{
  "best_metric": 0.633788037775446,
  "best_model_checkpoint": "/media/cse/HDD/Shawon/shawon/Timesformer_tuning_BdSLW60_divided/checkpoint-1754",
  "epoch": 2.1990867579908677,
  "eval_steps": 500,
  "global_step": 2190,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0045662100456621,
      "grad_norm": 18.209203720092773,
      "learning_rate": 2.2831050228310503e-06,
      "loss": 4.331,
      "step": 10
    },
    {
      "epoch": 0.0091324200913242,
      "grad_norm": 12.897218704223633,
      "learning_rate": 4.566210045662101e-06,
      "loss": 4.1036,
      "step": 20
    },
    {
      "epoch": 0.0136986301369863,
      "grad_norm": 12.286799430847168,
      "learning_rate": 6.849315068493151e-06,
      "loss": 4.2229,
      "step": 30
    },
    {
      "epoch": 0.0182648401826484,
      "grad_norm": 12.371978759765625,
      "learning_rate": 9.132420091324201e-06,
      "loss": 4.1291,
      "step": 40
    },
    {
      "epoch": 0.0228310502283105,
      "grad_norm": 12.65040397644043,
      "learning_rate": 1.1415525114155251e-05,
      "loss": 4.1588,
      "step": 50
    },
    {
      "epoch": 0.0273972602739726,
      "grad_norm": 12.903022766113281,
      "learning_rate": 1.3698630136986302e-05,
      "loss": 4.0512,
      "step": 60
    },
    {
      "epoch": 0.0319634703196347,
      "grad_norm": 12.025260925292969,
      "learning_rate": 1.5981735159817352e-05,
      "loss": 3.8854,
      "step": 70
    },
    {
      "epoch": 0.0365296803652968,
      "grad_norm": 12.173115730285645,
      "learning_rate": 1.8264840182648402e-05,
      "loss": 3.9969,
      "step": 80
    },
    {
      "epoch": 0.0410958904109589,
      "grad_norm": 13.235687255859375,
      "learning_rate": 2.0547945205479453e-05,
      "loss": 3.8276,
      "step": 90
    },
    {
      "epoch": 0.045662100456621,
      "grad_norm": 12.253507614135742,
      "learning_rate": 2.2831050228310503e-05,
      "loss": 3.7803,
      "step": 100
    },
    {
      "epoch": 0.0502283105022831,
      "grad_norm": 12.216194152832031,
      "learning_rate": 2.5114155251141553e-05,
      "loss": 3.6321,
      "step": 110
    },
    {
      "epoch": 0.0547945205479452,
      "grad_norm": 13.399759292602539,
      "learning_rate": 2.7397260273972603e-05,
      "loss": 3.4251,
      "step": 120
    },
    {
      "epoch": 0.0593607305936073,
      "grad_norm": 12.781786918640137,
      "learning_rate": 2.9680365296803654e-05,
      "loss": 3.2993,
      "step": 130
    },
    {
      "epoch": 0.0639269406392694,
      "grad_norm": 12.670400619506836,
      "learning_rate": 3.1963470319634704e-05,
      "loss": 3.2826,
      "step": 140
    },
    {
      "epoch": 0.0684931506849315,
      "grad_norm": 13.571783065795898,
      "learning_rate": 3.424657534246575e-05,
      "loss": 3.1383,
      "step": 150
    },
    {
      "epoch": 0.0730593607305936,
      "grad_norm": 13.101773262023926,
      "learning_rate": 3.6529680365296805e-05,
      "loss": 2.9979,
      "step": 160
    },
    {
      "epoch": 0.0776255707762557,
      "grad_norm": 12.823298454284668,
      "learning_rate": 3.881278538812785e-05,
      "loss": 2.8539,
      "step": 170
    },
    {
      "epoch": 0.0821917808219178,
      "grad_norm": 13.807844161987305,
      "learning_rate": 4.1095890410958905e-05,
      "loss": 2.9661,
      "step": 180
    },
    {
      "epoch": 0.0867579908675799,
      "grad_norm": 13.598432540893555,
      "learning_rate": 4.337899543378995e-05,
      "loss": 2.7239,
      "step": 190
    },
    {
      "epoch": 0.091324200913242,
      "grad_norm": 13.286529541015625,
      "learning_rate": 4.5662100456621006e-05,
      "loss": 2.4308,
      "step": 200
    },
    {
      "epoch": 0.0958904109589041,
      "grad_norm": 12.907256126403809,
      "learning_rate": 4.794520547945205e-05,
      "loss": 2.3219,
      "step": 210
    },
    {
      "epoch": 0.1004566210045662,
      "grad_norm": 12.69262409210205,
      "learning_rate": 4.997463216641299e-05,
      "loss": 2.0315,
      "step": 220
    },
    {
      "epoch": 0.1050228310502283,
      "grad_norm": 11.776590347290039,
      "learning_rate": 4.9720953830542875e-05,
      "loss": 2.4161,
      "step": 230
    },
    {
      "epoch": 0.1095890410958904,
      "grad_norm": 12.577189445495605,
      "learning_rate": 4.9467275494672755e-05,
      "loss": 1.9662,
      "step": 240
    },
    {
      "epoch": 0.1141552511415525,
      "grad_norm": 11.328509330749512,
      "learning_rate": 4.9213597158802636e-05,
      "loss": 1.9721,
      "step": 250
    },
    {
      "epoch": 0.1187214611872146,
      "grad_norm": 13.56044864654541,
      "learning_rate": 4.895991882293252e-05,
      "loss": 1.6344,
      "step": 260
    },
    {
      "epoch": 0.1232876712328767,
      "grad_norm": 10.372136116027832,
      "learning_rate": 4.8706240487062404e-05,
      "loss": 1.5725,
      "step": 270
    },
    {
      "epoch": 0.1278538812785388,
      "grad_norm": 9.894343376159668,
      "learning_rate": 4.8452562151192285e-05,
      "loss": 1.7684,
      "step": 280
    },
    {
      "epoch": 0.1324200913242009,
      "grad_norm": 9.916196823120117,
      "learning_rate": 4.819888381532217e-05,
      "loss": 1.7534,
      "step": 290
    },
    {
      "epoch": 0.136986301369863,
      "grad_norm": 10.576798439025879,
      "learning_rate": 4.794520547945205e-05,
      "loss": 1.5195,
      "step": 300
    },
    {
      "epoch": 0.1415525114155251,
      "grad_norm": 11.507007598876953,
      "learning_rate": 4.769152714358194e-05,
      "loss": 1.671,
      "step": 310
    },
    {
      "epoch": 0.1461187214611872,
      "grad_norm": 9.821493148803711,
      "learning_rate": 4.743784880771182e-05,
      "loss": 1.2848,
      "step": 320
    },
    {
      "epoch": 0.1506849315068493,
      "grad_norm": 8.048973083496094,
      "learning_rate": 4.71841704718417e-05,
      "loss": 1.3263,
      "step": 330
    },
    {
      "epoch": 0.1552511415525114,
      "grad_norm": 11.427051544189453,
      "learning_rate": 4.693049213597159e-05,
      "loss": 1.3725,
      "step": 340
    },
    {
      "epoch": 0.1598173515981735,
      "grad_norm": 11.1187744140625,
      "learning_rate": 4.667681380010147e-05,
      "loss": 1.3067,
      "step": 350
    },
    {
      "epoch": 0.1643835616438356,
      "grad_norm": 13.000616073608398,
      "learning_rate": 4.642313546423136e-05,
      "loss": 1.296,
      "step": 360
    },
    {
      "epoch": 0.1689497716894977,
      "grad_norm": 8.552903175354004,
      "learning_rate": 4.616945712836124e-05,
      "loss": 1.1536,
      "step": 370
    },
    {
      "epoch": 0.1735159817351598,
      "grad_norm": 8.447358131408691,
      "learning_rate": 4.591577879249112e-05,
      "loss": 1.1327,
      "step": 380
    },
    {
      "epoch": 0.1780821917808219,
      "grad_norm": 11.575722694396973,
      "learning_rate": 4.5662100456621006e-05,
      "loss": 1.1091,
      "step": 390
    },
    {
      "epoch": 0.182648401826484,
      "grad_norm": 10.289594650268555,
      "learning_rate": 4.5408422120750886e-05,
      "loss": 1.1887,
      "step": 400
    },
    {
      "epoch": 0.1872146118721461,
      "grad_norm": 10.607954025268555,
      "learning_rate": 4.5154743784880774e-05,
      "loss": 0.8785,
      "step": 410
    },
    {
      "epoch": 0.1917808219178082,
      "grad_norm": 8.302620887756348,
      "learning_rate": 4.4901065449010655e-05,
      "loss": 0.9114,
      "step": 420
    },
    {
      "epoch": 0.1963470319634703,
      "grad_norm": 10.900007247924805,
      "learning_rate": 4.4647387113140535e-05,
      "loss": 1.014,
      "step": 430
    },
    {
      "epoch": 0.2009132420091324,
      "grad_norm": 11.343582153320312,
      "learning_rate": 4.439370877727042e-05,
      "loss": 0.761,
      "step": 440
    },
    {
      "epoch": 0.2054794520547945,
      "grad_norm": 7.5472092628479,
      "learning_rate": 4.41400304414003e-05,
      "loss": 1.02,
      "step": 450
    },
    {
      "epoch": 0.2100456621004566,
      "grad_norm": 6.482777118682861,
      "learning_rate": 4.3886352105530184e-05,
      "loss": 0.7821,
      "step": 460
    },
    {
      "epoch": 0.2146118721461187,
      "grad_norm": 8.665425300598145,
      "learning_rate": 4.363267376966007e-05,
      "loss": 0.7712,
      "step": 470
    },
    {
      "epoch": 0.2191780821917808,
      "grad_norm": 6.963949203491211,
      "learning_rate": 4.337899543378995e-05,
      "loss": 0.741,
      "step": 480
    },
    {
      "epoch": 0.2237442922374429,
      "grad_norm": 7.155235290527344,
      "learning_rate": 4.312531709791984e-05,
      "loss": 0.8244,
      "step": 490
    },
    {
      "epoch": 0.228310502283105,
      "grad_norm": 7.2470197677612305,
      "learning_rate": 4.287163876204972e-05,
      "loss": 0.7631,
      "step": 500
    },
    {
      "epoch": 0.2328767123287671,
      "grad_norm": 7.359896659851074,
      "learning_rate": 4.26179604261796e-05,
      "loss": 0.6634,
      "step": 510
    },
    {
      "epoch": 0.2374429223744292,
      "grad_norm": 9.313497543334961,
      "learning_rate": 4.236428209030949e-05,
      "loss": 0.7052,
      "step": 520
    },
    {
      "epoch": 0.2420091324200913,
      "grad_norm": 7.674174785614014,
      "learning_rate": 4.211060375443937e-05,
      "loss": 0.5931,
      "step": 530
    },
    {
      "epoch": 0.2465753424657534,
      "grad_norm": 6.553604602813721,
      "learning_rate": 4.1856925418569256e-05,
      "loss": 0.6904,
      "step": 540
    },
    {
      "epoch": 0.2511415525114155,
      "grad_norm": 11.845129013061523,
      "learning_rate": 4.160324708269914e-05,
      "loss": 0.7885,
      "step": 550
    },
    {
      "epoch": 0.2557077625570776,
      "grad_norm": 9.967060089111328,
      "learning_rate": 4.134956874682902e-05,
      "loss": 0.9233,
      "step": 560
    },
    {
      "epoch": 0.2602739726027397,
      "grad_norm": 10.10986042022705,
      "learning_rate": 4.1095890410958905e-05,
      "loss": 0.4369,
      "step": 570
    },
    {
      "epoch": 0.2648401826484018,
      "grad_norm": 6.26952600479126,
      "learning_rate": 4.0842212075088786e-05,
      "loss": 0.5587,
      "step": 580
    },
    {
      "epoch": 0.2694063926940639,
      "grad_norm": 10.414834022521973,
      "learning_rate": 4.058853373921867e-05,
      "loss": 0.6551,
      "step": 590
    },
    {
      "epoch": 0.273972602739726,
      "grad_norm": 6.288766384124756,
      "learning_rate": 4.0334855403348554e-05,
      "loss": 0.6568,
      "step": 600
    },
    {
      "epoch": 0.2785388127853881,
      "grad_norm": 5.77100944519043,
      "learning_rate": 4.0081177067478435e-05,
      "loss": 0.5473,
      "step": 610
    },
    {
      "epoch": 0.2831050228310502,
      "grad_norm": 6.018027305603027,
      "learning_rate": 3.982749873160832e-05,
      "loss": 0.5149,
      "step": 620
    },
    {
      "epoch": 0.2876712328767123,
      "grad_norm": 8.372048377990723,
      "learning_rate": 3.95738203957382e-05,
      "loss": 0.4584,
      "step": 630
    },
    {
      "epoch": 0.2922374429223744,
      "grad_norm": 7.889549732208252,
      "learning_rate": 3.932014205986809e-05,
      "loss": 0.7308,
      "step": 640
    },
    {
      "epoch": 0.2968036529680365,
      "grad_norm": 9.812642097473145,
      "learning_rate": 3.906646372399797e-05,
      "loss": 0.3976,
      "step": 650
    },
    {
      "epoch": 0.3013698630136986,
      "grad_norm": 6.291619300842285,
      "learning_rate": 3.881278538812785e-05,
      "loss": 0.4715,
      "step": 660
    },
    {
      "epoch": 0.3059360730593607,
      "grad_norm": 11.111459732055664,
      "learning_rate": 3.855910705225774e-05,
      "loss": 0.5013,
      "step": 670
    },
    {
      "epoch": 0.3105022831050228,
      "grad_norm": 11.028929710388184,
      "learning_rate": 3.830542871638762e-05,
      "loss": 0.4116,
      "step": 680
    },
    {
      "epoch": 0.3150684931506849,
      "grad_norm": 10.91637897491455,
      "learning_rate": 3.80517503805175e-05,
      "loss": 0.6619,
      "step": 690
    },
    {
      "epoch": 0.319634703196347,
      "grad_norm": 5.09431791305542,
      "learning_rate": 3.779807204464739e-05,
      "loss": 0.5631,
      "step": 700
    },
    {
      "epoch": 0.3242009132420091,
      "grad_norm": 7.607274532318115,
      "learning_rate": 3.754439370877727e-05,
      "loss": 0.7068,
      "step": 710
    },
    {
      "epoch": 0.3287671232876712,
      "grad_norm": 5.097145080566406,
      "learning_rate": 3.7290715372907156e-05,
      "loss": 0.5534,
      "step": 720
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 4.283755779266357,
      "learning_rate": 3.7037037037037037e-05,
      "loss": 0.3874,
      "step": 730
    },
    {
      "epoch": 0.3378995433789954,
      "grad_norm": 8.169587135314941,
      "learning_rate": 3.678335870116692e-05,
      "loss": 0.3478,
      "step": 740
    },
    {
      "epoch": 0.3424657534246575,
      "grad_norm": 9.869154930114746,
      "learning_rate": 3.6529680365296805e-05,
      "loss": 0.5266,
      "step": 750
    },
    {
      "epoch": 0.3470319634703196,
      "grad_norm": 2.9827332496643066,
      "learning_rate": 3.6276002029426685e-05,
      "loss": 0.4598,
      "step": 760
    },
    {
      "epoch": 0.3515981735159817,
      "grad_norm": 5.095430850982666,
      "learning_rate": 3.602232369355657e-05,
      "loss": 0.4072,
      "step": 770
    },
    {
      "epoch": 0.3561643835616438,
      "grad_norm": 5.7128119468688965,
      "learning_rate": 3.5768645357686453e-05,
      "loss": 0.4988,
      "step": 780
    },
    {
      "epoch": 0.3607305936073059,
      "grad_norm": 7.71494722366333,
      "learning_rate": 3.5514967021816334e-05,
      "loss": 0.5598,
      "step": 790
    },
    {
      "epoch": 0.365296803652968,
      "grad_norm": 3.576751470565796,
      "learning_rate": 3.526128868594622e-05,
      "loss": 0.5077,
      "step": 800
    },
    {
      "epoch": 0.3698630136986301,
      "grad_norm": 3.606416702270508,
      "learning_rate": 3.50076103500761e-05,
      "loss": 0.3295,
      "step": 810
    },
    {
      "epoch": 0.3744292237442922,
      "grad_norm": 3.141601324081421,
      "learning_rate": 3.475393201420599e-05,
      "loss": 0.318,
      "step": 820
    },
    {
      "epoch": 0.3789954337899543,
      "grad_norm": 3.1620349884033203,
      "learning_rate": 3.450025367833587e-05,
      "loss": 0.4301,
      "step": 830
    },
    {
      "epoch": 0.3835616438356164,
      "grad_norm": 7.8440985679626465,
      "learning_rate": 3.424657534246575e-05,
      "loss": 0.3342,
      "step": 840
    },
    {
      "epoch": 0.3881278538812785,
      "grad_norm": 4.066171169281006,
      "learning_rate": 3.399289700659564e-05,
      "loss": 0.3352,
      "step": 850
    },
    {
      "epoch": 0.3926940639269406,
      "grad_norm": 4.03795051574707,
      "learning_rate": 3.373921867072552e-05,
      "loss": 0.5177,
      "step": 860
    },
    {
      "epoch": 0.3972602739726027,
      "grad_norm": 9.911918640136719,
      "learning_rate": 3.34855403348554e-05,
      "loss": 0.3063,
      "step": 870
    },
    {
      "epoch": 0.4004566210045662,
      "eval_accuracy": 0.6012591815320042,
      "eval_loss": 1.4254487752914429,
      "eval_runtime": 240.276,
      "eval_samples_per_second": 3.966,
      "eval_steps_per_second": 1.985,
      "step": 877
    },
    {
      "epoch": 1.0013698630136987,
      "grad_norm": 4.910429000854492,
      "learning_rate": 3.323186199898529e-05,
      "loss": 0.3396,
      "step": 880
    },
    {
      "epoch": 1.0059360730593607,
      "grad_norm": 11.615983963012695,
      "learning_rate": 3.297818366311517e-05,
      "loss": 0.2476,
      "step": 890
    },
    {
      "epoch": 1.0105022831050228,
      "grad_norm": 5.976046562194824,
      "learning_rate": 3.2724505327245055e-05,
      "loss": 0.267,
      "step": 900
    },
    {
      "epoch": 1.015068493150685,
      "grad_norm": 11.546513557434082,
      "learning_rate": 3.2470826991374936e-05,
      "loss": 0.5572,
      "step": 910
    },
    {
      "epoch": 1.019634703196347,
      "grad_norm": 4.033244609832764,
      "learning_rate": 3.221714865550482e-05,
      "loss": 0.2918,
      "step": 920
    },
    {
      "epoch": 1.0242009132420091,
      "grad_norm": 1.627201795578003,
      "learning_rate": 3.1963470319634704e-05,
      "loss": 0.2512,
      "step": 930
    },
    {
      "epoch": 1.0287671232876712,
      "grad_norm": 3.9416964054107666,
      "learning_rate": 3.1709791983764585e-05,
      "loss": 0.4496,
      "step": 940
    },
    {
      "epoch": 1.0333333333333334,
      "grad_norm": 8.648749351501465,
      "learning_rate": 3.145611364789447e-05,
      "loss": 0.4884,
      "step": 950
    },
    {
      "epoch": 1.0378995433789955,
      "grad_norm": 5.311582088470459,
      "learning_rate": 3.120243531202435e-05,
      "loss": 0.2819,
      "step": 960
    },
    {
      "epoch": 1.0424657534246575,
      "grad_norm": 5.9733757972717285,
      "learning_rate": 3.0948756976154234e-05,
      "loss": 0.372,
      "step": 970
    },
    {
      "epoch": 1.0470319634703196,
      "grad_norm": 10.992047309875488,
      "learning_rate": 3.069507864028412e-05,
      "loss": 0.4749,
      "step": 980
    },
    {
      "epoch": 1.0515981735159818,
      "grad_norm": 2.805821180343628,
      "learning_rate": 3.0441400304414e-05,
      "loss": 0.2255,
      "step": 990
    },
    {
      "epoch": 1.0561643835616439,
      "grad_norm": 1.54830002784729,
      "learning_rate": 3.0187721968543886e-05,
      "loss": 0.4999,
      "step": 1000
    },
    {
      "epoch": 1.060730593607306,
      "grad_norm": 4.090977191925049,
      "learning_rate": 2.993404363267377e-05,
      "loss": 0.3041,
      "step": 1010
    },
    {
      "epoch": 1.065296803652968,
      "grad_norm": 13.388101577758789,
      "learning_rate": 2.9680365296803654e-05,
      "loss": 0.3855,
      "step": 1020
    },
    {
      "epoch": 1.0698630136986302,
      "grad_norm": 1.3952845335006714,
      "learning_rate": 2.9426686960933534e-05,
      "loss": 0.2041,
      "step": 1030
    },
    {
      "epoch": 1.0744292237442923,
      "grad_norm": 1.3922096490859985,
      "learning_rate": 2.917300862506342e-05,
      "loss": 0.2754,
      "step": 1040
    },
    {
      "epoch": 1.0789954337899543,
      "grad_norm": 5.259091854095459,
      "learning_rate": 2.8919330289193303e-05,
      "loss": 0.4592,
      "step": 1050
    },
    {
      "epoch": 1.0835616438356164,
      "grad_norm": 8.525176048278809,
      "learning_rate": 2.8665651953323187e-05,
      "loss": 0.1942,
      "step": 1060
    },
    {
      "epoch": 1.0881278538812786,
      "grad_norm": 5.207125186920166,
      "learning_rate": 2.841197361745307e-05,
      "loss": 0.3025,
      "step": 1070
    },
    {
      "epoch": 1.0926940639269407,
      "grad_norm": 1.9372700452804565,
      "learning_rate": 2.815829528158295e-05,
      "loss": 0.3472,
      "step": 1080
    },
    {
      "epoch": 1.0972602739726027,
      "grad_norm": 12.879546165466309,
      "learning_rate": 2.7904616945712835e-05,
      "loss": 0.3996,
      "step": 1090
    },
    {
      "epoch": 1.1018264840182648,
      "grad_norm": 2.0068254470825195,
      "learning_rate": 2.765093860984272e-05,
      "loss": 0.3653,
      "step": 1100
    },
    {
      "epoch": 1.106392694063927,
      "grad_norm": 4.621579647064209,
      "learning_rate": 2.7397260273972603e-05,
      "loss": 0.3694,
      "step": 1110
    },
    {
      "epoch": 1.110958904109589,
      "grad_norm": 5.310023784637451,
      "learning_rate": 2.7143581938102484e-05,
      "loss": 0.2811,
      "step": 1120
    },
    {
      "epoch": 1.1155251141552511,
      "grad_norm": 2.6011314392089844,
      "learning_rate": 2.6889903602232368e-05,
      "loss": 0.2318,
      "step": 1130
    },
    {
      "epoch": 1.1200913242009132,
      "grad_norm": 16.11932945251465,
      "learning_rate": 2.6636225266362252e-05,
      "loss": 0.4196,
      "step": 1140
    },
    {
      "epoch": 1.1246575342465754,
      "grad_norm": 10.792712211608887,
      "learning_rate": 2.6382546930492136e-05,
      "loss": 0.3332,
      "step": 1150
    },
    {
      "epoch": 1.1292237442922375,
      "grad_norm": 0.7183528542518616,
      "learning_rate": 2.612886859462202e-05,
      "loss": 0.1916,
      "step": 1160
    },
    {
      "epoch": 1.1337899543378995,
      "grad_norm": 8.360386848449707,
      "learning_rate": 2.58751902587519e-05,
      "loss": 0.2133,
      "step": 1170
    },
    {
      "epoch": 1.1383561643835616,
      "grad_norm": 9.3360013961792,
      "learning_rate": 2.5621511922881785e-05,
      "loss": 0.1756,
      "step": 1180
    },
    {
      "epoch": 1.1429223744292236,
      "grad_norm": 3.6500494480133057,
      "learning_rate": 2.536783358701167e-05,
      "loss": 0.2249,
      "step": 1190
    },
    {
      "epoch": 1.1474885844748859,
      "grad_norm": 13.495393753051758,
      "learning_rate": 2.5114155251141553e-05,
      "loss": 0.4326,
      "step": 1200
    },
    {
      "epoch": 1.152054794520548,
      "grad_norm": 1.119050145149231,
      "learning_rate": 2.4860476915271437e-05,
      "loss": 0.2966,
      "step": 1210
    },
    {
      "epoch": 1.15662100456621,
      "grad_norm": 0.8353846073150635,
      "learning_rate": 2.4606798579401318e-05,
      "loss": 0.1481,
      "step": 1220
    },
    {
      "epoch": 1.1611872146118722,
      "grad_norm": 4.413615703582764,
      "learning_rate": 2.4353120243531202e-05,
      "loss": 0.1566,
      "step": 1230
    },
    {
      "epoch": 1.1657534246575343,
      "grad_norm": 3.4680347442626953,
      "learning_rate": 2.4099441907661086e-05,
      "loss": 0.2691,
      "step": 1240
    },
    {
      "epoch": 1.1703196347031963,
      "grad_norm": 2.294340133666992,
      "learning_rate": 2.384576357179097e-05,
      "loss": 0.2126,
      "step": 1250
    },
    {
      "epoch": 1.1748858447488584,
      "grad_norm": 0.6727573275566101,
      "learning_rate": 2.359208523592085e-05,
      "loss": 0.2691,
      "step": 1260
    },
    {
      "epoch": 1.1794520547945206,
      "grad_norm": 4.724236488342285,
      "learning_rate": 2.3338406900050735e-05,
      "loss": 0.242,
      "step": 1270
    },
    {
      "epoch": 1.1840182648401827,
      "grad_norm": 2.012953519821167,
      "learning_rate": 2.308472856418062e-05,
      "loss": 0.2324,
      "step": 1280
    },
    {
      "epoch": 1.1885844748858447,
      "grad_norm": 5.335368633270264,
      "learning_rate": 2.2831050228310503e-05,
      "loss": 0.2389,
      "step": 1290
    },
    {
      "epoch": 1.1931506849315068,
      "grad_norm": 8.303133010864258,
      "learning_rate": 2.2577371892440387e-05,
      "loss": 0.2827,
      "step": 1300
    },
    {
      "epoch": 1.197716894977169,
      "grad_norm": 4.6699299812316895,
      "learning_rate": 2.2323693556570268e-05,
      "loss": 0.2348,
      "step": 1310
    },
    {
      "epoch": 1.202283105022831,
      "grad_norm": 3.6652133464813232,
      "learning_rate": 2.207001522070015e-05,
      "loss": 0.2747,
      "step": 1320
    },
    {
      "epoch": 1.2068493150684931,
      "grad_norm": 5.8564934730529785,
      "learning_rate": 2.1816336884830036e-05,
      "loss": 0.3638,
      "step": 1330
    },
    {
      "epoch": 1.2114155251141552,
      "grad_norm": 2.1238956451416016,
      "learning_rate": 2.156265854895992e-05,
      "loss": 0.142,
      "step": 1340
    },
    {
      "epoch": 1.2159817351598174,
      "grad_norm": 7.869147777557373,
      "learning_rate": 2.13089802130898e-05,
      "loss": 0.1685,
      "step": 1350
    },
    {
      "epoch": 1.2205479452054795,
      "grad_norm": 4.023821830749512,
      "learning_rate": 2.1055301877219685e-05,
      "loss": 0.1917,
      "step": 1360
    },
    {
      "epoch": 1.2251141552511415,
      "grad_norm": 12.940045356750488,
      "learning_rate": 2.080162354134957e-05,
      "loss": 0.2935,
      "step": 1370
    },
    {
      "epoch": 1.2296803652968036,
      "grad_norm": 7.166123867034912,
      "learning_rate": 2.0547945205479453e-05,
      "loss": 0.2112,
      "step": 1380
    },
    {
      "epoch": 1.2342465753424658,
      "grad_norm": 16.69542121887207,
      "learning_rate": 2.0294266869609337e-05,
      "loss": 0.1751,
      "step": 1390
    },
    {
      "epoch": 1.238812785388128,
      "grad_norm": 7.8027448654174805,
      "learning_rate": 2.0040588533739217e-05,
      "loss": 0.2554,
      "step": 1400
    },
    {
      "epoch": 1.24337899543379,
      "grad_norm": 4.178590774536133,
      "learning_rate": 1.97869101978691e-05,
      "loss": 0.2078,
      "step": 1410
    },
    {
      "epoch": 1.247945205479452,
      "grad_norm": 7.469815731048584,
      "learning_rate": 1.9533231861998985e-05,
      "loss": 0.3719,
      "step": 1420
    },
    {
      "epoch": 1.252511415525114,
      "grad_norm": 8.0973482131958,
      "learning_rate": 1.927955352612887e-05,
      "loss": 0.2117,
      "step": 1430
    },
    {
      "epoch": 1.2570776255707763,
      "grad_norm": 7.601376533508301,
      "learning_rate": 1.902587519025875e-05,
      "loss": 0.3092,
      "step": 1440
    },
    {
      "epoch": 1.2616438356164383,
      "grad_norm": 0.7640622854232788,
      "learning_rate": 1.8772196854388634e-05,
      "loss": 0.0979,
      "step": 1450
    },
    {
      "epoch": 1.2662100456621004,
      "grad_norm": 2.080054521560669,
      "learning_rate": 1.8518518518518518e-05,
      "loss": 0.2393,
      "step": 1460
    },
    {
      "epoch": 1.2707762557077626,
      "grad_norm": 11.497714042663574,
      "learning_rate": 1.8264840182648402e-05,
      "loss": 0.2465,
      "step": 1470
    },
    {
      "epoch": 1.2753424657534247,
      "grad_norm": 4.990231513977051,
      "learning_rate": 1.8011161846778286e-05,
      "loss": 0.1007,
      "step": 1480
    },
    {
      "epoch": 1.2799086757990867,
      "grad_norm": 3.397449254989624,
      "learning_rate": 1.7757483510908167e-05,
      "loss": 0.2518,
      "step": 1490
    },
    {
      "epoch": 1.2844748858447488,
      "grad_norm": 6.128615379333496,
      "learning_rate": 1.750380517503805e-05,
      "loss": 0.1396,
      "step": 1500
    },
    {
      "epoch": 1.2890410958904108,
      "grad_norm": 10.955172538757324,
      "learning_rate": 1.7250126839167935e-05,
      "loss": 0.1437,
      "step": 1510
    },
    {
      "epoch": 1.293607305936073,
      "grad_norm": 3.4163191318511963,
      "learning_rate": 1.699644850329782e-05,
      "loss": 0.3169,
      "step": 1520
    },
    {
      "epoch": 1.2981735159817351,
      "grad_norm": 0.777817964553833,
      "learning_rate": 1.67427701674277e-05,
      "loss": 0.193,
      "step": 1530
    },
    {
      "epoch": 1.3027397260273972,
      "grad_norm": 6.872499465942383,
      "learning_rate": 1.6489091831557584e-05,
      "loss": 0.2019,
      "step": 1540
    },
    {
      "epoch": 1.3073059360730594,
      "grad_norm": 1.3268109560012817,
      "learning_rate": 1.6235413495687468e-05,
      "loss": 0.1058,
      "step": 1550
    },
    {
      "epoch": 1.3118721461187215,
      "grad_norm": 0.33468785881996155,
      "learning_rate": 1.5981735159817352e-05,
      "loss": 0.2684,
      "step": 1560
    },
    {
      "epoch": 1.3164383561643835,
      "grad_norm": 1.7843540906906128,
      "learning_rate": 1.5728056823947236e-05,
      "loss": 0.3024,
      "step": 1570
    },
    {
      "epoch": 1.3210045662100456,
      "grad_norm": 1.9103947877883911,
      "learning_rate": 1.5474378488077117e-05,
      "loss": 0.3474,
      "step": 1580
    },
    {
      "epoch": 1.3255707762557076,
      "grad_norm": 2.3000149726867676,
      "learning_rate": 1.5220700152207e-05,
      "loss": 0.1792,
      "step": 1590
    },
    {
      "epoch": 1.33013698630137,
      "grad_norm": 0.66684490442276,
      "learning_rate": 1.4967021816336885e-05,
      "loss": 0.239,
      "step": 1600
    },
    {
      "epoch": 1.334703196347032,
      "grad_norm": 0.592566967010498,
      "learning_rate": 1.4713343480466767e-05,
      "loss": 0.3183,
      "step": 1610
    },
    {
      "epoch": 1.339269406392694,
      "grad_norm": 6.709100246429443,
      "learning_rate": 1.4459665144596651e-05,
      "loss": 0.2196,
      "step": 1620
    },
    {
      "epoch": 1.3438356164383563,
      "grad_norm": 0.2524738907814026,
      "learning_rate": 1.4205986808726535e-05,
      "loss": 0.1112,
      "step": 1630
    },
    {
      "epoch": 1.3484018264840183,
      "grad_norm": 0.8381848931312561,
      "learning_rate": 1.3952308472856418e-05,
      "loss": 0.0951,
      "step": 1640
    },
    {
      "epoch": 1.3529680365296803,
      "grad_norm": 6.706734657287598,
      "learning_rate": 1.3698630136986302e-05,
      "loss": 0.2305,
      "step": 1650
    },
    {
      "epoch": 1.3575342465753424,
      "grad_norm": 0.4047660231590271,
      "learning_rate": 1.3444951801116184e-05,
      "loss": 0.2735,
      "step": 1660
    },
    {
      "epoch": 1.3621004566210044,
      "grad_norm": 10.571824073791504,
      "learning_rate": 1.3191273465246068e-05,
      "loss": 0.2175,
      "step": 1670
    },
    {
      "epoch": 1.3666666666666667,
      "grad_norm": 0.39465513825416565,
      "learning_rate": 1.293759512937595e-05,
      "loss": 0.1665,
      "step": 1680
    },
    {
      "epoch": 1.3712328767123287,
      "grad_norm": 5.5752739906311035,
      "learning_rate": 1.2683916793505835e-05,
      "loss": 0.2829,
      "step": 1690
    },
    {
      "epoch": 1.3757990867579908,
      "grad_norm": 0.23495937883853912,
      "learning_rate": 1.2430238457635719e-05,
      "loss": 0.1155,
      "step": 1700
    },
    {
      "epoch": 1.380365296803653,
      "grad_norm": 4.892249584197998,
      "learning_rate": 1.2176560121765601e-05,
      "loss": 0.1875,
      "step": 1710
    },
    {
      "epoch": 1.384931506849315,
      "grad_norm": 6.203292369842529,
      "learning_rate": 1.1922881785895485e-05,
      "loss": 0.198,
      "step": 1720
    },
    {
      "epoch": 1.3894977168949771,
      "grad_norm": 4.22658109664917,
      "learning_rate": 1.1669203450025367e-05,
      "loss": 0.2684,
      "step": 1730
    },
    {
      "epoch": 1.3940639269406392,
      "grad_norm": 1.199587106704712,
      "learning_rate": 1.1415525114155251e-05,
      "loss": 0.1568,
      "step": 1740
    },
    {
      "epoch": 1.3986301369863012,
      "grad_norm": 2.2809574604034424,
      "learning_rate": 1.1161846778285134e-05,
      "loss": 0.1077,
      "step": 1750
    },
    {
      "epoch": 1.4004566210045661,
      "eval_accuracy": 0.633788037775446,
      "eval_loss": 1.3121135234832764,
      "eval_runtime": 244.499,
      "eval_samples_per_second": 3.898,
      "eval_steps_per_second": 1.951,
      "step": 1754
    },
    {
      "epoch": 2.0027397260273974,
      "grad_norm": 1.9538928270339966,
      "learning_rate": 1.0908168442415018e-05,
      "loss": 0.14,
      "step": 1760
    },
    {
      "epoch": 2.007305936073059,
      "grad_norm": 0.5063276290893555,
      "learning_rate": 1.06544901065449e-05,
      "loss": 0.0791,
      "step": 1770
    },
    {
      "epoch": 2.0118721461187214,
      "grad_norm": 10.108781814575195,
      "learning_rate": 1.0400811770674784e-05,
      "loss": 0.1614,
      "step": 1780
    },
    {
      "epoch": 2.0164383561643837,
      "grad_norm": 2.624630928039551,
      "learning_rate": 1.0147133434804668e-05,
      "loss": 0.099,
      "step": 1790
    },
    {
      "epoch": 2.0210045662100455,
      "grad_norm": 0.6380520462989807,
      "learning_rate": 9.89345509893455e-06,
      "loss": 0.2079,
      "step": 1800
    },
    {
      "epoch": 2.025570776255708,
      "grad_norm": 1.0283434391021729,
      "learning_rate": 9.639776763064435e-06,
      "loss": 0.2092,
      "step": 1810
    },
    {
      "epoch": 2.03013698630137,
      "grad_norm": 9.170719146728516,
      "learning_rate": 9.386098427194317e-06,
      "loss": 0.2139,
      "step": 1820
    },
    {
      "epoch": 2.034703196347032,
      "grad_norm": 5.085514068603516,
      "learning_rate": 9.132420091324201e-06,
      "loss": 0.1782,
      "step": 1830
    },
    {
      "epoch": 2.039269406392694,
      "grad_norm": 7.392034530639648,
      "learning_rate": 8.878741755454084e-06,
      "loss": 0.1358,
      "step": 1840
    },
    {
      "epoch": 2.043835616438356,
      "grad_norm": 10.423238754272461,
      "learning_rate": 8.625063419583968e-06,
      "loss": 0.2036,
      "step": 1850
    },
    {
      "epoch": 2.0484018264840183,
      "grad_norm": 5.6207170486450195,
      "learning_rate": 8.37138508371385e-06,
      "loss": 0.089,
      "step": 1860
    },
    {
      "epoch": 2.0529680365296805,
      "grad_norm": 0.16385725140571594,
      "learning_rate": 8.117706747843734e-06,
      "loss": 0.138,
      "step": 1870
    },
    {
      "epoch": 2.0575342465753423,
      "grad_norm": 0.1996384710073471,
      "learning_rate": 7.864028411973618e-06,
      "loss": 0.2014,
      "step": 1880
    },
    {
      "epoch": 2.0621004566210046,
      "grad_norm": 0.25765588879585266,
      "learning_rate": 7.6103500761035e-06,
      "loss": 0.1384,
      "step": 1890
    },
    {
      "epoch": 2.066666666666667,
      "grad_norm": 11.252253532409668,
      "learning_rate": 7.356671740233384e-06,
      "loss": 0.1816,
      "step": 1900
    },
    {
      "epoch": 2.0712328767123287,
      "grad_norm": 10.344156265258789,
      "learning_rate": 7.102993404363268e-06,
      "loss": 0.2954,
      "step": 1910
    },
    {
      "epoch": 2.075799086757991,
      "grad_norm": 0.31745678186416626,
      "learning_rate": 6.849315068493151e-06,
      "loss": 0.0895,
      "step": 1920
    },
    {
      "epoch": 2.080365296803653,
      "grad_norm": 5.952996730804443,
      "learning_rate": 6.595636732623034e-06,
      "loss": 0.2875,
      "step": 1930
    },
    {
      "epoch": 2.084931506849315,
      "grad_norm": 2.1418561935424805,
      "learning_rate": 6.341958396752917e-06,
      "loss": 0.09,
      "step": 1940
    },
    {
      "epoch": 2.0894977168949773,
      "grad_norm": 8.547797203063965,
      "learning_rate": 6.0882800608828005e-06,
      "loss": 0.1073,
      "step": 1950
    },
    {
      "epoch": 2.094063926940639,
      "grad_norm": 0.28819894790649414,
      "learning_rate": 5.834601725012684e-06,
      "loss": 0.1083,
      "step": 1960
    },
    {
      "epoch": 2.0986301369863014,
      "grad_norm": 4.290701866149902,
      "learning_rate": 5.580923389142567e-06,
      "loss": 0.1421,
      "step": 1970
    },
    {
      "epoch": 2.1031963470319637,
      "grad_norm": 2.7559561729431152,
      "learning_rate": 5.32724505327245e-06,
      "loss": 0.1769,
      "step": 1980
    },
    {
      "epoch": 2.1077625570776255,
      "grad_norm": 0.0814567431807518,
      "learning_rate": 5.073566717402334e-06,
      "loss": 0.1929,
      "step": 1990
    },
    {
      "epoch": 2.1123287671232878,
      "grad_norm": 0.14855796098709106,
      "learning_rate": 4.819888381532217e-06,
      "loss": 0.222,
      "step": 2000
    },
    {
      "epoch": 2.1168949771689496,
      "grad_norm": 5.517074108123779,
      "learning_rate": 4.566210045662101e-06,
      "loss": 0.0973,
      "step": 2010
    },
    {
      "epoch": 2.121461187214612,
      "grad_norm": 5.795365810394287,
      "learning_rate": 4.312531709791984e-06,
      "loss": 0.2053,
      "step": 2020
    },
    {
      "epoch": 2.126027397260274,
      "grad_norm": 3.86086106300354,
      "learning_rate": 4.058853373921867e-06,
      "loss": 0.177,
      "step": 2030
    },
    {
      "epoch": 2.130593607305936,
      "grad_norm": 8.300323486328125,
      "learning_rate": 3.80517503805175e-06,
      "loss": 0.2199,
      "step": 2040
    },
    {
      "epoch": 2.135159817351598,
      "grad_norm": 12.849746704101562,
      "learning_rate": 3.551496702181634e-06,
      "loss": 0.2226,
      "step": 2050
    },
    {
      "epoch": 2.1397260273972605,
      "grad_norm": 9.738630294799805,
      "learning_rate": 3.297818366311517e-06,
      "loss": 0.0895,
      "step": 2060
    },
    {
      "epoch": 2.1442922374429223,
      "grad_norm": 8.472663879394531,
      "learning_rate": 3.0441400304414002e-06,
      "loss": 0.1436,
      "step": 2070
    },
    {
      "epoch": 2.1488584474885846,
      "grad_norm": 0.39250361919403076,
      "learning_rate": 2.7904616945712835e-06,
      "loss": 0.1693,
      "step": 2080
    },
    {
      "epoch": 2.1534246575342464,
      "grad_norm": 0.13123255968093872,
      "learning_rate": 2.536783358701167e-06,
      "loss": 0.0442,
      "step": 2090
    },
    {
      "epoch": 2.1579908675799087,
      "grad_norm": 0.15273036062717438,
      "learning_rate": 2.2831050228310503e-06,
      "loss": 0.0614,
      "step": 2100
    },
    {
      "epoch": 2.162557077625571,
      "grad_norm": 3.393217086791992,
      "learning_rate": 2.0294266869609335e-06,
      "loss": 0.0838,
      "step": 2110
    },
    {
      "epoch": 2.1671232876712327,
      "grad_norm": 0.510553240776062,
      "learning_rate": 1.775748351090817e-06,
      "loss": 0.0941,
      "step": 2120
    },
    {
      "epoch": 2.171689497716895,
      "grad_norm": 8.017914772033691,
      "learning_rate": 1.5220700152207001e-06,
      "loss": 0.111,
      "step": 2130
    },
    {
      "epoch": 2.1762557077625573,
      "grad_norm": 0.32205718755722046,
      "learning_rate": 1.2683916793505835e-06,
      "loss": 0.0666,
      "step": 2140
    },
    {
      "epoch": 2.180821917808219,
      "grad_norm": 1.1410584449768066,
      "learning_rate": 1.0147133434804667e-06,
      "loss": 0.0567,
      "step": 2150
    },
    {
      "epoch": 2.1853881278538814,
      "grad_norm": 0.22290684282779694,
      "learning_rate": 7.610350076103501e-07,
      "loss": 0.0834,
      "step": 2160
    },
    {
      "epoch": 2.189954337899543,
      "grad_norm": 5.116283416748047,
      "learning_rate": 5.073566717402334e-07,
      "loss": 0.1786,
      "step": 2170
    },
    {
      "epoch": 2.1945205479452055,
      "grad_norm": 1.6667741537094116,
      "learning_rate": 2.536783358701167e-07,
      "loss": 0.2055,
      "step": 2180
    },
    {
      "epoch": 2.1990867579908677,
      "grad_norm": 8.045380592346191,
      "learning_rate": 0.0,
      "loss": 0.2208,
      "step": 2190
    },
    {
      "epoch": 2.1990867579908677,
      "eval_accuracy": 0.5939139559286464,
      "eval_loss": 1.4424338340759277,
      "eval_runtime": 337.1936,
      "eval_samples_per_second": 2.826,
      "eval_steps_per_second": 1.415,
      "step": 2190
    },
    {
      "epoch": 2.1990867579908677,
      "step": 2190,
      "total_flos": 3.0707987980048663e+19,
      "train_loss": 0.7329474766504819,
      "train_runtime": 7235.1686,
      "train_samples_per_second": 2.422,
      "train_steps_per_second": 0.303
    },
    {
      "epoch": 2.1990867579908677,
      "eval_accuracy": 0.633788037775446,
      "eval_loss": 1.3121134042739868,
      "eval_runtime": 350.327,
      "eval_samples_per_second": 2.72,
      "eval_steps_per_second": 1.362,
      "step": 2190
    },
    {
      "epoch": 2.1990867579908677,
      "eval_accuracy": 0.7222659323367427,
      "eval_loss": 0.9983267188072205,
      "eval_runtime": 400.1224,
      "eval_samples_per_second": 3.177,
      "eval_steps_per_second": 1.59,
      "step": 2190
    }
  ],
  "logging_steps": 10,
  "max_steps": 2190,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 2,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 2
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.0707987980048663e+19,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}