{
  "best_metric": 0.8391068037456617,
  "best_model_checkpoint": "/home/ubuntu/utah/babylm-24/src/evaluation/results/finetune/DebertaV2-Base-10M_babylm-A/qqp/checkpoint-227405",
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 227405,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01099360172379675,
      "grad_norm": 4.236701011657715,
      "learning_rate": 2.993403838965722e-05,
      "loss": 0.6086,
      "step": 500
    },
    {
      "epoch": 0.0219872034475935,
      "grad_norm": 4.105135440826416,
      "learning_rate": 2.986807677931444e-05,
      "loss": 0.5618,
      "step": 1000
    },
    {
      "epoch": 0.03298080517139025,
      "grad_norm": 5.613928318023682,
      "learning_rate": 2.980211516897166e-05,
      "loss": 0.5252,
      "step": 1500
    },
    {
      "epoch": 0.043974406895187,
      "grad_norm": 12.754159927368164,
      "learning_rate": 2.973615355862888e-05,
      "loss": 0.5182,
      "step": 2000
    },
    {
      "epoch": 0.05496800861898375,
      "grad_norm": 5.488389015197754,
      "learning_rate": 2.9670191948286098e-05,
      "loss": 0.505,
      "step": 2500
    },
    {
      "epoch": 0.0659616103427805,
      "grad_norm": 5.46230936050415,
      "learning_rate": 2.9604230337943316e-05,
      "loss": 0.5044,
      "step": 3000
    },
    {
      "epoch": 0.07695521206657725,
      "grad_norm": 13.249972343444824,
      "learning_rate": 2.9538268727600537e-05,
      "loss": 0.498,
      "step": 3500
    },
    {
      "epoch": 0.087948813790374,
      "grad_norm": 13.043343544006348,
      "learning_rate": 2.9472307117257757e-05,
      "loss": 0.479,
      "step": 4000
    },
    {
      "epoch": 0.09894241551417075,
      "grad_norm": 14.147295951843262,
      "learning_rate": 2.9406345506914978e-05,
      "loss": 0.4692,
      "step": 4500
    },
    {
      "epoch": 0.1099360172379675,
      "grad_norm": 8.739255905151367,
      "learning_rate": 2.9340383896572196e-05,
      "loss": 0.4703,
      "step": 5000
    },
    {
      "epoch": 0.12092961896176425,
      "grad_norm": 11.371148109436035,
      "learning_rate": 2.9274422286229417e-05,
      "loss": 0.4557,
      "step": 5500
    },
    {
      "epoch": 0.131923220685561,
      "grad_norm": 8.44424057006836,
      "learning_rate": 2.9208460675886637e-05,
      "loss": 0.4635,
      "step": 6000
    },
    {
      "epoch": 0.14291682240935774,
      "grad_norm": 3.9929113388061523,
      "learning_rate": 2.914249906554385e-05,
      "loss": 0.4651,
      "step": 6500
    },
    {
      "epoch": 0.1539104241331545,
      "grad_norm": 3.4015634059906006,
      "learning_rate": 2.9076537455201072e-05,
      "loss": 0.4537,
      "step": 7000
    },
    {
      "epoch": 0.16490402585695124,
      "grad_norm": 4.32379674911499,
      "learning_rate": 2.9010575844858293e-05,
      "loss": 0.4445,
      "step": 7500
    },
    {
      "epoch": 0.175897627580748,
      "grad_norm": 5.210988521575928,
      "learning_rate": 2.8944614234515514e-05,
      "loss": 0.4326,
      "step": 8000
    },
    {
      "epoch": 0.18689122930454474,
      "grad_norm": 3.908278226852417,
      "learning_rate": 2.887865262417273e-05,
      "loss": 0.4454,
      "step": 8500
    },
    {
      "epoch": 0.1978848310283415,
      "grad_norm": 2.8256139755249023,
      "learning_rate": 2.8812691013829952e-05,
      "loss": 0.438,
      "step": 9000
    },
    {
      "epoch": 0.20887843275213824,
      "grad_norm": 7.012328624725342,
      "learning_rate": 2.8746729403487173e-05,
      "loss": 0.4333,
      "step": 9500
    },
    {
      "epoch": 0.219872034475935,
      "grad_norm": 5.030713081359863,
      "learning_rate": 2.8680767793144394e-05,
      "loss": 0.4588,
      "step": 10000
    },
    {
      "epoch": 0.23086563619973174,
      "grad_norm": 10.047335624694824,
      "learning_rate": 2.8614806182801608e-05,
      "loss": 0.4418,
      "step": 10500
    },
    {
      "epoch": 0.2418592379235285,
      "grad_norm": 5.562270164489746,
      "learning_rate": 2.854884457245883e-05,
      "loss": 0.4375,
      "step": 11000
    },
    {
      "epoch": 0.25285283964732524,
      "grad_norm": 14.433082580566406,
      "learning_rate": 2.848288296211605e-05,
      "loss": 0.4278,
      "step": 11500
    },
    {
      "epoch": 0.263846441371122,
      "grad_norm": 3.507910966873169,
      "learning_rate": 2.8416921351773267e-05,
      "loss": 0.4331,
      "step": 12000
    },
    {
      "epoch": 0.2748400430949188,
      "grad_norm": 13.184822082519531,
      "learning_rate": 2.8350959741430488e-05,
      "loss": 0.4325,
      "step": 12500
    },
    {
      "epoch": 0.2858336448187155,
      "grad_norm": 11.627528190612793,
      "learning_rate": 2.828499813108771e-05,
      "loss": 0.4283,
      "step": 13000
    },
    {
      "epoch": 0.29682724654251225,
      "grad_norm": 4.054474830627441,
      "learning_rate": 2.821903652074493e-05,
      "loss": 0.4361,
      "step": 13500
    },
    {
      "epoch": 0.307820848266309,
      "grad_norm": 7.193812847137451,
      "learning_rate": 2.8153074910402147e-05,
      "loss": 0.4251,
      "step": 14000
    },
    {
      "epoch": 0.3188144499901058,
      "grad_norm": 4.036972999572754,
      "learning_rate": 2.8087113300059365e-05,
      "loss": 0.4233,
      "step": 14500
    },
    {
      "epoch": 0.3298080517139025,
      "grad_norm": 6.451129913330078,
      "learning_rate": 2.8021151689716586e-05,
      "loss": 0.4407,
      "step": 15000
    },
    {
      "epoch": 0.34080165343769925,
      "grad_norm": 6.5472612380981445,
      "learning_rate": 2.7955190079373803e-05,
      "loss": 0.4158,
      "step": 15500
    },
    {
      "epoch": 0.351795255161496,
      "grad_norm": 16.589092254638672,
      "learning_rate": 2.7889228469031024e-05,
      "loss": 0.4261,
      "step": 16000
    },
    {
      "epoch": 0.3627888568852928,
      "grad_norm": 6.696326732635498,
      "learning_rate": 2.7823266858688245e-05,
      "loss": 0.4111,
      "step": 16500
    },
    {
      "epoch": 0.3737824586090895,
      "grad_norm": 4.396112442016602,
      "learning_rate": 2.7757305248345466e-05,
      "loss": 0.4236,
      "step": 17000
    },
    {
      "epoch": 0.38477606033288625,
      "grad_norm": 7.1260986328125,
      "learning_rate": 2.7691343638002683e-05,
      "loss": 0.413,
      "step": 17500
    },
    {
      "epoch": 0.395769662056683,
      "grad_norm": 8.553855895996094,
      "learning_rate": 2.7625382027659904e-05,
      "loss": 0.4105,
      "step": 18000
    },
    {
      "epoch": 0.4067632637804798,
      "grad_norm": 1.411335825920105,
      "learning_rate": 2.755942041731712e-05,
      "loss": 0.3977,
      "step": 18500
    },
    {
      "epoch": 0.4177568655042765,
      "grad_norm": 12.823638916015625,
      "learning_rate": 2.7493458806974342e-05,
      "loss": 0.4143,
      "step": 19000
    },
    {
      "epoch": 0.42875046722807325,
      "grad_norm": 11.823991775512695,
      "learning_rate": 2.742749719663156e-05,
      "loss": 0.4002,
      "step": 19500
    },
    {
      "epoch": 0.43974406895187,
      "grad_norm": 5.152065753936768,
      "learning_rate": 2.736153558628878e-05,
      "loss": 0.4001,
      "step": 20000
    },
    {
      "epoch": 0.4507376706756668,
      "grad_norm": 2.0797653198242188,
      "learning_rate": 2.7295573975946e-05,
      "loss": 0.4137,
      "step": 20500
    },
    {
      "epoch": 0.4617312723994635,
      "grad_norm": 5.874008655548096,
      "learning_rate": 2.722961236560322e-05,
      "loss": 0.4166,
      "step": 21000
    },
    {
      "epoch": 0.47272487412326025,
      "grad_norm": 20.658824920654297,
      "learning_rate": 2.716365075526044e-05,
      "loss": 0.4072,
      "step": 21500
    },
    {
      "epoch": 0.483718475847057,
      "grad_norm": 11.262660026550293,
      "learning_rate": 2.709768914491766e-05,
      "loss": 0.4048,
      "step": 22000
    },
    {
      "epoch": 0.4947120775708538,
      "grad_norm": 8.16556167602539,
      "learning_rate": 2.7031727534574878e-05,
      "loss": 0.4059,
      "step": 22500
    },
    {
      "epoch": 0.5057056792946505,
      "grad_norm": 12.176790237426758,
      "learning_rate": 2.6965765924232095e-05,
      "loss": 0.415,
      "step": 23000
    },
    {
      "epoch": 0.5166992810184473,
      "grad_norm": 8.630789756774902,
      "learning_rate": 2.6899804313889316e-05,
      "loss": 0.4059,
      "step": 23500
    },
    {
      "epoch": 0.527692882742244,
      "grad_norm": 11.014044761657715,
      "learning_rate": 2.6833842703546537e-05,
      "loss": 0.3804,
      "step": 24000
    },
    {
      "epoch": 0.5386864844660407,
      "grad_norm": 12.287202835083008,
      "learning_rate": 2.6767881093203758e-05,
      "loss": 0.4046,
      "step": 24500
    },
    {
      "epoch": 0.5496800861898375,
      "grad_norm": 6.118470668792725,
      "learning_rate": 2.6701919482860975e-05,
      "loss": 0.398,
      "step": 25000
    },
    {
      "epoch": 0.5606736879136343,
      "grad_norm": 17.233190536499023,
      "learning_rate": 2.6635957872518196e-05,
      "loss": 0.4013,
      "step": 25500
    },
    {
      "epoch": 0.571667289637431,
      "grad_norm": 15.902141571044922,
      "learning_rate": 2.6569996262175417e-05,
      "loss": 0.4158,
      "step": 26000
    },
    {
      "epoch": 0.5826608913612278,
      "grad_norm": 6.975952625274658,
      "learning_rate": 2.650403465183263e-05,
      "loss": 0.3939,
      "step": 26500
    },
    {
      "epoch": 0.5936544930850245,
      "grad_norm": 19.319835662841797,
      "learning_rate": 2.6438073041489852e-05,
      "loss": 0.4021,
      "step": 27000
    },
    {
      "epoch": 0.6046480948088213,
      "grad_norm": 3.9395010471343994,
      "learning_rate": 2.6372111431147073e-05,
      "loss": 0.4015,
      "step": 27500
    },
    {
      "epoch": 0.615641696532618,
      "grad_norm": 7.098001956939697,
      "learning_rate": 2.6306149820804294e-05,
      "loss": 0.3876,
      "step": 28000
    },
    {
      "epoch": 0.6266352982564147,
      "grad_norm": 3.967722177505493,
      "learning_rate": 2.624018821046151e-05,
      "loss": 0.3935,
      "step": 28500
    },
    {
      "epoch": 0.6376288999802115,
      "grad_norm": 5.257786273956299,
      "learning_rate": 2.6174226600118732e-05,
      "loss": 0.3959,
      "step": 29000
    },
    {
      "epoch": 0.6486225017040083,
      "grad_norm": 17.10645294189453,
      "learning_rate": 2.6108264989775953e-05,
      "loss": 0.3948,
      "step": 29500
    },
    {
      "epoch": 0.659616103427805,
      "grad_norm": 14.60950756072998,
      "learning_rate": 2.6042303379433174e-05,
      "loss": 0.4078,
      "step": 30000
    },
    {
      "epoch": 0.6706097051516018,
      "grad_norm": 6.0776166915893555,
      "learning_rate": 2.5976341769090388e-05,
      "loss": 0.3831,
      "step": 30500
    },
    {
      "epoch": 0.6816033068753985,
      "grad_norm": 1.830690622329712,
      "learning_rate": 2.591038015874761e-05,
      "loss": 0.3911,
      "step": 31000
    },
    {
      "epoch": 0.6925969085991953,
      "grad_norm": 16.717496871948242,
      "learning_rate": 2.584441854840483e-05,
      "loss": 0.3926,
      "step": 31500
    },
    {
      "epoch": 0.703590510322992,
      "grad_norm": 4.424517631530762,
      "learning_rate": 2.5778456938062047e-05,
      "loss": 0.3931,
      "step": 32000
    },
    {
      "epoch": 0.7145841120467887,
      "grad_norm": 6.995429992675781,
      "learning_rate": 2.5712495327719268e-05,
      "loss": 0.3912,
      "step": 32500
    },
    {
      "epoch": 0.7255777137705856,
      "grad_norm": 6.78953742980957,
      "learning_rate": 2.564653371737649e-05,
      "loss": 0.3858,
      "step": 33000
    },
    {
      "epoch": 0.7365713154943823,
      "grad_norm": 12.592966079711914,
      "learning_rate": 2.558057210703371e-05,
      "loss": 0.3822,
      "step": 33500
    },
    {
      "epoch": 0.747564917218179,
      "grad_norm": 1.8982641696929932,
      "learning_rate": 2.5514610496690927e-05,
      "loss": 0.3964,
      "step": 34000
    },
    {
      "epoch": 0.7585585189419758,
      "grad_norm": 3.2166171073913574,
      "learning_rate": 2.5448648886348144e-05,
      "loss": 0.3902,
      "step": 34500
    },
    {
      "epoch": 0.7695521206657725,
      "grad_norm": 10.258162498474121,
      "learning_rate": 2.5382687276005365e-05,
      "loss": 0.3691,
      "step": 35000
    },
    {
      "epoch": 0.7805457223895693,
      "grad_norm": 2.9836630821228027,
      "learning_rate": 2.5316725665662583e-05,
      "loss": 0.3937,
      "step": 35500
    },
    {
      "epoch": 0.791539324113366,
      "grad_norm": 10.085326194763184,
      "learning_rate": 2.5250764055319804e-05,
      "loss": 0.3897,
      "step": 36000
    },
    {
      "epoch": 0.8025329258371627,
      "grad_norm": 8.519903182983398,
      "learning_rate": 2.5184802444977024e-05,
      "loss": 0.361,
      "step": 36500
    },
    {
      "epoch": 0.8135265275609596,
      "grad_norm": 4.746450901031494,
      "learning_rate": 2.5118840834634245e-05,
      "loss": 0.3845,
      "step": 37000
    },
    {
      "epoch": 0.8245201292847563,
      "grad_norm": 13.058253288269043,
      "learning_rate": 2.5052879224291463e-05,
      "loss": 0.3871,
      "step": 37500
    },
    {
      "epoch": 0.835513731008553,
      "grad_norm": 2.8094441890716553,
      "learning_rate": 2.4986917613948684e-05,
      "loss": 0.4012,
      "step": 38000
    },
    {
      "epoch": 0.8465073327323498,
      "grad_norm": 4.1435770988464355,
      "learning_rate": 2.49209560036059e-05,
      "loss": 0.3802,
      "step": 38500
    },
    {
      "epoch": 0.8575009344561465,
      "grad_norm": 12.436211585998535,
      "learning_rate": 2.4854994393263122e-05,
      "loss": 0.3705,
      "step": 39000
    },
    {
      "epoch": 0.8684945361799433,
      "grad_norm": 16.29452133178711,
      "learning_rate": 2.478903278292034e-05,
      "loss": 0.3921,
      "step": 39500
    },
    {
      "epoch": 0.87948813790374,
      "grad_norm": 6.898037910461426,
      "learning_rate": 2.472307117257756e-05,
      "loss": 0.3925,
      "step": 40000
    },
    {
      "epoch": 0.8904817396275367,
      "grad_norm": 6.402541160583496,
      "learning_rate": 2.465710956223478e-05,
      "loss": 0.3771,
      "step": 40500
    },
    {
      "epoch": 0.9014753413513336,
      "grad_norm": 3.24283504486084,
      "learning_rate": 2.4591147951892e-05,
      "loss": 0.3698,
      "step": 41000
    },
    {
      "epoch": 0.9124689430751303,
      "grad_norm": 7.773282527923584,
      "learning_rate": 2.452518634154922e-05,
      "loss": 0.3823,
      "step": 41500
    },
    {
      "epoch": 0.923462544798927,
      "grad_norm": 4.645416736602783,
      "learning_rate": 2.445922473120644e-05,
      "loss": 0.3792,
      "step": 42000
    },
    {
      "epoch": 0.9344561465227238,
      "grad_norm": 8.859955787658691,
      "learning_rate": 2.4393263120863658e-05,
      "loss": 0.3988,
      "step": 42500
    },
    {
      "epoch": 0.9454497482465205,
      "grad_norm": 5.384950160980225,
      "learning_rate": 2.4327301510520875e-05,
      "loss": 0.367,
      "step": 43000
    },
    {
      "epoch": 0.9564433499703173,
      "grad_norm": 21.994779586791992,
      "learning_rate": 2.4261339900178096e-05,
      "loss": 0.3768,
      "step": 43500
    },
    {
      "epoch": 0.967436951694114,
      "grad_norm": 9.221137046813965,
      "learning_rate": 2.4195378289835317e-05,
      "loss": 0.3767,
      "step": 44000
    },
    {
      "epoch": 0.9784305534179107,
      "grad_norm": 1.9626529216766357,
      "learning_rate": 2.4129416679492538e-05,
      "loss": 0.3626,
      "step": 44500
    },
    {
      "epoch": 0.9894241551417076,
      "grad_norm": 19.04450798034668,
      "learning_rate": 2.4063455069149755e-05,
      "loss": 0.3754,
      "step": 45000
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8456591367721558,
      "eval_f1": 0.7902943944078505,
      "eval_loss": 0.37383726239204407,
      "eval_mcc": 0.6683280718139261,
      "eval_runtime": 29.7825,
      "eval_samples_per_second": 678.754,
      "eval_steps_per_second": 84.848,
      "step": 45481
    },
    {
      "epoch": 1.0004177568655044,
      "grad_norm": 14.85083293914795,
      "learning_rate": 2.3997493458806976e-05,
      "loss": 0.3783,
      "step": 45500
    },
    {
      "epoch": 1.011411358589301,
      "grad_norm": 7.004974365234375,
      "learning_rate": 2.3931531848464197e-05,
      "loss": 0.3307,
      "step": 46000
    },
    {
      "epoch": 1.0224049603130978,
      "grad_norm": 1.900647759437561,
      "learning_rate": 2.386557023812141e-05,
      "loss": 0.3454,
      "step": 46500
    },
    {
      "epoch": 1.0333985620368946,
      "grad_norm": 13.517301559448242,
      "learning_rate": 2.379960862777863e-05,
      "loss": 0.3404,
      "step": 47000
    },
    {
      "epoch": 1.0443921637606912,
      "grad_norm": 11.086533546447754,
      "learning_rate": 2.3733647017435853e-05,
      "loss": 0.344,
      "step": 47500
    },
    {
      "epoch": 1.055385765484488,
      "grad_norm": 1.5423752069473267,
      "learning_rate": 2.3667685407093073e-05,
      "loss": 0.348,
      "step": 48000
    },
    {
      "epoch": 1.0663793672082849,
      "grad_norm": 13.40974235534668,
      "learning_rate": 2.360172379675029e-05,
      "loss": 0.3353,
      "step": 48500
    },
    {
      "epoch": 1.0773729689320815,
      "grad_norm": 8.961548805236816,
      "learning_rate": 2.353576218640751e-05,
      "loss": 0.3442,
      "step": 49000
    },
    {
      "epoch": 1.0883665706558783,
      "grad_norm": 6.879663944244385,
      "learning_rate": 2.3469800576064733e-05,
      "loss": 0.331,
      "step": 49500
    },
    {
      "epoch": 1.099360172379675,
      "grad_norm": 10.819347381591797,
      "learning_rate": 2.3403838965721953e-05,
      "loss": 0.3431,
      "step": 50000
    },
    {
      "epoch": 1.1103537741034717,
      "grad_norm": 18.258974075317383,
      "learning_rate": 2.3337877355379167e-05,
      "loss": 0.3468,
      "step": 50500
    },
    {
      "epoch": 1.1213473758272685,
      "grad_norm": 11.075167655944824,
      "learning_rate": 2.3271915745036388e-05,
      "loss": 0.3313,
      "step": 51000
    },
    {
      "epoch": 1.1323409775510653,
      "grad_norm": 13.255118370056152,
      "learning_rate": 2.320595413469361e-05,
      "loss": 0.3379,
      "step": 51500
    },
    {
      "epoch": 1.1433345792748622,
      "grad_norm": 9.165648460388184,
      "learning_rate": 2.3139992524350827e-05,
      "loss": 0.3402,
      "step": 52000
    },
    {
      "epoch": 1.1543281809986587,
      "grad_norm": 20.563486099243164,
      "learning_rate": 2.3074030914008047e-05,
      "loss": 0.3429,
      "step": 52500
    },
    {
      "epoch": 1.1653217827224556,
      "grad_norm": 23.879840850830078,
      "learning_rate": 2.3008069303665268e-05,
      "loss": 0.3437,
      "step": 53000
    },
    {
      "epoch": 1.1763153844462524,
      "grad_norm": 16.95256996154785,
      "learning_rate": 2.294210769332249e-05,
      "loss": 0.3484,
      "step": 53500
    },
    {
      "epoch": 1.187308986170049,
      "grad_norm": 23.673189163208008,
      "learning_rate": 2.2876146082979707e-05,
      "loss": 0.3175,
      "step": 54000
    },
    {
      "epoch": 1.1983025878938458,
      "grad_norm": 12.443720817565918,
      "learning_rate": 2.2810184472636927e-05,
      "loss": 0.3471,
      "step": 54500
    },
    {
      "epoch": 1.2092961896176426,
      "grad_norm": 6.558742046356201,
      "learning_rate": 2.2744222862294145e-05,
      "loss": 0.3463,
      "step": 55000
    },
    {
      "epoch": 1.2202897913414392,
      "grad_norm": 3.0714826583862305,
      "learning_rate": 2.2678261251951362e-05,
      "loss": 0.3352,
      "step": 55500
    },
    {
      "epoch": 1.231283393065236,
      "grad_norm": 6.919187068939209,
      "learning_rate": 2.2612299641608583e-05,
      "loss": 0.335,
      "step": 56000
    },
    {
      "epoch": 1.2422769947890329,
      "grad_norm": 8.951086044311523,
      "learning_rate": 2.2546338031265804e-05,
      "loss": 0.3437,
      "step": 56500
    },
    {
      "epoch": 1.2532705965128295,
      "grad_norm": 81.6339111328125,
      "learning_rate": 2.2480376420923025e-05,
      "loss": 0.318,
      "step": 57000
    },
    {
      "epoch": 1.2642641982366263,
      "grad_norm": 10.197173118591309,
      "learning_rate": 2.2414414810580242e-05,
      "loss": 0.3478,
      "step": 57500
    },
    {
      "epoch": 1.275257799960423,
      "grad_norm": 3.3102078437805176,
      "learning_rate": 2.2348453200237463e-05,
      "loss": 0.3316,
      "step": 58000
    },
    {
      "epoch": 1.2862514016842197,
      "grad_norm": 7.871964454650879,
      "learning_rate": 2.2282491589894684e-05,
      "loss": 0.3329,
      "step": 58500
    },
    {
      "epoch": 1.2972450034080165,
      "grad_norm": 13.741714477539062,
      "learning_rate": 2.22165299795519e-05,
      "loss": 0.3285,
      "step": 59000
    },
    {
      "epoch": 1.3082386051318133,
      "grad_norm": 5.765045166015625,
      "learning_rate": 2.215056836920912e-05,
      "loss": 0.3363,
      "step": 59500
    },
    {
      "epoch": 1.31923220685561,
      "grad_norm": 21.365049362182617,
      "learning_rate": 2.208460675886634e-05,
      "loss": 0.3475,
      "step": 60000
    },
    {
      "epoch": 1.3302258085794068,
      "grad_norm": 16.869543075561523,
      "learning_rate": 2.201864514852356e-05,
      "loss": 0.3252,
      "step": 60500
    },
    {
      "epoch": 1.3412194103032036,
      "grad_norm": 31.089399337768555,
      "learning_rate": 2.1952683538180778e-05,
      "loss": 0.3344,
      "step": 61000
    },
    {
      "epoch": 1.3522130120270002,
      "grad_norm": 11.333529472351074,
      "learning_rate": 2.1886721927838e-05,
      "loss": 0.3641,
      "step": 61500
    },
    {
      "epoch": 1.363206613750797,
      "grad_norm": 9.713915824890137,
      "learning_rate": 2.182076031749522e-05,
      "loss": 0.3415,
      "step": 62000
    },
    {
      "epoch": 1.3742002154745938,
      "grad_norm": 8.068568229675293,
      "learning_rate": 2.175479870715244e-05,
      "loss": 0.3591,
      "step": 62500
    },
    {
      "epoch": 1.3851938171983904,
      "grad_norm": 19.801572799682617,
      "learning_rate": 2.1688837096809655e-05,
      "loss": 0.335,
      "step": 63000
    },
    {
      "epoch": 1.3961874189221872,
      "grad_norm": 28.160655975341797,
      "learning_rate": 2.1622875486466876e-05,
      "loss": 0.3374,
      "step": 63500
    },
    {
      "epoch": 1.407181020645984,
      "grad_norm": 2.872919797897339,
      "learning_rate": 2.1556913876124096e-05,
      "loss": 0.3403,
      "step": 64000
    },
    {
      "epoch": 1.4181746223697809,
      "grad_norm": 2.8728220462799072,
      "learning_rate": 2.1490952265781317e-05,
      "loss": 0.3384,
      "step": 64500
    },
    {
      "epoch": 1.4291682240935775,
      "grad_norm": 5.093236923217773,
      "learning_rate": 2.1424990655438535e-05,
      "loss": 0.3305,
      "step": 65000
    },
    {
      "epoch": 1.4401618258173743,
      "grad_norm": 17.081308364868164,
      "learning_rate": 2.1359029045095756e-05,
      "loss": 0.3589,
      "step": 65500
    },
    {
      "epoch": 1.451155427541171,
      "grad_norm": 2.3248064517974854,
      "learning_rate": 2.1293067434752976e-05,
      "loss": 0.3313,
      "step": 66000
    },
    {
      "epoch": 1.462149029264968,
      "grad_norm": 24.70163917541504,
      "learning_rate": 2.1227105824410194e-05,
      "loss": 0.3338,
      "step": 66500
    },
    {
      "epoch": 1.4731426309887645,
      "grad_norm": 19.80680274963379,
      "learning_rate": 2.116114421406741e-05,
      "loss": 0.3487,
      "step": 67000
    },
    {
      "epoch": 1.4841362327125613,
      "grad_norm": 2.365659713745117,
      "learning_rate": 2.1095182603724632e-05,
      "loss": 0.3479,
      "step": 67500
    },
    {
      "epoch": 1.4951298344363582,
      "grad_norm": 2.9082655906677246,
      "learning_rate": 2.1029220993381853e-05,
      "loss": 0.3427,
      "step": 68000
    },
    {
      "epoch": 1.5061234361601548,
      "grad_norm": 1.5247036218643188,
      "learning_rate": 2.096325938303907e-05,
      "loss": 0.321,
      "step": 68500
    },
    {
      "epoch": 1.5171170378839516,
      "grad_norm": 13.082464218139648,
      "learning_rate": 2.089729777269629e-05,
      "loss": 0.347,
      "step": 69000
    },
    {
      "epoch": 1.5281106396077484,
      "grad_norm": 32.83438491821289,
      "learning_rate": 2.0831336162353512e-05,
      "loss": 0.3382,
      "step": 69500
    },
    {
      "epoch": 1.539104241331545,
      "grad_norm": 19.514705657958984,
      "learning_rate": 2.0765374552010733e-05,
      "loss": 0.334,
      "step": 70000
    },
    {
      "epoch": 1.5500978430553418,
      "grad_norm": 11.129077911376953,
      "learning_rate": 2.069941294166795e-05,
      "loss": 0.3376,
      "step": 70500
    },
    {
      "epoch": 1.5610914447791386,
      "grad_norm": 14.449658393859863,
      "learning_rate": 2.0633451331325168e-05,
      "loss": 0.3297,
      "step": 71000
    },
    {
      "epoch": 1.5720850465029352,
      "grad_norm": 4.129580974578857,
      "learning_rate": 2.056748972098239e-05,
      "loss": 0.3405,
      "step": 71500
    },
    {
      "epoch": 1.583078648226732,
      "grad_norm": 4.104194164276123,
      "learning_rate": 2.0501528110639606e-05,
      "loss": 0.3327,
      "step": 72000
    },
    {
      "epoch": 1.5940722499505289,
      "grad_norm": 12.376803398132324,
      "learning_rate": 2.0435566500296827e-05,
      "loss": 0.3178,
      "step": 72500
    },
    {
      "epoch": 1.6050658516743255,
      "grad_norm": 18.45488739013672,
      "learning_rate": 2.0369604889954048e-05,
      "loss": 0.3457,
      "step": 73000
    },
    {
      "epoch": 1.6160594533981223,
      "grad_norm": 6.8812174797058105,
      "learning_rate": 2.030364327961127e-05,
      "loss": 0.3452,
      "step": 73500
    },
    {
      "epoch": 1.6270530551219191,
      "grad_norm": 3.4659981727600098,
      "learning_rate": 2.0237681669268486e-05,
      "loss": 0.3407,
      "step": 74000
    },
    {
      "epoch": 1.6380466568457157,
      "grad_norm": 21.697237014770508,
      "learning_rate": 2.0171720058925707e-05,
      "loss": 0.3493,
      "step": 74500
    },
    {
      "epoch": 1.6490402585695125,
      "grad_norm": 20.997262954711914,
      "learning_rate": 2.0105758448582925e-05,
      "loss": 0.3453,
      "step": 75000
    },
    {
      "epoch": 1.6600338602933093,
      "grad_norm": 8.582404136657715,
      "learning_rate": 2.0039796838240142e-05,
      "loss": 0.344,
      "step": 75500
    },
    {
      "epoch": 1.671027462017106,
      "grad_norm": 15.028887748718262,
      "learning_rate": 1.9973835227897363e-05,
      "loss": 0.3428,
      "step": 76000
    },
    {
      "epoch": 1.6820210637409028,
      "grad_norm": 15.200948715209961,
      "learning_rate": 1.9907873617554584e-05,
      "loss": 0.3531,
      "step": 76500
    },
    {
      "epoch": 1.6930146654646996,
      "grad_norm": 12.243021011352539,
      "learning_rate": 1.9841912007211805e-05,
      "loss": 0.3288,
      "step": 77000
    },
    {
      "epoch": 1.7040082671884962,
      "grad_norm": 0.7417749166488647,
      "learning_rate": 1.9775950396869022e-05,
      "loss": 0.3305,
      "step": 77500
    },
    {
      "epoch": 1.715001868912293,
      "grad_norm": 12.099386215209961,
      "learning_rate": 1.9709988786526243e-05,
      "loss": 0.3521,
      "step": 78000
    },
    {
      "epoch": 1.7259954706360898,
      "grad_norm": 10.566434860229492,
      "learning_rate": 1.9644027176183464e-05,
      "loss": 0.3466,
      "step": 78500
    },
    {
      "epoch": 1.7369890723598864,
      "grad_norm": 1.6488581895828247,
      "learning_rate": 1.957806556584068e-05,
      "loss": 0.3453,
      "step": 79000
    },
    {
      "epoch": 1.7479826740836832,
      "grad_norm": 15.446043968200684,
      "learning_rate": 1.95121039554979e-05,
      "loss": 0.3547,
      "step": 79500
    },
    {
      "epoch": 1.75897627580748,
      "grad_norm": 3.6907153129577637,
      "learning_rate": 1.944614234515512e-05,
      "loss": 0.3346,
      "step": 80000
    },
    {
      "epoch": 1.7699698775312767,
      "grad_norm": 13.95593547821045,
      "learning_rate": 1.938018073481234e-05,
      "loss": 0.3325,
      "step": 80500
    },
    {
      "epoch": 1.7809634792550737,
      "grad_norm": 7.613198757171631,
      "learning_rate": 1.9314219124469558e-05,
      "loss": 0.3302,
      "step": 81000
    },
    {
      "epoch": 1.7919570809788703,
      "grad_norm": 17.56180191040039,
      "learning_rate": 1.924825751412678e-05,
      "loss": 0.3303,
      "step": 81500
    },
    {
      "epoch": 1.802950682702667,
      "grad_norm": 30.033525466918945,
      "learning_rate": 1.9182295903784e-05,
      "loss": 0.327,
      "step": 82000
    },
    {
      "epoch": 1.813944284426464,
      "grad_norm": 2.658094644546509,
      "learning_rate": 1.911633429344122e-05,
      "loss": 0.3462,
      "step": 82500
    },
    {
      "epoch": 1.8249378861502605,
      "grad_norm": 8.311567306518555,
      "learning_rate": 1.9050372683098434e-05,
      "loss": 0.3347,
      "step": 83000
    },
    {
      "epoch": 1.8359314878740571,
      "grad_norm": 17.150461196899414,
      "learning_rate": 1.8984411072755655e-05,
      "loss": 0.3282,
      "step": 83500
    },
    {
      "epoch": 1.8469250895978542,
      "grad_norm": 10.157220840454102,
      "learning_rate": 1.8918449462412876e-05,
      "loss": 0.3193,
      "step": 84000
    },
    {
      "epoch": 1.8579186913216508,
      "grad_norm": 4.80257511138916,
      "learning_rate": 1.8852487852070097e-05,
      "loss": 0.3541,
      "step": 84500
    },
    {
      "epoch": 1.8689122930454476,
      "grad_norm": 3.8798446655273438,
      "learning_rate": 1.8786526241727314e-05,
      "loss": 0.3483,
      "step": 85000
    },
    {
      "epoch": 1.8799058947692444,
      "grad_norm": 8.881115913391113,
      "learning_rate": 1.8720564631384535e-05,
      "loss": 0.3302,
      "step": 85500
    },
    {
      "epoch": 1.890899496493041,
      "grad_norm": 0.945717453956604,
      "learning_rate": 1.8654603021041756e-05,
      "loss": 0.336,
      "step": 86000
    },
    {
      "epoch": 1.9018930982168378,
      "grad_norm": 5.144163608551025,
      "learning_rate": 1.8588641410698974e-05,
      "loss": 0.3148,
      "step": 86500
    },
    {
      "epoch": 1.9128866999406346,
      "grad_norm": 1.5305918455123901,
      "learning_rate": 1.852267980035619e-05,
      "loss": 0.3395,
      "step": 87000
    },
    {
      "epoch": 1.9238803016644312,
      "grad_norm": 15.06664752960205,
      "learning_rate": 1.8456718190013412e-05,
      "loss": 0.341,
      "step": 87500
    },
    {
      "epoch": 1.934873903388228,
      "grad_norm": 33.20983123779297,
      "learning_rate": 1.8390756579670633e-05,
      "loss": 0.3619,
      "step": 88000
    },
    {
      "epoch": 1.9458675051120249,
      "grad_norm": 11.427024841308594,
      "learning_rate": 1.832479496932785e-05,
      "loss": 0.3474,
      "step": 88500
    },
    {
      "epoch": 1.9568611068358215,
      "grad_norm": 23.793506622314453,
      "learning_rate": 1.825883335898507e-05,
      "loss": 0.321,
      "step": 89000
    },
    {
      "epoch": 1.9678547085596183,
      "grad_norm": 3.9075679779052734,
      "learning_rate": 1.8192871748642292e-05,
      "loss": 0.33,
      "step": 89500
    },
    {
      "epoch": 1.9788483102834151,
      "grad_norm": 1.7106132507324219,
      "learning_rate": 1.8126910138299513e-05,
      "loss": 0.3562,
      "step": 90000
    },
    {
      "epoch": 1.9898419120072117,
      "grad_norm": 1.6460707187652588,
      "learning_rate": 1.806094852795673e-05,
      "loss": 0.3378,
      "step": 90500
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8638139963150024,
      "eval_f1": 0.8192264758027448,
      "eval_loss": 0.3989393711090088,
      "eval_mcc": 0.7101352338901957,
      "eval_runtime": 29.2332,
      "eval_samples_per_second": 691.508,
      "eval_steps_per_second": 86.443,
      "step": 90962
    },
    {
      "epoch": 2.0008355137310088,
      "grad_norm": 0.9355267882347107,
      "learning_rate": 1.7994986917613948e-05,
      "loss": 0.3212,
      "step": 91000
    },
    {
      "epoch": 2.0118291154548054,
      "grad_norm": 11.889479637145996,
      "learning_rate": 1.792902530727117e-05,
      "loss": 0.3003,
      "step": 91500
    },
    {
      "epoch": 2.022822717178602,
      "grad_norm": 5.802761077880859,
      "learning_rate": 1.7863063696928386e-05,
      "loss": 0.2923,
      "step": 92000
    },
    {
      "epoch": 2.033816318902399,
      "grad_norm": 7.432724475860596,
      "learning_rate": 1.7797102086585607e-05,
      "loss": 0.293,
      "step": 92500
    },
    {
      "epoch": 2.0448099206261956,
      "grad_norm": 4.792222499847412,
      "learning_rate": 1.7731140476242828e-05,
      "loss": 0.3043,
      "step": 93000
    },
    {
      "epoch": 2.055803522349992,
      "grad_norm": 8.417468070983887,
      "learning_rate": 1.766517886590005e-05,
      "loss": 0.2916,
      "step": 93500
    },
    {
      "epoch": 2.0667971240737892,
      "grad_norm": 1.0828003883361816,
      "learning_rate": 1.7599217255557266e-05,
      "loss": 0.3112,
      "step": 94000
    },
    {
      "epoch": 2.077790725797586,
      "grad_norm": 3.0800647735595703,
      "learning_rate": 1.7533255645214487e-05,
      "loss": 0.2804,
      "step": 94500
    },
    {
      "epoch": 2.0887843275213824,
      "grad_norm": 3.254809617996216,
      "learning_rate": 1.7467294034871704e-05,
      "loss": 0.2909,
      "step": 95000
    },
    {
      "epoch": 2.0997779292451795,
      "grad_norm": 2.6388871669769287,
      "learning_rate": 1.7401332424528925e-05,
      "loss": 0.305,
      "step": 95500
    },
    {
      "epoch": 2.110771530968976,
      "grad_norm": 16.700735092163086,
      "learning_rate": 1.7335370814186143e-05,
      "loss": 0.2901,
      "step": 96000
    },
    {
      "epoch": 2.1217651326927727,
      "grad_norm": 17.04293441772461,
      "learning_rate": 1.7269409203843363e-05,
      "loss": 0.3217,
      "step": 96500
    },
    {
      "epoch": 2.1327587344165697,
      "grad_norm": 1.1329630613327026,
      "learning_rate": 1.7203447593500584e-05,
      "loss": 0.2898,
      "step": 97000
    },
    {
      "epoch": 2.1437523361403663,
      "grad_norm": 36.23415756225586,
      "learning_rate": 1.7137485983157802e-05,
      "loss": 0.2864,
      "step": 97500
    },
    {
      "epoch": 2.154745937864163,
      "grad_norm": 106.94963836669922,
      "learning_rate": 1.7071524372815023e-05,
      "loss": 0.3083,
      "step": 98000
    },
    {
      "epoch": 2.16573953958796,
      "grad_norm": 0.37686920166015625,
      "learning_rate": 1.7005562762472243e-05,
      "loss": 0.3238,
      "step": 98500
    },
    {
      "epoch": 2.1767331413117565,
      "grad_norm": 36.30667495727539,
      "learning_rate": 1.693960115212946e-05,
      "loss": 0.2917,
      "step": 99000
    },
    {
      "epoch": 2.187726743035553,
      "grad_norm": 35.45988082885742,
      "learning_rate": 1.687363954178668e-05,
      "loss": 0.3009,
      "step": 99500
    },
    {
      "epoch": 2.19872034475935,
      "grad_norm": 0.25279441475868225,
      "learning_rate": 1.68076779314439e-05,
      "loss": 0.3158,
      "step": 100000
    },
    {
      "epoch": 2.209713946483147,
      "grad_norm": 10.676984786987305,
      "learning_rate": 1.674171632110112e-05,
      "loss": 0.3005,
      "step": 100500
    },
    {
      "epoch": 2.2207075482069434,
      "grad_norm": 19.04880714416504,
      "learning_rate": 1.6675754710758337e-05,
      "loss": 0.3018,
      "step": 101000
    },
    {
      "epoch": 2.2317011499307404,
      "grad_norm": 35.431583404541016,
      "learning_rate": 1.660979310041556e-05,
      "loss": 0.3135,
      "step": 101500
    },
    {
      "epoch": 2.242694751654537,
      "grad_norm": 0.3378468155860901,
      "learning_rate": 1.654383149007278e-05,
      "loss": 0.2779,
      "step": 102000
    },
    {
      "epoch": 2.2536883533783336,
      "grad_norm": 46.81476974487305,
      "learning_rate": 1.647786987973e-05,
      "loss": 0.279,
      "step": 102500
    },
    {
      "epoch": 2.2646819551021307,
      "grad_norm": 36.49277877807617,
      "learning_rate": 1.6411908269387214e-05,
      "loss": 0.3184,
      "step": 103000
    },
    {
      "epoch": 2.2756755568259273,
      "grad_norm": 12.877152442932129,
      "learning_rate": 1.6345946659044435e-05,
      "loss": 0.3024,
      "step": 103500
    },
    {
      "epoch": 2.2866691585497243,
      "grad_norm": 4.798713684082031,
      "learning_rate": 1.6279985048701656e-05,
      "loss": 0.3014,
      "step": 104000
    },
    {
      "epoch": 2.297662760273521,
      "grad_norm": 6.63606071472168,
      "learning_rate": 1.6214023438358877e-05,
      "loss": 0.3214,
      "step": 104500
    },
    {
      "epoch": 2.3086563619973175,
      "grad_norm": 13.403897285461426,
      "learning_rate": 1.6148061828016094e-05,
      "loss": 0.2943,
      "step": 105000
    },
    {
      "epoch": 2.3196499637211145,
      "grad_norm": 33.9350471496582,
      "learning_rate": 1.6082100217673315e-05,
      "loss": 0.302,
      "step": 105500
    },
    {
      "epoch": 2.330643565444911,
      "grad_norm": 3.330829620361328,
      "learning_rate": 1.6016138607330536e-05,
      "loss": 0.3087,
      "step": 106000
    },
    {
      "epoch": 2.3416371671687077,
      "grad_norm": 1.2686516046524048,
      "learning_rate": 1.5950176996987753e-05,
      "loss": 0.3007,
      "step": 106500
    },
    {
      "epoch": 2.3526307688925048,
      "grad_norm": 20.976926803588867,
      "learning_rate": 1.5884215386644974e-05,
      "loss": 0.3094,
      "step": 107000
    },
    {
      "epoch": 2.3636243706163014,
      "grad_norm": 0.7142143249511719,
      "learning_rate": 1.581825377630219e-05,
      "loss": 0.3169,
      "step": 107500
    },
    {
      "epoch": 2.374617972340098,
      "grad_norm": 6.738494873046875,
      "learning_rate": 1.5752292165959412e-05,
      "loss": 0.3101,
      "step": 108000
    },
    {
      "epoch": 2.385611574063895,
      "grad_norm": 0.8053629398345947,
      "learning_rate": 1.568633055561663e-05,
      "loss": 0.3208,
      "step": 108500
    },
    {
      "epoch": 2.3966051757876916,
      "grad_norm": 0.35285481810569763,
      "learning_rate": 1.562036894527385e-05,
      "loss": 0.2903,
      "step": 109000
    },
    {
      "epoch": 2.407598777511488,
      "grad_norm": 0.9598795771598816,
      "learning_rate": 1.555440733493107e-05,
      "loss": 0.3189,
      "step": 109500
    },
    {
      "epoch": 2.4185923792352853,
      "grad_norm": 8.283425331115723,
      "learning_rate": 1.5488445724588292e-05,
      "loss": 0.2922,
      "step": 110000
    },
    {
      "epoch": 2.429585980959082,
      "grad_norm": 2.2365481853485107,
      "learning_rate": 1.542248411424551e-05,
      "loss": 0.2865,
      "step": 110500
    },
    {
      "epoch": 2.4405795826828784,
      "grad_norm": 22.584705352783203,
      "learning_rate": 1.535652250390273e-05,
      "loss": 0.2883,
      "step": 111000
    },
    {
      "epoch": 2.4515731844066755,
      "grad_norm": 1.3138020038604736,
      "learning_rate": 1.5290560893559948e-05,
      "loss": 0.3233,
      "step": 111500
    },
    {
      "epoch": 2.462566786130472,
      "grad_norm": 17.076557159423828,
      "learning_rate": 1.5224599283217167e-05,
      "loss": 0.3017,
      "step": 112000
    },
    {
      "epoch": 2.4735603878542687,
      "grad_norm": 139.9231719970703,
      "learning_rate": 1.5158637672874386e-05,
      "loss": 0.3033,
      "step": 112500
    },
    {
      "epoch": 2.4845539895780657,
      "grad_norm": 8.334077835083008,
      "learning_rate": 1.5092676062531607e-05,
      "loss": 0.2925,
      "step": 113000
    },
    {
      "epoch": 2.4955475913018623,
      "grad_norm": 0.4488193094730377,
      "learning_rate": 1.5026714452188828e-05,
      "loss": 0.314,
      "step": 113500
    },
    {
      "epoch": 2.506541193025659,
      "grad_norm": 18.986644744873047,
      "learning_rate": 1.4960752841846047e-05,
      "loss": 0.3011,
      "step": 114000
    },
    {
      "epoch": 2.517534794749456,
      "grad_norm": 0.16863927245140076,
      "learning_rate": 1.4894791231503265e-05,
      "loss": 0.2845,
      "step": 114500
    },
    {
      "epoch": 2.5285283964732526,
      "grad_norm": 19.12157440185547,
      "learning_rate": 1.4828829621160486e-05,
      "loss": 0.3092,
      "step": 115000
    },
    {
      "epoch": 2.539521998197049,
      "grad_norm": 6.872998237609863,
      "learning_rate": 1.4762868010817705e-05,
      "loss": 0.3072,
      "step": 115500
    },
    {
      "epoch": 2.550515599920846,
      "grad_norm": 0.4193851947784424,
      "learning_rate": 1.4696906400474924e-05,
      "loss": 0.3004,
      "step": 116000
    },
    {
      "epoch": 2.561509201644643,
      "grad_norm": 0.6917738318443298,
      "learning_rate": 1.4630944790132143e-05,
      "loss": 0.2967,
      "step": 116500
    },
    {
      "epoch": 2.5725028033684394,
      "grad_norm": 10.825478553771973,
      "learning_rate": 1.4564983179789362e-05,
      "loss": 0.3086,
      "step": 117000
    },
    {
      "epoch": 2.5834964050922364,
      "grad_norm": 68.07927703857422,
      "learning_rate": 1.4499021569446583e-05,
      "loss": 0.2964,
      "step": 117500
    },
    {
      "epoch": 2.594490006816033,
      "grad_norm": 5.563518047332764,
      "learning_rate": 1.4433059959103802e-05,
      "loss": 0.3067,
      "step": 118000
    },
    {
      "epoch": 2.6054836085398296,
      "grad_norm": 4.1622633934021,
      "learning_rate": 1.4367098348761021e-05,
      "loss": 0.3019,
      "step": 118500
    },
    {
      "epoch": 2.6164772102636267,
      "grad_norm": 20.468860626220703,
      "learning_rate": 1.430113673841824e-05,
      "loss": 0.2857,
      "step": 119000
    },
    {
      "epoch": 2.6274708119874233,
      "grad_norm": 19.43634605407715,
      "learning_rate": 1.4235175128075461e-05,
      "loss": 0.3079,
      "step": 119500
    },
    {
      "epoch": 2.63846441371122,
      "grad_norm": 6.021149158477783,
      "learning_rate": 1.416921351773268e-05,
      "loss": 0.2901,
      "step": 120000
    },
    {
      "epoch": 2.649458015435017,
      "grad_norm": 8.589285850524902,
      "learning_rate": 1.41032519073899e-05,
      "loss": 0.3109,
      "step": 120500
    },
    {
      "epoch": 2.6604516171588135,
      "grad_norm": 16.921823501586914,
      "learning_rate": 1.4037290297047119e-05,
      "loss": 0.3002,
      "step": 121000
    },
    {
      "epoch": 2.67144521888261,
      "grad_norm": 16.486186981201172,
      "learning_rate": 1.3971328686704338e-05,
      "loss": 0.306,
      "step": 121500
    },
    {
      "epoch": 2.682438820606407,
      "grad_norm": 8.290379524230957,
      "learning_rate": 1.3905367076361559e-05,
      "loss": 0.3005,
      "step": 122000
    },
    {
      "epoch": 2.6934324223302037,
      "grad_norm": 0.8587543964385986,
      "learning_rate": 1.3839405466018776e-05,
      "loss": 0.29,
      "step": 122500
    },
    {
      "epoch": 2.7044260240540003,
      "grad_norm": 45.68854904174805,
      "learning_rate": 1.3773443855675997e-05,
      "loss": 0.3037,
      "step": 123000
    },
    {
      "epoch": 2.7154196257777974,
      "grad_norm": 13.316100120544434,
      "learning_rate": 1.3707482245333216e-05,
      "loss": 0.2717,
      "step": 123500
    },
    {
      "epoch": 2.726413227501594,
      "grad_norm": 5.796350479125977,
      "learning_rate": 1.3641520634990437e-05,
      "loss": 0.3116,
      "step": 124000
    },
    {
      "epoch": 2.7374068292253906,
      "grad_norm": 10.975761413574219,
      "learning_rate": 1.3575559024647655e-05,
      "loss": 0.3089,
      "step": 124500
    },
    {
      "epoch": 2.7484004309491876,
      "grad_norm": 99.86316680908203,
      "learning_rate": 1.3509597414304875e-05,
      "loss": 0.3071,
      "step": 125000
    },
    {
      "epoch": 2.759394032672984,
      "grad_norm": 92.33716583251953,
      "learning_rate": 1.3443635803962095e-05,
      "loss": 0.2881,
      "step": 125500
    },
    {
      "epoch": 2.770387634396781,
      "grad_norm": 20.75370979309082,
      "learning_rate": 1.3377674193619314e-05,
      "loss": 0.2922,
      "step": 126000
    },
    {
      "epoch": 2.781381236120578,
      "grad_norm": 63.51997756958008,
      "learning_rate": 1.3311712583276533e-05,
      "loss": 0.3016,
      "step": 126500
    },
    {
      "epoch": 2.7923748378443745,
      "grad_norm": 12.819772720336914,
      "learning_rate": 1.3245750972933752e-05,
      "loss": 0.3005,
      "step": 127000
    },
    {
      "epoch": 2.803368439568171,
      "grad_norm": 14.167094230651855,
      "learning_rate": 1.3179789362590973e-05,
      "loss": 0.3099,
      "step": 127500
    },
    {
      "epoch": 2.814362041291968,
      "grad_norm": 14.828591346740723,
      "learning_rate": 1.3113827752248192e-05,
      "loss": 0.2759,
      "step": 128000
    },
    {
      "epoch": 2.8253556430157647,
      "grad_norm": 9.91226577758789,
      "learning_rate": 1.3047866141905411e-05,
      "loss": 0.3025,
      "step": 128500
    },
    {
      "epoch": 2.8363492447395617,
      "grad_norm": 38.544525146484375,
      "learning_rate": 1.298190453156263e-05,
      "loss": 0.3038,
      "step": 129000
    },
    {
      "epoch": 2.8473428464633583,
      "grad_norm": 5.008056640625,
      "learning_rate": 1.2915942921219851e-05,
      "loss": 0.2947,
      "step": 129500
    },
    {
      "epoch": 2.858336448187155,
      "grad_norm": 14.466870307922363,
      "learning_rate": 1.284998131087707e-05,
      "loss": 0.2989,
      "step": 130000
    },
    {
      "epoch": 2.869330049910952,
      "grad_norm": 0.3647148907184601,
      "learning_rate": 1.278401970053429e-05,
      "loss": 0.2864,
      "step": 130500
    },
    {
      "epoch": 2.8803236516347486,
      "grad_norm": 0.18057258427143097,
      "learning_rate": 1.2718058090191509e-05,
      "loss": 0.2894,
      "step": 131000
    },
    {
      "epoch": 2.891317253358545,
      "grad_norm": 0.5057438015937805,
      "learning_rate": 1.2652096479848728e-05,
      "loss": 0.302,
      "step": 131500
    },
    {
      "epoch": 2.902310855082342,
      "grad_norm": 10.934133529663086,
      "learning_rate": 1.2586134869505949e-05,
      "loss": 0.3047,
      "step": 132000
    },
    {
      "epoch": 2.913304456806139,
      "grad_norm": 2.3341269493103027,
      "learning_rate": 1.2520173259163166e-05,
      "loss": 0.3011,
      "step": 132500
    },
    {
      "epoch": 2.924298058529936,
      "grad_norm": 0.3195688724517822,
      "learning_rate": 1.2454211648820387e-05,
      "loss": 0.2939,
      "step": 133000
    },
    {
      "epoch": 2.9352916602537324,
      "grad_norm": 8.257743835449219,
      "learning_rate": 1.2388250038477606e-05,
      "loss": 0.3057,
      "step": 133500
    },
    {
      "epoch": 2.946285261977529,
      "grad_norm": 20.577478408813477,
      "learning_rate": 1.2322288428134827e-05,
      "loss": 0.2966,
      "step": 134000
    },
    {
      "epoch": 2.957278863701326,
      "grad_norm": 0.777562141418457,
      "learning_rate": 1.2256326817792044e-05,
      "loss": 0.3013,
      "step": 134500
    },
    {
      "epoch": 2.9682724654251227,
      "grad_norm": 58.6212158203125,
      "learning_rate": 1.2190365207449265e-05,
      "loss": 0.2874,
      "step": 135000
    },
    {
      "epoch": 2.9792660671489193,
      "grad_norm": 69.42217254638672,
      "learning_rate": 1.2124403597106484e-05,
      "loss": 0.293,
      "step": 135500
    },
    {
      "epoch": 2.9902596688727163,
      "grad_norm": 4.408263683319092,
      "learning_rate": 1.2058441986763704e-05,
      "loss": 0.3005,
      "step": 136000
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8718278408050537,
      "eval_f1": 0.8299757201916136,
      "eval_loss": 0.42042940855026245,
      "eval_mcc": 0.7272909835972381,
      "eval_runtime": 28.09,
      "eval_samples_per_second": 719.65,
      "eval_steps_per_second": 89.961,
      "step": 136443
    },
    {
      "epoch": 3.001253270596513,
      "grad_norm": 6.014369010925293,
      "learning_rate": 1.1992480376420923e-05,
      "loss": 0.281,
      "step": 136500
    },
    {
      "epoch": 3.0122468723203095,
      "grad_norm": 10.639359474182129,
      "learning_rate": 1.1926518766078142e-05,
      "loss": 0.2556,
      "step": 137000
    },
    {
      "epoch": 3.023240474044106,
      "grad_norm": 3.0724806785583496,
      "learning_rate": 1.1860557155735363e-05,
      "loss": 0.2414,
      "step": 137500
    },
    {
      "epoch": 3.034234075767903,
      "grad_norm": 0.3316449522972107,
      "learning_rate": 1.1794595545392582e-05,
      "loss": 0.2539,
      "step": 138000
    },
    {
      "epoch": 3.0452276774916998,
      "grad_norm": 23.327177047729492,
      "learning_rate": 1.1728633935049801e-05,
      "loss": 0.2374,
      "step": 138500
    },
    {
      "epoch": 3.056221279215497,
      "grad_norm": 0.7128089666366577,
      "learning_rate": 1.166267232470702e-05,
      "loss": 0.2476,
      "step": 139000
    },
    {
      "epoch": 3.0672148809392934,
      "grad_norm": 189.15638732910156,
      "learning_rate": 1.1596710714364241e-05,
      "loss": 0.2431,
      "step": 139500
    },
    {
      "epoch": 3.07820848266309,
      "grad_norm": 17.80859375,
      "learning_rate": 1.153074910402146e-05,
      "loss": 0.225,
      "step": 140000
    },
    {
      "epoch": 3.089202084386887,
      "grad_norm": 0.14886409044265747,
      "learning_rate": 1.146478749367868e-05,
      "loss": 0.2495,
      "step": 140500
    },
    {
      "epoch": 3.1001956861106836,
      "grad_norm": 0.5925188064575195,
      "learning_rate": 1.1398825883335899e-05,
      "loss": 0.2541,
      "step": 141000
    },
    {
      "epoch": 3.1111892878344802,
      "grad_norm": 9.857983589172363,
      "learning_rate": 1.1332864272993118e-05,
      "loss": 0.2565,
      "step": 141500
    },
    {
      "epoch": 3.1221828895582773,
      "grad_norm": 0.7951391339302063,
      "learning_rate": 1.1266902662650339e-05,
      "loss": 0.2561,
      "step": 142000
    },
    {
      "epoch": 3.133176491282074,
      "grad_norm": 61.58017349243164,
      "learning_rate": 1.1200941052307556e-05,
      "loss": 0.2361,
      "step": 142500
    },
    {
      "epoch": 3.1441700930058705,
      "grad_norm": 9.199590682983398,
      "learning_rate": 1.1134979441964777e-05,
      "loss": 0.2559,
      "step": 143000
    },
    {
      "epoch": 3.1551636947296675,
      "grad_norm": 1.7396503686904907,
      "learning_rate": 1.1069017831621996e-05,
      "loss": 0.2543,
      "step": 143500
    },
    {
      "epoch": 3.166157296453464,
      "grad_norm": 185.94760131835938,
      "learning_rate": 1.1003056221279217e-05,
      "loss": 0.2309,
      "step": 144000
    },
    {
      "epoch": 3.1771508981772607,
      "grad_norm": 42.58454132080078,
      "learning_rate": 1.0937094610936434e-05,
      "loss": 0.2634,
      "step": 144500
    },
    {
      "epoch": 3.1881444999010577,
      "grad_norm": 0.19487299025058746,
      "learning_rate": 1.0871133000593655e-05,
      "loss": 0.2668,
      "step": 145000
    },
    {
      "epoch": 3.1991381016248543,
      "grad_norm": 0.11774999648332596,
      "learning_rate": 1.0805171390250874e-05,
      "loss": 0.2531,
      "step": 145500
    },
    {
      "epoch": 3.210131703348651,
      "grad_norm": 0.8709030747413635,
      "learning_rate": 1.0739209779908093e-05,
      "loss": 0.2579,
      "step": 146000
    },
    {
      "epoch": 3.221125305072448,
      "grad_norm": 0.26620733737945557,
      "learning_rate": 1.0673248169565313e-05,
      "loss": 0.2456,
      "step": 146500
    },
    {
      "epoch": 3.2321189067962446,
      "grad_norm": 30.161376953125,
      "learning_rate": 1.0607286559222532e-05,
      "loss": 0.2588,
      "step": 147000
    },
    {
      "epoch": 3.243112508520041,
      "grad_norm": 14.72189998626709,
      "learning_rate": 1.0541324948879753e-05,
      "loss": 0.2538,
      "step": 147500
    },
    {
      "epoch": 3.2541061102438382,
      "grad_norm": 22.82953453063965,
      "learning_rate": 1.0475363338536972e-05,
      "loss": 0.2398,
      "step": 148000
    },
    {
      "epoch": 3.265099711967635,
      "grad_norm": 1.3340407609939575,
      "learning_rate": 1.0409401728194191e-05,
      "loss": 0.2619,
      "step": 148500
    },
    {
      "epoch": 3.2760933136914314,
      "grad_norm": 4.700684070587158,
      "learning_rate": 1.034344011785141e-05,
      "loss": 0.2563,
      "step": 149000
    },
    {
      "epoch": 3.2870869154152285,
      "grad_norm": 49.13290786743164,
      "learning_rate": 1.0277478507508631e-05,
      "loss": 0.2756,
      "step": 149500
    },
    {
      "epoch": 3.298080517139025,
      "grad_norm": 0.5340966582298279,
      "learning_rate": 1.021151689716585e-05,
      "loss": 0.249,
      "step": 150000
    },
    {
      "epoch": 3.3090741188628217,
      "grad_norm": 156.9650115966797,
      "learning_rate": 1.014555528682307e-05,
      "loss": 0.237,
      "step": 150500
    },
    {
      "epoch": 3.3200677205866187,
      "grad_norm": 0.09667583554983139,
      "learning_rate": 1.0079593676480288e-05,
      "loss": 0.2621,
      "step": 151000
    },
    {
      "epoch": 3.3310613223104153,
      "grad_norm": 9.222663879394531,
      "learning_rate": 1.0013632066137508e-05,
      "loss": 0.2457,
      "step": 151500
    },
    {
      "epoch": 3.342054924034212,
      "grad_norm": 14.612710952758789,
      "learning_rate": 9.947670455794728e-06,
      "loss": 0.2555,
      "step": 152000
    },
    {
      "epoch": 3.353048525758009,
      "grad_norm": 50.92832565307617,
      "learning_rate": 9.881708845451946e-06,
      "loss": 0.2394,
      "step": 152500
    },
    {
      "epoch": 3.3640421274818055,
      "grad_norm": 1.5905165672302246,
      "learning_rate": 9.815747235109167e-06,
      "loss": 0.2408,
      "step": 153000
    },
    {
      "epoch": 3.375035729205602,
      "grad_norm": 34.99452209472656,
      "learning_rate": 9.749785624766386e-06,
      "loss": 0.2351,
      "step": 153500
    },
    {
      "epoch": 3.386029330929399,
      "grad_norm": 1.3218666315078735,
      "learning_rate": 9.683824014423607e-06,
      "loss": 0.2494,
      "step": 154000
    },
    {
      "epoch": 3.3970229326531958,
      "grad_norm": 19.163127899169922,
      "learning_rate": 9.617862404080824e-06,
      "loss": 0.2716,
      "step": 154500
    },
    {
      "epoch": 3.4080165343769924,
      "grad_norm": 0.3468831479549408,
      "learning_rate": 9.551900793738045e-06,
      "loss": 0.2617,
      "step": 155000
    },
    {
      "epoch": 3.4190101361007894,
      "grad_norm": 19.626012802124023,
      "learning_rate": 9.485939183395264e-06,
      "loss": 0.2651,
      "step": 155500
    },
    {
      "epoch": 3.430003737824586,
      "grad_norm": 17.755313873291016,
      "learning_rate": 9.419977573052483e-06,
      "loss": 0.2487,
      "step": 156000
    },
    {
      "epoch": 3.4409973395483826,
      "grad_norm": 14.6954984664917,
      "learning_rate": 9.354015962709702e-06,
      "loss": 0.2579,
      "step": 156500
    },
    {
      "epoch": 3.4519909412721796,
      "grad_norm": 24.834569931030273,
      "learning_rate": 9.288054352366922e-06,
      "loss": 0.2343,
      "step": 157000
    },
    {
      "epoch": 3.4629845429959762,
      "grad_norm": 11.748075485229492,
      "learning_rate": 9.222092742024142e-06,
      "loss": 0.2493,
      "step": 157500
    },
    {
      "epoch": 3.473978144719773,
      "grad_norm": 0.26250067353248596,
      "learning_rate": 9.156131131681362e-06,
      "loss": 0.2518,
      "step": 158000
    },
    {
      "epoch": 3.48497174644357,
      "grad_norm": 0.31238773465156555,
      "learning_rate": 9.09016952133858e-06,
      "loss": 0.2563,
      "step": 158500
    },
    {
      "epoch": 3.4959653481673665,
      "grad_norm": 0.38414067029953003,
      "learning_rate": 9.0242079109958e-06,
      "loss": 0.2605,
      "step": 159000
    },
    {
      "epoch": 3.506958949891163,
      "grad_norm": 0.26335904002189636,
      "learning_rate": 8.95824630065302e-06,
      "loss": 0.25,
      "step": 159500
    },
    {
      "epoch": 3.51795255161496,
      "grad_norm": 11.267284393310547,
      "learning_rate": 8.89228469031024e-06,
      "loss": 0.2446,
      "step": 160000
    },
    {
      "epoch": 3.5289461533387567,
      "grad_norm": 13.036714553833008,
      "learning_rate": 8.826323079967459e-06,
      "loss": 0.2341,
      "step": 160500
    },
    {
      "epoch": 3.5399397550625533,
      "grad_norm": 9.3615140914917,
      "learning_rate": 8.760361469624678e-06,
      "loss": 0.2632,
      "step": 161000
    },
    {
      "epoch": 3.5509333567863504,
      "grad_norm": 0.3549996018409729,
      "learning_rate": 8.694399859281897e-06,
      "loss": 0.2507,
      "step": 161500
    },
    {
      "epoch": 3.561926958510147,
      "grad_norm": 0.46619582176208496,
      "learning_rate": 8.628438248939118e-06,
      "loss": 0.2543,
      "step": 162000
    },
    {
      "epoch": 3.572920560233944,
      "grad_norm": 0.9738485217094421,
      "learning_rate": 8.562476638596337e-06,
      "loss": 0.2672,
      "step": 162500
    },
    {
      "epoch": 3.5839141619577406,
      "grad_norm": 12.682645797729492,
      "learning_rate": 8.496515028253557e-06,
      "loss": 0.2422,
      "step": 163000
    },
    {
      "epoch": 3.594907763681537,
      "grad_norm": 0.33584246039390564,
      "learning_rate": 8.430553417910776e-06,
      "loss": 0.2597,
      "step": 163500
    },
    {
      "epoch": 3.6059013654053342,
      "grad_norm": 8.06340217590332,
      "learning_rate": 8.364591807567997e-06,
      "loss": 0.271,
      "step": 164000
    },
    {
      "epoch": 3.616894967129131,
      "grad_norm": 0.4343748688697815,
      "learning_rate": 8.298630197225216e-06,
      "loss": 0.2686,
      "step": 164500
    },
    {
      "epoch": 3.6278885688529274,
      "grad_norm": 38.23839569091797,
      "learning_rate": 8.232668586882435e-06,
      "loss": 0.2605,
      "step": 165000
    },
    {
      "epoch": 3.6388821705767245,
      "grad_norm": 20.085224151611328,
      "learning_rate": 8.166706976539654e-06,
      "loss": 0.2449,
      "step": 165500
    },
    {
      "epoch": 3.649875772300521,
      "grad_norm": 0.3538534641265869,
      "learning_rate": 8.100745366196875e-06,
      "loss": 0.2436,
      "step": 166000
    },
    {
      "epoch": 3.660869374024318,
      "grad_norm": 0.3242553174495697,
      "learning_rate": 8.034783755854094e-06,
      "loss": 0.2468,
      "step": 166500
    },
    {
      "epoch": 3.6718629757481147,
      "grad_norm": 28.696617126464844,
      "learning_rate": 7.968822145511311e-06,
      "loss": 0.2621,
      "step": 167000
    },
    {
      "epoch": 3.6828565774719113,
      "grad_norm": 0.26111406087875366,
      "learning_rate": 7.902860535168532e-06,
      "loss": 0.2519,
      "step": 167500
    },
    {
      "epoch": 3.6938501791957083,
      "grad_norm": 0.24540553987026215,
      "learning_rate": 7.836898924825751e-06,
      "loss": 0.2515,
      "step": 168000
    },
    {
      "epoch": 3.704843780919505,
      "grad_norm": 0.4676073491573334,
      "learning_rate": 7.770937314482972e-06,
      "loss": 0.256,
      "step": 168500
    },
    {
      "epoch": 3.7158373826433015,
      "grad_norm": 87.25594329833984,
      "learning_rate": 7.70497570414019e-06,
      "loss": 0.2354,
      "step": 169000
    },
    {
      "epoch": 3.7268309843670986,
      "grad_norm": 1.1010403633117676,
      "learning_rate": 7.63901409379741e-06,
      "loss": 0.2394,
      "step": 169500
    },
    {
      "epoch": 3.737824586090895,
      "grad_norm": 0.1542312502861023,
      "learning_rate": 7.57305248345463e-06,
      "loss": 0.2491,
      "step": 170000
    },
    {
      "epoch": 3.748818187814692,
      "grad_norm": 2.4090046882629395,
      "learning_rate": 7.50709087311185e-06,
      "loss": 0.2337,
      "step": 170500
    },
    {
      "epoch": 3.759811789538489,
      "grad_norm": 6.501917362213135,
      "learning_rate": 7.441129262769069e-06,
      "loss": 0.2406,
      "step": 171000
    },
    {
      "epoch": 3.7708053912622854,
      "grad_norm": 19.246479034423828,
      "learning_rate": 7.375167652426288e-06,
      "loss": 0.2547,
      "step": 171500
    },
    {
      "epoch": 3.781798992986082,
      "grad_norm": 0.10231161117553711,
      "learning_rate": 7.309206042083508e-06,
      "loss": 0.2379,
      "step": 172000
    },
    {
      "epoch": 3.792792594709879,
      "grad_norm": 10.832609176635742,
      "learning_rate": 7.243244431740727e-06,
      "loss": 0.2711,
      "step": 172500
    },
    {
      "epoch": 3.8037861964336757,
      "grad_norm": 18.556346893310547,
      "learning_rate": 7.177282821397947e-06,
      "loss": 0.2265,
      "step": 173000
    },
    {
      "epoch": 3.8147797981574723,
      "grad_norm": 1.0246055126190186,
      "learning_rate": 7.111321211055166e-06,
      "loss": 0.2334,
      "step": 173500
    },
    {
      "epoch": 3.8257733998812693,
      "grad_norm": 1.1454087495803833,
      "learning_rate": 7.0453596007123855e-06,
      "loss": 0.2491,
      "step": 174000
    },
    {
      "epoch": 3.836767001605066,
      "grad_norm": 11.247049331665039,
      "learning_rate": 6.979397990369605e-06,
      "loss": 0.2632,
      "step": 174500
    },
    {
      "epoch": 3.8477606033288625,
      "grad_norm": 14.466601371765137,
      "learning_rate": 6.913436380026825e-06,
      "loss": 0.2639,
      "step": 175000
    },
    {
      "epoch": 3.8587542050526595,
      "grad_norm": 7.91213321685791,
      "learning_rate": 6.847474769684044e-06,
      "loss": 0.2527,
      "step": 175500
    },
    {
      "epoch": 3.869747806776456,
      "grad_norm": 26.528411865234375,
      "learning_rate": 6.781513159341264e-06,
      "loss": 0.2438,
      "step": 176000
    },
    {
      "epoch": 3.8807414085002527,
      "grad_norm": 0.7833952903747559,
      "learning_rate": 6.715551548998483e-06,
      "loss": 0.2585,
      "step": 176500
    },
    {
      "epoch": 3.8917350102240498,
      "grad_norm": 53.77830123901367,
      "learning_rate": 6.649589938655703e-06,
      "loss": 0.2615,
      "step": 177000
    },
    {
      "epoch": 3.9027286119478464,
      "grad_norm": 0.6139953136444092,
      "learning_rate": 6.583628328312922e-06,
      "loss": 0.2644,
      "step": 177500
    },
    {
      "epoch": 3.913722213671643,
      "grad_norm": 1.4486163854599,
      "learning_rate": 6.517666717970142e-06,
      "loss": 0.2302,
      "step": 178000
    },
    {
      "epoch": 3.92471581539544,
      "grad_norm": 5.287415504455566,
      "learning_rate": 6.451705107627361e-06,
      "loss": 0.2727,
      "step": 178500
    },
    {
      "epoch": 3.9357094171192366,
      "grad_norm": 26.611614227294922,
      "learning_rate": 6.3857434972845804e-06,
      "loss": 0.2514,
      "step": 179000
    },
    {
      "epoch": 3.946703018843033,
      "grad_norm": 14.361977577209473,
      "learning_rate": 6.3197818869418e-06,
      "loss": 0.2476,
      "step": 179500
    },
    {
      "epoch": 3.9576966205668302,
      "grad_norm": 1.3597434759140015,
      "learning_rate": 6.25382027659902e-06,
      "loss": 0.2406,
      "step": 180000
    },
    {
      "epoch": 3.968690222290627,
      "grad_norm": 10.218100547790527,
      "learning_rate": 6.187858666256239e-06,
      "loss": 0.254,
      "step": 180500
    },
    {
      "epoch": 3.9796838240144234,
      "grad_norm": 0.7666225433349609,
      "learning_rate": 6.121897055913459e-06,
      "loss": 0.2741,
      "step": 181000
    },
    {
      "epoch": 3.9906774257382205,
      "grad_norm": 36.5604133605957,
      "learning_rate": 6.055935445570678e-06,
      "loss": 0.2374,
      "step": 181500
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8765273094177246,
      "eval_f1": 0.8365850464842216,
      "eval_loss": 0.5333936214447021,
      "eval_mcc": 0.7376033359055921,
      "eval_runtime": 27.772,
      "eval_samples_per_second": 727.891,
      "eval_steps_per_second": 90.991,
      "step": 181924
    },
    {
      "epoch": 4.0016710274620175,
      "grad_norm": 0.4658304750919342,
      "learning_rate": 5.989973835227898e-06,
      "loss": 0.2447,
      "step": 182000
    },
    {
      "epoch": 4.012664629185814,
      "grad_norm": 0.2597205340862274,
      "learning_rate": 5.924012224885117e-06,
      "loss": 0.1824,
      "step": 182500
    },
    {
      "epoch": 4.023658230909611,
      "grad_norm": 0.4755733907222748,
      "learning_rate": 5.858050614542337e-06,
      "loss": 0.2266,
      "step": 183000
    },
    {
      "epoch": 4.034651832633408,
      "grad_norm": 0.3996201753616333,
      "learning_rate": 5.792089004199556e-06,
      "loss": 0.1901,
      "step": 183500
    },
    {
      "epoch": 4.045645434357204,
      "grad_norm": 0.8636412620544434,
      "learning_rate": 5.726127393856775e-06,
      "loss": 0.1955,
      "step": 184000
    },
    {
      "epoch": 4.056639036081001,
      "grad_norm": 0.3265284597873688,
      "learning_rate": 5.6601657835139945e-06,
      "loss": 0.1979,
      "step": 184500
    },
    {
      "epoch": 4.067632637804798,
      "grad_norm": 0.1188616007566452,
      "learning_rate": 5.5942041731712145e-06,
      "loss": 0.2066,
      "step": 185000
    },
    {
      "epoch": 4.078626239528594,
      "grad_norm": 0.7546807527542114,
      "learning_rate": 5.528242562828434e-06,
      "loss": 0.1959,
      "step": 185500
    },
    {
      "epoch": 4.089619841252391,
      "grad_norm": 0.12139397114515305,
      "learning_rate": 5.462280952485654e-06,
      "loss": 0.1969,
      "step": 186000
    },
    {
      "epoch": 4.100613442976188,
      "grad_norm": 0.13494807481765747,
      "learning_rate": 5.396319342142873e-06,
      "loss": 0.1814,
      "step": 186500
    },
    {
      "epoch": 4.111607044699984,
      "grad_norm": 0.023194080218672752,
      "learning_rate": 5.330357731800093e-06,
      "loss": 0.1877,
      "step": 187000
    },
    {
      "epoch": 4.122600646423781,
      "grad_norm": 0.07898598164319992,
      "learning_rate": 5.264396121457312e-06,
      "loss": 0.209,
      "step": 187500
    },
    {
      "epoch": 4.1335942481475785,
      "grad_norm": 0.11429109424352646,
      "learning_rate": 5.198434511114532e-06,
      "loss": 0.1794,
      "step": 188000
    },
    {
      "epoch": 4.144587849871375,
      "grad_norm": 0.1126711368560791,
      "learning_rate": 5.132472900771751e-06,
      "loss": 0.2143,
      "step": 188500
    },
    {
      "epoch": 4.155581451595172,
      "grad_norm": 36.669212341308594,
      "learning_rate": 5.06651129042897e-06,
      "loss": 0.1995,
      "step": 189000
    },
    {
      "epoch": 4.166575053318969,
      "grad_norm": 0.0557066835463047,
      "learning_rate": 5.0005496800861894e-06,
      "loss": 0.1977,
      "step": 189500
    },
    {
      "epoch": 4.177568655042765,
      "grad_norm": 0.11854979395866394,
      "learning_rate": 4.9345880697434094e-06,
      "loss": 0.203,
      "step": 190000
    },
    {
      "epoch": 4.188562256766562,
      "grad_norm": 47.56736755371094,
      "learning_rate": 4.868626459400629e-06,
      "loss": 0.1673,
      "step": 190500
    },
    {
      "epoch": 4.199555858490359,
      "grad_norm": 1.5440220832824707,
      "learning_rate": 4.802664849057849e-06,
      "loss": 0.1998,
      "step": 191000
    },
    {
      "epoch": 4.210549460214155,
      "grad_norm": 0.11512400209903717,
      "learning_rate": 4.736703238715068e-06,
      "loss": 0.2027,
      "step": 191500
    },
    {
      "epoch": 4.221543061937952,
      "grad_norm": 13.000309944152832,
      "learning_rate": 4.670741628372288e-06,
      "loss": 0.2268,
      "step": 192000
    },
    {
      "epoch": 4.232536663661749,
      "grad_norm": 0.053511910140514374,
      "learning_rate": 4.604780018029507e-06,
      "loss": 0.2119,
      "step": 192500
    },
    {
      "epoch": 4.243530265385545,
      "grad_norm": 1.0577130317687988,
      "learning_rate": 4.538818407686727e-06,
      "loss": 0.207,
      "step": 193000
    },
    {
      "epoch": 4.254523867109342,
      "grad_norm": 0.5129163861274719,
      "learning_rate": 4.472856797343946e-06,
      "loss": 0.1971,
      "step": 193500
    },
    {
      "epoch": 4.265517468833139,
      "grad_norm": 38.92678451538086,
      "learning_rate": 4.406895187001165e-06,
      "loss": 0.1915,
      "step": 194000
    },
    {
      "epoch": 4.276511070556936,
      "grad_norm": 0.2508489489555359,
      "learning_rate": 4.340933576658384e-06,
      "loss": 0.2076,
      "step": 194500
    },
    {
      "epoch": 4.287504672280733,
      "grad_norm": 0.8289797306060791,
      "learning_rate": 4.274971966315604e-06,
      "loss": 0.1906,
      "step": 195000
    },
    {
      "epoch": 4.29849827400453,
      "grad_norm": 0.1511843502521515,
      "learning_rate": 4.2090103559728235e-06,
      "loss": 0.1908,
      "step": 195500
    },
    {
      "epoch": 4.309491875728326,
      "grad_norm": 36.47195053100586,
      "learning_rate": 4.1430487456300435e-06,
      "loss": 0.2198,
      "step": 196000
    },
    {
      "epoch": 4.320485477452123,
      "grad_norm": 59.39978790283203,
      "learning_rate": 4.077087135287263e-06,
      "loss": 0.1958,
      "step": 196500
    },
    {
      "epoch": 4.33147907917592,
      "grad_norm": 25.194355010986328,
      "learning_rate": 4.011125524944483e-06,
      "loss": 0.185,
      "step": 197000
    },
    {
      "epoch": 4.342472680899716,
      "grad_norm": 20.661163330078125,
      "learning_rate": 3.945163914601702e-06,
      "loss": 0.2032,
      "step": 197500
    },
    {
      "epoch": 4.353466282623513,
      "grad_norm": 0.04815911129117012,
      "learning_rate": 3.879202304258922e-06,
      "loss": 0.194,
      "step": 198000
    },
    {
      "epoch": 4.36445988434731,
      "grad_norm": 0.18730510771274567,
      "learning_rate": 3.8132406939161414e-06,
      "loss": 0.236,
      "step": 198500
    },
    {
      "epoch": 4.375453486071106,
      "grad_norm": 15.972749710083008,
      "learning_rate": 3.7472790835733606e-06,
      "loss": 0.196,
      "step": 199000
    },
    {
      "epoch": 4.386447087794903,
      "grad_norm": 0.25309285521507263,
      "learning_rate": 3.68131747323058e-06,
      "loss": 0.2161,
      "step": 199500
    },
    {
      "epoch": 4.3974406895187,
      "grad_norm": 0.17074181139469147,
      "learning_rate": 3.6153558628877997e-06,
      "loss": 0.1836,
      "step": 200000
    },
    {
      "epoch": 4.4084342912424965,
      "grad_norm": 20.413162231445312,
      "learning_rate": 3.549394252545019e-06,
      "loss": 0.1911,
      "step": 200500
    },
    {
      "epoch": 4.419427892966294,
      "grad_norm": 0.5931562781333923,
      "learning_rate": 3.4834326422022384e-06,
      "loss": 0.1847,
      "step": 201000
    },
    {
      "epoch": 4.430421494690091,
      "grad_norm": 37.10576248168945,
      "learning_rate": 3.417471031859458e-06,
      "loss": 0.1657,
      "step": 201500
    },
    {
      "epoch": 4.441415096413887,
      "grad_norm": 0.06108024716377258,
      "learning_rate": 3.3515094215166776e-06,
      "loss": 0.1949,
      "step": 202000
    },
    {
      "epoch": 4.452408698137684,
      "grad_norm": 14.7476224899292,
      "learning_rate": 3.285547811173897e-06,
      "loss": 0.2052,
      "step": 202500
    },
    {
      "epoch": 4.463402299861481,
      "grad_norm": 0.4280465841293335,
      "learning_rate": 3.2195862008311163e-06,
      "loss": 0.2127,
      "step": 203000
    },
    {
      "epoch": 4.474395901585277,
      "grad_norm": 0.38156208395957947,
      "learning_rate": 3.153624590488336e-06,
      "loss": 0.1949,
      "step": 203500
    },
    {
      "epoch": 4.485389503309074,
      "grad_norm": 325.33026123046875,
      "learning_rate": 3.0876629801455555e-06,
      "loss": 0.1976,
      "step": 204000
    },
    {
      "epoch": 4.496383105032871,
      "grad_norm": 99.75337982177734,
      "learning_rate": 3.021701369802775e-06,
      "loss": 0.2031,
      "step": 204500
    },
    {
      "epoch": 4.507376706756667,
      "grad_norm": 0.17061945796012878,
      "learning_rate": 2.9557397594599946e-06,
      "loss": 0.1964,
      "step": 205000
    },
    {
      "epoch": 4.518370308480464,
      "grad_norm": 25.07261085510254,
      "learning_rate": 2.8897781491172138e-06,
      "loss": 0.1875,
      "step": 205500
    },
    {
      "epoch": 4.529363910204261,
      "grad_norm": 0.2692670226097107,
      "learning_rate": 2.8238165387744334e-06,
      "loss": 0.1978,
      "step": 206000
    },
    {
      "epoch": 4.5403575119280575,
      "grad_norm": 2.56193470954895,
      "learning_rate": 2.757854928431653e-06,
      "loss": 0.207,
      "step": 206500
    },
    {
      "epoch": 4.5513511136518545,
      "grad_norm": 1.1347905397415161,
      "learning_rate": 2.6918933180888725e-06,
      "loss": 0.2049,
      "step": 207000
    },
    {
      "epoch": 4.562344715375652,
      "grad_norm": 0.9405034184455872,
      "learning_rate": 2.625931707746092e-06,
      "loss": 0.182,
      "step": 207500
    },
    {
      "epoch": 4.573338317099449,
      "grad_norm": 0.10386385023593903,
      "learning_rate": 2.5599700974033112e-06,
      "loss": 0.1816,
      "step": 208000
    },
    {
      "epoch": 4.584331918823245,
      "grad_norm": 1.0305184125900269,
      "learning_rate": 2.494008487060531e-06,
      "loss": 0.208,
      "step": 208500
    },
    {
      "epoch": 4.595325520547042,
      "grad_norm": 5.062295913696289,
      "learning_rate": 2.4280468767177504e-06,
      "loss": 0.1918,
      "step": 209000
    },
    {
      "epoch": 4.606319122270838,
      "grad_norm": 13.542932510375977,
      "learning_rate": 2.36208526637497e-06,
      "loss": 0.2016,
      "step": 209500
    },
    {
      "epoch": 4.617312723994635,
      "grad_norm": 28.13912582397461,
      "learning_rate": 2.2961236560321896e-06,
      "loss": 0.2065,
      "step": 210000
    },
    {
      "epoch": 4.628306325718432,
      "grad_norm": 38.89891052246094,
      "learning_rate": 2.2301620456894087e-06,
      "loss": 0.1951,
      "step": 210500
    },
    {
      "epoch": 4.639299927442229,
      "grad_norm": 220.25010681152344,
      "learning_rate": 2.1642004353466283e-06,
      "loss": 0.2243,
      "step": 211000
    },
    {
      "epoch": 4.650293529166025,
      "grad_norm": 0.11063925921916962,
      "learning_rate": 2.098238825003848e-06,
      "loss": 0.1862,
      "step": 211500
    },
    {
      "epoch": 4.661287130889822,
      "grad_norm": 0.9656747579574585,
      "learning_rate": 2.0322772146610674e-06,
      "loss": 0.1796,
      "step": 212000
    },
    {
      "epoch": 4.672280732613618,
      "grad_norm": 0.03588191047310829,
      "learning_rate": 1.966315604318287e-06,
      "loss": 0.1936,
      "step": 212500
    },
    {
      "epoch": 4.6832743343374155,
      "grad_norm": 25.791149139404297,
      "learning_rate": 1.9003539939755062e-06,
      "loss": 0.2102,
      "step": 213000
    },
    {
      "epoch": 4.6942679360612125,
      "grad_norm": 1.5398284196853638,
      "learning_rate": 1.8343923836327257e-06,
      "loss": 0.1941,
      "step": 213500
    },
    {
      "epoch": 4.7052615377850096,
      "grad_norm": 0.8514572978019714,
      "learning_rate": 1.7684307732899453e-06,
      "loss": 0.2124,
      "step": 214000
    },
    {
      "epoch": 4.716255139508806,
      "grad_norm": 0.1109534353017807,
      "learning_rate": 1.7024691629471647e-06,
      "loss": 0.1896,
      "step": 214500
    },
    {
      "epoch": 4.727248741232603,
      "grad_norm": 33.977500915527344,
      "learning_rate": 1.6365075526043843e-06,
      "loss": 0.1868,
      "step": 215000
    },
    {
      "epoch": 4.738242342956399,
      "grad_norm": 0.09221459925174713,
      "learning_rate": 1.5705459422616038e-06,
      "loss": 0.2147,
      "step": 215500
    },
    {
      "epoch": 4.749235944680196,
      "grad_norm": 0.13753363490104675,
      "learning_rate": 1.5045843319188232e-06,
      "loss": 0.1805,
      "step": 216000
    },
    {
      "epoch": 4.760229546403993,
      "grad_norm": 0.03300468996167183,
      "learning_rate": 1.4386227215760428e-06,
      "loss": 0.195,
      "step": 216500
    },
    {
      "epoch": 4.77122314812779,
      "grad_norm": 52.13509750366211,
      "learning_rate": 1.3726611112332624e-06,
      "loss": 0.1934,
      "step": 217000
    },
    {
      "epoch": 4.782216749851586,
      "grad_norm": 0.05201047658920288,
      "learning_rate": 1.3066995008904817e-06,
      "loss": 0.1986,
      "step": 217500
    },
    {
      "epoch": 4.793210351575383,
      "grad_norm": 0.15796062350273132,
      "learning_rate": 1.2407378905477013e-06,
      "loss": 0.189,
      "step": 218000
    },
    {
      "epoch": 4.80420395329918,
      "grad_norm": 0.603727400302887,
      "learning_rate": 1.1747762802049207e-06,
      "loss": 0.194,
      "step": 218500
    },
    {
      "epoch": 4.815197555022976,
      "grad_norm": 19.412994384765625,
      "learning_rate": 1.1088146698621402e-06,
      "loss": 0.2022,
      "step": 219000
    },
    {
      "epoch": 4.8261911567467735,
      "grad_norm": 0.08504907041788101,
      "learning_rate": 1.0428530595193598e-06,
      "loss": 0.1858,
      "step": 219500
    },
    {
      "epoch": 4.8371847584705705,
      "grad_norm": 0.07863516360521317,
      "learning_rate": 9.768914491765792e-07,
      "loss": 0.1942,
      "step": 220000
    },
    {
      "epoch": 4.848178360194367,
      "grad_norm": 23.51129722595215,
      "learning_rate": 9.109298388337987e-07,
      "loss": 0.1912,
      "step": 220500
    },
    {
      "epoch": 4.859171961918164,
      "grad_norm": 5.780854225158691,
      "learning_rate": 8.449682284910183e-07,
      "loss": 0.2185,
      "step": 221000
    },
    {
      "epoch": 4.870165563641961,
      "grad_norm": 0.05857408419251442,
      "learning_rate": 7.790066181482378e-07,
      "loss": 0.1914,
      "step": 221500
    },
    {
      "epoch": 4.881159165365757,
      "grad_norm": 0.18249481916427612,
      "learning_rate": 7.130450078054573e-07,
      "loss": 0.1738,
      "step": 222000
    },
    {
      "epoch": 4.892152767089554,
      "grad_norm": 418.6382751464844,
      "learning_rate": 6.470833974626767e-07,
      "loss": 0.2041,
      "step": 222500
    },
    {
      "epoch": 4.903146368813351,
      "grad_norm": 0.7230046987533569,
      "learning_rate": 5.811217871198962e-07,
      "loss": 0.1957,
      "step": 223000
    },
    {
      "epoch": 4.914139970537147,
      "grad_norm": 0.05270848050713539,
      "learning_rate": 5.151601767771158e-07,
      "loss": 0.1835,
      "step": 223500
    },
    {
      "epoch": 4.925133572260944,
      "grad_norm": 1709.6539306640625,
      "learning_rate": 4.491985664343352e-07,
      "loss": 0.1921,
      "step": 224000
    },
    {
      "epoch": 4.936127173984741,
      "grad_norm": 602.2431640625,
      "learning_rate": 3.8323695609155474e-07,
      "loss": 0.1727,
      "step": 224500
    },
    {
      "epoch": 4.947120775708537,
      "grad_norm": 0.297931432723999,
      "learning_rate": 3.1727534574877426e-07,
      "loss": 0.1915,
      "step": 225000
    },
    {
      "epoch": 4.958114377432334,
      "grad_norm": 0.3502364456653595,
      "learning_rate": 2.5131373540599373e-07,
      "loss": 0.1776,
      "step": 225500
    },
    {
      "epoch": 4.9691079791561314,
      "grad_norm": 0.023652415722608566,
      "learning_rate": 1.8535212506321323e-07,
      "loss": 0.1978,
      "step": 226000
    },
    {
      "epoch": 4.980101580879928,
      "grad_norm": 0.24965056777000427,
      "learning_rate": 1.1939051472043273e-07,
      "loss": 0.1924,
      "step": 226500
    },
    {
      "epoch": 4.991095182603725,
      "grad_norm": 0.07366069406270981,
      "learning_rate": 5.342890437765221e-08,
      "loss": 0.1959,
      "step": 227000
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8784565925598145,
      "eval_f1": 0.8391068037456617,
      "eval_loss": 0.6183628439903259,
      "eval_mcc": 0.7416813202825935,
      "eval_runtime": 27.9497,
      "eval_samples_per_second": 723.264,
      "eval_steps_per_second": 90.412,
      "step": 227405
    },
    {
      "epoch": 5.0,
      "step": 227405,
      "total_flos": 1.0733045580407808e+17,
      "train_loss": 0.30088049875882883,
      "train_runtime": 13193.5971,
      "train_samples_per_second": 137.887,
      "train_steps_per_second": 17.236
    }
  ],
  "logging_steps": 500,
  "max_steps": 227405,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 3,
        "early_stopping_threshold": 0.001
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.0733045580407808e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}