{
  "best_metric": 1.064771056175232,
  "best_model_checkpoint": "sbert-ru-huawei-sentiment-fine-up/checkpoint-4868",
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 4868,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.010271158586688579,
      "grad_norm": 31.50909423828125,
      "learning_rate": 5.476951163852123e-07,
      "loss": 5.7521,
      "step": 25
    },
    {
      "epoch": 0.020542317173377157,
      "grad_norm": 65.67906951904297,
      "learning_rate": 1.1182108626198083e-06,
      "loss": 5.3431,
      "step": 50
    },
    {
      "epoch": 0.030813475760065736,
      "grad_norm": 47.638511657714844,
      "learning_rate": 1.6659059790050208e-06,
      "loss": 3.9751,
      "step": 75
    },
    {
      "epoch": 0.041084634346754315,
      "grad_norm": 23.86779022216797,
      "learning_rate": 2.2364217252396165e-06,
      "loss": 1.9721,
      "step": 100
    },
    {
      "epoch": 0.05135579293344289,
      "grad_norm": 17.054542541503906,
      "learning_rate": 2.806937471474213e-06,
      "loss": 1.3281,
      "step": 125
    },
    {
      "epoch": 0.06162695152013147,
      "grad_norm": 10.087335586547852,
      "learning_rate": 3.3774532177088093e-06,
      "loss": 1.236,
      "step": 150
    },
    {
      "epoch": 0.07189811010682005,
      "grad_norm": 22.066057205200195,
      "learning_rate": 3.9479689639434044e-06,
      "loss": 1.0576,
      "step": 175
    },
    {
      "epoch": 0.08216926869350863,
      "grad_norm": 16.504131317138672,
      "learning_rate": 4.518484710178001e-06,
      "loss": 0.9854,
      "step": 200
    },
    {
      "epoch": 0.0924404272801972,
      "grad_norm": 16.981975555419922,
      "learning_rate": 5.089000456412597e-06,
      "loss": 0.9098,
      "step": 225
    },
    {
      "epoch": 0.10271158586688578,
      "grad_norm": 16.270986557006836,
      "learning_rate": 5.659516202647193e-06,
      "loss": 0.9133,
      "step": 250
    },
    {
      "epoch": 0.11298274445357437,
      "grad_norm": 13.069902420043945,
      "learning_rate": 6.230031948881789e-06,
      "loss": 0.8994,
      "step": 275
    },
    {
      "epoch": 0.12325390304026294,
      "grad_norm": 12.201191902160645,
      "learning_rate": 6.800547695116385e-06,
      "loss": 0.8731,
      "step": 300
    },
    {
      "epoch": 0.13352506162695152,
      "grad_norm": 7.924577236175537,
      "learning_rate": 7.371063441350981e-06,
      "loss": 0.9483,
      "step": 325
    },
    {
      "epoch": 0.1437962202136401,
      "grad_norm": 16.856870651245117,
      "learning_rate": 7.941579187585578e-06,
      "loss": 0.7615,
      "step": 350
    },
    {
      "epoch": 0.15406737880032867,
      "grad_norm": 17.550830841064453,
      "learning_rate": 8.512094933820173e-06,
      "loss": 0.8699,
      "step": 375
    },
    {
      "epoch": 0.16433853738701726,
      "grad_norm": 8.01524543762207,
      "learning_rate": 9.08261068005477e-06,
      "loss": 0.9946,
      "step": 400
    },
    {
      "epoch": 0.17460969597370585,
      "grad_norm": 12.508491516113281,
      "learning_rate": 9.653126426289367e-06,
      "loss": 0.8598,
      "step": 425
    },
    {
      "epoch": 0.1848808545603944,
      "grad_norm": 11.055355072021484,
      "learning_rate": 1.0223642172523962e-05,
      "loss": 0.8902,
      "step": 450
    },
    {
      "epoch": 0.195152013147083,
      "grad_norm": 12.723591804504395,
      "learning_rate": 1.0794157918758559e-05,
      "loss": 0.9183,
      "step": 475
    },
    {
      "epoch": 0.20542317173377156,
      "grad_norm": 29.50878143310547,
      "learning_rate": 1.1364673664993154e-05,
      "loss": 0.9221,
      "step": 500
    },
    {
      "epoch": 0.21569433032046015,
      "grad_norm": 15.626035690307617,
      "learning_rate": 1.193518941122775e-05,
      "loss": 0.8211,
      "step": 525
    },
    {
      "epoch": 0.22596548890714874,
      "grad_norm": 10.57879638671875,
      "learning_rate": 1.2505705157462347e-05,
      "loss": 0.9007,
      "step": 550
    },
    {
      "epoch": 0.2362366474938373,
      "grad_norm": 10.496281623840332,
      "learning_rate": 1.3076220903696942e-05,
      "loss": 0.8799,
      "step": 575
    },
    {
      "epoch": 0.2465078060805259,
      "grad_norm": 12.28962516784668,
      "learning_rate": 1.3646736649931538e-05,
      "loss": 0.8934,
      "step": 600
    },
    {
      "epoch": 0.25677896466721445,
      "grad_norm": 10.393588066101074,
      "learning_rate": 1.4217252396166134e-05,
      "loss": 0.8276,
      "step": 625
    },
    {
      "epoch": 0.26705012325390304,
      "grad_norm": 11.521328926086426,
      "learning_rate": 1.4787768142400731e-05,
      "loss": 0.8975,
      "step": 650
    },
    {
      "epoch": 0.2773212818405916,
      "grad_norm": 10.707319259643555,
      "learning_rate": 1.5358283888635328e-05,
      "loss": 0.8862,
      "step": 675
    },
    {
      "epoch": 0.2875924404272802,
      "grad_norm": 11.027580261230469,
      "learning_rate": 1.5928799634869923e-05,
      "loss": 0.839,
      "step": 700
    },
    {
      "epoch": 0.29786359901396875,
      "grad_norm": 8.072233200073242,
      "learning_rate": 1.6499315381104518e-05,
      "loss": 0.8836,
      "step": 725
    },
    {
      "epoch": 0.30813475760065734,
      "grad_norm": 5.254186153411865,
      "learning_rate": 1.7069831127339113e-05,
      "loss": 0.8102,
      "step": 750
    },
    {
      "epoch": 0.3184059161873459,
      "grad_norm": 9.644079208374023,
      "learning_rate": 1.7640346873573712e-05,
      "loss": 0.8363,
      "step": 775
    },
    {
      "epoch": 0.3286770747740345,
      "grad_norm": 17.169221878051758,
      "learning_rate": 1.8210862619808307e-05,
      "loss": 0.7916,
      "step": 800
    },
    {
      "epoch": 0.3389482333607231,
      "grad_norm": 12.09295654296875,
      "learning_rate": 1.8781378366042902e-05,
      "loss": 0.8846,
      "step": 825
    },
    {
      "epoch": 0.3492193919474117,
      "grad_norm": 20.629968643188477,
      "learning_rate": 1.93518941122775e-05,
      "loss": 0.9433,
      "step": 850
    },
    {
      "epoch": 0.35949055053410023,
      "grad_norm": 12.02403736114502,
      "learning_rate": 1.9922409858512096e-05,
      "loss": 0.7863,
      "step": 875
    },
    {
      "epoch": 0.3697617091207888,
      "grad_norm": 7.93563985824585,
      "learning_rate": 2.0492925604746694e-05,
      "loss": 0.8346,
      "step": 900
    },
    {
      "epoch": 0.3800328677074774,
      "grad_norm": 10.790715217590332,
      "learning_rate": 2.106344135098129e-05,
      "loss": 0.8584,
      "step": 925
    },
    {
      "epoch": 0.390304026294166,
      "grad_norm": 13.152490615844727,
      "learning_rate": 2.1633957097215885e-05,
      "loss": 0.8276,
      "step": 950
    },
    {
      "epoch": 0.4005751848808546,
      "grad_norm": 8.368447303771973,
      "learning_rate": 2.220447284345048e-05,
      "loss": 0.9162,
      "step": 975
    },
    {
      "epoch": 0.4108463434675431,
      "grad_norm": 5.231266021728516,
      "learning_rate": 2.2774988589685075e-05,
      "loss": 0.8999,
      "step": 1000
    },
    {
      "epoch": 0.4211175020542317,
      "grad_norm": 4.503598690032959,
      "learning_rate": 2.334550433591967e-05,
      "loss": 0.8748,
      "step": 1025
    },
    {
      "epoch": 0.4313886606409203,
      "grad_norm": 5.98704719543457,
      "learning_rate": 2.391602008215427e-05,
      "loss": 0.9061,
      "step": 1050
    },
    {
      "epoch": 0.4416598192276089,
      "grad_norm": 9.622344970703125,
      "learning_rate": 2.4486535828388864e-05,
      "loss": 0.9336,
      "step": 1075
    },
    {
      "epoch": 0.4519309778142975,
      "grad_norm": 7.320569038391113,
      "learning_rate": 2.5057051574623462e-05,
      "loss": 0.8711,
      "step": 1100
    },
    {
      "epoch": 0.462202136400986,
      "grad_norm": 6.236270427703857,
      "learning_rate": 2.562756732085806e-05,
      "loss": 0.8359,
      "step": 1125
    },
    {
      "epoch": 0.4724732949876746,
      "grad_norm": 9.492266654968262,
      "learning_rate": 2.6198083067092656e-05,
      "loss": 0.8248,
      "step": 1150
    },
    {
      "epoch": 0.4827444535743632,
      "grad_norm": 9.270378112792969,
      "learning_rate": 2.676859881332725e-05,
      "loss": 0.8539,
      "step": 1175
    },
    {
      "epoch": 0.4930156121610518,
      "grad_norm": 8.744667053222656,
      "learning_rate": 2.7316293929712462e-05,
      "loss": 0.9035,
      "step": 1200
    },
    {
      "epoch": 0.5032867707477403,
      "grad_norm": 7.469476222991943,
      "learning_rate": 2.7886809675947057e-05,
      "loss": 0.8951,
      "step": 1225
    },
    {
      "epoch": 0.5135579293344289,
      "grad_norm": 6.202262878417969,
      "learning_rate": 2.8457325422181653e-05,
      "loss": 0.9175,
      "step": 1250
    },
    {
      "epoch": 0.5238290879211175,
      "grad_norm": 6.080565452575684,
      "learning_rate": 2.9027841168416248e-05,
      "loss": 0.8948,
      "step": 1275
    },
    {
      "epoch": 0.5341002465078061,
      "grad_norm": 10.15191650390625,
      "learning_rate": 2.9598356914650843e-05,
      "loss": 0.8631,
      "step": 1300
    },
    {
      "epoch": 0.5443714050944947,
      "grad_norm": 13.219573020935059,
      "learning_rate": 3.0168872660885445e-05,
      "loss": 0.9074,
      "step": 1325
    },
    {
      "epoch": 0.5546425636811833,
      "grad_norm": 7.929615497589111,
      "learning_rate": 3.073938840712004e-05,
      "loss": 0.8134,
      "step": 1350
    },
    {
      "epoch": 0.5649137222678718,
      "grad_norm": 9.197999954223633,
      "learning_rate": 3.130990415335464e-05,
      "loss": 0.8826,
      "step": 1375
    },
    {
      "epoch": 0.5751848808545604,
      "grad_norm": 9.510801315307617,
      "learning_rate": 3.1880419899589233e-05,
      "loss": 0.865,
      "step": 1400
    },
    {
      "epoch": 0.585456039441249,
      "grad_norm": 14.126985549926758,
      "learning_rate": 3.245093564582383e-05,
      "loss": 0.9076,
      "step": 1425
    },
    {
      "epoch": 0.5957271980279375,
      "grad_norm": 13.595422744750977,
      "learning_rate": 3.3021451392058424e-05,
      "loss": 0.8754,
      "step": 1450
    },
    {
      "epoch": 0.6059983566146261,
      "grad_norm": 8.661046028137207,
      "learning_rate": 3.359196713829302e-05,
      "loss": 0.8723,
      "step": 1475
    },
    {
      "epoch": 0.6162695152013147,
      "grad_norm": 13.609139442443848,
      "learning_rate": 3.4162482884527614e-05,
      "loss": 0.8326,
      "step": 1500
    },
    {
      "epoch": 0.6265406737880033,
      "grad_norm": 14.61794662475586,
      "learning_rate": 3.473299863076221e-05,
      "loss": 0.8022,
      "step": 1525
    },
    {
      "epoch": 0.6368118323746919,
      "grad_norm": 10.622493743896484,
      "learning_rate": 3.5303514376996804e-05,
      "loss": 0.8726,
      "step": 1550
    },
    {
      "epoch": 0.6470829909613804,
      "grad_norm": 5.068788051605225,
      "learning_rate": 3.585120949338202e-05,
      "loss": 0.8936,
      "step": 1575
    },
    {
      "epoch": 0.657354149548069,
      "grad_norm": 6.1167402267456055,
      "learning_rate": 3.6421725239616614e-05,
      "loss": 0.8834,
      "step": 1600
    },
    {
      "epoch": 0.6676253081347576,
      "grad_norm": 6.667758941650391,
      "learning_rate": 3.699224098585121e-05,
      "loss": 0.925,
      "step": 1625
    },
    {
      "epoch": 0.6778964667214462,
      "grad_norm": 11.814676284790039,
      "learning_rate": 3.7562756732085804e-05,
      "loss": 0.8439,
      "step": 1650
    },
    {
      "epoch": 0.6881676253081348,
      "grad_norm": 13.39101505279541,
      "learning_rate": 3.8133272478320406e-05,
      "loss": 0.7441,
      "step": 1675
    },
    {
      "epoch": 0.6984387838948234,
      "grad_norm": 7.23984956741333,
      "learning_rate": 3.8703788224555e-05,
      "loss": 0.939,
      "step": 1700
    },
    {
      "epoch": 0.7087099424815119,
      "grad_norm": 16.45111846923828,
      "learning_rate": 3.9274303970789596e-05,
      "loss": 0.9217,
      "step": 1725
    },
    {
      "epoch": 0.7189811010682005,
      "grad_norm": 13.015716552734375,
      "learning_rate": 3.984481971702419e-05,
      "loss": 0.9648,
      "step": 1750
    },
    {
      "epoch": 0.729252259654889,
      "grad_norm": 4.811708927154541,
      "learning_rate": 4.041533546325879e-05,
      "loss": 0.854,
      "step": 1775
    },
    {
      "epoch": 0.7395234182415776,
      "grad_norm": 12.734000205993652,
      "learning_rate": 4.098585120949339e-05,
      "loss": 0.8663,
      "step": 1800
    },
    {
      "epoch": 0.7497945768282662,
      "grad_norm": 5.706826210021973,
      "learning_rate": 4.1556366955727984e-05,
      "loss": 0.9436,
      "step": 1825
    },
    {
      "epoch": 0.7600657354149548,
      "grad_norm": 10.904698371887207,
      "learning_rate": 4.212688270196258e-05,
      "loss": 0.9147,
      "step": 1850
    },
    {
      "epoch": 0.7703368940016434,
      "grad_norm": 13.604780197143555,
      "learning_rate": 4.2697398448197174e-05,
      "loss": 0.9512,
      "step": 1875
    },
    {
      "epoch": 0.780608052588332,
      "grad_norm": 11.89741325378418,
      "learning_rate": 4.326791419443177e-05,
      "loss": 0.9356,
      "step": 1900
    },
    {
      "epoch": 0.7908792111750206,
      "grad_norm": 9.004523277282715,
      "learning_rate": 4.3838429940666364e-05,
      "loss": 0.927,
      "step": 1925
    },
    {
      "epoch": 0.8011503697617092,
      "grad_norm": 11.654400825500488,
      "learning_rate": 4.440894568690096e-05,
      "loss": 0.8424,
      "step": 1950
    },
    {
      "epoch": 0.8114215283483976,
      "grad_norm": 8.863574028015137,
      "learning_rate": 4.4979461433135554e-05,
      "loss": 0.8867,
      "step": 1975
    },
    {
      "epoch": 0.8216926869350862,
      "grad_norm": 6.209610462188721,
      "learning_rate": 4.554997717937015e-05,
      "loss": 0.8153,
      "step": 2000
    },
    {
      "epoch": 0.8319638455217748,
      "grad_norm": 7.504034996032715,
      "learning_rate": 4.6120492925604745e-05,
      "loss": 0.855,
      "step": 2025
    },
    {
      "epoch": 0.8422350041084634,
      "grad_norm": 12.570643424987793,
      "learning_rate": 4.669100867183934e-05,
      "loss": 0.9577,
      "step": 2050
    },
    {
      "epoch": 0.852506162695152,
      "grad_norm": 9.295639038085938,
      "learning_rate": 4.726152441807394e-05,
      "loss": 0.84,
      "step": 2075
    },
    {
      "epoch": 0.8627773212818406,
      "grad_norm": 8.446712493896484,
      "learning_rate": 4.783204016430854e-05,
      "loss": 0.9405,
      "step": 2100
    },
    {
      "epoch": 0.8730484798685292,
      "grad_norm": 9.992650032043457,
      "learning_rate": 4.840255591054313e-05,
      "loss": 1.0059,
      "step": 2125
    },
    {
      "epoch": 0.8833196384552178,
      "grad_norm": 9.760039329528809,
      "learning_rate": 4.897307165677773e-05,
      "loss": 0.8583,
      "step": 2150
    },
    {
      "epoch": 0.8935907970419064,
      "grad_norm": 17.948223114013672,
      "learning_rate": 4.954358740301233e-05,
      "loss": 0.9798,
      "step": 2175
    },
    {
      "epoch": 0.903861955628595,
      "grad_norm": 10.00885009765625,
      "learning_rate": 4.9999992064830994e-05,
      "loss": 1.1331,
      "step": 2200
    },
    {
      "epoch": 0.9141331142152835,
      "grad_norm": 9.577177047729492,
      "learning_rate": 4.9999714334444495e-05,
      "loss": 1.1493,
      "step": 2225
    },
    {
      "epoch": 0.924404272801972,
      "grad_norm": 12.528915405273438,
      "learning_rate": 4.999903985064472e-05,
      "loss": 1.1891,
      "step": 2250
    },
    {
      "epoch": 0.9346754313886606,
      "grad_norm": 21.42876625061035,
      "learning_rate": 4.9997968624135974e-05,
      "loss": 1.1592,
      "step": 2275
    },
    {
      "epoch": 0.9449465899753492,
      "grad_norm": 4.62033748626709,
      "learning_rate": 4.9996500671918936e-05,
      "loss": 1.1924,
      "step": 2300
    },
    {
      "epoch": 0.9552177485620378,
      "grad_norm": 12.126282691955566,
      "learning_rate": 4.999463601729047e-05,
      "loss": 1.1318,
      "step": 2325
    },
    {
      "epoch": 0.9654889071487264,
      "grad_norm": 7.639413356781006,
      "learning_rate": 4.999237468984326e-05,
      "loss": 1.1486,
      "step": 2350
    },
    {
      "epoch": 0.975760065735415,
      "grad_norm": 7.761475563049316,
      "learning_rate": 4.998971672546527e-05,
      "loss": 1.1463,
      "step": 2375
    },
    {
      "epoch": 0.9860312243221035,
      "grad_norm": 9.19140911102295,
      "learning_rate": 4.998666216633926e-05,
      "loss": 1.0832,
      "step": 2400
    },
    {
      "epoch": 0.9963023829087921,
      "grad_norm": 6.729145526885986,
      "learning_rate": 4.998321106094204e-05,
      "loss": 1.0749,
      "step": 2425
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.5863556971129148,
      "eval_f1_macro": 0.29454395983476717,
      "eval_f1_micro": 0.5863556971129148,
      "eval_f1_weighted": 0.47959265391783995,
      "eval_loss": 1.1192505359649658,
      "eval_precision_macro": 0.4105745703977606,
      "eval_precision_micro": 0.5863556971129148,
      "eval_precision_weighted": 0.5196212057725517,
      "eval_recall_macro": 0.3058262541410274,
      "eval_recall_micro": 0.5863556971129148,
      "eval_recall_weighted": 0.5863556971129148,
      "eval_runtime": 58.2052,
      "eval_samples_per_second": 167.219,
      "eval_steps_per_second": 5.24,
      "step": 2434
    },
    {
      "epoch": 1.0065735414954806,
      "grad_norm": 12.559377670288086,
      "learning_rate": 4.997936346404375e-05,
      "loss": 1.2554,
      "step": 2450
    },
    {
      "epoch": 1.0168447000821692,
      "grad_norm": 5.791057586669922,
      "learning_rate": 4.997511943670697e-05,
      "loss": 1.084,
      "step": 2475
    },
    {
      "epoch": 1.0271158586688578,
      "grad_norm": 13.0701322555542,
      "learning_rate": 4.9970479046285785e-05,
      "loss": 1.1482,
      "step": 2500
    },
    {
      "epoch": 1.0373870172555464,
      "grad_norm": 51.21843338012695,
      "learning_rate": 4.9965442366424646e-05,
      "loss": 1.1803,
      "step": 2525
    },
    {
      "epoch": 1.047658175842235,
      "grad_norm": 8.89924430847168,
      "learning_rate": 4.996000947705727e-05,
      "loss": 1.1225,
      "step": 2550
    },
    {
      "epoch": 1.0579293344289236,
      "grad_norm": 9.04704761505127,
      "learning_rate": 4.9954180464405345e-05,
      "loss": 1.1617,
      "step": 2575
    },
    {
      "epoch": 1.0682004930156122,
      "grad_norm": 7.195584297180176,
      "learning_rate": 4.9947955420977144e-05,
      "loss": 1.0882,
      "step": 2600
    },
    {
      "epoch": 1.0784716516023007,
      "grad_norm": 16.638315200805664,
      "learning_rate": 4.994133444556609e-05,
      "loss": 1.1194,
      "step": 2625
    },
    {
      "epoch": 1.0887428101889893,
      "grad_norm": 4.319007396697998,
      "learning_rate": 4.993431764324916e-05,
      "loss": 1.1418,
      "step": 2650
    },
    {
      "epoch": 1.099013968775678,
      "grad_norm": 17.74847984313965,
      "learning_rate": 4.992690512538525e-05,
      "loss": 1.0342,
      "step": 2675
    },
    {
      "epoch": 1.1092851273623665,
      "grad_norm": 10.52570915222168,
      "learning_rate": 4.991909700961337e-05,
      "loss": 1.1285,
      "step": 2700
    },
    {
      "epoch": 1.119556285949055,
      "grad_norm": 5.708019733428955,
      "learning_rate": 4.991089341985079e-05,
      "loss": 1.0229,
      "step": 2725
    },
    {
      "epoch": 1.1298274445357437,
      "grad_norm": 21.305837631225586,
      "learning_rate": 4.9902294486291104e-05,
      "loss": 1.0598,
      "step": 2750
    },
    {
      "epoch": 1.1400986031224323,
      "grad_norm": 23.91814422607422,
      "learning_rate": 4.9893300345402105e-05,
      "loss": 1.1705,
      "step": 2775
    },
    {
      "epoch": 1.1503697617091209,
      "grad_norm": 8.894378662109375,
      "learning_rate": 4.988391113992368e-05,
      "loss": 1.0257,
      "step": 2800
    },
    {
      "epoch": 1.1606409202958095,
      "grad_norm": 12.799375534057617,
      "learning_rate": 4.987412701886549e-05,
      "loss": 1.1463,
      "step": 2825
    },
    {
      "epoch": 1.170912078882498,
      "grad_norm": 12.658759117126465,
      "learning_rate": 4.9863948137504666e-05,
      "loss": 1.1462,
      "step": 2850
    },
    {
      "epoch": 1.1811832374691864,
      "grad_norm": 6.2793707847595215,
      "learning_rate": 4.9853374657383275e-05,
      "loss": 1.0928,
      "step": 2875
    },
    {
      "epoch": 1.191454396055875,
      "grad_norm": 10.470609664916992,
      "learning_rate": 4.9842406746305814e-05,
      "loss": 1.0454,
      "step": 2900
    },
    {
      "epoch": 1.2017255546425636,
      "grad_norm": 46.87442398071289,
      "learning_rate": 4.9831044578336524e-05,
      "loss": 1.1075,
      "step": 2925
    },
    {
      "epoch": 1.2119967132292522,
      "grad_norm": 9.962231636047363,
      "learning_rate": 4.981928833379661e-05,
      "loss": 1.1,
      "step": 2950
    },
    {
      "epoch": 1.2222678718159408,
      "grad_norm": 4.477356433868408,
      "learning_rate": 4.9807138199261414e-05,
      "loss": 1.0811,
      "step": 2975
    },
    {
      "epoch": 1.2325390304026294,
      "grad_norm": 11.651799201965332,
      "learning_rate": 4.9794594367557435e-05,
      "loss": 1.024,
      "step": 3000
    },
    {
      "epoch": 1.242810188989318,
      "grad_norm": 7.755206108093262,
      "learning_rate": 4.978165703775927e-05,
      "loss": 1.067,
      "step": 3025
    },
    {
      "epoch": 1.2530813475760065,
      "grad_norm": 15.455122947692871,
      "learning_rate": 4.9768326415186454e-05,
      "loss": 1.0311,
      "step": 3050
    },
    {
      "epoch": 1.2633525061626951,
      "grad_norm": 9.131962776184082,
      "learning_rate": 4.975460271140021e-05,
      "loss": 1.0401,
      "step": 3075
    },
    {
      "epoch": 1.2736236647493837,
      "grad_norm": 29.119171142578125,
      "learning_rate": 4.974048614420007e-05,
      "loss": 1.0969,
      "step": 3100
    },
    {
      "epoch": 1.2838948233360723,
      "grad_norm": 6.163785457611084,
      "learning_rate": 4.972597693762045e-05,
      "loss": 1.1278,
      "step": 3125
    },
    {
      "epoch": 1.2941659819227609,
      "grad_norm": 10.012632369995117,
      "learning_rate": 4.971107532192707e-05,
      "loss": 1.0993,
      "step": 3150
    },
    {
      "epoch": 1.3044371405094495,
      "grad_norm": 6.5291666984558105,
      "learning_rate": 4.96957815336133e-05,
      "loss": 1.0934,
      "step": 3175
    },
    {
      "epoch": 1.314708299096138,
      "grad_norm": 18.37449836730957,
      "learning_rate": 4.9680095815396435e-05,
      "loss": 1.104,
      "step": 3200
    },
    {
      "epoch": 1.3249794576828267,
      "grad_norm": 8.442584991455078,
      "learning_rate": 4.9664018416213796e-05,
      "loss": 1.0918,
      "step": 3225
    },
    {
      "epoch": 1.3352506162695152,
      "grad_norm": 5.685943603515625,
      "learning_rate": 4.9647549591218835e-05,
      "loss": 1.0535,
      "step": 3250
    },
    {
      "epoch": 1.3455217748562038,
      "grad_norm": 5.7307448387146,
      "learning_rate": 4.963068960177701e-05,
      "loss": 1.1604,
      "step": 3275
    },
    {
      "epoch": 1.3557929334428924,
      "grad_norm": 4.045636177062988,
      "learning_rate": 4.961343871546175e-05,
      "loss": 1.0484,
      "step": 3300
    },
    {
      "epoch": 1.366064092029581,
      "grad_norm": 6.996286869049072,
      "learning_rate": 4.959579720605009e-05,
      "loss": 1.1055,
      "step": 3325
    },
    {
      "epoch": 1.3763352506162696,
      "grad_norm": 10.471301078796387,
      "learning_rate": 4.957776535351838e-05,
      "loss": 1.119,
      "step": 3350
    },
    {
      "epoch": 1.3866064092029582,
      "grad_norm": 6.319746017456055,
      "learning_rate": 4.955934344403786e-05,
      "loss": 1.127,
      "step": 3375
    },
    {
      "epoch": 1.3968775677896468,
      "grad_norm": 6.542300224304199,
      "learning_rate": 4.954053176997009e-05,
      "loss": 0.9722,
      "step": 3400
    },
    {
      "epoch": 1.4071487263763354,
      "grad_norm": 7.353511810302734,
      "learning_rate": 4.952133062986232e-05,
      "loss": 1.0545,
      "step": 3425
    },
    {
      "epoch": 1.417419884963024,
      "grad_norm": 6.178640842437744,
      "learning_rate": 4.950174032844274e-05,
      "loss": 1.022,
      "step": 3450
    },
    {
      "epoch": 1.4276910435497125,
      "grad_norm": 7.4956955909729,
      "learning_rate": 4.948176117661567e-05,
      "loss": 1.0411,
      "step": 3475
    },
    {
      "epoch": 1.437962202136401,
      "grad_norm": 16.34415626525879,
      "learning_rate": 4.9461393491456595e-05,
      "loss": 1.0635,
      "step": 3500
    },
    {
      "epoch": 1.4482333607230895,
      "grad_norm": 16.984901428222656,
      "learning_rate": 4.944063759620715e-05,
      "loss": 1.1581,
      "step": 3525
    },
    {
      "epoch": 1.458504519309778,
      "grad_norm": 17.63494300842285,
      "learning_rate": 4.941949382027e-05,
      "loss": 1.0962,
      "step": 3550
    },
    {
      "epoch": 1.4687756778964667,
      "grad_norm": 5.165820598602295,
      "learning_rate": 4.939796249920358e-05,
      "loss": 1.0976,
      "step": 3575
    },
    {
      "epoch": 1.4790468364831553,
      "grad_norm": 6.0279436111450195,
      "learning_rate": 4.93760439747168e-05,
      "loss": 1.0734,
      "step": 3600
    },
    {
      "epoch": 1.4893179950698439,
      "grad_norm": 4.7154717445373535,
      "learning_rate": 4.935373859466361e-05,
      "loss": 1.0507,
      "step": 3625
    },
    {
      "epoch": 1.4995891536565324,
      "grad_norm": 11.337488174438477,
      "learning_rate": 4.93310467130375e-05,
      "loss": 1.1746,
      "step": 3650
    },
    {
      "epoch": 1.509860312243221,
      "grad_norm": 18.193342208862305,
      "learning_rate": 4.930796868996582e-05,
      "loss": 1.0982,
      "step": 3675
    },
    {
      "epoch": 1.5201314708299096,
      "grad_norm": 7.017096996307373,
      "learning_rate": 4.928450489170415e-05,
      "loss": 1.1556,
      "step": 3700
    },
    {
      "epoch": 1.5304026294165982,
      "grad_norm": 2.5547895431518555,
      "learning_rate": 4.926065569063041e-05,
      "loss": 1.0685,
      "step": 3725
    },
    {
      "epoch": 1.5406737880032868,
      "grad_norm": 8.476696014404297,
      "learning_rate": 4.9236421465239025e-05,
      "loss": 1.0737,
      "step": 3750
    },
    {
      "epoch": 1.5509449465899754,
      "grad_norm": 8.107565879821777,
      "learning_rate": 4.921180260013483e-05,
      "loss": 1.086,
      "step": 3775
    },
    {
      "epoch": 1.561216105176664,
      "grad_norm": 8.92119312286377,
      "learning_rate": 4.9186799486027057e-05,
      "loss": 1.1056,
      "step": 3800
    },
    {
      "epoch": 1.5714872637633523,
      "grad_norm": 9.719097137451172,
      "learning_rate": 4.916141251972307e-05,
      "loss": 0.9651,
      "step": 3825
    },
    {
      "epoch": 1.581758422350041,
      "grad_norm": 16.80173110961914,
      "learning_rate": 4.9135642104122093e-05,
      "loss": 0.9173,
      "step": 3850
    },
    {
      "epoch": 1.5920295809367295,
      "grad_norm": 15.660393714904785,
      "learning_rate": 4.910948864820882e-05,
      "loss": 1.0374,
      "step": 3875
    },
    {
      "epoch": 1.6023007395234181,
      "grad_norm": 17.720876693725586,
      "learning_rate": 4.90829525670469e-05,
      "loss": 0.9644,
      "step": 3900
    },
    {
      "epoch": 1.6125718981101067,
      "grad_norm": 31.2305965423584,
      "learning_rate": 4.905603428177238e-05,
      "loss": 1.0033,
      "step": 3925
    },
    {
      "epoch": 1.6228430566967953,
      "grad_norm": 7.728428840637207,
      "learning_rate": 4.9028734219587e-05,
      "loss": 1.0823,
      "step": 3950
    },
    {
      "epoch": 1.6331142152834839,
      "grad_norm": 15.458768844604492,
      "learning_rate": 4.900105281375143e-05,
      "loss": 1.0676,
      "step": 3975
    },
    {
      "epoch": 1.6433853738701725,
      "grad_norm": 18.69976806640625,
      "learning_rate": 4.8972990503578386e-05,
      "loss": 1.0445,
      "step": 4000
    },
    {
      "epoch": 1.653656532456861,
      "grad_norm": 8.92398452758789,
      "learning_rate": 4.894454773442564e-05,
      "loss": 1.0722,
      "step": 4025
    },
    {
      "epoch": 1.6639276910435497,
      "grad_norm": 12.593587875366211,
      "learning_rate": 4.891572495768898e-05,
      "loss": 1.0598,
      "step": 4050
    },
    {
      "epoch": 1.6741988496302382,
      "grad_norm": 4.575959205627441,
      "learning_rate": 4.8886522630795045e-05,
      "loss": 1.0429,
      "step": 4075
    },
    {
      "epoch": 1.6844700082169268,
      "grad_norm": 14.307016372680664,
      "learning_rate": 4.8856941217194035e-05,
      "loss": 0.9699,
      "step": 4100
    },
    {
      "epoch": 1.6947411668036154,
      "grad_norm": 6.749273300170898,
      "learning_rate": 4.882698118635239e-05,
      "loss": 1.0919,
      "step": 4125
    },
    {
      "epoch": 1.705012325390304,
      "grad_norm": 9.383212089538574,
      "learning_rate": 4.879664301374532e-05,
      "loss": 1.1693,
      "step": 4150
    },
    {
      "epoch": 1.7152834839769926,
      "grad_norm": 4.1049909591674805,
      "learning_rate": 4.876592718084928e-05,
      "loss": 1.159,
      "step": 4175
    },
    {
      "epoch": 1.7255546425636812,
      "grad_norm": 8.32151985168457,
      "learning_rate": 4.873483417513429e-05,
      "loss": 1.0969,
      "step": 4200
    },
    {
      "epoch": 1.7358258011503698,
      "grad_norm": 6.129182815551758,
      "learning_rate": 4.8703364490056245e-05,
      "loss": 1.0632,
      "step": 4225
    },
    {
      "epoch": 1.7460969597370584,
      "grad_norm": 5.181979179382324,
      "learning_rate": 4.8671518625049036e-05,
      "loss": 0.9523,
      "step": 4250
    },
    {
      "epoch": 1.756368118323747,
      "grad_norm": 6.566084384918213,
      "learning_rate": 4.863929708551667e-05,
      "loss": 1.0336,
      "step": 4275
    },
    {
      "epoch": 1.7666392769104355,
      "grad_norm": 14.47562313079834,
      "learning_rate": 4.8606700382825224e-05,
      "loss": 1.0493,
      "step": 4300
    },
    {
      "epoch": 1.7769104354971241,
      "grad_norm": 3.870394706726074,
      "learning_rate": 4.8573729034294736e-05,
      "loss": 1.1012,
      "step": 4325
    },
    {
      "epoch": 1.7871815940838127,
      "grad_norm": 21.563486099243164,
      "learning_rate": 4.854038356319098e-05,
      "loss": 1.0629,
      "step": 4350
    },
    {
      "epoch": 1.7974527526705013,
      "grad_norm": 4.163707256317139,
      "learning_rate": 4.850666449871718e-05,
      "loss": 1.0252,
      "step": 4375
    },
    {
      "epoch": 1.80772391125719,
      "grad_norm": 7.478860378265381,
      "learning_rate": 4.8472572376005614e-05,
      "loss": 1.1695,
      "step": 4400
    },
    {
      "epoch": 1.8179950698438785,
      "grad_norm": 8.34506893157959,
      "learning_rate": 4.843810773610911e-05,
      "loss": 1.0471,
      "step": 4425
    },
    {
      "epoch": 1.828266228430567,
      "grad_norm": 7.001756191253662,
      "learning_rate": 4.840327112599244e-05,
      "loss": 1.0156,
      "step": 4450
    },
    {
      "epoch": 1.8385373870172557,
      "grad_norm": 4.876439094543457,
      "learning_rate": 4.836806309852371e-05,
      "loss": 1.022,
      "step": 4475
    },
    {
      "epoch": 1.8488085456039443,
      "grad_norm": 5.670014381408691,
      "learning_rate": 4.833248421246548e-05,
      "loss": 1.1103,
      "step": 4500
    },
    {
      "epoch": 1.8590797041906328,
      "grad_norm": 5.291884422302246,
      "learning_rate": 4.8296535032466e-05,
      "loss": 1.0795,
      "step": 4525
    },
    {
      "epoch": 1.8693508627773214,
      "grad_norm": 7.388947486877441,
      "learning_rate": 4.826021612905017e-05,
      "loss": 1.1053,
      "step": 4550
    },
    {
      "epoch": 1.87962202136401,
      "grad_norm": 5.867747783660889,
      "learning_rate": 4.822352807861054e-05,
      "loss": 1.0175,
      "step": 4575
    },
    {
      "epoch": 1.8898931799506984,
      "grad_norm": 11.747688293457031,
      "learning_rate": 4.818647146339813e-05,
      "loss": 1.0419,
      "step": 4600
    },
    {
      "epoch": 1.900164338537387,
      "grad_norm": 7.364708423614502,
      "learning_rate": 4.814904687151321e-05,
      "loss": 1.0909,
      "step": 4625
    },
    {
      "epoch": 1.9104354971240756,
      "grad_norm": 10.691391944885254,
      "learning_rate": 4.811125489689594e-05,
      "loss": 1.1139,
      "step": 4650
    },
    {
      "epoch": 1.9207066557107642,
      "grad_norm": 61.839820861816406,
      "learning_rate": 4.807309613931698e-05,
      "loss": 1.0478,
      "step": 4675
    },
    {
      "epoch": 1.9309778142974527,
      "grad_norm": 3.956354856491089,
      "learning_rate": 4.803457120436792e-05,
      "loss": 1.1058,
      "step": 4700
    },
    {
      "epoch": 1.9412489728841413,
      "grad_norm": 8.450248718261719,
      "learning_rate": 4.799568070345175e-05,
      "loss": 1.1035,
      "step": 4725
    },
    {
      "epoch": 1.95152013147083,
      "grad_norm": 9.974756240844727,
      "learning_rate": 4.795642525377307e-05,
      "loss": 1.0522,
      "step": 4750
    },
    {
      "epoch": 1.9617912900575185,
      "grad_norm": 23.98746109008789,
      "learning_rate": 4.791680547832835e-05,
      "loss": 1.0571,
      "step": 4775
    },
    {
      "epoch": 1.972062448644207,
      "grad_norm": 10.666019439697266,
      "learning_rate": 4.7876822005896016e-05,
      "loss": 1.1004,
      "step": 4800
    },
    {
      "epoch": 1.9823336072308957,
      "grad_norm": 8.37453842163086,
      "learning_rate": 4.7836475471026484e-05,
      "loss": 1.106,
      "step": 4825
    },
    {
      "epoch": 1.9926047658175843,
      "grad_norm": 6.877459526062012,
      "learning_rate": 4.779576651403209e-05,
      "loss": 1.0404,
      "step": 4850
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.6005342648720846,
      "eval_f1_macro": 0.34074175503832355,
      "eval_f1_micro": 0.6005342648720846,
      "eval_f1_weighted": 0.5316262488920555,
      "eval_loss": 1.064771056175232,
      "eval_precision_macro": 0.42047577765386696,
      "eval_precision_micro": 0.6005342648720846,
      "eval_precision_weighted": 0.5362395289892545,
      "eval_recall_macro": 0.32992208872704937,
      "eval_recall_micro": 0.6005342648720846,
      "eval_recall_weighted": 0.6005342648720846,
      "eval_runtime": 58.1251,
      "eval_samples_per_second": 167.449,
      "eval_steps_per_second": 5.247,
      "step": 4868
    }
  ],
  "logging_steps": 25,
  "max_steps": 21906,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.01
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.629350229586739e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}