{
  "best_metric": 0.881407804131599,
  "best_model_checkpoint": "distilbert-base-multilingual-cased-aoe-hyper/run-4/checkpoint-1308",
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 1308,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01529051987767584,
      "grad_norm": 3.7934694290161133,
      "learning_rate": 6.268194604781005e-06,
      "loss": 0.6321,
      "step": 10
    },
    {
      "epoch": 0.03058103975535168,
      "grad_norm": 2.5848259925842285,
      "learning_rate": 6.244141670764194e-06,
      "loss": 0.6228,
      "step": 20
    },
    {
      "epoch": 0.045871559633027525,
      "grad_norm": 2.499722957611084,
      "learning_rate": 6.220088736747383e-06,
      "loss": 0.5166,
      "step": 30
    },
    {
      "epoch": 0.06116207951070336,
      "grad_norm": 3.433131456375122,
      "learning_rate": 6.1960358027305715e-06,
      "loss": 0.5895,
      "step": 40
    },
    {
      "epoch": 0.0764525993883792,
      "grad_norm": 2.0335094928741455,
      "learning_rate": 6.17198286871376e-06,
      "loss": 0.5482,
      "step": 50
    },
    {
      "epoch": 0.09174311926605505,
      "grad_norm": 2.239384174346924,
      "learning_rate": 6.1479299346969484e-06,
      "loss": 0.5125,
      "step": 60
    },
    {
      "epoch": 0.10703363914373089,
      "grad_norm": 2.9248292446136475,
      "learning_rate": 6.123877000680137e-06,
      "loss": 0.4949,
      "step": 70
    },
    {
      "epoch": 0.12232415902140673,
      "grad_norm": 2.9950616359710693,
      "learning_rate": 6.099824066663326e-06,
      "loss": 0.4823,
      "step": 80
    },
    {
      "epoch": 0.13761467889908258,
      "grad_norm": 3.120652914047241,
      "learning_rate": 6.075771132646515e-06,
      "loss": 0.3806,
      "step": 90
    },
    {
      "epoch": 0.1529051987767584,
      "grad_norm": 2.4965806007385254,
      "learning_rate": 6.051718198629704e-06,
      "loss": 0.4585,
      "step": 100
    },
    {
      "epoch": 0.16819571865443425,
      "grad_norm": 2.0298407077789307,
      "learning_rate": 6.027665264612893e-06,
      "loss": 0.288,
      "step": 110
    },
    {
      "epoch": 0.1834862385321101,
      "grad_norm": 2.880314350128174,
      "learning_rate": 6.003612330596082e-06,
      "loss": 0.4147,
      "step": 120
    },
    {
      "epoch": 0.19877675840978593,
      "grad_norm": 3.3571255207061768,
      "learning_rate": 5.97955939657927e-06,
      "loss": 0.4601,
      "step": 130
    },
    {
      "epoch": 0.21406727828746178,
      "grad_norm": 7.866518020629883,
      "learning_rate": 5.955506462562459e-06,
      "loss": 0.4556,
      "step": 140
    },
    {
      "epoch": 0.22935779816513763,
      "grad_norm": 5.3301472663879395,
      "learning_rate": 5.9314535285456475e-06,
      "loss": 0.473,
      "step": 150
    },
    {
      "epoch": 0.24464831804281345,
      "grad_norm": 3.0050952434539795,
      "learning_rate": 5.907400594528836e-06,
      "loss": 0.383,
      "step": 160
    },
    {
      "epoch": 0.2599388379204893,
      "grad_norm": 8.946796417236328,
      "learning_rate": 5.883347660512025e-06,
      "loss": 0.3306,
      "step": 170
    },
    {
      "epoch": 0.27522935779816515,
      "grad_norm": 12.456232070922852,
      "learning_rate": 5.859294726495214e-06,
      "loss": 0.3614,
      "step": 180
    },
    {
      "epoch": 0.290519877675841,
      "grad_norm": 20.59422492980957,
      "learning_rate": 5.835241792478403e-06,
      "loss": 0.4216,
      "step": 190
    },
    {
      "epoch": 0.3058103975535168,
      "grad_norm": 3.627934694290161,
      "learning_rate": 5.811188858461592e-06,
      "loss": 0.451,
      "step": 200
    },
    {
      "epoch": 0.3211009174311927,
      "grad_norm": 4.5968852043151855,
      "learning_rate": 5.78713592444478e-06,
      "loss": 0.3596,
      "step": 210
    },
    {
      "epoch": 0.3363914373088685,
      "grad_norm": 8.742961883544922,
      "learning_rate": 5.763082990427969e-06,
      "loss": 0.4125,
      "step": 220
    },
    {
      "epoch": 0.3516819571865443,
      "grad_norm": 4.3628644943237305,
      "learning_rate": 5.739030056411158e-06,
      "loss": 0.3956,
      "step": 230
    },
    {
      "epoch": 0.3669724770642202,
      "grad_norm": 6.9421210289001465,
      "learning_rate": 5.7149771223943465e-06,
      "loss": 0.4196,
      "step": 240
    },
    {
      "epoch": 0.382262996941896,
      "grad_norm": 5.0060906410217285,
      "learning_rate": 5.690924188377535e-06,
      "loss": 0.3415,
      "step": 250
    },
    {
      "epoch": 0.39755351681957185,
      "grad_norm": 3.756944417953491,
      "learning_rate": 5.666871254360724e-06,
      "loss": 0.3433,
      "step": 260
    },
    {
      "epoch": 0.41284403669724773,
      "grad_norm": 8.966843605041504,
      "learning_rate": 5.642818320343913e-06,
      "loss": 0.326,
      "step": 270
    },
    {
      "epoch": 0.42813455657492355,
      "grad_norm": 13.75231647491455,
      "learning_rate": 5.618765386327102e-06,
      "loss": 0.2412,
      "step": 280
    },
    {
      "epoch": 0.4434250764525994,
      "grad_norm": 9.244980812072754,
      "learning_rate": 5.594712452310291e-06,
      "loss": 0.3949,
      "step": 290
    },
    {
      "epoch": 0.45871559633027525,
      "grad_norm": 7.885373115539551,
      "learning_rate": 5.57065951829348e-06,
      "loss": 0.2693,
      "step": 300
    },
    {
      "epoch": 0.4740061162079511,
      "grad_norm": 14.435099601745605,
      "learning_rate": 5.546606584276669e-06,
      "loss": 0.4065,
      "step": 310
    },
    {
      "epoch": 0.4892966360856269,
      "grad_norm": 27.61042022705078,
      "learning_rate": 5.5225536502598575e-06,
      "loss": 0.3138,
      "step": 320
    },
    {
      "epoch": 0.5045871559633027,
      "grad_norm": 6.691371440887451,
      "learning_rate": 5.4985007162430455e-06,
      "loss": 0.2268,
      "step": 330
    },
    {
      "epoch": 0.5198776758409785,
      "grad_norm": 6.305394172668457,
      "learning_rate": 5.474447782226234e-06,
      "loss": 0.2565,
      "step": 340
    },
    {
      "epoch": 0.5351681957186545,
      "grad_norm": 5.162963390350342,
      "learning_rate": 5.450394848209423e-06,
      "loss": 0.3489,
      "step": 350
    },
    {
      "epoch": 0.5504587155963303,
      "grad_norm": 12.863113403320312,
      "learning_rate": 5.426341914192612e-06,
      "loss": 0.4205,
      "step": 360
    },
    {
      "epoch": 0.5657492354740061,
      "grad_norm": 8.720885276794434,
      "learning_rate": 5.402288980175801e-06,
      "loss": 0.2633,
      "step": 370
    },
    {
      "epoch": 0.581039755351682,
      "grad_norm": 1.3078076839447021,
      "learning_rate": 5.378236046158989e-06,
      "loss": 0.2705,
      "step": 380
    },
    {
      "epoch": 0.5963302752293578,
      "grad_norm": 19.819198608398438,
      "learning_rate": 5.354183112142178e-06,
      "loss": 0.3007,
      "step": 390
    },
    {
      "epoch": 0.6116207951070336,
      "grad_norm": 9.876558303833008,
      "learning_rate": 5.330130178125367e-06,
      "loss": 0.2945,
      "step": 400
    },
    {
      "epoch": 0.6269113149847095,
      "grad_norm": 4.9784770011901855,
      "learning_rate": 5.306077244108556e-06,
      "loss": 0.3746,
      "step": 410
    },
    {
      "epoch": 0.6422018348623854,
      "grad_norm": 21.62326431274414,
      "learning_rate": 5.2820243100917446e-06,
      "loss": 0.291,
      "step": 420
    },
    {
      "epoch": 0.6574923547400612,
      "grad_norm": 10.322495460510254,
      "learning_rate": 5.2579713760749334e-06,
      "loss": 0.2793,
      "step": 430
    },
    {
      "epoch": 0.672782874617737,
      "grad_norm": 27.78510856628418,
      "learning_rate": 5.233918442058122e-06,
      "loss": 0.4447,
      "step": 440
    },
    {
      "epoch": 0.6880733944954128,
      "grad_norm": 8.71343994140625,
      "learning_rate": 5.209865508041311e-06,
      "loss": 0.169,
      "step": 450
    },
    {
      "epoch": 0.7033639143730887,
      "grad_norm": 7.693337917327881,
      "learning_rate": 5.1858125740245e-06,
      "loss": 0.394,
      "step": 460
    },
    {
      "epoch": 0.7186544342507645,
      "grad_norm": 8.130308151245117,
      "learning_rate": 5.161759640007688e-06,
      "loss": 0.1776,
      "step": 470
    },
    {
      "epoch": 0.7339449541284404,
      "grad_norm": 2.7745859622955322,
      "learning_rate": 5.137706705990877e-06,
      "loss": 0.216,
      "step": 480
    },
    {
      "epoch": 0.7492354740061162,
      "grad_norm": 8.477354049682617,
      "learning_rate": 5.113653771974066e-06,
      "loss": 0.3386,
      "step": 490
    },
    {
      "epoch": 0.764525993883792,
      "grad_norm": 8.951611518859863,
      "learning_rate": 5.089600837957255e-06,
      "loss": 0.2797,
      "step": 500
    },
    {
      "epoch": 0.7798165137614679,
      "grad_norm": 11.43747615814209,
      "learning_rate": 5.065547903940444e-06,
      "loss": 0.4733,
      "step": 510
    },
    {
      "epoch": 0.7951070336391437,
      "grad_norm": 19.584590911865234,
      "learning_rate": 5.0414949699236325e-06,
      "loss": 0.4629,
      "step": 520
    },
    {
      "epoch": 0.8103975535168195,
      "grad_norm": 11.580733299255371,
      "learning_rate": 5.017442035906821e-06,
      "loss": 0.1946,
      "step": 530
    },
    {
      "epoch": 0.8256880733944955,
      "grad_norm": 3.2449963092803955,
      "learning_rate": 4.99338910189001e-06,
      "loss": 0.282,
      "step": 540
    },
    {
      "epoch": 0.8409785932721713,
      "grad_norm": 20.366369247436523,
      "learning_rate": 4.969336167873199e-06,
      "loss": 0.3433,
      "step": 550
    },
    {
      "epoch": 0.8562691131498471,
      "grad_norm": 9.700865745544434,
      "learning_rate": 4.945283233856388e-06,
      "loss": 0.3337,
      "step": 560
    },
    {
      "epoch": 0.8715596330275229,
      "grad_norm": 8.121973037719727,
      "learning_rate": 4.921230299839577e-06,
      "loss": 0.3658,
      "step": 570
    },
    {
      "epoch": 0.8868501529051988,
      "grad_norm": 8.352603912353516,
      "learning_rate": 4.897177365822766e-06,
      "loss": 0.4038,
      "step": 580
    },
    {
      "epoch": 0.9021406727828746,
      "grad_norm": 6.9203290939331055,
      "learning_rate": 4.873124431805954e-06,
      "loss": 0.3139,
      "step": 590
    },
    {
      "epoch": 0.9174311926605505,
      "grad_norm": 9.17638111114502,
      "learning_rate": 4.849071497789143e-06,
      "loss": 0.37,
      "step": 600
    },
    {
      "epoch": 0.9327217125382263,
      "grad_norm": 17.2983341217041,
      "learning_rate": 4.8250185637723315e-06,
      "loss": 0.3072,
      "step": 610
    },
    {
      "epoch": 0.9480122324159022,
      "grad_norm": 5.322959899902344,
      "learning_rate": 4.80096562975552e-06,
      "loss": 0.323,
      "step": 620
    },
    {
      "epoch": 0.963302752293578,
      "grad_norm": 19.547361373901367,
      "learning_rate": 4.776912695738708e-06,
      "loss": 0.4587,
      "step": 630
    },
    {
      "epoch": 0.9785932721712538,
      "grad_norm": 8.637187004089355,
      "learning_rate": 4.752859761721897e-06,
      "loss": 0.2532,
      "step": 640
    },
    {
      "epoch": 0.9938837920489296,
      "grad_norm": 11.7959566116333,
      "learning_rate": 4.728806827705086e-06,
      "loss": 0.2689,
      "step": 650
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8706962509563887,
      "eval_f1": 0.6932849364791288,
      "eval_loss": 0.3414114713668823,
      "eval_precision": 0.799163179916318,
      "eval_recall": 0.6121794871794872,
      "eval_runtime": 4.8917,
      "eval_samples_per_second": 267.189,
      "eval_steps_per_second": 16.763,
      "step": 654
    },
    {
      "epoch": 1.0091743119266054,
      "grad_norm": 2.8421857357025146,
      "learning_rate": 4.704753893688275e-06,
      "loss": 0.3789,
      "step": 660
    },
    {
      "epoch": 1.0244648318042813,
      "grad_norm": 8.888960838317871,
      "learning_rate": 4.680700959671464e-06,
      "loss": 0.2548,
      "step": 670
    },
    {
      "epoch": 1.039755351681957,
      "grad_norm": 6.103824615478516,
      "learning_rate": 4.656648025654653e-06,
      "loss": 0.3478,
      "step": 680
    },
    {
      "epoch": 1.0550458715596331,
      "grad_norm": 22.265275955200195,
      "learning_rate": 4.632595091637842e-06,
      "loss": 0.3016,
      "step": 690
    },
    {
      "epoch": 1.070336391437309,
      "grad_norm": 9.670002937316895,
      "learning_rate": 4.6085421576210305e-06,
      "loss": 0.2558,
      "step": 700
    },
    {
      "epoch": 1.0856269113149848,
      "grad_norm": 1.852484941482544,
      "learning_rate": 4.584489223604219e-06,
      "loss": 0.2439,
      "step": 710
    },
    {
      "epoch": 1.1009174311926606,
      "grad_norm": 2.102276563644409,
      "learning_rate": 4.560436289587408e-06,
      "loss": 0.2946,
      "step": 720
    },
    {
      "epoch": 1.1162079510703364,
      "grad_norm": 5.1781439781188965,
      "learning_rate": 4.536383355570596e-06,
      "loss": 0.2724,
      "step": 730
    },
    {
      "epoch": 1.1314984709480123,
      "grad_norm": 5.4545722007751465,
      "learning_rate": 4.512330421553785e-06,
      "loss": 0.2599,
      "step": 740
    },
    {
      "epoch": 1.146788990825688,
      "grad_norm": 10.079143524169922,
      "learning_rate": 4.488277487536974e-06,
      "loss": 0.2584,
      "step": 750
    },
    {
      "epoch": 1.162079510703364,
      "grad_norm": 4.362283229827881,
      "learning_rate": 4.464224553520163e-06,
      "loss": 0.2489,
      "step": 760
    },
    {
      "epoch": 1.1773700305810397,
      "grad_norm": 7.355106353759766,
      "learning_rate": 4.440171619503352e-06,
      "loss": 0.3536,
      "step": 770
    },
    {
      "epoch": 1.1926605504587156,
      "grad_norm": 2.0413646697998047,
      "learning_rate": 4.416118685486541e-06,
      "loss": 0.1874,
      "step": 780
    },
    {
      "epoch": 1.2079510703363914,
      "grad_norm": 2.6824581623077393,
      "learning_rate": 4.3920657514697295e-06,
      "loss": 0.2813,
      "step": 790
    },
    {
      "epoch": 1.2232415902140672,
      "grad_norm": 1.147255539894104,
      "learning_rate": 4.368012817452918e-06,
      "loss": 0.2321,
      "step": 800
    },
    {
      "epoch": 1.238532110091743,
      "grad_norm": 8.774001121520996,
      "learning_rate": 4.343959883436107e-06,
      "loss": 0.2953,
      "step": 810
    },
    {
      "epoch": 1.2538226299694188,
      "grad_norm": 25.203166961669922,
      "learning_rate": 4.319906949419296e-06,
      "loss": 0.266,
      "step": 820
    },
    {
      "epoch": 1.2691131498470947,
      "grad_norm": 1.5749539136886597,
      "learning_rate": 4.295854015402485e-06,
      "loss": 0.2924,
      "step": 830
    },
    {
      "epoch": 1.2844036697247707,
      "grad_norm": 15.727303504943848,
      "learning_rate": 4.271801081385674e-06,
      "loss": 0.2302,
      "step": 840
    },
    {
      "epoch": 1.2996941896024465,
      "grad_norm": 35.48811721801758,
      "learning_rate": 4.247748147368862e-06,
      "loss": 0.2971,
      "step": 850
    },
    {
      "epoch": 1.3149847094801224,
      "grad_norm": 16.261396408081055,
      "learning_rate": 4.223695213352051e-06,
      "loss": 0.3347,
      "step": 860
    },
    {
      "epoch": 1.3302752293577982,
      "grad_norm": 1.1547783613204956,
      "learning_rate": 4.19964227933524e-06,
      "loss": 0.1146,
      "step": 870
    },
    {
      "epoch": 1.345565749235474,
      "grad_norm": 23.506380081176758,
      "learning_rate": 4.1755893453184286e-06,
      "loss": 0.2384,
      "step": 880
    },
    {
      "epoch": 1.3608562691131498,
      "grad_norm": 9.628520011901855,
      "learning_rate": 4.151536411301617e-06,
      "loss": 0.3036,
      "step": 890
    },
    {
      "epoch": 1.3761467889908257,
      "grad_norm": 36.97438049316406,
      "learning_rate": 4.1274834772848055e-06,
      "loss": 0.2811,
      "step": 900
    },
    {
      "epoch": 1.3914373088685015,
      "grad_norm": 1.8695220947265625,
      "learning_rate": 4.103430543267994e-06,
      "loss": 0.2825,
      "step": 910
    },
    {
      "epoch": 1.4067278287461773,
      "grad_norm": 23.051288604736328,
      "learning_rate": 4.079377609251183e-06,
      "loss": 0.194,
      "step": 920
    },
    {
      "epoch": 1.4220183486238533,
      "grad_norm": 10.35480785369873,
      "learning_rate": 4.055324675234372e-06,
      "loss": 0.2528,
      "step": 930
    },
    {
      "epoch": 1.4373088685015292,
      "grad_norm": 4.950748443603516,
      "learning_rate": 4.031271741217561e-06,
      "loss": 0.3389,
      "step": 940
    },
    {
      "epoch": 1.452599388379205,
      "grad_norm": 0.8531694412231445,
      "learning_rate": 4.00721880720075e-06,
      "loss": 0.2006,
      "step": 950
    },
    {
      "epoch": 1.4678899082568808,
      "grad_norm": 5.678575038909912,
      "learning_rate": 3.983165873183939e-06,
      "loss": 0.1901,
      "step": 960
    },
    {
      "epoch": 1.4831804281345566,
      "grad_norm": 15.318537712097168,
      "learning_rate": 3.959112939167128e-06,
      "loss": 0.6674,
      "step": 970
    },
    {
      "epoch": 1.4984709480122325,
      "grad_norm": 17.652233123779297,
      "learning_rate": 3.9350600051503165e-06,
      "loss": 0.1986,
      "step": 980
    },
    {
      "epoch": 1.5137614678899083,
      "grad_norm": 0.26174286007881165,
      "learning_rate": 3.9110070711335045e-06,
      "loss": 0.267,
      "step": 990
    },
    {
      "epoch": 1.529051987767584,
      "grad_norm": 9.704527854919434,
      "learning_rate": 3.886954137116693e-06,
      "loss": 0.3589,
      "step": 1000
    },
    {
      "epoch": 1.54434250764526,
      "grad_norm": 22.791322708129883,
      "learning_rate": 3.862901203099882e-06,
      "loss": 0.3213,
      "step": 1010
    },
    {
      "epoch": 1.5596330275229358,
      "grad_norm": 7.6159281730651855,
      "learning_rate": 3.838848269083071e-06,
      "loss": 0.3337,
      "step": 1020
    },
    {
      "epoch": 1.5749235474006116,
      "grad_norm": 28.031082153320312,
      "learning_rate": 3.81479533506626e-06,
      "loss": 0.3376,
      "step": 1030
    },
    {
      "epoch": 1.5902140672782874,
      "grad_norm": 20.402673721313477,
      "learning_rate": 3.790742401049449e-06,
      "loss": 0.3356,
      "step": 1040
    },
    {
      "epoch": 1.6055045871559632,
      "grad_norm": 2.4617958068847656,
      "learning_rate": 3.7666894670326378e-06,
      "loss": 0.2407,
      "step": 1050
    },
    {
      "epoch": 1.620795107033639,
      "grad_norm": 25.68115234375,
      "learning_rate": 3.7426365330158266e-06,
      "loss": 0.2576,
      "step": 1060
    },
    {
      "epoch": 1.6360856269113149,
      "grad_norm": 0.6402398347854614,
      "learning_rate": 3.7185835989990155e-06,
      "loss": 0.2803,
      "step": 1070
    },
    {
      "epoch": 1.6513761467889907,
      "grad_norm": 2.904221296310425,
      "learning_rate": 3.694530664982204e-06,
      "loss": 0.2053,
      "step": 1080
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 29.700824737548828,
      "learning_rate": 3.670477730965393e-06,
      "loss": 0.463,
      "step": 1090
    },
    {
      "epoch": 1.6819571865443423,
      "grad_norm": 5.498286724090576,
      "learning_rate": 3.6464247969485817e-06,
      "loss": 0.2744,
      "step": 1100
    },
    {
      "epoch": 1.6972477064220184,
      "grad_norm": 15.300117492675781,
      "learning_rate": 3.6223718629317706e-06,
      "loss": 0.2365,
      "step": 1110
    },
    {
      "epoch": 1.7125382262996942,
      "grad_norm": 17.502981185913086,
      "learning_rate": 3.5983189289149595e-06,
      "loss": 0.5628,
      "step": 1120
    },
    {
      "epoch": 1.72782874617737,
      "grad_norm": 9.557957649230957,
      "learning_rate": 3.5742659948981483e-06,
      "loss": 0.2042,
      "step": 1130
    },
    {
      "epoch": 1.7431192660550459,
      "grad_norm": 13.202752113342285,
      "learning_rate": 3.5502130608813364e-06,
      "loss": 0.2895,
      "step": 1140
    },
    {
      "epoch": 1.7584097859327217,
      "grad_norm": 10.660998344421387,
      "learning_rate": 3.5261601268645252e-06,
      "loss": 0.2851,
      "step": 1150
    },
    {
      "epoch": 1.7737003058103975,
      "grad_norm": 38.87590408325195,
      "learning_rate": 3.5021071928477137e-06,
      "loss": 0.2375,
      "step": 1160
    },
    {
      "epoch": 1.7889908256880735,
      "grad_norm": 13.82657241821289,
      "learning_rate": 3.4780542588309026e-06,
      "loss": 0.3832,
      "step": 1170
    },
    {
      "epoch": 1.8042813455657494,
      "grad_norm": 28.141618728637695,
      "learning_rate": 3.4540013248140914e-06,
      "loss": 0.2534,
      "step": 1180
    },
    {
      "epoch": 1.8195718654434252,
      "grad_norm": 2.6695139408111572,
      "learning_rate": 3.4299483907972803e-06,
      "loss": 0.2649,
      "step": 1190
    },
    {
      "epoch": 1.834862385321101,
      "grad_norm": 1.363698959350586,
      "learning_rate": 3.405895456780469e-06,
      "loss": 0.2242,
      "step": 1200
    },
    {
      "epoch": 1.8501529051987768,
      "grad_norm": 28.01278305053711,
      "learning_rate": 3.381842522763658e-06,
      "loss": 0.3021,
      "step": 1210
    },
    {
      "epoch": 1.8654434250764527,
      "grad_norm": 31.1824951171875,
      "learning_rate": 3.3577895887468465e-06,
      "loss": 0.2872,
      "step": 1220
    },
    {
      "epoch": 1.8807339449541285,
      "grad_norm": 24.482133865356445,
      "learning_rate": 3.3337366547300354e-06,
      "loss": 0.3046,
      "step": 1230
    },
    {
      "epoch": 1.8960244648318043,
      "grad_norm": 17.4123592376709,
      "learning_rate": 3.3096837207132243e-06,
      "loss": 0.2265,
      "step": 1240
    },
    {
      "epoch": 1.9113149847094801,
      "grad_norm": 5.79738187789917,
      "learning_rate": 3.285630786696413e-06,
      "loss": 0.2958,
      "step": 1250
    },
    {
      "epoch": 1.926605504587156,
      "grad_norm": 12.928561210632324,
      "learning_rate": 3.261577852679602e-06,
      "loss": 0.3954,
      "step": 1260
    },
    {
      "epoch": 1.9418960244648318,
      "grad_norm": 0.2753223180770874,
      "learning_rate": 3.237524918662791e-06,
      "loss": 0.1626,
      "step": 1270
    },
    {
      "epoch": 1.9571865443425076,
      "grad_norm": 1.013130784034729,
      "learning_rate": 3.2134719846459793e-06,
      "loss": 0.3307,
      "step": 1280
    },
    {
      "epoch": 1.9724770642201834,
      "grad_norm": 8.603392601013184,
      "learning_rate": 3.1894190506291682e-06,
      "loss": 0.1127,
      "step": 1290
    },
    {
      "epoch": 1.9877675840978593,
      "grad_norm": 17.115859985351562,
      "learning_rate": 3.165366116612357e-06,
      "loss": 0.4232,
      "step": 1300
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.881407804131599,
      "eval_f1": 0.7543581616481775,
      "eval_loss": 0.35450226068496704,
      "eval_precision": 0.7460815047021944,
      "eval_recall": 0.7628205128205128,
      "eval_runtime": 4.8961,
      "eval_samples_per_second": 266.945,
      "eval_steps_per_second": 16.748,
      "step": 1308
    }
  ],
  "logging_steps": 10,
  "max_steps": 2616,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1377131076427776.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": {
    "learning_rate": 6.292247538797816e-06,
    "num_train_epochs": 4,
    "per_device_train_batch_size": 8,
    "seed": 16
  }
}