{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 1665,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0,
      "eval_f1": 0.35766423357664234,
      "eval_fn": 1724,
      "eval_fp": 740,
      "eval_loss": 0.6976209878921509,
      "eval_precision": 0.48106591865357645,
      "eval_recall": 0.2846473029045643,
      "eval_runtime": 68.4179,
      "eval_samples_per_second": 68.301,
      "eval_steps_per_second": 4.283,
      "eval_tn": 1523,
      "eval_tp": 686,
      "step": 0
    },
    {
      "epoch": 0.018026137899954935,
      "grad_norm": 9.306479454040527,
      "learning_rate": 4.981949458483755e-05,
      "loss": 0.6855,
      "step": 10
    },
    {
      "epoch": 0.03605227579990987,
      "grad_norm": 3.271188497543335,
      "learning_rate": 4.963898916967509e-05,
      "loss": 0.667,
      "step": 20
    },
    {
      "epoch": 0.054078413699864804,
      "grad_norm": 7.420863628387451,
      "learning_rate": 4.945848375451264e-05,
      "loss": 0.6415,
      "step": 30
    },
    {
      "epoch": 0.07210455159981974,
      "grad_norm": 3.4994685649871826,
      "learning_rate": 4.927797833935018e-05,
      "loss": 0.6188,
      "step": 40
    },
    {
      "epoch": 0.09013068949977468,
      "grad_norm": 12.319598197937012,
      "learning_rate": 4.909747292418773e-05,
      "loss": 0.6083,
      "step": 50
    },
    {
      "epoch": 0.10815682739972961,
      "grad_norm": 5.15464448928833,
      "learning_rate": 4.891696750902527e-05,
      "loss": 0.5685,
      "step": 60
    },
    {
      "epoch": 0.12618296529968454,
      "grad_norm": 8.32410717010498,
      "learning_rate": 4.873646209386282e-05,
      "loss": 0.5153,
      "step": 70
    },
    {
      "epoch": 0.14420910319963948,
      "grad_norm": 26.5758113861084,
      "learning_rate": 4.855595667870036e-05,
      "loss": 0.518,
      "step": 80
    },
    {
      "epoch": 0.16223524109959442,
      "grad_norm": 10.276503562927246,
      "learning_rate": 4.837545126353791e-05,
      "loss": 0.4902,
      "step": 90
    },
    {
      "epoch": 0.18026137899954936,
      "grad_norm": 8.947182655334473,
      "learning_rate": 4.819494584837546e-05,
      "loss": 0.4889,
      "step": 100
    },
    {
      "epoch": 0.19828751689950427,
      "grad_norm": 8.54738998413086,
      "learning_rate": 4.8014440433213e-05,
      "loss": 0.4671,
      "step": 110
    },
    {
      "epoch": 0.21631365479945922,
      "grad_norm": 10.866387367248535,
      "learning_rate": 4.783393501805055e-05,
      "loss": 0.4543,
      "step": 120
    },
    {
      "epoch": 0.23433979269941416,
      "grad_norm": 38.420108795166016,
      "learning_rate": 4.765342960288809e-05,
      "loss": 0.4354,
      "step": 130
    },
    {
      "epoch": 0.25236593059936907,
      "grad_norm": 13.113662719726562,
      "learning_rate": 4.747292418772563e-05,
      "loss": 0.5018,
      "step": 140
    },
    {
      "epoch": 0.270392068499324,
      "grad_norm": 36.51860809326172,
      "learning_rate": 4.7292418772563177e-05,
      "loss": 0.4681,
      "step": 150
    },
    {
      "epoch": 0.28841820639927895,
      "grad_norm": 40.15293502807617,
      "learning_rate": 4.711191335740072e-05,
      "loss": 0.4798,
      "step": 160
    },
    {
      "epoch": 0.3064443442992339,
      "grad_norm": 12.45967960357666,
      "learning_rate": 4.693140794223827e-05,
      "loss": 0.4849,
      "step": 170
    },
    {
      "epoch": 0.32447048219918884,
      "grad_norm": 10.393397331237793,
      "learning_rate": 4.675090252707581e-05,
      "loss": 0.4496,
      "step": 180
    },
    {
      "epoch": 0.3424966200991438,
      "grad_norm": 8.426735877990723,
      "learning_rate": 4.657039711191336e-05,
      "loss": 0.4799,
      "step": 190
    },
    {
      "epoch": 0.3605227579990987,
      "grad_norm": 10.40811824798584,
      "learning_rate": 4.63898916967509e-05,
      "loss": 0.45,
      "step": 200
    },
    {
      "epoch": 0.3785488958990536,
      "grad_norm": 10.157293319702148,
      "learning_rate": 4.620938628158845e-05,
      "loss": 0.4516,
      "step": 210
    },
    {
      "epoch": 0.39657503379900855,
      "grad_norm": 9.582480430603027,
      "learning_rate": 4.602888086642599e-05,
      "loss": 0.4491,
      "step": 220
    },
    {
      "epoch": 0.4146011716989635,
      "grad_norm": 17.83095359802246,
      "learning_rate": 4.584837545126354e-05,
      "loss": 0.4575,
      "step": 230
    },
    {
      "epoch": 0.43262730959891843,
      "grad_norm": 10.945122718811035,
      "learning_rate": 4.566787003610109e-05,
      "loss": 0.4324,
      "step": 240
    },
    {
      "epoch": 0.45065344749887337,
      "grad_norm": 5.155959606170654,
      "learning_rate": 4.548736462093863e-05,
      "loss": 0.421,
      "step": 250
    },
    {
      "epoch": 0.4686795853988283,
      "grad_norm": 12.092293739318848,
      "learning_rate": 4.530685920577618e-05,
      "loss": 0.4406,
      "step": 260
    },
    {
      "epoch": 0.48670572329878325,
      "grad_norm": 12.724702835083008,
      "learning_rate": 4.5126353790613716e-05,
      "loss": 0.3947,
      "step": 270
    },
    {
      "epoch": 0.5047318611987381,
      "grad_norm": 26.710527420043945,
      "learning_rate": 4.494584837545127e-05,
      "loss": 0.4068,
      "step": 280
    },
    {
      "epoch": 0.5227579990986931,
      "grad_norm": 8.311524391174316,
      "learning_rate": 4.4765342960288806e-05,
      "loss": 0.4483,
      "step": 290
    },
    {
      "epoch": 0.540784136998648,
      "grad_norm": 12.505125045776367,
      "learning_rate": 4.458483754512636e-05,
      "loss": 0.3945,
      "step": 300
    },
    {
      "epoch": 0.558810274898603,
      "grad_norm": 10.924736976623535,
      "learning_rate": 4.44043321299639e-05,
      "loss": 0.4386,
      "step": 310
    },
    {
      "epoch": 0.5768364127985579,
      "grad_norm": 9.005159378051758,
      "learning_rate": 4.422382671480145e-05,
      "loss": 0.4157,
      "step": 320
    },
    {
      "epoch": 0.5948625506985128,
      "grad_norm": 13.740028381347656,
      "learning_rate": 4.404332129963899e-05,
      "loss": 0.4392,
      "step": 330
    },
    {
      "epoch": 0.6128886885984678,
      "grad_norm": 58.44659423828125,
      "learning_rate": 4.386281588447654e-05,
      "loss": 0.4345,
      "step": 340
    },
    {
      "epoch": 0.6309148264984227,
      "grad_norm": 8.443741798400879,
      "learning_rate": 4.368231046931408e-05,
      "loss": 0.4439,
      "step": 350
    },
    {
      "epoch": 0.6489409643983777,
      "grad_norm": 7.2340521812438965,
      "learning_rate": 4.350180505415163e-05,
      "loss": 0.4026,
      "step": 360
    },
    {
      "epoch": 0.6669671022983326,
      "grad_norm": 7.140610218048096,
      "learning_rate": 4.332129963898917e-05,
      "loss": 0.4284,
      "step": 370
    },
    {
      "epoch": 0.6849932401982876,
      "grad_norm": 26.290523529052734,
      "learning_rate": 4.314079422382672e-05,
      "loss": 0.3861,
      "step": 380
    },
    {
      "epoch": 0.7030193780982424,
      "grad_norm": 18.140789031982422,
      "learning_rate": 4.296028880866426e-05,
      "loss": 0.4209,
      "step": 390
    },
    {
      "epoch": 0.7210455159981974,
      "grad_norm": 9.561443328857422,
      "learning_rate": 4.277978339350181e-05,
      "loss": 0.4432,
      "step": 400
    },
    {
      "epoch": 0.7390716538981523,
      "grad_norm": 11.68012809753418,
      "learning_rate": 4.259927797833935e-05,
      "loss": 0.4329,
      "step": 410
    },
    {
      "epoch": 0.7570977917981072,
      "grad_norm": 13.150470733642578,
      "learning_rate": 4.24187725631769e-05,
      "loss": 0.4157,
      "step": 420
    },
    {
      "epoch": 0.7751239296980622,
      "grad_norm": 11.43278980255127,
      "learning_rate": 4.223826714801444e-05,
      "loss": 0.4084,
      "step": 430
    },
    {
      "epoch": 0.7931500675980171,
      "grad_norm": 6.537344455718994,
      "learning_rate": 4.205776173285199e-05,
      "loss": 0.4095,
      "step": 440
    },
    {
      "epoch": 0.8111762054979721,
      "grad_norm": 8.772361755371094,
      "learning_rate": 4.187725631768953e-05,
      "loss": 0.3763,
      "step": 450
    },
    {
      "epoch": 0.829202343397927,
      "grad_norm": 7.250331878662109,
      "learning_rate": 4.169675090252708e-05,
      "loss": 0.4262,
      "step": 460
    },
    {
      "epoch": 0.847228481297882,
      "grad_norm": 8.936566352844238,
      "learning_rate": 4.151624548736462e-05,
      "loss": 0.3947,
      "step": 470
    },
    {
      "epoch": 0.8652546191978369,
      "grad_norm": 8.440159797668457,
      "learning_rate": 4.1335740072202167e-05,
      "loss": 0.4056,
      "step": 480
    },
    {
      "epoch": 0.8832807570977917,
      "grad_norm": 9.320460319519043,
      "learning_rate": 4.115523465703972e-05,
      "loss": 0.4019,
      "step": 490
    },
    {
      "epoch": 0.9013068949977467,
      "grad_norm": 6.329681873321533,
      "learning_rate": 4.0974729241877256e-05,
      "loss": 0.4092,
      "step": 500
    },
    {
      "epoch": 0.9193330328977016,
      "grad_norm": 16.446720123291016,
      "learning_rate": 4.079422382671481e-05,
      "loss": 0.4132,
      "step": 510
    },
    {
      "epoch": 0.9373591707976566,
      "grad_norm": 7.657299041748047,
      "learning_rate": 4.0613718411552346e-05,
      "loss": 0.4091,
      "step": 520
    },
    {
      "epoch": 0.9553853086976115,
      "grad_norm": 8.183615684509277,
      "learning_rate": 4.043321299638989e-05,
      "loss": 0.3985,
      "step": 530
    },
    {
      "epoch": 0.9734114465975665,
      "grad_norm": 7.5413384437561035,
      "learning_rate": 4.0252707581227436e-05,
      "loss": 0.4404,
      "step": 540
    },
    {
      "epoch": 0.9914375844975214,
      "grad_norm": 10.21881103515625,
      "learning_rate": 4.007220216606498e-05,
      "loss": 0.421,
      "step": 550
    },
    {
      "epoch": 1.0,
      "eval_f1": 0.8386043664558254,
      "eval_fn": 355,
      "eval_fp": 436,
      "eval_loss": 0.3710826635360718,
      "eval_precision": 0.8249698916097953,
      "eval_recall": 0.8526970954356846,
      "eval_runtime": 78.3531,
      "eval_samples_per_second": 59.64,
      "eval_steps_per_second": 3.739,
      "eval_tn": 1827,
      "eval_tp": 2055,
      "step": 555
    },
    {
      "epoch": 1.0090130689499774,
      "grad_norm": 7.165682792663574,
      "learning_rate": 3.989169675090253e-05,
      "loss": 0.3664,
      "step": 560
    },
    {
      "epoch": 1.0270392068499323,
      "grad_norm": 7.220385551452637,
      "learning_rate": 3.971119133574007e-05,
      "loss": 0.3672,
      "step": 570
    },
    {
      "epoch": 1.0450653447498874,
      "grad_norm": 51.91734313964844,
      "learning_rate": 3.953068592057762e-05,
      "loss": 0.3674,
      "step": 580
    },
    {
      "epoch": 1.0630914826498423,
      "grad_norm": 9.907829284667969,
      "learning_rate": 3.935018050541516e-05,
      "loss": 0.3315,
      "step": 590
    },
    {
      "epoch": 1.0811176205497972,
      "grad_norm": 13.948915481567383,
      "learning_rate": 3.916967509025271e-05,
      "loss": 0.4092,
      "step": 600
    },
    {
      "epoch": 1.099143758449752,
      "grad_norm": 8.972203254699707,
      "learning_rate": 3.898916967509025e-05,
      "loss": 0.3407,
      "step": 610
    },
    {
      "epoch": 1.117169896349707,
      "grad_norm": 6.633499622344971,
      "learning_rate": 3.88086642599278e-05,
      "loss": 0.3509,
      "step": 620
    },
    {
      "epoch": 1.135196034249662,
      "grad_norm": 8.704663276672363,
      "learning_rate": 3.862815884476535e-05,
      "loss": 0.3468,
      "step": 630
    },
    {
      "epoch": 1.153222172149617,
      "grad_norm": 16.486459732055664,
      "learning_rate": 3.844765342960289e-05,
      "loss": 0.3504,
      "step": 640
    },
    {
      "epoch": 1.1712483100495719,
      "grad_norm": 36.337745666503906,
      "learning_rate": 3.826714801444044e-05,
      "loss": 0.3634,
      "step": 650
    },
    {
      "epoch": 1.1892744479495267,
      "grad_norm": 6.744643211364746,
      "learning_rate": 3.8086642599277976e-05,
      "loss": 0.3729,
      "step": 660
    },
    {
      "epoch": 1.2073005858494819,
      "grad_norm": 5.880044460296631,
      "learning_rate": 3.790613718411553e-05,
      "loss": 0.3685,
      "step": 670
    },
    {
      "epoch": 1.2253267237494367,
      "grad_norm": 10.722400665283203,
      "learning_rate": 3.7725631768953066e-05,
      "loss": 0.3134,
      "step": 680
    },
    {
      "epoch": 1.2433528616493916,
      "grad_norm": 20.80872917175293,
      "learning_rate": 3.754512635379062e-05,
      "loss": 0.3843,
      "step": 690
    },
    {
      "epoch": 1.2613789995493465,
      "grad_norm": 7.892972469329834,
      "learning_rate": 3.7364620938628155e-05,
      "loss": 0.3312,
      "step": 700
    },
    {
      "epoch": 1.2794051374493014,
      "grad_norm": 8.907954216003418,
      "learning_rate": 3.718411552346571e-05,
      "loss": 0.3645,
      "step": 710
    },
    {
      "epoch": 1.2974312753492563,
      "grad_norm": 8.709657669067383,
      "learning_rate": 3.700361010830325e-05,
      "loss": 0.3575,
      "step": 720
    },
    {
      "epoch": 1.3154574132492114,
      "grad_norm": 12.098897933959961,
      "learning_rate": 3.68231046931408e-05,
      "loss": 0.3509,
      "step": 730
    },
    {
      "epoch": 1.3334835511491663,
      "grad_norm": 6.648235321044922,
      "learning_rate": 3.664259927797834e-05,
      "loss": 0.3539,
      "step": 740
    },
    {
      "epoch": 1.3515096890491212,
      "grad_norm": 7.914617538452148,
      "learning_rate": 3.646209386281589e-05,
      "loss": 0.3765,
      "step": 750
    },
    {
      "epoch": 1.3695358269490763,
      "grad_norm": 7.850524425506592,
      "learning_rate": 3.628158844765343e-05,
      "loss": 0.3653,
      "step": 760
    },
    {
      "epoch": 1.3875619648490312,
      "grad_norm": 13.03383731842041,
      "learning_rate": 3.610108303249098e-05,
      "loss": 0.3917,
      "step": 770
    },
    {
      "epoch": 1.405588102748986,
      "grad_norm": 6.959196090698242,
      "learning_rate": 3.592057761732852e-05,
      "loss": 0.3633,
      "step": 780
    },
    {
      "epoch": 1.423614240648941,
      "grad_norm": 6.3111252784729,
      "learning_rate": 3.574007220216607e-05,
      "loss": 0.3652,
      "step": 790
    },
    {
      "epoch": 1.4416403785488958,
      "grad_norm": 40.7597541809082,
      "learning_rate": 3.555956678700361e-05,
      "loss": 0.3568,
      "step": 800
    },
    {
      "epoch": 1.4596665164488507,
      "grad_norm": 7.647876262664795,
      "learning_rate": 3.537906137184116e-05,
      "loss": 0.329,
      "step": 810
    },
    {
      "epoch": 1.4776926543488058,
      "grad_norm": 23.181804656982422,
      "learning_rate": 3.51985559566787e-05,
      "loss": 0.3216,
      "step": 820
    },
    {
      "epoch": 1.4957187922487607,
      "grad_norm": 26.149192810058594,
      "learning_rate": 3.5018050541516247e-05,
      "loss": 0.3415,
      "step": 830
    },
    {
      "epoch": 1.5137449301487156,
      "grad_norm": 11.924449920654297,
      "learning_rate": 3.483754512635379e-05,
      "loss": 0.3539,
      "step": 840
    },
    {
      "epoch": 1.5317710680486707,
      "grad_norm": 8.165074348449707,
      "learning_rate": 3.4657039711191336e-05,
      "loss": 0.3407,
      "step": 850
    },
    {
      "epoch": 1.5497972059486256,
      "grad_norm": 6.361219882965088,
      "learning_rate": 3.447653429602888e-05,
      "loss": 0.3412,
      "step": 860
    },
    {
      "epoch": 1.5678233438485805,
      "grad_norm": 8.405501365661621,
      "learning_rate": 3.4296028880866426e-05,
      "loss": 0.3115,
      "step": 870
    },
    {
      "epoch": 1.5858494817485354,
      "grad_norm": 8.599916458129883,
      "learning_rate": 3.411552346570397e-05,
      "loss": 0.339,
      "step": 880
    },
    {
      "epoch": 1.6038756196484902,
      "grad_norm": 6.046282768249512,
      "learning_rate": 3.3935018050541516e-05,
      "loss": 0.3503,
      "step": 890
    },
    {
      "epoch": 1.6219017575484451,
      "grad_norm": 19.992610931396484,
      "learning_rate": 3.375451263537907e-05,
      "loss": 0.3483,
      "step": 900
    },
    {
      "epoch": 1.6399278954484002,
      "grad_norm": 10.666455268859863,
      "learning_rate": 3.3574007220216606e-05,
      "loss": 0.3637,
      "step": 910
    },
    {
      "epoch": 1.6579540333483551,
      "grad_norm": 19.68842124938965,
      "learning_rate": 3.339350180505416e-05,
      "loss": 0.3554,
      "step": 920
    },
    {
      "epoch": 1.67598017124831,
      "grad_norm": 15.999527931213379,
      "learning_rate": 3.3212996389891696e-05,
      "loss": 0.345,
      "step": 930
    },
    {
      "epoch": 1.694006309148265,
      "grad_norm": 8.567536354064941,
      "learning_rate": 3.303249097472924e-05,
      "loss": 0.3611,
      "step": 940
    },
    {
      "epoch": 1.71203244704822,
      "grad_norm": 13.945199012756348,
      "learning_rate": 3.2851985559566786e-05,
      "loss": 0.3527,
      "step": 950
    },
    {
      "epoch": 1.7300585849481749,
      "grad_norm": 12.041361808776855,
      "learning_rate": 3.267148014440433e-05,
      "loss": 0.3543,
      "step": 960
    },
    {
      "epoch": 1.7480847228481298,
      "grad_norm": 8.660093307495117,
      "learning_rate": 3.249097472924188e-05,
      "loss": 0.3211,
      "step": 970
    },
    {
      "epoch": 1.7661108607480847,
      "grad_norm": 19.359636306762695,
      "learning_rate": 3.231046931407942e-05,
      "loss": 0.354,
      "step": 980
    },
    {
      "epoch": 1.7841369986480395,
      "grad_norm": 14.446808815002441,
      "learning_rate": 3.212996389891697e-05,
      "loss": 0.4002,
      "step": 990
    },
    {
      "epoch": 1.8021631365479944,
      "grad_norm": 8.190558433532715,
      "learning_rate": 3.194945848375451e-05,
      "loss": 0.3757,
      "step": 1000
    },
    {
      "epoch": 1.8201892744479495,
      "grad_norm": 7.9925055503845215,
      "learning_rate": 3.176895306859206e-05,
      "loss": 0.3397,
      "step": 1010
    },
    {
      "epoch": 1.8382154123479044,
      "grad_norm": 18.883665084838867,
      "learning_rate": 3.15884476534296e-05,
      "loss": 0.3367,
      "step": 1020
    },
    {
      "epoch": 1.8562415502478595,
      "grad_norm": 8.38142204284668,
      "learning_rate": 3.140794223826715e-05,
      "loss": 0.363,
      "step": 1030
    },
    {
      "epoch": 1.8742676881478144,
      "grad_norm": 11.258013725280762,
      "learning_rate": 3.12274368231047e-05,
      "loss": 0.3584,
      "step": 1040
    },
    {
      "epoch": 1.8922938260477693,
      "grad_norm": 9.164728164672852,
      "learning_rate": 3.104693140794224e-05,
      "loss": 0.313,
      "step": 1050
    },
    {
      "epoch": 1.9103199639477242,
      "grad_norm": 7.581974983215332,
      "learning_rate": 3.086642599277979e-05,
      "loss": 0.3294,
      "step": 1060
    },
    {
      "epoch": 1.928346101847679,
      "grad_norm": 45.30561828613281,
      "learning_rate": 3.0685920577617325e-05,
      "loss": 0.318,
      "step": 1070
    },
    {
      "epoch": 1.946372239747634,
      "grad_norm": 5.765126705169678,
      "learning_rate": 3.0505415162454877e-05,
      "loss": 0.3459,
      "step": 1080
    },
    {
      "epoch": 1.9643983776475888,
      "grad_norm": 8.358948707580566,
      "learning_rate": 3.032490974729242e-05,
      "loss": 0.3022,
      "step": 1090
    },
    {
      "epoch": 1.982424515547544,
      "grad_norm": 9.03178596496582,
      "learning_rate": 3.0144404332129967e-05,
      "loss": 0.3374,
      "step": 1100
    },
    {
      "epoch": 2.0,
      "grad_norm": 9.664074897766113,
      "learning_rate": 2.996389891696751e-05,
      "loss": 0.3068,
      "step": 1110
    },
    {
      "epoch": 2.0,
      "eval_f1": 0.8450475997569374,
      "eval_fn": 324,
      "eval_fp": 441,
      "eval_loss": 0.3717796802520752,
      "eval_precision": 0.8254847645429363,
      "eval_recall": 0.8655601659751038,
      "eval_runtime": 72.1654,
      "eval_samples_per_second": 64.754,
      "eval_steps_per_second": 4.06,
      "eval_tn": 1822,
      "eval_tp": 2086,
      "step": 1110
    },
    {
      "epoch": 2.018026137899955,
      "grad_norm": 7.0051350593566895,
      "learning_rate": 2.9783393501805057e-05,
      "loss": 0.2963,
      "step": 1120
    },
    {
      "epoch": 2.0360522757999098,
      "grad_norm": 33.976837158203125,
      "learning_rate": 2.9602888086642598e-05,
      "loss": 0.2376,
      "step": 1130
    },
    {
      "epoch": 2.0540784136998647,
      "grad_norm": 8.37226390838623,
      "learning_rate": 2.9422382671480147e-05,
      "loss": 0.2855,
      "step": 1140
    },
    {
      "epoch": 2.0721045515998195,
      "grad_norm": 28.015409469604492,
      "learning_rate": 2.924187725631769e-05,
      "loss": 0.2624,
      "step": 1150
    },
    {
      "epoch": 2.090130689499775,
      "grad_norm": 81.69696807861328,
      "learning_rate": 2.906137184115524e-05,
      "loss": 0.3098,
      "step": 1160
    },
    {
      "epoch": 2.1081568273997298,
      "grad_norm": 12.006569862365723,
      "learning_rate": 2.888086642599278e-05,
      "loss": 0.2975,
      "step": 1170
    },
    {
      "epoch": 2.1261829652996846,
      "grad_norm": 12.094599723815918,
      "learning_rate": 2.870036101083033e-05,
      "loss": 0.2865,
      "step": 1180
    },
    {
      "epoch": 2.1442091031996395,
      "grad_norm": 10.78286361694336,
      "learning_rate": 2.851985559566787e-05,
      "loss": 0.2778,
      "step": 1190
    },
    {
      "epoch": 2.1622352410995944,
      "grad_norm": 7.721689701080322,
      "learning_rate": 2.8339350180505413e-05,
      "loss": 0.2603,
      "step": 1200
    },
    {
      "epoch": 2.1802613789995493,
      "grad_norm": 8.49467945098877,
      "learning_rate": 2.815884476534296e-05,
      "loss": 0.2722,
      "step": 1210
    },
    {
      "epoch": 2.198287516899504,
      "grad_norm": 10.548264503479004,
      "learning_rate": 2.7978339350180506e-05,
      "loss": 0.2625,
      "step": 1220
    },
    {
      "epoch": 2.216313654799459,
      "grad_norm": 13.317939758300781,
      "learning_rate": 2.779783393501805e-05,
      "loss": 0.3161,
      "step": 1230
    },
    {
      "epoch": 2.234339792699414,
      "grad_norm": 8.207959175109863,
      "learning_rate": 2.7617328519855596e-05,
      "loss": 0.2829,
      "step": 1240
    },
    {
      "epoch": 2.2523659305993693,
      "grad_norm": 10.530780792236328,
      "learning_rate": 2.7436823104693144e-05,
      "loss": 0.2604,
      "step": 1250
    },
    {
      "epoch": 2.270392068499324,
      "grad_norm": 12.912242889404297,
      "learning_rate": 2.7256317689530686e-05,
      "loss": 0.2925,
      "step": 1260
    },
    {
      "epoch": 2.288418206399279,
      "grad_norm": 9.53242015838623,
      "learning_rate": 2.7075812274368234e-05,
      "loss": 0.2566,
      "step": 1270
    },
    {
      "epoch": 2.306444344299234,
      "grad_norm": 8.965003967285156,
      "learning_rate": 2.6895306859205776e-05,
      "loss": 0.2734,
      "step": 1280
    },
    {
      "epoch": 2.324470482199189,
      "grad_norm": 11.137091636657715,
      "learning_rate": 2.6714801444043324e-05,
      "loss": 0.2906,
      "step": 1290
    },
    {
      "epoch": 2.3424966200991437,
      "grad_norm": 11.334556579589844,
      "learning_rate": 2.6534296028880866e-05,
      "loss": 0.2951,
      "step": 1300
    },
    {
      "epoch": 2.3605227579990986,
      "grad_norm": 9.768026351928711,
      "learning_rate": 2.6353790613718414e-05,
      "loss": 0.2779,
      "step": 1310
    },
    {
      "epoch": 2.3785488958990535,
      "grad_norm": 10.248305320739746,
      "learning_rate": 2.617328519855596e-05,
      "loss": 0.2836,
      "step": 1320
    },
    {
      "epoch": 2.3965750337990084,
      "grad_norm": 10.527518272399902,
      "learning_rate": 2.59927797833935e-05,
      "loss": 0.2864,
      "step": 1330
    },
    {
      "epoch": 2.4146011716989637,
      "grad_norm": 7.96708869934082,
      "learning_rate": 2.581227436823105e-05,
      "loss": 0.2618,
      "step": 1340
    },
    {
      "epoch": 2.4326273095989186,
      "grad_norm": 10.488567352294922,
      "learning_rate": 2.563176895306859e-05,
      "loss": 0.2543,
      "step": 1350
    },
    {
      "epoch": 2.4506534474988735,
      "grad_norm": 47.32187271118164,
      "learning_rate": 2.545126353790614e-05,
      "loss": 0.2538,
      "step": 1360
    },
    {
      "epoch": 2.4686795853988284,
      "grad_norm": 12.458937644958496,
      "learning_rate": 2.527075812274368e-05,
      "loss": 0.2416,
      "step": 1370
    },
    {
      "epoch": 2.4867057232987833,
      "grad_norm": 19.352401733398438,
      "learning_rate": 2.509025270758123e-05,
      "loss": 0.2431,
      "step": 1380
    },
    {
      "epoch": 2.504731861198738,
      "grad_norm": 11.721567153930664,
      "learning_rate": 2.4909747292418774e-05,
      "loss": 0.3046,
      "step": 1390
    },
    {
      "epoch": 2.522757999098693,
      "grad_norm": 8.958959579467773,
      "learning_rate": 2.472924187725632e-05,
      "loss": 0.3058,
      "step": 1400
    },
    {
      "epoch": 2.540784136998648,
      "grad_norm": 11.200770378112793,
      "learning_rate": 2.4548736462093864e-05,
      "loss": 0.2726,
      "step": 1410
    },
    {
      "epoch": 2.558810274898603,
      "grad_norm": 20.568017959594727,
      "learning_rate": 2.436823104693141e-05,
      "loss": 0.2762,
      "step": 1420
    },
    {
      "epoch": 2.576836412798558,
      "grad_norm": 12.618547439575195,
      "learning_rate": 2.4187725631768953e-05,
      "loss": 0.2784,
      "step": 1430
    },
    {
      "epoch": 2.5948625506985126,
      "grad_norm": 10.008194923400879,
      "learning_rate": 2.40072202166065e-05,
      "loss": 0.294,
      "step": 1440
    },
    {
      "epoch": 2.612888688598468,
      "grad_norm": 9.454615592956543,
      "learning_rate": 2.3826714801444043e-05,
      "loss": 0.3102,
      "step": 1450
    },
    {
      "epoch": 2.630914826498423,
      "grad_norm": 15.94392204284668,
      "learning_rate": 2.3646209386281588e-05,
      "loss": 0.2378,
      "step": 1460
    },
    {
      "epoch": 2.6489409643983777,
      "grad_norm": 10.787693977355957,
      "learning_rate": 2.3465703971119137e-05,
      "loss": 0.2617,
      "step": 1470
    },
    {
      "epoch": 2.6669671022983326,
      "grad_norm": 17.901201248168945,
      "learning_rate": 2.328519855595668e-05,
      "loss": 0.3069,
      "step": 1480
    },
    {
      "epoch": 2.6849932401982874,
      "grad_norm": 38.81023406982422,
      "learning_rate": 2.3104693140794227e-05,
      "loss": 0.2764,
      "step": 1490
    },
    {
      "epoch": 2.7030193780982423,
      "grad_norm": 14.45818042755127,
      "learning_rate": 2.292418772563177e-05,
      "loss": 0.291,
      "step": 1500
    },
    {
      "epoch": 2.721045515998197,
      "grad_norm": 7.874146461486816,
      "learning_rate": 2.2743682310469316e-05,
      "loss": 0.2743,
      "step": 1510
    },
    {
      "epoch": 2.7390716538981525,
      "grad_norm": 9.052064895629883,
      "learning_rate": 2.2563176895306858e-05,
      "loss": 0.3207,
      "step": 1520
    },
    {
      "epoch": 2.757097791798107,
      "grad_norm": 9.65465259552002,
      "learning_rate": 2.2382671480144403e-05,
      "loss": 0.2747,
      "step": 1530
    },
    {
      "epoch": 2.7751239296980623,
      "grad_norm": 8.10863208770752,
      "learning_rate": 2.220216606498195e-05,
      "loss": 0.2883,
      "step": 1540
    },
    {
      "epoch": 2.793150067598017,
      "grad_norm": 9.309361457824707,
      "learning_rate": 2.2021660649819496e-05,
      "loss": 0.3039,
      "step": 1550
    },
    {
      "epoch": 2.811176205497972,
      "grad_norm": 54.50463104248047,
      "learning_rate": 2.184115523465704e-05,
      "loss": 0.2912,
      "step": 1560
    },
    {
      "epoch": 2.829202343397927,
      "grad_norm": 11.151042938232422,
      "learning_rate": 2.1660649819494586e-05,
      "loss": 0.3161,
      "step": 1570
    },
    {
      "epoch": 2.847228481297882,
      "grad_norm": 11.726816177368164,
      "learning_rate": 2.148014440433213e-05,
      "loss": 0.2485,
      "step": 1580
    },
    {
      "epoch": 2.8652546191978367,
      "grad_norm": 9.904781341552734,
      "learning_rate": 2.1299638989169676e-05,
      "loss": 0.3208,
      "step": 1590
    },
    {
      "epoch": 2.8832807570977916,
      "grad_norm": 13.534920692443848,
      "learning_rate": 2.111913357400722e-05,
      "loss": 0.3127,
      "step": 1600
    },
    {
      "epoch": 2.901306894997747,
      "grad_norm": 8.898674964904785,
      "learning_rate": 2.0938628158844766e-05,
      "loss": 0.2989,
      "step": 1610
    },
    {
      "epoch": 2.9193330328977014,
      "grad_norm": 14.875420570373535,
      "learning_rate": 2.075812274368231e-05,
      "loss": 0.3275,
      "step": 1620
    },
    {
      "epoch": 2.9373591707976567,
      "grad_norm": 7.563508033752441,
      "learning_rate": 2.057761732851986e-05,
      "loss": 0.2747,
      "step": 1630
    },
    {
      "epoch": 2.9553853086976116,
      "grad_norm": 15.946221351623535,
      "learning_rate": 2.0397111913357404e-05,
      "loss": 0.2661,
      "step": 1640
    },
    {
      "epoch": 2.9734114465975665,
      "grad_norm": 14.538371086120605,
      "learning_rate": 2.0216606498194946e-05,
      "loss": 0.2767,
      "step": 1650
    },
    {
      "epoch": 2.9914375844975214,
      "grad_norm": 14.545721054077148,
      "learning_rate": 2.003610108303249e-05,
      "loss": 0.2721,
      "step": 1660
    },
    {
      "epoch": 3.0,
      "eval_f1": 0.8490452261306533,
      "eval_fn": 298,
      "eval_fp": 453,
      "eval_loss": 0.39444267749786377,
      "eval_precision": 0.8233918128654971,
      "eval_recall": 0.8763485477178423,
      "eval_runtime": 72.7061,
      "eval_samples_per_second": 64.272,
      "eval_steps_per_second": 4.03,
      "eval_tn": 1810,
      "eval_tp": 2112,
      "step": 1665
    }
  ],
  "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": 1.154738112933888e+17,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}