch": 0.91,
      "learning_rate": 6.975285872371818e-05,
      "loss": 0.8655,
      "step": 2460
    },
    {
      "epoch": 0.91,
      "learning_rate": 6.950694700602484e-05,
      "loss": 0.8659,
      "step": 2480
    },
    {
      "epoch": 0.92,
      "learning_rate": 6.926103528833149e-05,
      "loss": 0.8581,
      "step": 2500
    },
    {
      "epoch": 0.93,
      "learning_rate": 6.901512357063814e-05,
      "loss": 0.8608,
      "step": 2520
    },
    {
      "epoch": 0.94,
      "learning_rate": 6.876921185294479e-05,
      "loss": 0.8623,
      "step": 2540
    },
    {
      "epoch": 0.94,
      "learning_rate": 6.852330013525144e-05,
      "loss": 0.8644,
      "step": 2560
    },
    {
      "epoch": 0.95,
      "learning_rate": 6.82773884175581e-05,
      "loss": 0.8604,
      "step": 2580
    },
    {
      "epoch": 0.96,
      "learning_rate": 6.803147669986475e-05,
      "loss": 0.8708,
      "step": 2600
    },
    {
      "epoch": 0.97,
      "learning_rate": 6.77855649821714e-05,
      "loss": 0.8585,
      "step": 2620
    },
    {
      "epoch": 0.97,
      "learning_rate": 6.753965326447805e-05,
      "loss": 0.8596,
      "step": 2640
    },
    {
      "epoch": 0.98,
      "learning_rate": 6.72937415467847e-05,
      "loss": 0.8582,
      "step": 2660
    },
    {
      "epoch": 0.99,
      "learning_rate": 6.704782982909136e-05,
      "loss": 0.8587,
      "step": 2680
    },
    {
      "epoch": 1.0,
      "learning_rate": 6.680191811139801e-05,
      "loss": 0.8598,
      "step": 2700
    },
    {
      "epoch": 1.0,
      "learning_rate": 6.655600639370466e-05,
      "loss": 0.8617,
      "step": 2720
    },
    {
      "epoch": 1.01,
      "learning_rate": 6.631009467601131e-05,
      "loss": 0.8589,
      "step": 2740
    },
    {
      "epoch": 1.02,
      "learning_rate": 6.606418295831797e-05,
      "loss": 0.8582,
      "step": 2760
    },
    {
      "epoch": 1.03,
      "learning_rate": 6.581827124062462e-05,
      "loss": 0.8576,
      "step": 2780
    },
    {
      "epoch": 1.03,
      "learning_rate": 6.557235952293127e-05,
      "loss": 0.8613,
      "step": 2800
    },
    {
      "epoch": 1.04,
      "learning_rate": 6.532644780523792e-05,
      "loss": 0.8517,
      "step": 2820
    },
    {
      "epoch": 1.05,
      "learning_rate": 6.508053608754458e-05,
      "loss": 0.8579,
      "step": 2840
    },
    {
      "epoch": 1.05,
      "learning_rate": 6.483462436985123e-05,
      "loss": 0.8588,
      "step": 2860
    },
    {
      "epoch": 1.06,
      "learning_rate": 6.458871265215788e-05,
      "loss": 0.8554,
      "step": 2880
    },
    {
      "epoch": 1.07,
      "learning_rate": 6.434280093446453e-05,
      "loss": 0.8542,
      "step": 2900
    },
    {
      "epoch": 1.08,
      "learning_rate": 6.409688921677118e-05,
      "loss": 0.8501,
      "step": 2920
    },
    {
      "epoch": 1.08,
      "learning_rate": 6.385097749907784e-05,
      "loss": 0.8523,
      "step": 2940
    },
    {
      "epoch": 1.09,
      "learning_rate": 6.360506578138449e-05,
      "loss": 0.8506,
      "step": 2960
    },
    {
      "epoch": 1.1,
      "learning_rate": 6.335915406369114e-05,
      "loss": 0.8613,
      "step": 2980
    },
    {
      "epoch": 1.11,
      "learning_rate": 6.311324234599779e-05,
      "loss": 0.8468,
      "step": 3000
    }
  ],
  "max_steps": 8133,
  "num_train_epochs": 3,
  "total_flos": 1.5168022953677292e+19,
  "trial_name": null,
  "trial_params": null
}