{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 693,
  "global_step": 1110,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0,
      "eval_f1": 0.14772727272727273,
      "eval_loss": 0.8850772380828857,
      "eval_precision": 0.08080808080808081,
      "eval_recall": 0.859504132231405,
      "eval_runtime": 99.3343,
      "eval_samples_per_second": 47.043,
      "eval_steps_per_second": 5.889,
      "step": 0
    },
    {
      "epoch": 0.018026137899954935,
      "grad_norm": 6.000438690185547,
      "learning_rate": 4.981949458483755e-05,
      "loss": 0.3931,
      "step": 10
    },
    {
      "epoch": 0.03605227579990987,
      "grad_norm": 47.497467041015625,
      "learning_rate": 4.963898916967509e-05,
      "loss": 0.2607,
      "step": 20
    },
    {
      "epoch": 0.054078413699864804,
      "grad_norm": 13.464141845703125,
      "learning_rate": 4.945848375451264e-05,
      "loss": 0.1813,
      "step": 30
    },
    {
      "epoch": 0.07210455159981974,
      "grad_norm": 15.57222843170166,
      "learning_rate": 4.927797833935018e-05,
      "loss": 0.2618,
      "step": 40
    },
    {
      "epoch": 0.09013068949977468,
      "grad_norm": 11.892234802246094,
      "learning_rate": 4.909747292418773e-05,
      "loss": 0.1559,
      "step": 50
    },
    {
      "epoch": 0.10815682739972961,
      "grad_norm": 6.340595722198486,
      "learning_rate": 4.891696750902527e-05,
      "loss": 0.1648,
      "step": 60
    },
    {
      "epoch": 0.12618296529968454,
      "grad_norm": 10.196683883666992,
      "learning_rate": 4.873646209386282e-05,
      "loss": 0.1472,
      "step": 70
    },
    {
      "epoch": 0.14420910319963948,
      "grad_norm": 11.511515617370605,
      "learning_rate": 4.855595667870036e-05,
      "loss": 0.2108,
      "step": 80
    },
    {
      "epoch": 0.16223524109959442,
      "grad_norm": 8.681073188781738,
      "learning_rate": 4.837545126353791e-05,
      "loss": 0.1561,
      "step": 90
    },
    {
      "epoch": 0.18026137899954936,
      "grad_norm": 41.062801361083984,
      "learning_rate": 4.819494584837546e-05,
      "loss": 0.1563,
      "step": 100
    },
    {
      "epoch": 0.19828751689950427,
      "grad_norm": 19.93891716003418,
      "learning_rate": 4.8014440433213e-05,
      "loss": 0.1881,
      "step": 110
    },
    {
      "epoch": 0.21631365479945922,
      "grad_norm": 21.90877914428711,
      "learning_rate": 4.783393501805055e-05,
      "loss": 0.1262,
      "step": 120
    },
    {
      "epoch": 0.23433979269941416,
      "grad_norm": 14.123740196228027,
      "learning_rate": 4.765342960288809e-05,
      "loss": 0.1535,
      "step": 130
    },
    {
      "epoch": 0.25236593059936907,
      "grad_norm": 30.671157836914062,
      "learning_rate": 4.747292418772563e-05,
      "loss": 0.1458,
      "step": 140
    },
    {
      "epoch": 0.270392068499324,
      "grad_norm": 7.560745716094971,
      "learning_rate": 4.7292418772563177e-05,
      "loss": 0.1797,
      "step": 150
    },
    {
      "epoch": 0.28841820639927895,
      "grad_norm": 12.448927879333496,
      "learning_rate": 4.711191335740072e-05,
      "loss": 0.1898,
      "step": 160
    },
    {
      "epoch": 0.3064443442992339,
      "grad_norm": 20.040008544921875,
      "learning_rate": 4.693140794223827e-05,
      "loss": 0.1698,
      "step": 170
    },
    {
      "epoch": 0.32447048219918884,
      "grad_norm": 7.592446804046631,
      "learning_rate": 4.675090252707581e-05,
      "loss": 0.156,
      "step": 180
    },
    {
      "epoch": 0.3424966200991438,
      "grad_norm": 18.830904006958008,
      "learning_rate": 4.657039711191336e-05,
      "loss": 0.1536,
      "step": 190
    },
    {
      "epoch": 0.3605227579990987,
      "grad_norm": 3.684938430786133,
      "learning_rate": 4.63898916967509e-05,
      "loss": 0.1152,
      "step": 200
    },
    {
      "epoch": 0.3785488958990536,
      "grad_norm": 13.293309211730957,
      "learning_rate": 4.620938628158845e-05,
      "loss": 0.2116,
      "step": 210
    },
    {
      "epoch": 0.39657503379900855,
      "grad_norm": 14.630786895751953,
      "learning_rate": 4.602888086642599e-05,
      "loss": 0.1549,
      "step": 220
    },
    {
      "epoch": 0.4146011716989635,
      "grad_norm": 20.37642478942871,
      "learning_rate": 4.584837545126354e-05,
      "loss": 0.1679,
      "step": 230
    },
    {
      "epoch": 0.43262730959891843,
      "grad_norm": 13.183340072631836,
      "learning_rate": 4.566787003610109e-05,
      "loss": 0.1581,
      "step": 240
    },
    {
      "epoch": 0.45065344749887337,
      "grad_norm": 21.777263641357422,
      "learning_rate": 4.548736462093863e-05,
      "loss": 0.1618,
      "step": 250
    },
    {
      "epoch": 0.4686795853988283,
      "grad_norm": 20.683944702148438,
      "learning_rate": 4.530685920577618e-05,
      "loss": 0.1606,
      "step": 260
    },
    {
      "epoch": 0.48670572329878325,
      "grad_norm": 30.506752014160156,
      "learning_rate": 4.5126353790613716e-05,
      "loss": 0.1831,
      "step": 270
    },
    {
      "epoch": 0.5047318611987381,
      "grad_norm": 22.953815460205078,
      "learning_rate": 4.494584837545127e-05,
      "loss": 0.1565,
      "step": 280
    },
    {
      "epoch": 0.5227579990986931,
      "grad_norm": 6.177365303039551,
      "learning_rate": 4.4765342960288806e-05,
      "loss": 0.1287,
      "step": 290
    },
    {
      "epoch": 0.540784136998648,
      "grad_norm": 8.373098373413086,
      "learning_rate": 4.458483754512636e-05,
      "loss": 0.1553,
      "step": 300
    },
    {
      "epoch": 0.558810274898603,
      "grad_norm": 10.07934284210205,
      "learning_rate": 4.44043321299639e-05,
      "loss": 0.1868,
      "step": 310
    },
    {
      "epoch": 0.5768364127985579,
      "grad_norm": 18.686702728271484,
      "learning_rate": 4.422382671480145e-05,
      "loss": 0.1576,
      "step": 320
    },
    {
      "epoch": 0.5948625506985128,
      "grad_norm": 10.679413795471191,
      "learning_rate": 4.404332129963899e-05,
      "loss": 0.1409,
      "step": 330
    },
    {
      "epoch": 0.6128886885984678,
      "grad_norm": 11.381563186645508,
      "learning_rate": 4.386281588447654e-05,
      "loss": 0.2043,
      "step": 340
    },
    {
      "epoch": 0.6309148264984227,
      "grad_norm": 40.68292236328125,
      "learning_rate": 4.368231046931408e-05,
      "loss": 0.1288,
      "step": 350
    },
    {
      "epoch": 0.6489409643983777,
      "grad_norm": 17.46027946472168,
      "learning_rate": 4.350180505415163e-05,
      "loss": 0.1537,
      "step": 360
    },
    {
      "epoch": 0.6669671022983326,
      "grad_norm": 13.656320571899414,
      "learning_rate": 4.332129963898917e-05,
      "loss": 0.1201,
      "step": 370
    },
    {
      "epoch": 0.6849932401982876,
      "grad_norm": 20.412939071655273,
      "learning_rate": 4.314079422382672e-05,
      "loss": 0.1321,
      "step": 380
    },
    {
      "epoch": 0.7030193780982424,
      "grad_norm": 9.957433700561523,
      "learning_rate": 4.296028880866426e-05,
      "loss": 0.141,
      "step": 390
    },
    {
      "epoch": 0.7210455159981974,
      "grad_norm": 9.682558059692383,
      "learning_rate": 4.277978339350181e-05,
      "loss": 0.1607,
      "step": 400
    },
    {
      "epoch": 0.7390716538981523,
      "grad_norm": 9.733131408691406,
      "learning_rate": 4.259927797833935e-05,
      "loss": 0.1037,
      "step": 410
    },
    {
      "epoch": 0.7570977917981072,
      "grad_norm": 15.76230525970459,
      "learning_rate": 4.24187725631769e-05,
      "loss": 0.2101,
      "step": 420
    },
    {
      "epoch": 0.7751239296980622,
      "grad_norm": 22.166545867919922,
      "learning_rate": 4.223826714801444e-05,
      "loss": 0.1459,
      "step": 430
    },
    {
      "epoch": 0.7931500675980171,
      "grad_norm": 17.87278175354004,
      "learning_rate": 4.205776173285199e-05,
      "loss": 0.147,
      "step": 440
    },
    {
      "epoch": 0.8111762054979721,
      "grad_norm": 5.201169490814209,
      "learning_rate": 4.187725631768953e-05,
      "loss": 0.1247,
      "step": 450
    },
    {
      "epoch": 0.829202343397927,
      "grad_norm": 8.68251895904541,
      "learning_rate": 4.169675090252708e-05,
      "loss": 0.1172,
      "step": 460
    },
    {
      "epoch": 0.847228481297882,
      "grad_norm": 14.086528778076172,
      "learning_rate": 4.151624548736462e-05,
      "loss": 0.111,
      "step": 470
    },
    {
      "epoch": 0.8652546191978369,
      "grad_norm": 5.421584606170654,
      "learning_rate": 4.1335740072202167e-05,
      "loss": 0.1264,
      "step": 480
    },
    {
      "epoch": 0.8832807570977917,
      "grad_norm": 7.700821399688721,
      "learning_rate": 4.115523465703972e-05,
      "loss": 0.1278,
      "step": 490
    },
    {
      "epoch": 0.9013068949977467,
      "grad_norm": 9.820999145507812,
      "learning_rate": 4.0974729241877256e-05,
      "loss": 0.1294,
      "step": 500
    },
    {
      "epoch": 0.9193330328977016,
      "grad_norm": 15.016349792480469,
      "learning_rate": 4.079422382671481e-05,
      "loss": 0.1607,
      "step": 510
    },
    {
      "epoch": 0.9373591707976566,
      "grad_norm": 10.43041706085205,
      "learning_rate": 4.0613718411552346e-05,
      "loss": 0.1091,
      "step": 520
    },
    {
      "epoch": 0.9553853086976115,
      "grad_norm": 26.99540901184082,
      "learning_rate": 4.043321299638989e-05,
      "loss": 0.1442,
      "step": 530
    },
    {
      "epoch": 0.9734114465975665,
      "grad_norm": 8.318906784057617,
      "learning_rate": 4.0252707581227436e-05,
      "loss": 0.1233,
      "step": 540
    },
    {
      "epoch": 0.9914375844975214,
      "grad_norm": 8.85923957824707,
      "learning_rate": 4.007220216606498e-05,
      "loss": 0.1466,
      "step": 550
    },
    {
      "epoch": 1.0090130689499774,
      "grad_norm": 4.69356107711792,
      "learning_rate": 3.989169675090253e-05,
      "loss": 0.1433,
      "step": 560
    },
    {
      "epoch": 1.0270392068499323,
      "grad_norm": 17.66107177734375,
      "learning_rate": 3.971119133574007e-05,
      "loss": 0.0944,
      "step": 570
    },
    {
      "epoch": 1.0450653447498874,
      "grad_norm": 4.721229553222656,
      "learning_rate": 3.953068592057762e-05,
      "loss": 0.1372,
      "step": 580
    },
    {
      "epoch": 1.0630914826498423,
      "grad_norm": 6.2178955078125,
      "learning_rate": 3.935018050541516e-05,
      "loss": 0.1412,
      "step": 590
    },
    {
      "epoch": 1.0811176205497972,
      "grad_norm": 6.32495641708374,
      "learning_rate": 3.916967509025271e-05,
      "loss": 0.0892,
      "step": 600
    },
    {
      "epoch": 1.099143758449752,
      "grad_norm": 12.221674919128418,
      "learning_rate": 3.898916967509025e-05,
      "loss": 0.1333,
      "step": 610
    },
    {
      "epoch": 1.117169896349707,
      "grad_norm": 1.7088432312011719,
      "learning_rate": 3.88086642599278e-05,
      "loss": 0.1077,
      "step": 620
    },
    {
      "epoch": 1.135196034249662,
      "grad_norm": 17.136337280273438,
      "learning_rate": 3.862815884476535e-05,
      "loss": 0.1381,
      "step": 630
    },
    {
      "epoch": 1.153222172149617,
      "grad_norm": 5.424814224243164,
      "learning_rate": 3.844765342960289e-05,
      "loss": 0.1228,
      "step": 640
    },
    {
      "epoch": 1.1712483100495719,
      "grad_norm": 18.775466918945312,
      "learning_rate": 3.826714801444044e-05,
      "loss": 0.099,
      "step": 650
    },
    {
      "epoch": 1.1892744479495267,
      "grad_norm": 8.731025695800781,
      "learning_rate": 3.8086642599277976e-05,
      "loss": 0.1324,
      "step": 660
    },
    {
      "epoch": 1.2073005858494819,
      "grad_norm": 9.53122615814209,
      "learning_rate": 3.790613718411553e-05,
      "loss": 0.1085,
      "step": 670
    },
    {
      "epoch": 1.2253267237494367,
      "grad_norm": 17.20035743713379,
      "learning_rate": 3.7725631768953066e-05,
      "loss": 0.0774,
      "step": 680
    },
    {
      "epoch": 1.2433528616493916,
      "grad_norm": 18.292253494262695,
      "learning_rate": 3.754512635379062e-05,
      "loss": 0.0965,
      "step": 690
    },
    {
      "epoch": 1.2487607030193781,
      "eval_f1": 0.7416879795396419,
      "eval_loss": 0.10558413714170456,
      "eval_precision": 0.6921241050119332,
      "eval_recall": 0.7988980716253443,
      "eval_runtime": 115.001,
      "eval_samples_per_second": 40.634,
      "eval_steps_per_second": 5.087,
      "step": 693
    },
    {
      "epoch": 1.2613789995493465,
      "grad_norm": 6.922005653381348,
      "learning_rate": 3.7364620938628155e-05,
      "loss": 0.0906,
      "step": 700
    },
    {
      "epoch": 1.2794051374493014,
      "grad_norm": 33.214107513427734,
      "learning_rate": 3.718411552346571e-05,
      "loss": 0.072,
      "step": 710
    },
    {
      "epoch": 1.2974312753492563,
      "grad_norm": 9.25637149810791,
      "learning_rate": 3.700361010830325e-05,
      "loss": 0.1597,
      "step": 720
    },
    {
      "epoch": 1.3154574132492114,
      "grad_norm": 10.336318016052246,
      "learning_rate": 3.68231046931408e-05,
      "loss": 0.1063,
      "step": 730
    },
    {
      "epoch": 1.3334835511491663,
      "grad_norm": 11.632000923156738,
      "learning_rate": 3.664259927797834e-05,
      "loss": 0.1175,
      "step": 740
    },
    {
      "epoch": 1.3515096890491212,
      "grad_norm": 12.909192085266113,
      "learning_rate": 3.646209386281589e-05,
      "loss": 0.1285,
      "step": 750
    },
    {
      "epoch": 1.3695358269490763,
      "grad_norm": 7.545075416564941,
      "learning_rate": 3.628158844765343e-05,
      "loss": 0.0864,
      "step": 760
    },
    {
      "epoch": 1.3875619648490312,
      "grad_norm": 25.248464584350586,
      "learning_rate": 3.610108303249098e-05,
      "loss": 0.0769,
      "step": 770
    },
    {
      "epoch": 1.405588102748986,
      "grad_norm": 10.5347318649292,
      "learning_rate": 3.592057761732852e-05,
      "loss": 0.088,
      "step": 780
    },
    {
      "epoch": 1.423614240648941,
      "grad_norm": 13.101734161376953,
      "learning_rate": 3.574007220216607e-05,
      "loss": 0.1084,
      "step": 790
    },
    {
      "epoch": 1.4416403785488958,
      "grad_norm": 12.188843727111816,
      "learning_rate": 3.555956678700361e-05,
      "loss": 0.1101,
      "step": 800
    },
    {
      "epoch": 1.4596665164488507,
      "grad_norm": 11.552268981933594,
      "learning_rate": 3.537906137184116e-05,
      "loss": 0.0785,
      "step": 810
    },
    {
      "epoch": 1.4776926543488058,
      "grad_norm": 10.269721031188965,
      "learning_rate": 3.51985559566787e-05,
      "loss": 0.1161,
      "step": 820
    },
    {
      "epoch": 1.4957187922487607,
      "grad_norm": 12.09333610534668,
      "learning_rate": 3.5018050541516247e-05,
      "loss": 0.1113,
      "step": 830
    },
    {
      "epoch": 1.5137449301487156,
      "grad_norm": 5.4690141677856445,
      "learning_rate": 3.483754512635379e-05,
      "loss": 0.1046,
      "step": 840
    },
    {
      "epoch": 1.5317710680486707,
      "grad_norm": 6.296072959899902,
      "learning_rate": 3.4657039711191336e-05,
      "loss": 0.1202,
      "step": 850
    },
    {
      "epoch": 1.5497972059486256,
      "grad_norm": 8.854303359985352,
      "learning_rate": 3.447653429602888e-05,
      "loss": 0.0929,
      "step": 860
    },
    {
      "epoch": 1.5678233438485805,
      "grad_norm": 23.987024307250977,
      "learning_rate": 3.4296028880866426e-05,
      "loss": 0.1326,
      "step": 870
    },
    {
      "epoch": 1.5858494817485354,
      "grad_norm": 7.229393005371094,
      "learning_rate": 3.411552346570397e-05,
      "loss": 0.1169,
      "step": 880
    },
    {
      "epoch": 1.6038756196484902,
      "grad_norm": 5.463610649108887,
      "learning_rate": 3.3935018050541516e-05,
      "loss": 0.0809,
      "step": 890
    },
    {
      "epoch": 1.6219017575484451,
      "grad_norm": 5.9659624099731445,
      "learning_rate": 3.375451263537907e-05,
      "loss": 0.1265,
      "step": 900
    },
    {
      "epoch": 1.6399278954484002,
      "grad_norm": 17.582469940185547,
      "learning_rate": 3.3574007220216606e-05,
      "loss": 0.0846,
      "step": 910
    },
    {
      "epoch": 1.6579540333483551,
      "grad_norm": 13.633155822753906,
      "learning_rate": 3.339350180505416e-05,
      "loss": 0.0999,
      "step": 920
    },
    {
      "epoch": 1.67598017124831,
      "grad_norm": 14.496593475341797,
      "learning_rate": 3.3212996389891696e-05,
      "loss": 0.0903,
      "step": 930
    },
    {
      "epoch": 1.694006309148265,
      "grad_norm": 7.594356536865234,
      "learning_rate": 3.303249097472924e-05,
      "loss": 0.1021,
      "step": 940
    },
    {
      "epoch": 1.71203244704822,
      "grad_norm": 20.628849029541016,
      "learning_rate": 3.2851985559566786e-05,
      "loss": 0.1131,
      "step": 950
    },
    {
      "epoch": 1.7300585849481749,
      "grad_norm": 8.25106430053711,
      "learning_rate": 3.267148014440433e-05,
      "loss": 0.0872,
      "step": 960
    },
    {
      "epoch": 1.7480847228481298,
      "grad_norm": 17.08070182800293,
      "learning_rate": 3.249097472924188e-05,
      "loss": 0.0663,
      "step": 970
    },
    {
      "epoch": 1.7661108607480847,
      "grad_norm": 38.33162307739258,
      "learning_rate": 3.231046931407942e-05,
      "loss": 0.142,
      "step": 980
    },
    {
      "epoch": 1.7841369986480395,
      "grad_norm": 17.89957618713379,
      "learning_rate": 3.212996389891697e-05,
      "loss": 0.0942,
      "step": 990
    },
    {
      "epoch": 1.8021631365479944,
      "grad_norm": 9.691226959228516,
      "learning_rate": 3.194945848375451e-05,
      "loss": 0.0902,
      "step": 1000
    },
    {
      "epoch": 1.8201892744479495,
      "grad_norm": 9.764283180236816,
      "learning_rate": 3.176895306859206e-05,
      "loss": 0.1108,
      "step": 1010
    },
    {
      "epoch": 1.8382154123479044,
      "grad_norm": 13.829035758972168,
      "learning_rate": 3.15884476534296e-05,
      "loss": 0.0881,
      "step": 1020
    },
    {
      "epoch": 1.8562415502478595,
      "grad_norm": 8.372889518737793,
      "learning_rate": 3.140794223826715e-05,
      "loss": 0.1272,
      "step": 1030
    },
    {
      "epoch": 1.8742676881478144,
      "grad_norm": 12.576041221618652,
      "learning_rate": 3.12274368231047e-05,
      "loss": 0.1076,
      "step": 1040
    },
    {
      "epoch": 1.8922938260477693,
      "grad_norm": 12.995355606079102,
      "learning_rate": 3.104693140794224e-05,
      "loss": 0.0932,
      "step": 1050
    },
    {
      "epoch": 1.9103199639477242,
      "grad_norm": 6.626877784729004,
      "learning_rate": 3.086642599277979e-05,
      "loss": 0.105,
      "step": 1060
    },
    {
      "epoch": 1.928346101847679,
      "grad_norm": 11.209733009338379,
      "learning_rate": 3.0685920577617325e-05,
      "loss": 0.1331,
      "step": 1070
    },
    {
      "epoch": 1.946372239747634,
      "grad_norm": 7.832246780395508,
      "learning_rate": 3.0505415162454877e-05,
      "loss": 0.087,
      "step": 1080
    },
    {
      "epoch": 1.9643983776475888,
      "grad_norm": 9.489190101623535,
      "learning_rate": 3.032490974729242e-05,
      "loss": 0.0904,
      "step": 1090
    },
    {
      "epoch": 1.982424515547544,
      "grad_norm": 14.940125465393066,
      "learning_rate": 3.0144404332129967e-05,
      "loss": 0.0959,
      "step": 1100
    },
    {
      "epoch": 2.0,
      "grad_norm": 8.826193809509277,
      "learning_rate": 2.996389891696751e-05,
      "loss": 0.1276,
      "step": 1110
    }
  ],
  "logging_steps": 10,
  "max_steps": 2770,
  "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": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.858514256176742e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}