{
  "best_metric": 0.9832303618711385,
  "best_model_checkpoint": "wav2vec2-base-ft-keyword-spotting/checkpoint-1995",
  "epoch": 4.996869129618034,
  "eval_steps": 500,
  "global_step": 1995,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.025046963055729492,
      "grad_norm": 1.278334379196167,
      "learning_rate": 1.5e-06,
      "loss": 2.4699,
      "step": 10
    },
    {
      "epoch": 0.050093926111458985,
      "grad_norm": 1.3655153512954712,
      "learning_rate": 3e-06,
      "loss": 2.4539,
      "step": 20
    },
    {
      "epoch": 0.07514088916718847,
      "grad_norm": 1.3241969347000122,
      "learning_rate": 4.5e-06,
      "loss": 2.4044,
      "step": 30
    },
    {
      "epoch": 0.10018785222291797,
      "grad_norm": 1.7512290477752686,
      "learning_rate": 6e-06,
      "loss": 2.3261,
      "step": 40
    },
    {
      "epoch": 0.12523481527864747,
      "grad_norm": 2.1144192218780518,
      "learning_rate": 7.5e-06,
      "loss": 2.2156,
      "step": 50
    },
    {
      "epoch": 0.15028177833437695,
      "grad_norm": 2.416064739227295,
      "learning_rate": 9e-06,
      "loss": 2.0569,
      "step": 60
    },
    {
      "epoch": 0.17532874139010646,
      "grad_norm": 2.3332056999206543,
      "learning_rate": 1.05e-05,
      "loss": 1.8437,
      "step": 70
    },
    {
      "epoch": 0.20037570444583594,
      "grad_norm": 2.037410259246826,
      "learning_rate": 1.2e-05,
      "loss": 1.7693,
      "step": 80
    },
    {
      "epoch": 0.22542266750156542,
      "grad_norm": 2.00655198097229,
      "learning_rate": 1.3500000000000001e-05,
      "loss": 1.6565,
      "step": 90
    },
    {
      "epoch": 0.25046963055729493,
      "grad_norm": 1.5333060026168823,
      "learning_rate": 1.5e-05,
      "loss": 1.6115,
      "step": 100
    },
    {
      "epoch": 0.27551659361302444,
      "grad_norm": 0.9409968852996826,
      "learning_rate": 1.65e-05,
      "loss": 1.6253,
      "step": 110
    },
    {
      "epoch": 0.3005635566687539,
      "grad_norm": 1.4992973804473877,
      "learning_rate": 1.8e-05,
      "loss": 1.5118,
      "step": 120
    },
    {
      "epoch": 0.3256105197244834,
      "grad_norm": 0.5201308727264404,
      "learning_rate": 1.95e-05,
      "loss": 1.503,
      "step": 130
    },
    {
      "epoch": 0.3506574827802129,
      "grad_norm": 1.1067086458206177,
      "learning_rate": 2.1e-05,
      "loss": 1.5415,
      "step": 140
    },
    {
      "epoch": 0.37570444583594237,
      "grad_norm": 1.8773168325424194,
      "learning_rate": 2.25e-05,
      "loss": 1.4933,
      "step": 150
    },
    {
      "epoch": 0.4007514088916719,
      "grad_norm": 2.180652618408203,
      "learning_rate": 2.4e-05,
      "loss": 1.449,
      "step": 160
    },
    {
      "epoch": 0.4257983719474014,
      "grad_norm": 1.6153920888900757,
      "learning_rate": 2.55e-05,
      "loss": 1.4281,
      "step": 170
    },
    {
      "epoch": 0.45084533500313084,
      "grad_norm": 3.9366486072540283,
      "learning_rate": 2.7000000000000002e-05,
      "loss": 1.3706,
      "step": 180
    },
    {
      "epoch": 0.47589229805886035,
      "grad_norm": 8.680933952331543,
      "learning_rate": 2.8499999999999998e-05,
      "loss": 1.3632,
      "step": 190
    },
    {
      "epoch": 0.5009392611145899,
      "grad_norm": 1.6787041425704956,
      "learning_rate": 3e-05,
      "loss": 1.3052,
      "step": 200
    },
    {
      "epoch": 0.5259862241703194,
      "grad_norm": 3.0554661750793457,
      "learning_rate": 2.9832869080779945e-05,
      "loss": 1.2246,
      "step": 210
    },
    {
      "epoch": 0.5510331872260489,
      "grad_norm": 2.5610270500183105,
      "learning_rate": 2.9665738161559886e-05,
      "loss": 1.1641,
      "step": 220
    },
    {
      "epoch": 0.5760801502817783,
      "grad_norm": 4.368018627166748,
      "learning_rate": 2.9498607242339834e-05,
      "loss": 1.1112,
      "step": 230
    },
    {
      "epoch": 0.6011271133375078,
      "grad_norm": 3.1300201416015625,
      "learning_rate": 2.933147632311978e-05,
      "loss": 1.1034,
      "step": 240
    },
    {
      "epoch": 0.6261740763932373,
      "grad_norm": 3.441697597503662,
      "learning_rate": 2.916434540389972e-05,
      "loss": 0.9748,
      "step": 250
    },
    {
      "epoch": 0.6512210394489668,
      "grad_norm": 2.930171251296997,
      "learning_rate": 2.8997214484679665e-05,
      "loss": 0.9908,
      "step": 260
    },
    {
      "epoch": 0.6762680025046963,
      "grad_norm": 4.085419178009033,
      "learning_rate": 2.8830083565459613e-05,
      "loss": 0.8302,
      "step": 270
    },
    {
      "epoch": 0.7013149655604258,
      "grad_norm": 3.5097601413726807,
      "learning_rate": 2.8662952646239554e-05,
      "loss": 0.788,
      "step": 280
    },
    {
      "epoch": 0.7263619286161553,
      "grad_norm": 4.0369062423706055,
      "learning_rate": 2.84958217270195e-05,
      "loss": 0.766,
      "step": 290
    },
    {
      "epoch": 0.7514088916718847,
      "grad_norm": 2.472437858581543,
      "learning_rate": 2.8328690807799443e-05,
      "loss": 0.742,
      "step": 300
    },
    {
      "epoch": 0.7764558547276142,
      "grad_norm": 3.6972556114196777,
      "learning_rate": 2.8161559888579388e-05,
      "loss": 0.7595,
      "step": 310
    },
    {
      "epoch": 0.8015028177833438,
      "grad_norm": 3.38409161567688,
      "learning_rate": 2.7994428969359332e-05,
      "loss": 0.6941,
      "step": 320
    },
    {
      "epoch": 0.8265497808390733,
      "grad_norm": 2.243342399597168,
      "learning_rate": 2.7827298050139277e-05,
      "loss": 0.6347,
      "step": 330
    },
    {
      "epoch": 0.8515967438948028,
      "grad_norm": 4.12843132019043,
      "learning_rate": 2.7660167130919218e-05,
      "loss": 0.6031,
      "step": 340
    },
    {
      "epoch": 0.8766437069505323,
      "grad_norm": 2.2556419372558594,
      "learning_rate": 2.7493036211699166e-05,
      "loss": 0.5767,
      "step": 350
    },
    {
      "epoch": 0.9016906700062617,
      "grad_norm": 3.9663896560668945,
      "learning_rate": 2.732590529247911e-05,
      "loss": 0.5565,
      "step": 360
    },
    {
      "epoch": 0.9267376330619912,
      "grad_norm": 3.594534397125244,
      "learning_rate": 2.7158774373259055e-05,
      "loss": 0.5529,
      "step": 370
    },
    {
      "epoch": 0.9517845961177207,
      "grad_norm": 2.7648637294769287,
      "learning_rate": 2.6991643454038996e-05,
      "loss": 0.4842,
      "step": 380
    },
    {
      "epoch": 0.9768315591734502,
      "grad_norm": 3.66634202003479,
      "learning_rate": 2.6824512534818944e-05,
      "loss": 0.5065,
      "step": 390
    },
    {
      "epoch": 0.9993738259236068,
      "eval_accuracy": 0.9713150926743159,
      "eval_loss": 0.3426418602466583,
      "eval_runtime": 10.5903,
      "eval_samples_per_second": 641.906,
      "eval_steps_per_second": 20.113,
      "step": 399
    },
    {
      "epoch": 1.0018785222291797,
      "grad_norm": 2.2331535816192627,
      "learning_rate": 2.665738161559889e-05,
      "loss": 0.4929,
      "step": 400
    },
    {
      "epoch": 1.0269254852849092,
      "grad_norm": 2.8309195041656494,
      "learning_rate": 2.649025069637883e-05,
      "loss": 0.413,
      "step": 410
    },
    {
      "epoch": 1.0519724483406387,
      "grad_norm": 3.615835428237915,
      "learning_rate": 2.6323119777158774e-05,
      "loss": 0.4205,
      "step": 420
    },
    {
      "epoch": 1.0770194113963683,
      "grad_norm": 4.138455867767334,
      "learning_rate": 2.6155988857938722e-05,
      "loss": 0.3937,
      "step": 430
    },
    {
      "epoch": 1.1020663744520978,
      "grad_norm": 3.2305445671081543,
      "learning_rate": 2.5988857938718663e-05,
      "loss": 0.4097,
      "step": 440
    },
    {
      "epoch": 1.127113337507827,
      "grad_norm": 3.1868467330932617,
      "learning_rate": 2.5821727019498608e-05,
      "loss": 0.3661,
      "step": 450
    },
    {
      "epoch": 1.1521603005635566,
      "grad_norm": 1.8689473867416382,
      "learning_rate": 2.5654596100278553e-05,
      "loss": 0.3327,
      "step": 460
    },
    {
      "epoch": 1.177207263619286,
      "grad_norm": 1.8282161951065063,
      "learning_rate": 2.5487465181058497e-05,
      "loss": 0.3211,
      "step": 470
    },
    {
      "epoch": 1.2022542266750156,
      "grad_norm": 3.445858955383301,
      "learning_rate": 2.532033426183844e-05,
      "loss": 0.3468,
      "step": 480
    },
    {
      "epoch": 1.227301189730745,
      "grad_norm": 3.0723907947540283,
      "learning_rate": 2.5153203342618386e-05,
      "loss": 0.3533,
      "step": 490
    },
    {
      "epoch": 1.2523481527864746,
      "grad_norm": 4.13520622253418,
      "learning_rate": 2.4986072423398327e-05,
      "loss": 0.2817,
      "step": 500
    },
    {
      "epoch": 1.277395115842204,
      "grad_norm": 6.625748634338379,
      "learning_rate": 2.4818941504178275e-05,
      "loss": 0.3295,
      "step": 510
    },
    {
      "epoch": 1.3024420788979336,
      "grad_norm": 1.7438695430755615,
      "learning_rate": 2.465181058495822e-05,
      "loss": 0.2915,
      "step": 520
    },
    {
      "epoch": 1.3274890419536631,
      "grad_norm": 4.603183746337891,
      "learning_rate": 2.448467966573816e-05,
      "loss": 0.3447,
      "step": 530
    },
    {
      "epoch": 1.3525360050093926,
      "grad_norm": 2.3780734539031982,
      "learning_rate": 2.4317548746518106e-05,
      "loss": 0.2678,
      "step": 540
    },
    {
      "epoch": 1.3775829680651221,
      "grad_norm": 2.449040174484253,
      "learning_rate": 2.415041782729805e-05,
      "loss": 0.3029,
      "step": 550
    },
    {
      "epoch": 1.4026299311208517,
      "grad_norm": 3.899897336959839,
      "learning_rate": 2.3983286908077995e-05,
      "loss": 0.2609,
      "step": 560
    },
    {
      "epoch": 1.4276768941765812,
      "grad_norm": 4.0847978591918945,
      "learning_rate": 2.381615598885794e-05,
      "loss": 0.2979,
      "step": 570
    },
    {
      "epoch": 1.4527238572323107,
      "grad_norm": 3.0735533237457275,
      "learning_rate": 2.3649025069637884e-05,
      "loss": 0.2612,
      "step": 580
    },
    {
      "epoch": 1.4777708202880402,
      "grad_norm": 4.23598051071167,
      "learning_rate": 2.3481894150417825e-05,
      "loss": 0.2657,
      "step": 590
    },
    {
      "epoch": 1.5028177833437697,
      "grad_norm": 2.6922895908355713,
      "learning_rate": 2.3314763231197773e-05,
      "loss": 0.2332,
      "step": 600
    },
    {
      "epoch": 1.5278647463994992,
      "grad_norm": 2.444784641265869,
      "learning_rate": 2.3147632311977718e-05,
      "loss": 0.2646,
      "step": 610
    },
    {
      "epoch": 1.5529117094552287,
      "grad_norm": 3.595010757446289,
      "learning_rate": 2.298050139275766e-05,
      "loss": 0.2258,
      "step": 620
    },
    {
      "epoch": 1.577958672510958,
      "grad_norm": 4.990175724029541,
      "learning_rate": 2.2813370473537603e-05,
      "loss": 0.2477,
      "step": 630
    },
    {
      "epoch": 1.6030056355666875,
      "grad_norm": 2.4451816082000732,
      "learning_rate": 2.264623955431755e-05,
      "loss": 0.2268,
      "step": 640
    },
    {
      "epoch": 1.628052598622417,
      "grad_norm": 5.3694233894348145,
      "learning_rate": 2.2479108635097492e-05,
      "loss": 0.226,
      "step": 650
    },
    {
      "epoch": 1.6530995616781465,
      "grad_norm": 4.013739585876465,
      "learning_rate": 2.2311977715877437e-05,
      "loss": 0.2519,
      "step": 660
    },
    {
      "epoch": 1.678146524733876,
      "grad_norm": 2.913640022277832,
      "learning_rate": 2.214484679665738e-05,
      "loss": 0.2723,
      "step": 670
    },
    {
      "epoch": 1.7031934877896056,
      "grad_norm": 3.3302664756774902,
      "learning_rate": 2.1977715877437326e-05,
      "loss": 0.2412,
      "step": 680
    },
    {
      "epoch": 1.7282404508453348,
      "grad_norm": 3.4415056705474854,
      "learning_rate": 2.181058495821727e-05,
      "loss": 0.2463,
      "step": 690
    },
    {
      "epoch": 1.7532874139010644,
      "grad_norm": 2.424220085144043,
      "learning_rate": 2.1643454038997215e-05,
      "loss": 0.198,
      "step": 700
    },
    {
      "epoch": 1.7783343769567939,
      "grad_norm": 3.404729127883911,
      "learning_rate": 2.147632311977716e-05,
      "loss": 0.2263,
      "step": 710
    },
    {
      "epoch": 1.8033813400125234,
      "grad_norm": 2.495832681655884,
      "learning_rate": 2.1309192200557104e-05,
      "loss": 0.2146,
      "step": 720
    },
    {
      "epoch": 1.8284283030682529,
      "grad_norm": 3.8898746967315674,
      "learning_rate": 2.114206128133705e-05,
      "loss": 0.2289,
      "step": 730
    },
    {
      "epoch": 1.8534752661239824,
      "grad_norm": 2.768136978149414,
      "learning_rate": 2.0974930362116993e-05,
      "loss": 0.2144,
      "step": 740
    },
    {
      "epoch": 1.878522229179712,
      "grad_norm": 2.6077802181243896,
      "learning_rate": 2.0807799442896935e-05,
      "loss": 0.2269,
      "step": 750
    },
    {
      "epoch": 1.9035691922354414,
      "grad_norm": 2.223156452178955,
      "learning_rate": 2.0640668523676883e-05,
      "loss": 0.1809,
      "step": 760
    },
    {
      "epoch": 1.928616155291171,
      "grad_norm": 2.7288057804107666,
      "learning_rate": 2.0473537604456827e-05,
      "loss": 0.1833,
      "step": 770
    },
    {
      "epoch": 1.9536631183469004,
      "grad_norm": 2.93996000289917,
      "learning_rate": 2.0306406685236768e-05,
      "loss": 0.2276,
      "step": 780
    },
    {
      "epoch": 1.97871008140263,
      "grad_norm": 1.5655947923660278,
      "learning_rate": 2.0139275766016713e-05,
      "loss": 0.2089,
      "step": 790
    },
    {
      "epoch": 1.9987476518472134,
      "eval_accuracy": 0.9780817887614004,
      "eval_loss": 0.12669295072555542,
      "eval_runtime": 10.6875,
      "eval_samples_per_second": 636.068,
      "eval_steps_per_second": 19.93,
      "step": 798
    },
    {
      "epoch": 2.0037570444583594,
      "grad_norm": 2.9249160289764404,
      "learning_rate": 1.997214484679666e-05,
      "loss": 0.201,
      "step": 800
    },
    {
      "epoch": 2.028804007514089,
      "grad_norm": 2.023646116256714,
      "learning_rate": 1.9805013927576602e-05,
      "loss": 0.1868,
      "step": 810
    },
    {
      "epoch": 2.0538509705698185,
      "grad_norm": 1.4999943971633911,
      "learning_rate": 1.9637883008356546e-05,
      "loss": 0.1885,
      "step": 820
    },
    {
      "epoch": 2.078897933625548,
      "grad_norm": 3.2979843616485596,
      "learning_rate": 1.947075208913649e-05,
      "loss": 0.1904,
      "step": 830
    },
    {
      "epoch": 2.1039448966812775,
      "grad_norm": 1.8831233978271484,
      "learning_rate": 1.9303621169916436e-05,
      "loss": 0.185,
      "step": 840
    },
    {
      "epoch": 2.128991859737007,
      "grad_norm": 1.3918356895446777,
      "learning_rate": 1.913649025069638e-05,
      "loss": 0.1829,
      "step": 850
    },
    {
      "epoch": 2.1540388227927365,
      "grad_norm": 1.2814128398895264,
      "learning_rate": 1.8969359331476325e-05,
      "loss": 0.1667,
      "step": 860
    },
    {
      "epoch": 2.179085785848466,
      "grad_norm": 3.2199411392211914,
      "learning_rate": 1.8802228412256266e-05,
      "loss": 0.179,
      "step": 870
    },
    {
      "epoch": 2.2041327489041955,
      "grad_norm": 3.293250560760498,
      "learning_rate": 1.863509749303621e-05,
      "loss": 0.213,
      "step": 880
    },
    {
      "epoch": 2.229179711959925,
      "grad_norm": 2.887782335281372,
      "learning_rate": 1.846796657381616e-05,
      "loss": 0.1872,
      "step": 890
    },
    {
      "epoch": 2.254226675015654,
      "grad_norm": 3.306884288787842,
      "learning_rate": 1.83008356545961e-05,
      "loss": 0.1762,
      "step": 900
    },
    {
      "epoch": 2.279273638071384,
      "grad_norm": 3.3350651264190674,
      "learning_rate": 1.8133704735376044e-05,
      "loss": 0.2153,
      "step": 910
    },
    {
      "epoch": 2.304320601127113,
      "grad_norm": 1.8800674676895142,
      "learning_rate": 1.796657381615599e-05,
      "loss": 0.2069,
      "step": 920
    },
    {
      "epoch": 2.329367564182843,
      "grad_norm": 3.2760367393493652,
      "learning_rate": 1.7799442896935933e-05,
      "loss": 0.1756,
      "step": 930
    },
    {
      "epoch": 2.354414527238572,
      "grad_norm": 4.827606678009033,
      "learning_rate": 1.7632311977715878e-05,
      "loss": 0.1673,
      "step": 940
    },
    {
      "epoch": 2.3794614902943017,
      "grad_norm": 4.115577697753906,
      "learning_rate": 1.7465181058495822e-05,
      "loss": 0.1503,
      "step": 950
    },
    {
      "epoch": 2.404508453350031,
      "grad_norm": 2.163761854171753,
      "learning_rate": 1.7298050139275764e-05,
      "loss": 0.2029,
      "step": 960
    },
    {
      "epoch": 2.4295554164057607,
      "grad_norm": 3.1453959941864014,
      "learning_rate": 1.713091922005571e-05,
      "loss": 0.1748,
      "step": 970
    },
    {
      "epoch": 2.45460237946149,
      "grad_norm": 0.9899351596832275,
      "learning_rate": 1.6963788300835656e-05,
      "loss": 0.1589,
      "step": 980
    },
    {
      "epoch": 2.4796493425172197,
      "grad_norm": 2.109621524810791,
      "learning_rate": 1.6796657381615597e-05,
      "loss": 0.1709,
      "step": 990
    },
    {
      "epoch": 2.504696305572949,
      "grad_norm": 4.260733604431152,
      "learning_rate": 1.6629526462395542e-05,
      "loss": 0.1776,
      "step": 1000
    },
    {
      "epoch": 2.5297432686286787,
      "grad_norm": 1.1494958400726318,
      "learning_rate": 1.646239554317549e-05,
      "loss": 0.1651,
      "step": 1010
    },
    {
      "epoch": 2.554790231684408,
      "grad_norm": 4.246566295623779,
      "learning_rate": 1.6295264623955434e-05,
      "loss": 0.1941,
      "step": 1020
    },
    {
      "epoch": 2.5798371947401377,
      "grad_norm": 2.06291127204895,
      "learning_rate": 1.6128133704735375e-05,
      "loss": 0.1276,
      "step": 1030
    },
    {
      "epoch": 2.6048841577958672,
      "grad_norm": 2.7648720741271973,
      "learning_rate": 1.596100278551532e-05,
      "loss": 0.1725,
      "step": 1040
    },
    {
      "epoch": 2.6299311208515967,
      "grad_norm": 1.355832815170288,
      "learning_rate": 1.5793871866295268e-05,
      "loss": 0.1585,
      "step": 1050
    },
    {
      "epoch": 2.6549780839073263,
      "grad_norm": 3.2558505535125732,
      "learning_rate": 1.562674094707521e-05,
      "loss": 0.1778,
      "step": 1060
    },
    {
      "epoch": 2.6800250469630558,
      "grad_norm": 6.209780693054199,
      "learning_rate": 1.5459610027855154e-05,
      "loss": 0.1789,
      "step": 1070
    },
    {
      "epoch": 2.7050720100187853,
      "grad_norm": 2.586533784866333,
      "learning_rate": 1.5292479108635098e-05,
      "loss": 0.1648,
      "step": 1080
    },
    {
      "epoch": 2.730118973074515,
      "grad_norm": 3.024934768676758,
      "learning_rate": 1.5125348189415043e-05,
      "loss": 0.1428,
      "step": 1090
    },
    {
      "epoch": 2.7551659361302443,
      "grad_norm": 3.368821382522583,
      "learning_rate": 1.4958217270194987e-05,
      "loss": 0.1499,
      "step": 1100
    },
    {
      "epoch": 2.780212899185974,
      "grad_norm": 2.3457772731781006,
      "learning_rate": 1.479108635097493e-05,
      "loss": 0.1733,
      "step": 1110
    },
    {
      "epoch": 2.8052598622417033,
      "grad_norm": 1.2713210582733154,
      "learning_rate": 1.4623955431754876e-05,
      "loss": 0.1533,
      "step": 1120
    },
    {
      "epoch": 2.830306825297433,
      "grad_norm": 2.4018795490264893,
      "learning_rate": 1.445682451253482e-05,
      "loss": 0.1713,
      "step": 1130
    },
    {
      "epoch": 2.8553537883531623,
      "grad_norm": 2.1352932453155518,
      "learning_rate": 1.4289693593314764e-05,
      "loss": 0.1723,
      "step": 1140
    },
    {
      "epoch": 2.8804007514088914,
      "grad_norm": 1.5749989748001099,
      "learning_rate": 1.4122562674094708e-05,
      "loss": 0.1965,
      "step": 1150
    },
    {
      "epoch": 2.9054477144646214,
      "grad_norm": 3.0269646644592285,
      "learning_rate": 1.3955431754874653e-05,
      "loss": 0.1965,
      "step": 1160
    },
    {
      "epoch": 2.9304946775203504,
      "grad_norm": 2.5006277561187744,
      "learning_rate": 1.3788300835654596e-05,
      "loss": 0.163,
      "step": 1170
    },
    {
      "epoch": 2.9555416405760804,
      "grad_norm": 1.5534733533859253,
      "learning_rate": 1.362116991643454e-05,
      "loss": 0.1569,
      "step": 1180
    },
    {
      "epoch": 2.9805886036318094,
      "grad_norm": 2.774616241455078,
      "learning_rate": 1.3454038997214485e-05,
      "loss": 0.1834,
      "step": 1190
    },
    {
      "epoch": 2.9981214777708205,
      "eval_accuracy": 0.980435422182995,
      "eval_loss": 0.08890607953071594,
      "eval_runtime": 10.7458,
      "eval_samples_per_second": 632.617,
      "eval_steps_per_second": 19.822,
      "step": 1197
    },
    {
      "epoch": 3.005635566687539,
      "grad_norm": 3.6320059299468994,
      "learning_rate": 1.3286908077994428e-05,
      "loss": 0.1771,
      "step": 1200
    },
    {
      "epoch": 3.0306825297432685,
      "grad_norm": 1.4830687046051025,
      "learning_rate": 1.3119777158774374e-05,
      "loss": 0.171,
      "step": 1210
    },
    {
      "epoch": 3.055729492798998,
      "grad_norm": 1.7326056957244873,
      "learning_rate": 1.2952646239554317e-05,
      "loss": 0.1576,
      "step": 1220
    },
    {
      "epoch": 3.0807764558547275,
      "grad_norm": 2.301881790161133,
      "learning_rate": 1.2785515320334262e-05,
      "loss": 0.1675,
      "step": 1230
    },
    {
      "epoch": 3.105823418910457,
      "grad_norm": 3.6545441150665283,
      "learning_rate": 1.2618384401114206e-05,
      "loss": 0.1707,
      "step": 1240
    },
    {
      "epoch": 3.1308703819661865,
      "grad_norm": 1.9802820682525635,
      "learning_rate": 1.245125348189415e-05,
      "loss": 0.1744,
      "step": 1250
    },
    {
      "epoch": 3.155917345021916,
      "grad_norm": 2.116532564163208,
      "learning_rate": 1.2284122562674095e-05,
      "loss": 0.1592,
      "step": 1260
    },
    {
      "epoch": 3.1809643080776455,
      "grad_norm": 2.7397706508636475,
      "learning_rate": 1.211699164345404e-05,
      "loss": 0.1567,
      "step": 1270
    },
    {
      "epoch": 3.206011271133375,
      "grad_norm": 2.8573620319366455,
      "learning_rate": 1.1949860724233983e-05,
      "loss": 0.1673,
      "step": 1280
    },
    {
      "epoch": 3.2310582341891045,
      "grad_norm": 3.157711982727051,
      "learning_rate": 1.1782729805013929e-05,
      "loss": 0.1345,
      "step": 1290
    },
    {
      "epoch": 3.256105197244834,
      "grad_norm": 4.402627944946289,
      "learning_rate": 1.1615598885793872e-05,
      "loss": 0.1689,
      "step": 1300
    },
    {
      "epoch": 3.2811521603005636,
      "grad_norm": 1.3651645183563232,
      "learning_rate": 1.1448467966573816e-05,
      "loss": 0.0912,
      "step": 1310
    },
    {
      "epoch": 3.306199123356293,
      "grad_norm": 3.8698039054870605,
      "learning_rate": 1.1281337047353761e-05,
      "loss": 0.1594,
      "step": 1320
    },
    {
      "epoch": 3.3312460864120226,
      "grad_norm": 3.035409450531006,
      "learning_rate": 1.1114206128133705e-05,
      "loss": 0.1601,
      "step": 1330
    },
    {
      "epoch": 3.356293049467752,
      "grad_norm": 1.8605338335037231,
      "learning_rate": 1.0947075208913648e-05,
      "loss": 0.1713,
      "step": 1340
    },
    {
      "epoch": 3.3813400125234816,
      "grad_norm": 3.120457887649536,
      "learning_rate": 1.0779944289693595e-05,
      "loss": 0.1542,
      "step": 1350
    },
    {
      "epoch": 3.406386975579211,
      "grad_norm": 3.9332990646362305,
      "learning_rate": 1.0612813370473537e-05,
      "loss": 0.1451,
      "step": 1360
    },
    {
      "epoch": 3.4314339386349406,
      "grad_norm": 3.2169320583343506,
      "learning_rate": 1.0445682451253482e-05,
      "loss": 0.1417,
      "step": 1370
    },
    {
      "epoch": 3.45648090169067,
      "grad_norm": 1.290043592453003,
      "learning_rate": 1.0278551532033427e-05,
      "loss": 0.1672,
      "step": 1380
    },
    {
      "epoch": 3.4815278647463996,
      "grad_norm": 1.9304981231689453,
      "learning_rate": 1.0111420612813371e-05,
      "loss": 0.1558,
      "step": 1390
    },
    {
      "epoch": 3.506574827802129,
      "grad_norm": 1.4201843738555908,
      "learning_rate": 9.944289693593314e-06,
      "loss": 0.1535,
      "step": 1400
    },
    {
      "epoch": 3.5316217908578587,
      "grad_norm": 1.5534597635269165,
      "learning_rate": 9.77715877437326e-06,
      "loss": 0.1275,
      "step": 1410
    },
    {
      "epoch": 3.5566687539135877,
      "grad_norm": 3.183288097381592,
      "learning_rate": 9.610027855153203e-06,
      "loss": 0.1588,
      "step": 1420
    },
    {
      "epoch": 3.5817157169693177,
      "grad_norm": 3.252284288406372,
      "learning_rate": 9.44289693593315e-06,
      "loss": 0.1266,
      "step": 1430
    },
    {
      "epoch": 3.6067626800250467,
      "grad_norm": 3.5988216400146484,
      "learning_rate": 9.275766016713092e-06,
      "loss": 0.1484,
      "step": 1440
    },
    {
      "epoch": 3.6318096430807767,
      "grad_norm": 0.8129429221153259,
      "learning_rate": 9.108635097493037e-06,
      "loss": 0.1249,
      "step": 1450
    },
    {
      "epoch": 3.6568566061365058,
      "grad_norm": 1.1750004291534424,
      "learning_rate": 8.941504178272981e-06,
      "loss": 0.1552,
      "step": 1460
    },
    {
      "epoch": 3.6819035691922357,
      "grad_norm": 2.3609495162963867,
      "learning_rate": 8.774373259052926e-06,
      "loss": 0.1506,
      "step": 1470
    },
    {
      "epoch": 3.706950532247965,
      "grad_norm": 3.020759344100952,
      "learning_rate": 8.607242339832869e-06,
      "loss": 0.1492,
      "step": 1480
    },
    {
      "epoch": 3.7319974953036943,
      "grad_norm": 2.7013983726501465,
      "learning_rate": 8.440111420612815e-06,
      "loss": 0.1045,
      "step": 1490
    },
    {
      "epoch": 3.757044458359424,
      "grad_norm": 3.6517138481140137,
      "learning_rate": 8.272980501392758e-06,
      "loss": 0.1354,
      "step": 1500
    },
    {
      "epoch": 3.7820914214151533,
      "grad_norm": 1.283496618270874,
      "learning_rate": 8.1058495821727e-06,
      "loss": 0.1376,
      "step": 1510
    },
    {
      "epoch": 3.807138384470883,
      "grad_norm": 4.068831443786621,
      "learning_rate": 7.938718662952647e-06,
      "loss": 0.147,
      "step": 1520
    },
    {
      "epoch": 3.8321853475266123,
      "grad_norm": 1.8731778860092163,
      "learning_rate": 7.77158774373259e-06,
      "loss": 0.134,
      "step": 1530
    },
    {
      "epoch": 3.857232310582342,
      "grad_norm": 2.750255823135376,
      "learning_rate": 7.604456824512535e-06,
      "loss": 0.1398,
      "step": 1540
    },
    {
      "epoch": 3.8822792736380713,
      "grad_norm": 2.020031213760376,
      "learning_rate": 7.43732590529248e-06,
      "loss": 0.1236,
      "step": 1550
    },
    {
      "epoch": 3.907326236693801,
      "grad_norm": 3.4288525581359863,
      "learning_rate": 7.2701949860724235e-06,
      "loss": 0.1223,
      "step": 1560
    },
    {
      "epoch": 3.9323731997495304,
      "grad_norm": 2.974663019180298,
      "learning_rate": 7.103064066852368e-06,
      "loss": 0.1596,
      "step": 1570
    },
    {
      "epoch": 3.95742016280526,
      "grad_norm": 2.157737970352173,
      "learning_rate": 6.935933147632313e-06,
      "loss": 0.1277,
      "step": 1580
    },
    {
      "epoch": 3.9824671258609894,
      "grad_norm": 2.2225987911224365,
      "learning_rate": 6.768802228412256e-06,
      "loss": 0.142,
      "step": 1590
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.981318034716093,
      "eval_loss": 0.08539879322052002,
      "eval_runtime": 10.6636,
      "eval_samples_per_second": 637.494,
      "eval_steps_per_second": 19.974,
      "step": 1597
    },
    {
      "epoch": 4.007514088916719,
      "grad_norm": 1.7441836595535278,
      "learning_rate": 6.601671309192201e-06,
      "loss": 0.1491,
      "step": 1600
    },
    {
      "epoch": 4.032561051972448,
      "grad_norm": 2.0195930004119873,
      "learning_rate": 6.4345403899721455e-06,
      "loss": 0.1357,
      "step": 1610
    },
    {
      "epoch": 4.057608015028178,
      "grad_norm": 2.4423089027404785,
      "learning_rate": 6.267409470752089e-06,
      "loss": 0.1438,
      "step": 1620
    },
    {
      "epoch": 4.082654978083907,
      "grad_norm": 2.6411936283111572,
      "learning_rate": 6.100278551532034e-06,
      "loss": 0.131,
      "step": 1630
    },
    {
      "epoch": 4.107701941139637,
      "grad_norm": 0.8893925547599792,
      "learning_rate": 5.933147632311978e-06,
      "loss": 0.1249,
      "step": 1640
    },
    {
      "epoch": 4.132748904195366,
      "grad_norm": 2.2715697288513184,
      "learning_rate": 5.766016713091923e-06,
      "loss": 0.1642,
      "step": 1650
    },
    {
      "epoch": 4.157795867251096,
      "grad_norm": 1.903985619544983,
      "learning_rate": 5.598885793871867e-06,
      "loss": 0.1418,
      "step": 1660
    },
    {
      "epoch": 4.182842830306825,
      "grad_norm": 2.41510272026062,
      "learning_rate": 5.43175487465181e-06,
      "loss": 0.1535,
      "step": 1670
    },
    {
      "epoch": 4.207889793362555,
      "grad_norm": 3.837315320968628,
      "learning_rate": 5.264623955431755e-06,
      "loss": 0.1433,
      "step": 1680
    },
    {
      "epoch": 4.232936756418284,
      "grad_norm": 2.4811267852783203,
      "learning_rate": 5.0974930362116986e-06,
      "loss": 0.121,
      "step": 1690
    },
    {
      "epoch": 4.257983719474014,
      "grad_norm": 4.462907791137695,
      "learning_rate": 4.930362116991643e-06,
      "loss": 0.1267,
      "step": 1700
    },
    {
      "epoch": 4.283030682529743,
      "grad_norm": 3.019132137298584,
      "learning_rate": 4.763231197771588e-06,
      "loss": 0.1151,
      "step": 1710
    },
    {
      "epoch": 4.308077645585473,
      "grad_norm": 3.2697532176971436,
      "learning_rate": 4.596100278551532e-06,
      "loss": 0.1068,
      "step": 1720
    },
    {
      "epoch": 4.333124608641202,
      "grad_norm": 3.3036513328552246,
      "learning_rate": 4.428969359331476e-06,
      "loss": 0.1009,
      "step": 1730
    },
    {
      "epoch": 4.358171571696932,
      "grad_norm": 2.5975863933563232,
      "learning_rate": 4.2618384401114205e-06,
      "loss": 0.1168,
      "step": 1740
    },
    {
      "epoch": 4.383218534752661,
      "grad_norm": 1.6975170373916626,
      "learning_rate": 4.094707520891365e-06,
      "loss": 0.1549,
      "step": 1750
    },
    {
      "epoch": 4.408265497808391,
      "grad_norm": 2.671894073486328,
      "learning_rate": 3.927576601671309e-06,
      "loss": 0.1211,
      "step": 1760
    },
    {
      "epoch": 4.43331246086412,
      "grad_norm": 4.581605911254883,
      "learning_rate": 3.7604456824512533e-06,
      "loss": 0.1406,
      "step": 1770
    },
    {
      "epoch": 4.45835942391985,
      "grad_norm": 3.0454249382019043,
      "learning_rate": 3.593314763231198e-06,
      "loss": 0.137,
      "step": 1780
    },
    {
      "epoch": 4.483406386975579,
      "grad_norm": 2.0320184230804443,
      "learning_rate": 3.426183844011142e-06,
      "loss": 0.1105,
      "step": 1790
    },
    {
      "epoch": 4.508453350031308,
      "grad_norm": 1.5015639066696167,
      "learning_rate": 3.259052924791086e-06,
      "loss": 0.1282,
      "step": 1800
    },
    {
      "epoch": 4.533500313087038,
      "grad_norm": 2.1987569332122803,
      "learning_rate": 3.0919220055710307e-06,
      "loss": 0.1111,
      "step": 1810
    },
    {
      "epoch": 4.558547276142768,
      "grad_norm": 1.3748631477355957,
      "learning_rate": 2.924791086350975e-06,
      "loss": 0.1326,
      "step": 1820
    },
    {
      "epoch": 4.583594239198497,
      "grad_norm": 1.8067643642425537,
      "learning_rate": 2.7576601671309194e-06,
      "loss": 0.1112,
      "step": 1830
    },
    {
      "epoch": 4.608641202254226,
      "grad_norm": 2.4993462562561035,
      "learning_rate": 2.5905292479108636e-06,
      "loss": 0.1014,
      "step": 1840
    },
    {
      "epoch": 4.633688165309956,
      "grad_norm": 3.3623006343841553,
      "learning_rate": 2.4233983286908077e-06,
      "loss": 0.1319,
      "step": 1850
    },
    {
      "epoch": 4.658735128365686,
      "grad_norm": 2.678271770477295,
      "learning_rate": 2.2562674094707523e-06,
      "loss": 0.1391,
      "step": 1860
    },
    {
      "epoch": 4.683782091421415,
      "grad_norm": 1.3857640027999878,
      "learning_rate": 2.0891364902506964e-06,
      "loss": 0.1149,
      "step": 1870
    },
    {
      "epoch": 4.708829054477144,
      "grad_norm": 2.552626371383667,
      "learning_rate": 1.922005571030641e-06,
      "loss": 0.119,
      "step": 1880
    },
    {
      "epoch": 4.733876017532874,
      "grad_norm": 1.3172489404678345,
      "learning_rate": 1.7548746518105849e-06,
      "loss": 0.1226,
      "step": 1890
    },
    {
      "epoch": 4.758922980588603,
      "grad_norm": 4.114231586456299,
      "learning_rate": 1.5877437325905292e-06,
      "loss": 0.1191,
      "step": 1900
    },
    {
      "epoch": 4.783969943644333,
      "grad_norm": 3.1994917392730713,
      "learning_rate": 1.4206128133704736e-06,
      "loss": 0.1064,
      "step": 1910
    },
    {
      "epoch": 4.809016906700062,
      "grad_norm": 2.785364866256714,
      "learning_rate": 1.253481894150418e-06,
      "loss": 0.1092,
      "step": 1920
    },
    {
      "epoch": 4.834063869755792,
      "grad_norm": 1.6309309005737305,
      "learning_rate": 1.0863509749303623e-06,
      "loss": 0.1355,
      "step": 1930
    },
    {
      "epoch": 4.859110832811521,
      "grad_norm": 3.174809694290161,
      "learning_rate": 9.192200557103064e-07,
      "loss": 0.1211,
      "step": 1940
    },
    {
      "epoch": 4.884157795867251,
      "grad_norm": 2.4154794216156006,
      "learning_rate": 7.520891364902508e-07,
      "loss": 0.1135,
      "step": 1950
    },
    {
      "epoch": 4.90920475892298,
      "grad_norm": 2.3144145011901855,
      "learning_rate": 5.84958217270195e-07,
      "loss": 0.1121,
      "step": 1960
    },
    {
      "epoch": 4.93425172197871,
      "grad_norm": 2.8474748134613037,
      "learning_rate": 4.178272980501393e-07,
      "loss": 0.143,
      "step": 1970
    },
    {
      "epoch": 4.959298685034439,
      "grad_norm": 2.4842255115509033,
      "learning_rate": 2.506963788300836e-07,
      "loss": 0.116,
      "step": 1980
    },
    {
      "epoch": 4.984345648090169,
      "grad_norm": 2.9249684810638428,
      "learning_rate": 8.356545961002785e-08,
      "loss": 0.1472,
      "step": 1990
    },
    {
      "epoch": 4.996869129618034,
      "eval_accuracy": 0.9832303618711385,
      "eval_loss": 0.07819115370512009,
      "eval_runtime": 10.7075,
      "eval_samples_per_second": 634.883,
      "eval_steps_per_second": 19.893,
      "step": 1995
    },
    {
      "epoch": 4.996869129618034,
      "step": 1995,
      "total_flos": 2.3180194781952e+18,
      "train_loss": 0.40026830736557045,
      "train_runtime": 1087.6247,
      "train_samples_per_second": 234.888,
      "train_steps_per_second": 1.834
    }
  ],
  "logging_steps": 10,
  "max_steps": 1995,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.3180194781952e+18,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}