{
  "best_global_step": 2500,
  "best_metric": 0.9688026421468009,
  "best_model_checkpoint": "artifacts/layoutlmv3-large-model2b-radiology-lab-operative/checkpoint-2500",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 3093,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.019405700424499697,
      "grad_norm": 6.008031368255615,
      "learning_rate": 1.2258064516129033e-06,
      "loss": 1.0255,
      "step": 20
    },
    {
      "epoch": 0.038811400848999394,
      "grad_norm": 5.029001712799072,
      "learning_rate": 2.5161290322580645e-06,
      "loss": 1.0011,
      "step": 40
    },
    {
      "epoch": 0.05821710127349909,
      "grad_norm": 33.76041030883789,
      "learning_rate": 3.8064516129032257e-06,
      "loss": 0.6529,
      "step": 60
    },
    {
      "epoch": 0.07762280169799879,
      "grad_norm": 20.754009246826172,
      "learning_rate": 5.096774193548387e-06,
      "loss": 0.2957,
      "step": 80
    },
    {
      "epoch": 0.09702850212249849,
      "grad_norm": 22.04238510131836,
      "learning_rate": 6.3870967741935485e-06,
      "loss": 0.2382,
      "step": 100
    },
    {
      "epoch": 0.11643420254699818,
      "grad_norm": 14.927919387817383,
      "learning_rate": 7.67741935483871e-06,
      "loss": 0.212,
      "step": 120
    },
    {
      "epoch": 0.13583990297149787,
      "grad_norm": 46.08305740356445,
      "learning_rate": 8.967741935483871e-06,
      "loss": 0.1463,
      "step": 140
    },
    {
      "epoch": 0.15524560339599758,
      "grad_norm": 27.541467666625977,
      "learning_rate": 1.0258064516129034e-05,
      "loss": 0.262,
      "step": 160
    },
    {
      "epoch": 0.17465130382049726,
      "grad_norm": 31.197599411010742,
      "learning_rate": 1.1548387096774194e-05,
      "loss": 0.1866,
      "step": 180
    },
    {
      "epoch": 0.19405700424499697,
      "grad_norm": 25.980010986328125,
      "learning_rate": 1.2838709677419356e-05,
      "loss": 0.2072,
      "step": 200
    },
    {
      "epoch": 0.21346270466949666,
      "grad_norm": 7.316959381103516,
      "learning_rate": 1.4129032258064517e-05,
      "loss": 0.2019,
      "step": 220
    },
    {
      "epoch": 0.23286840509399637,
      "grad_norm": 5.203572750091553,
      "learning_rate": 1.541935483870968e-05,
      "loss": 0.2868,
      "step": 240
    },
    {
      "epoch": 0.25227410551849605,
      "grad_norm": 21.817180633544922,
      "learning_rate": 1.670967741935484e-05,
      "loss": 0.3223,
      "step": 260
    },
    {
      "epoch": 0.27167980594299573,
      "grad_norm": 0.11316101253032684,
      "learning_rate": 1.8e-05,
      "loss": 0.2355,
      "step": 280
    },
    {
      "epoch": 0.2910855063674955,
      "grad_norm": 14.177412033081055,
      "learning_rate": 1.9290322580645162e-05,
      "loss": 0.373,
      "step": 300
    },
    {
      "epoch": 0.31049120679199516,
      "grad_norm": 0.7051326036453247,
      "learning_rate": 1.9935321595400647e-05,
      "loss": 0.2878,
      "step": 320
    },
    {
      "epoch": 0.32989690721649484,
      "grad_norm": 52.49479293823242,
      "learning_rate": 1.9791591807402086e-05,
      "loss": 0.2077,
      "step": 340
    },
    {
      "epoch": 0.3493026076409945,
      "grad_norm": 3.4517762660980225,
      "learning_rate": 1.9647862019403524e-05,
      "loss": 0.1491,
      "step": 360
    },
    {
      "epoch": 0.36870830806549426,
      "grad_norm": 6.324164867401123,
      "learning_rate": 1.950413223140496e-05,
      "loss": 0.1541,
      "step": 380
    },
    {
      "epoch": 0.38811400848999394,
      "grad_norm": 18.803449630737305,
      "learning_rate": 1.93604024434064e-05,
      "loss": 0.3205,
      "step": 400
    },
    {
      "epoch": 0.40751970891449363,
      "grad_norm": 54.76909637451172,
      "learning_rate": 1.9216672655407835e-05,
      "loss": 0.1982,
      "step": 420
    },
    {
      "epoch": 0.4269254093389933,
      "grad_norm": 87.57189178466797,
      "learning_rate": 1.9072942867409273e-05,
      "loss": 0.2861,
      "step": 440
    },
    {
      "epoch": 0.44633110976349305,
      "grad_norm": 29.666324615478516,
      "learning_rate": 1.8929213079410708e-05,
      "loss": 0.1775,
      "step": 460
    },
    {
      "epoch": 0.46573681018799273,
      "grad_norm": 45.25077819824219,
      "learning_rate": 1.8785483291412146e-05,
      "loss": 0.1778,
      "step": 480
    },
    {
      "epoch": 0.4851425106124924,
      "grad_norm": 0.07991417497396469,
      "learning_rate": 1.8641753503413585e-05,
      "loss": 0.2022,
      "step": 500
    },
    {
      "epoch": 0.4851425106124924,
      "eval_accuracy": 0.9618320610687023,
      "eval_f1_lab_report": 0.9573770491803278,
      "eval_f1_operative_report": 0.9512605042016806,
      "eval_f1_radiology_report": 0.9681597409606044,
      "eval_loss": 0.18711775541305542,
      "eval_macro_f1": 0.9589324314475377,
      "eval_macro_precision": 0.9581257833148703,
      "eval_macro_recall": 0.9598566998218754,
      "eval_precision_lab_report": 0.946515397082658,
      "eval_precision_operative_report": 0.9528619528619529,
      "eval_precision_radiology_report": 0.975,
      "eval_recall_lab_report": 0.9684908789386402,
      "eval_recall_operative_report": 0.9496644295302014,
      "eval_recall_radiology_report": 0.9614147909967846,
      "eval_runtime": 76.0141,
      "eval_samples_per_second": 24.127,
      "eval_steps_per_second": 24.127,
      "eval_weighted_f1": 0.9618686091734364,
      "step": 500
    },
    {
      "epoch": 0.5045482110369921,
      "grad_norm": 0.17595137655735016,
      "learning_rate": 1.849802371541502e-05,
      "loss": 0.1835,
      "step": 520
    },
    {
      "epoch": 0.5239539114614918,
      "grad_norm": 0.2895595133304596,
      "learning_rate": 1.8354293927416458e-05,
      "loss": 0.1614,
      "step": 540
    },
    {
      "epoch": 0.5433596118859915,
      "grad_norm": 12.082067489624023,
      "learning_rate": 1.8210564139417896e-05,
      "loss": 0.237,
      "step": 560
    },
    {
      "epoch": 0.5627653123104912,
      "grad_norm": 27.726442337036133,
      "learning_rate": 1.8066834351419334e-05,
      "loss": 0.1608,
      "step": 580
    },
    {
      "epoch": 0.582171012734991,
      "grad_norm": 0.3235521614551544,
      "learning_rate": 1.792310456342077e-05,
      "loss": 0.1046,
      "step": 600
    },
    {
      "epoch": 0.6015767131594906,
      "grad_norm": 0.10597099363803864,
      "learning_rate": 1.7779374775422207e-05,
      "loss": 0.1611,
      "step": 620
    },
    {
      "epoch": 0.6209824135839903,
      "grad_norm": 0.024578498676419258,
      "learning_rate": 1.7635644987423645e-05,
      "loss": 0.1923,
      "step": 640
    },
    {
      "epoch": 0.6403881140084899,
      "grad_norm": 9.07615852355957,
      "learning_rate": 1.7491915199425083e-05,
      "loss": 0.2025,
      "step": 660
    },
    {
      "epoch": 0.6597938144329897,
      "grad_norm": 0.38979506492614746,
      "learning_rate": 1.7348185411426518e-05,
      "loss": 0.224,
      "step": 680
    },
    {
      "epoch": 0.6791995148574894,
      "grad_norm": 19.371118545532227,
      "learning_rate": 1.7204455623427956e-05,
      "loss": 0.2593,
      "step": 700
    },
    {
      "epoch": 0.698605215281989,
      "grad_norm": 57.02224349975586,
      "learning_rate": 1.7060725835429395e-05,
      "loss": 0.1404,
      "step": 720
    },
    {
      "epoch": 0.7180109157064888,
      "grad_norm": 34.49756622314453,
      "learning_rate": 1.6916996047430833e-05,
      "loss": 0.1934,
      "step": 740
    },
    {
      "epoch": 0.7374166161309885,
      "grad_norm": 11.088144302368164,
      "learning_rate": 1.6773266259432268e-05,
      "loss": 0.159,
      "step": 760
    },
    {
      "epoch": 0.7568223165554882,
      "grad_norm": 10.713658332824707,
      "learning_rate": 1.6629536471433706e-05,
      "loss": 0.0968,
      "step": 780
    },
    {
      "epoch": 0.7762280169799879,
      "grad_norm": 0.7861451506614685,
      "learning_rate": 1.6485806683435144e-05,
      "loss": 0.1733,
      "step": 800
    },
    {
      "epoch": 0.7956337174044875,
      "grad_norm": 0.05248471349477768,
      "learning_rate": 1.634207689543658e-05,
      "loss": 0.1023,
      "step": 820
    },
    {
      "epoch": 0.8150394178289873,
      "grad_norm": 4.544727802276611,
      "learning_rate": 1.6198347107438017e-05,
      "loss": 0.1223,
      "step": 840
    },
    {
      "epoch": 0.834445118253487,
      "grad_norm": 2.436984062194824,
      "learning_rate": 1.6054617319439455e-05,
      "loss": 0.1791,
      "step": 860
    },
    {
      "epoch": 0.8538508186779866,
      "grad_norm": 0.05013319104909897,
      "learning_rate": 1.5910887531440894e-05,
      "loss": 0.1434,
      "step": 880
    },
    {
      "epoch": 0.8732565191024864,
      "grad_norm": 0.08662869036197662,
      "learning_rate": 1.576715774344233e-05,
      "loss": 0.1201,
      "step": 900
    },
    {
      "epoch": 0.8926622195269861,
      "grad_norm": 16.299718856811523,
      "learning_rate": 1.5623427955443767e-05,
      "loss": 0.1434,
      "step": 920
    },
    {
      "epoch": 0.9120679199514857,
      "grad_norm": 0.05646131932735443,
      "learning_rate": 1.5479698167445205e-05,
      "loss": 0.1717,
      "step": 940
    },
    {
      "epoch": 0.9314736203759855,
      "grad_norm": 0.09558133780956268,
      "learning_rate": 1.5335968379446643e-05,
      "loss": 0.1053,
      "step": 960
    },
    {
      "epoch": 0.9508793208004851,
      "grad_norm": 6.447268962860107,
      "learning_rate": 1.5192238591448078e-05,
      "loss": 0.2034,
      "step": 980
    },
    {
      "epoch": 0.9702850212249848,
      "grad_norm": 0.0330624096095562,
      "learning_rate": 1.5048508803449516e-05,
      "loss": 0.0799,
      "step": 1000
    },
    {
      "epoch": 0.9702850212249848,
      "eval_accuracy": 0.9618320610687023,
      "eval_f1_lab_report": 0.9590163934426229,
      "eval_f1_operative_report": 0.9489291598023064,
      "eval_f1_radiology_report": 0.9679521998913634,
      "eval_loss": 0.19497999548912048,
      "eval_macro_f1": 0.9586325843787643,
      "eval_macro_precision": 0.9538175035389038,
      "eval_macro_recall": 0.9638587098603532,
      "eval_precision_lab_report": 0.9481361426256077,
      "eval_precision_operative_report": 0.9320388349514563,
      "eval_precision_radiology_report": 0.9812775330396476,
      "eval_recall_lab_report": 0.9701492537313433,
      "eval_recall_operative_report": 0.9664429530201343,
      "eval_recall_radiology_report": 0.954983922829582,
      "eval_runtime": 74.5556,
      "eval_samples_per_second": 24.599,
      "eval_steps_per_second": 24.599,
      "eval_weighted_f1": 0.9619232155755894,
      "step": 1000
    },
    {
      "epoch": 0.9896907216494846,
      "grad_norm": 0.07646194845438004,
      "learning_rate": 1.4904779015450954e-05,
      "loss": 0.1185,
      "step": 1020
    },
    {
      "epoch": 1.0087325651910248,
      "grad_norm": 3.177273750305176,
      "learning_rate": 1.476104922745239e-05,
      "loss": 0.1305,
      "step": 1040
    },
    {
      "epoch": 1.0281382656155245,
      "grad_norm": 0.3256101608276367,
      "learning_rate": 1.4617319439453827e-05,
      "loss": 0.1422,
      "step": 1060
    },
    {
      "epoch": 1.0475439660400243,
      "grad_norm": 0.05191199854016304,
      "learning_rate": 1.4473589651455264e-05,
      "loss": 0.0818,
      "step": 1080
    },
    {
      "epoch": 1.066949666464524,
      "grad_norm": 0.156155526638031,
      "learning_rate": 1.4329859863456704e-05,
      "loss": 0.0902,
      "step": 1100
    },
    {
      "epoch": 1.0863553668890236,
      "grad_norm": 12.914316177368164,
      "learning_rate": 1.418613007545814e-05,
      "loss": 0.1257,
      "step": 1120
    },
    {
      "epoch": 1.1057610673135234,
      "grad_norm": 3.1470491886138916,
      "learning_rate": 1.4042400287459577e-05,
      "loss": 0.0983,
      "step": 1140
    },
    {
      "epoch": 1.125166767738023,
      "grad_norm": 0.03118334338068962,
      "learning_rate": 1.3898670499461015e-05,
      "loss": 0.12,
      "step": 1160
    },
    {
      "epoch": 1.1445724681625227,
      "grad_norm": 7.371583461761475,
      "learning_rate": 1.3754940711462453e-05,
      "loss": 0.1793,
      "step": 1180
    },
    {
      "epoch": 1.1639781685870225,
      "grad_norm": 0.16912613809108734,
      "learning_rate": 1.361121092346389e-05,
      "loss": 0.1069,
      "step": 1200
    },
    {
      "epoch": 1.1833838690115221,
      "grad_norm": 3.693732738494873,
      "learning_rate": 1.3467481135465326e-05,
      "loss": 0.0895,
      "step": 1220
    },
    {
      "epoch": 1.2027895694360218,
      "grad_norm": 0.022687703371047974,
      "learning_rate": 1.3323751347466764e-05,
      "loss": 0.1099,
      "step": 1240
    },
    {
      "epoch": 1.2221952698605216,
      "grad_norm": 0.05721860006451607,
      "learning_rate": 1.31800215594682e-05,
      "loss": 0.1145,
      "step": 1260
    },
    {
      "epoch": 1.2416009702850213,
      "grad_norm": 0.08816768229007721,
      "learning_rate": 1.3036291771469637e-05,
      "loss": 0.1281,
      "step": 1280
    },
    {
      "epoch": 1.2610066707095209,
      "grad_norm": 0.1484384387731552,
      "learning_rate": 1.2892561983471074e-05,
      "loss": 0.122,
      "step": 1300
    },
    {
      "epoch": 1.2804123711340205,
      "grad_norm": 0.05721456557512283,
      "learning_rate": 1.2748832195472514e-05,
      "loss": 0.0823,
      "step": 1320
    },
    {
      "epoch": 1.2998180715585204,
      "grad_norm": 0.025458386167883873,
      "learning_rate": 1.260510240747395e-05,
      "loss": 0.1565,
      "step": 1340
    },
    {
      "epoch": 1.31922377198302,
      "grad_norm": 0.035696808248758316,
      "learning_rate": 1.2461372619475387e-05,
      "loss": 0.0849,
      "step": 1360
    },
    {
      "epoch": 1.3386294724075198,
      "grad_norm": 77.40007781982422,
      "learning_rate": 1.2317642831476823e-05,
      "loss": 0.1149,
      "step": 1380
    },
    {
      "epoch": 1.3580351728320195,
      "grad_norm": 3.5960588455200195,
      "learning_rate": 1.2173913043478263e-05,
      "loss": 0.0673,
      "step": 1400
    },
    {
      "epoch": 1.377440873256519,
      "grad_norm": 4.798350811004639,
      "learning_rate": 1.20301832554797e-05,
      "loss": 0.1357,
      "step": 1420
    },
    {
      "epoch": 1.3968465736810187,
      "grad_norm": 0.013127906247973442,
      "learning_rate": 1.1886453467481136e-05,
      "loss": 0.0861,
      "step": 1440
    },
    {
      "epoch": 1.4162522741055186,
      "grad_norm": 3.35602068901062,
      "learning_rate": 1.1742723679482573e-05,
      "loss": 0.0908,
      "step": 1460
    },
    {
      "epoch": 1.4356579745300182,
      "grad_norm": 14.540184020996094,
      "learning_rate": 1.1598993891484011e-05,
      "loss": 0.1224,
      "step": 1480
    },
    {
      "epoch": 1.4550636749545178,
      "grad_norm": 0.014716535806655884,
      "learning_rate": 1.1455264103485449e-05,
      "loss": 0.0367,
      "step": 1500
    },
    {
      "epoch": 1.4550636749545178,
      "eval_accuracy": 0.9634678298800436,
      "eval_f1_lab_report": 0.9541734860883797,
      "eval_f1_operative_report": 0.9632107023411371,
      "eval_f1_radiology_report": 0.9696969696969697,
      "eval_loss": 0.22640103101730347,
      "eval_macro_f1": 0.9623603860421621,
      "eval_macro_precision": 0.9603588842689454,
      "eval_macro_recall": 0.9645394789338851,
      "eval_precision_lab_report": 0.9418416801292407,
      "eval_precision_operative_report": 0.96,
      "eval_precision_radiology_report": 0.9792349726775956,
      "eval_recall_lab_report": 0.966832504145937,
      "eval_recall_operative_report": 0.9664429530201343,
      "eval_recall_radiology_report": 0.9603429796355841,
      "eval_runtime": 74.4003,
      "eval_samples_per_second": 24.65,
      "eval_steps_per_second": 24.65,
      "eval_weighted_f1": 0.9635390807722053,
      "step": 1500
    },
    {
      "epoch": 1.4744693753790177,
      "grad_norm": 0.09435494244098663,
      "learning_rate": 1.1311534315486886e-05,
      "loss": 0.1736,
      "step": 1520
    },
    {
      "epoch": 1.4938750758035173,
      "grad_norm": 4.139671325683594,
      "learning_rate": 1.1167804527488322e-05,
      "loss": 0.106,
      "step": 1540
    },
    {
      "epoch": 1.513280776228017,
      "grad_norm": 0.10734705626964569,
      "learning_rate": 1.102407473948976e-05,
      "loss": 0.1164,
      "step": 1560
    },
    {
      "epoch": 1.5326864766525166,
      "grad_norm": 38.01471710205078,
      "learning_rate": 1.0880344951491197e-05,
      "loss": 0.1739,
      "step": 1580
    },
    {
      "epoch": 1.5520921770770164,
      "grad_norm": 4.102553844451904,
      "learning_rate": 1.0736615163492633e-05,
      "loss": 0.1108,
      "step": 1600
    },
    {
      "epoch": 1.571497877501516,
      "grad_norm": 60.5077018737793,
      "learning_rate": 1.0592885375494073e-05,
      "loss": 0.1988,
      "step": 1620
    },
    {
      "epoch": 1.5909035779260159,
      "grad_norm": 31.832502365112305,
      "learning_rate": 1.044915558749551e-05,
      "loss": 0.0927,
      "step": 1640
    },
    {
      "epoch": 1.6103092783505155,
      "grad_norm": 12.829935073852539,
      "learning_rate": 1.0305425799496946e-05,
      "loss": 0.1653,
      "step": 1660
    },
    {
      "epoch": 1.6297149787750151,
      "grad_norm": 0.03310922160744667,
      "learning_rate": 1.0161696011498383e-05,
      "loss": 0.1074,
      "step": 1680
    },
    {
      "epoch": 1.6491206791995148,
      "grad_norm": 2.218942642211914,
      "learning_rate": 1.0017966223499823e-05,
      "loss": 0.1057,
      "step": 1700
    },
    {
      "epoch": 1.6685263796240144,
      "grad_norm": 0.02302134409546852,
      "learning_rate": 9.87423643550126e-06,
      "loss": 0.1304,
      "step": 1720
    },
    {
      "epoch": 1.6879320800485142,
      "grad_norm": 0.18017438054084778,
      "learning_rate": 9.730506647502696e-06,
      "loss": 0.1198,
      "step": 1740
    },
    {
      "epoch": 1.707337780473014,
      "grad_norm": 9.171842575073242,
      "learning_rate": 9.586776859504134e-06,
      "loss": 0.072,
      "step": 1760
    },
    {
      "epoch": 1.7267434808975137,
      "grad_norm": 0.02474987506866455,
      "learning_rate": 9.44304707150557e-06,
      "loss": 0.0802,
      "step": 1780
    },
    {
      "epoch": 1.7461491813220134,
      "grad_norm": 35.86277389526367,
      "learning_rate": 9.299317283507007e-06,
      "loss": 0.1525,
      "step": 1800
    },
    {
      "epoch": 1.765554881746513,
      "grad_norm": 0.5513366460800171,
      "learning_rate": 9.155587495508445e-06,
      "loss": 0.109,
      "step": 1820
    },
    {
      "epoch": 1.7849605821710126,
      "grad_norm": 0.4787786602973938,
      "learning_rate": 9.011857707509882e-06,
      "loss": 0.0881,
      "step": 1840
    },
    {
      "epoch": 1.8043662825955125,
      "grad_norm": 58.9631462097168,
      "learning_rate": 8.868127919511318e-06,
      "loss": 0.093,
      "step": 1860
    },
    {
      "epoch": 1.823771983020012,
      "grad_norm": 1.1750720739364624,
      "learning_rate": 8.724398131512756e-06,
      "loss": 0.1067,
      "step": 1880
    },
    {
      "epoch": 1.843177683444512,
      "grad_norm": 0.019888682290911674,
      "learning_rate": 8.580668343514193e-06,
      "loss": 0.1207,
      "step": 1900
    },
    {
      "epoch": 1.8625833838690116,
      "grad_norm": 27.15701675415039,
      "learning_rate": 8.436938555515631e-06,
      "loss": 0.0722,
      "step": 1920
    },
    {
      "epoch": 1.8819890842935112,
      "grad_norm": 0.017322596162557602,
      "learning_rate": 8.293208767517068e-06,
      "loss": 0.0825,
      "step": 1940
    },
    {
      "epoch": 1.9013947847180108,
      "grad_norm": 9.127324104309082,
      "learning_rate": 8.149478979518506e-06,
      "loss": 0.0302,
      "step": 1960
    },
    {
      "epoch": 1.9208004851425104,
      "grad_norm": 0.03801129013299942,
      "learning_rate": 8.005749191519944e-06,
      "loss": 0.1329,
      "step": 1980
    },
    {
      "epoch": 1.9402061855670103,
      "grad_norm": 0.4433538019657135,
      "learning_rate": 7.86201940352138e-06,
      "loss": 0.0602,
      "step": 2000
    },
    {
      "epoch": 1.9402061855670103,
      "eval_accuracy": 0.9640130861504908,
      "eval_f1_lab_report": 0.9573070607553367,
      "eval_f1_operative_report": 0.9613445378151261,
      "eval_f1_radiology_report": 0.9692722371967655,
      "eval_loss": 0.18627716600894928,
      "eval_macro_f1": 0.962641278589076,
      "eval_macro_precision": 0.9619948908575613,
      "eval_macro_recall": 0.9633741538297612,
      "eval_precision_lab_report": 0.9479674796747968,
      "eval_precision_operative_report": 0.9629629629629629,
      "eval_precision_radiology_report": 0.9750542299349241,
      "eval_recall_lab_report": 0.966832504145937,
      "eval_recall_operative_report": 0.959731543624161,
      "eval_recall_radiology_report": 0.9635584137191854,
      "eval_runtime": 76.3508,
      "eval_samples_per_second": 24.021,
      "eval_steps_per_second": 24.021,
      "eval_weighted_f1": 0.9640500693614819,
      "step": 2000
    },
    {
      "epoch": 1.9596118859915101,
      "grad_norm": 0.8890208601951599,
      "learning_rate": 7.718289615522819e-06,
      "loss": 0.0708,
      "step": 2020
    },
    {
      "epoch": 1.9790175864160098,
      "grad_norm": 20.68431854248047,
      "learning_rate": 7.574559827524255e-06,
      "loss": 0.1728,
      "step": 2040
    },
    {
      "epoch": 1.9984232868405094,
      "grad_norm": 0.05337516963481903,
      "learning_rate": 7.430830039525693e-06,
      "loss": 0.0552,
      "step": 2060
    },
    {
      "epoch": 2.0174651303820497,
      "grad_norm": 0.43338507413864136,
      "learning_rate": 7.287100251527129e-06,
      "loss": 0.056,
      "step": 2080
    },
    {
      "epoch": 2.0368708308065493,
      "grad_norm": 0.017596101388335228,
      "learning_rate": 7.143370463528567e-06,
      "loss": 0.0264,
      "step": 2100
    },
    {
      "epoch": 2.056276531231049,
      "grad_norm": 0.009210024029016495,
      "learning_rate": 6.999640675530004e-06,
      "loss": 0.0629,
      "step": 2120
    },
    {
      "epoch": 2.075682231655549,
      "grad_norm": 0.020559094846248627,
      "learning_rate": 6.855910887531442e-06,
      "loss": 0.0847,
      "step": 2140
    },
    {
      "epoch": 2.0950879320800486,
      "grad_norm": 9.2818603515625,
      "learning_rate": 6.712181099532879e-06,
      "loss": 0.076,
      "step": 2160
    },
    {
      "epoch": 2.1144936325045482,
      "grad_norm": 1.3512309789657593,
      "learning_rate": 6.568451311534316e-06,
      "loss": 0.0855,
      "step": 2180
    },
    {
      "epoch": 2.133899332929048,
      "grad_norm": 0.3617396652698517,
      "learning_rate": 6.424721523535753e-06,
      "loss": 0.1217,
      "step": 2200
    },
    {
      "epoch": 2.1533050333535475,
      "grad_norm": 0.13735735416412354,
      "learning_rate": 6.280991735537191e-06,
      "loss": 0.0461,
      "step": 2220
    },
    {
      "epoch": 2.172710733778047,
      "grad_norm": 0.0184041578322649,
      "learning_rate": 6.137261947538627e-06,
      "loss": 0.0747,
      "step": 2240
    },
    {
      "epoch": 2.192116434202547,
      "grad_norm": 1.5114548206329346,
      "learning_rate": 5.993532159540065e-06,
      "loss": 0.0036,
      "step": 2260
    },
    {
      "epoch": 2.211522134627047,
      "grad_norm": 0.04163361340761185,
      "learning_rate": 5.849802371541502e-06,
      "loss": 0.0558,
      "step": 2280
    },
    {
      "epoch": 2.2309278350515465,
      "grad_norm": 0.010899573564529419,
      "learning_rate": 5.70607258354294e-06,
      "loss": 0.1196,
      "step": 2300
    },
    {
      "epoch": 2.250333535476046,
      "grad_norm": 0.016753580421209335,
      "learning_rate": 5.562342795544377e-06,
      "loss": 0.1146,
      "step": 2320
    },
    {
      "epoch": 2.2697392359005457,
      "grad_norm": 0.016113819554448128,
      "learning_rate": 5.418613007545814e-06,
      "loss": 0.0876,
      "step": 2340
    },
    {
      "epoch": 2.2891449363250453,
      "grad_norm": 2.43402099609375,
      "learning_rate": 5.274883219547252e-06,
      "loss": 0.1042,
      "step": 2360
    },
    {
      "epoch": 2.308550636749545,
      "grad_norm": 3.1326165199279785,
      "learning_rate": 5.131153431548689e-06,
      "loss": 0.0646,
      "step": 2380
    },
    {
      "epoch": 2.327956337174045,
      "grad_norm": 50.91073989868164,
      "learning_rate": 4.987423643550126e-06,
      "loss": 0.0952,
      "step": 2400
    },
    {
      "epoch": 2.3473620375985447,
      "grad_norm": 24.86041259765625,
      "learning_rate": 4.8436938555515634e-06,
      "loss": 0.0743,
      "step": 2420
    },
    {
      "epoch": 2.3667677380230443,
      "grad_norm": 0.04265863448381424,
      "learning_rate": 4.699964067553001e-06,
      "loss": 0.0568,
      "step": 2440
    },
    {
      "epoch": 2.386173438447544,
      "grad_norm": 0.01936446689069271,
      "learning_rate": 4.556234279554438e-06,
      "loss": 0.0785,
      "step": 2460
    },
    {
      "epoch": 2.4055791388720436,
      "grad_norm": 10.52826976776123,
      "learning_rate": 4.4125044915558755e-06,
      "loss": 0.0975,
      "step": 2480
    },
    {
      "epoch": 2.424984839296543,
      "grad_norm": 0.04131530225276947,
      "learning_rate": 4.268774703557312e-06,
      "loss": 0.0604,
      "step": 2500
    },
    {
      "epoch": 2.424984839296543,
      "eval_accuracy": 0.9689203925845147,
      "eval_f1_lab_report": 0.9628406275805119,
      "eval_f1_operative_report": 0.9715242881072027,
      "eval_f1_radiology_report": 0.9720430107526882,
      "eval_loss": 0.1429206132888794,
      "eval_macro_f1": 0.9688026421468009,
      "eval_macro_precision": 0.9679900085044105,
      "eval_macro_recall": 0.9696347790290774,
      "eval_precision_lab_report": 0.9588815789473685,
      "eval_precision_operative_report": 0.9698996655518395,
      "eval_precision_radiology_report": 0.9751887810140237,
      "eval_recall_lab_report": 0.966832504145937,
      "eval_recall_operative_report": 0.9731543624161074,
      "eval_recall_radiology_report": 0.9689174705251876,
      "eval_runtime": 76.4838,
      "eval_samples_per_second": 23.979,
      "eval_steps_per_second": 23.979,
      "eval_weighted_f1": 0.9689330781457215,
      "step": 2500
    },
    {
      "epoch": 2.4443905397210433,
      "grad_norm": 0.013441434130072594,
      "learning_rate": 4.125044915558749e-06,
      "loss": 0.0245,
      "step": 2520
    },
    {
      "epoch": 2.463796240145543,
      "grad_norm": 0.011124825105071068,
      "learning_rate": 3.981315127560187e-06,
      "loss": 0.0639,
      "step": 2540
    },
    {
      "epoch": 2.4832019405700425,
      "grad_norm": 0.14957989752292633,
      "learning_rate": 3.837585339561625e-06,
      "loss": 0.0563,
      "step": 2560
    },
    {
      "epoch": 2.502607640994542,
      "grad_norm": 0.04595116898417473,
      "learning_rate": 3.693855551563062e-06,
      "loss": 0.1041,
      "step": 2580
    },
    {
      "epoch": 2.5220133414190418,
      "grad_norm": 0.014421135187149048,
      "learning_rate": 3.5501257635644993e-06,
      "loss": 0.0271,
      "step": 2600
    },
    {
      "epoch": 2.5414190418435414,
      "grad_norm": 0.4144510328769684,
      "learning_rate": 3.4063959755659366e-06,
      "loss": 0.113,
      "step": 2620
    },
    {
      "epoch": 2.560824742268041,
      "grad_norm": 0.07063297182321548,
      "learning_rate": 3.2626661875673735e-06,
      "loss": 0.0172,
      "step": 2640
    },
    {
      "epoch": 2.580230442692541,
      "grad_norm": 0.00817196350544691,
      "learning_rate": 3.118936399568811e-06,
      "loss": 0.0459,
      "step": 2660
    },
    {
      "epoch": 2.5996361431170407,
      "grad_norm": 0.019881105050444603,
      "learning_rate": 2.9752066115702483e-06,
      "loss": 0.1082,
      "step": 2680
    },
    {
      "epoch": 2.6190418435415403,
      "grad_norm": 0.01721939630806446,
      "learning_rate": 2.8314768235716856e-06,
      "loss": 0.0233,
      "step": 2700
    },
    {
      "epoch": 2.63844754396604,
      "grad_norm": 0.01037369854748249,
      "learning_rate": 2.687747035573123e-06,
      "loss": 0.0232,
      "step": 2720
    },
    {
      "epoch": 2.6578532443905396,
      "grad_norm": 24.733327865600586,
      "learning_rate": 2.54401724757456e-06,
      "loss": 0.0983,
      "step": 2740
    },
    {
      "epoch": 2.6772589448150397,
      "grad_norm": 0.025556540116667747,
      "learning_rate": 2.4002874595759973e-06,
      "loss": 0.014,
      "step": 2760
    },
    {
      "epoch": 2.6966646452395393,
      "grad_norm": 0.006412121932953596,
      "learning_rate": 2.2565576715774346e-06,
      "loss": 0.0413,
      "step": 2780
    },
    {
      "epoch": 2.716070345664039,
      "grad_norm": 0.4874618947505951,
      "learning_rate": 2.112827883578872e-06,
      "loss": 0.0336,
      "step": 2800
    },
    {
      "epoch": 2.7354760460885386,
      "grad_norm": 0.45429569482803345,
      "learning_rate": 1.969098095580309e-06,
      "loss": 0.0433,
      "step": 2820
    },
    {
      "epoch": 2.754881746513038,
      "grad_norm": 11.270548820495605,
      "learning_rate": 1.8253683075817465e-06,
      "loss": 0.0362,
      "step": 2840
    },
    {
      "epoch": 2.774287446937538,
      "grad_norm": 5.483437538146973,
      "learning_rate": 1.6816385195831839e-06,
      "loss": 0.0068,
      "step": 2860
    },
    {
      "epoch": 2.7936931473620374,
      "grad_norm": 0.021955380216240883,
      "learning_rate": 1.537908731584621e-06,
      "loss": 0.0389,
      "step": 2880
    },
    {
      "epoch": 2.813098847786537,
      "grad_norm": 0.082429438829422,
      "learning_rate": 1.3941789435860584e-06,
      "loss": 0.1418,
      "step": 2900
    },
    {
      "epoch": 2.832504548211037,
      "grad_norm": 0.024992460384964943,
      "learning_rate": 1.2504491555874957e-06,
      "loss": 0.0418,
      "step": 2920
    },
    {
      "epoch": 2.8519102486355368,
      "grad_norm": 16.65825080871582,
      "learning_rate": 1.1067193675889329e-06,
      "loss": 0.1293,
      "step": 2940
    },
    {
      "epoch": 2.8713159490600364,
      "grad_norm": 0.013967704959213734,
      "learning_rate": 9.629895795903702e-07,
      "loss": 0.0556,
      "step": 2960
    },
    {
      "epoch": 2.890721649484536,
      "grad_norm": 3.084559917449951,
      "learning_rate": 8.192597915918074e-07,
      "loss": 0.0314,
      "step": 2980
    },
    {
      "epoch": 2.9101273499090357,
      "grad_norm": 0.006931495852768421,
      "learning_rate": 6.755300035932449e-07,
      "loss": 0.0447,
      "step": 3000
    },
    {
      "epoch": 2.9101273499090357,
      "eval_accuracy": 0.9672846237731734,
      "eval_f1_lab_report": 0.9615699100572364,
      "eval_f1_operative_report": 0.9695945945945946,
      "eval_f1_radiology_report": 0.9703184025903939,
      "eval_loss": 0.18174326419830322,
      "eval_macro_f1": 0.9671609690807417,
      "eval_macro_precision": 0.9672504953360493,
      "eval_macro_recall": 0.9672566800502619,
      "eval_precision_lab_report": 0.9483870967741935,
      "eval_precision_operative_report": 0.9761904761904762,
      "eval_precision_radiology_report": 0.9771739130434782,
      "eval_recall_lab_report": 0.9751243781094527,
      "eval_recall_operative_report": 0.9630872483221476,
      "eval_recall_radiology_report": 0.9635584137191854,
      "eval_runtime": 75.8211,
      "eval_samples_per_second": 24.189,
      "eval_steps_per_second": 24.189,
      "eval_weighted_f1": 0.9673243808999674,
      "step": 3000
    },
    {
      "epoch": 2.9295330503335357,
      "grad_norm": 0.009754876606166363,
      "learning_rate": 5.318002155946821e-07,
      "loss": 0.0418,
      "step": 3020
    },
    {
      "epoch": 2.9489387507580354,
      "grad_norm": 0.00972274225205183,
      "learning_rate": 3.880704275961193e-07,
      "loss": 0.0475,
      "step": 3040
    },
    {
      "epoch": 2.968344451182535,
      "grad_norm": 0.5922208428382874,
      "learning_rate": 2.443406395975566e-07,
      "loss": 0.0214,
      "step": 3060
    },
    {
      "epoch": 2.9877501516070346,
      "grad_norm": 0.08816200494766235,
      "learning_rate": 1.0061085159899391e-07,
      "loss": 0.0484,
      "step": 3080
    },
    {
      "epoch": 3.0,
      "step": 3093,
      "total_flos": 4.625429557128192e+16,
      "train_loss": 0.13503855510683765,
      "train_runtime": 5596.7286,
      "train_samples_per_second": 8.839,
      "train_steps_per_second": 0.553
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9689203925845147,
      "eval_f1_lab_report": 0.9628406275805119,
      "eval_f1_operative_report": 0.9715242881072027,
      "eval_f1_radiology_report": 0.9720430107526882,
      "eval_loss": 0.1427270472049713,
      "eval_macro_f1": 0.9688026421468009,
      "eval_macro_precision": 0.9679900085044105,
      "eval_macro_recall": 0.9696347790290774,
      "eval_precision_lab_report": 0.9588815789473685,
      "eval_precision_operative_report": 0.9698996655518395,
      "eval_precision_radiology_report": 0.9751887810140237,
      "eval_recall_lab_report": 0.966832504145937,
      "eval_recall_operative_report": 0.9731543624161074,
      "eval_recall_radiology_report": 0.9689174705251876,
      "eval_runtime": 74.9471,
      "eval_samples_per_second": 24.471,
      "eval_steps_per_second": 24.471,
      "eval_weighted_f1": 0.9689330781457215,
      "step": 3093
    }
  ],
  "logging_steps": 20,
  "max_steps": 3093,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "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": 4.625429557128192e+16,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}