{
  "best_metric": 0.041273970156908035,
  "best_model_checkpoint": "./microsoft_dit/checkpoint-16835",
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 16835,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00297000297000297,
      "grad_norm": 1.4537379741668701,
      "learning_rate": 2.9982179982179983e-05,
      "loss": 0.3085,
      "step": 10
    },
    {
      "epoch": 0.00594000594000594,
      "grad_norm": 1.9534900188446045,
      "learning_rate": 2.9964359964359965e-05,
      "loss": 0.2158,
      "step": 20
    },
    {
      "epoch": 0.00891000891000891,
      "grad_norm": 2.2834959030151367,
      "learning_rate": 2.9946539946539947e-05,
      "loss": 0.1555,
      "step": 30
    },
    {
      "epoch": 0.01188001188001188,
      "grad_norm": 2.4054393768310547,
      "learning_rate": 2.992871992871993e-05,
      "loss": 0.1099,
      "step": 40
    },
    {
      "epoch": 0.01485001485001485,
      "grad_norm": 1.9460220336914062,
      "learning_rate": 2.991089991089991e-05,
      "loss": 0.1214,
      "step": 50
    },
    {
      "epoch": 0.01782001782001782,
      "grad_norm": 1.1060905456542969,
      "learning_rate": 2.9893079893079894e-05,
      "loss": 0.1046,
      "step": 60
    },
    {
      "epoch": 0.02079002079002079,
      "grad_norm": 1.0159635543823242,
      "learning_rate": 2.9875259875259876e-05,
      "loss": 0.1055,
      "step": 70
    },
    {
      "epoch": 0.02376002376002376,
      "grad_norm": 2.2551214694976807,
      "learning_rate": 2.9857439857439858e-05,
      "loss": 0.099,
      "step": 80
    },
    {
      "epoch": 0.02673002673002673,
      "grad_norm": 1.3134405612945557,
      "learning_rate": 2.983961983961984e-05,
      "loss": 0.0975,
      "step": 90
    },
    {
      "epoch": 0.0297000297000297,
      "grad_norm": 1.055510401725769,
      "learning_rate": 2.9821799821799822e-05,
      "loss": 0.0945,
      "step": 100
    },
    {
      "epoch": 0.03267003267003267,
      "grad_norm": 1.4456548690795898,
      "learning_rate": 2.9803979803979805e-05,
      "loss": 0.1058,
      "step": 110
    },
    {
      "epoch": 0.03564003564003564,
      "grad_norm": 1.4712682962417603,
      "learning_rate": 2.9786159786159787e-05,
      "loss": 0.0901,
      "step": 120
    },
    {
      "epoch": 0.03861003861003861,
      "grad_norm": 2.145982265472412,
      "learning_rate": 2.976833976833977e-05,
      "loss": 0.1149,
      "step": 130
    },
    {
      "epoch": 0.04158004158004158,
      "grad_norm": 2.104418992996216,
      "learning_rate": 2.975051975051975e-05,
      "loss": 0.102,
      "step": 140
    },
    {
      "epoch": 0.04455004455004455,
      "grad_norm": 1.3641546964645386,
      "learning_rate": 2.9732699732699733e-05,
      "loss": 0.0996,
      "step": 150
    },
    {
      "epoch": 0.04752004752004752,
      "grad_norm": 2.1951277256011963,
      "learning_rate": 2.9714879714879715e-05,
      "loss": 0.092,
      "step": 160
    },
    {
      "epoch": 0.05049005049005049,
      "grad_norm": 1.2363626956939697,
      "learning_rate": 2.9697059697059698e-05,
      "loss": 0.0971,
      "step": 170
    },
    {
      "epoch": 0.05346005346005346,
      "grad_norm": 1.299441933631897,
      "learning_rate": 2.967923967923968e-05,
      "loss": 0.1,
      "step": 180
    },
    {
      "epoch": 0.05643005643005643,
      "grad_norm": 0.9083341956138611,
      "learning_rate": 2.9661419661419662e-05,
      "loss": 0.0997,
      "step": 190
    },
    {
      "epoch": 0.0594000594000594,
      "grad_norm": 1.1018556356430054,
      "learning_rate": 2.9643599643599644e-05,
      "loss": 0.0916,
      "step": 200
    },
    {
      "epoch": 0.062370062370062374,
      "grad_norm": 1.025562047958374,
      "learning_rate": 2.9625779625779626e-05,
      "loss": 0.0995,
      "step": 210
    },
    {
      "epoch": 0.06534006534006534,
      "grad_norm": 1.1609880924224854,
      "learning_rate": 2.960795960795961e-05,
      "loss": 0.0844,
      "step": 220
    },
    {
      "epoch": 0.0683100683100683,
      "grad_norm": 1.7311973571777344,
      "learning_rate": 2.959013959013959e-05,
      "loss": 0.0862,
      "step": 230
    },
    {
      "epoch": 0.07128007128007129,
      "grad_norm": 1.5643036365509033,
      "learning_rate": 2.9572319572319573e-05,
      "loss": 0.0999,
      "step": 240
    },
    {
      "epoch": 0.07425007425007425,
      "grad_norm": 1.468729019165039,
      "learning_rate": 2.9554499554499555e-05,
      "loss": 0.0989,
      "step": 250
    },
    {
      "epoch": 0.07722007722007722,
      "grad_norm": 0.8540601134300232,
      "learning_rate": 2.9536679536679537e-05,
      "loss": 0.0822,
      "step": 260
    },
    {
      "epoch": 0.08019008019008018,
      "grad_norm": 1.0522710084915161,
      "learning_rate": 2.951885951885952e-05,
      "loss": 0.0818,
      "step": 270
    },
    {
      "epoch": 0.08316008316008316,
      "grad_norm": 1.6779365539550781,
      "learning_rate": 2.95010395010395e-05,
      "loss": 0.0804,
      "step": 280
    },
    {
      "epoch": 0.08613008613008613,
      "grad_norm": 0.6255664825439453,
      "learning_rate": 2.9483219483219484e-05,
      "loss": 0.0958,
      "step": 290
    },
    {
      "epoch": 0.0891000891000891,
      "grad_norm": 0.9004175662994385,
      "learning_rate": 2.9465399465399466e-05,
      "loss": 0.0924,
      "step": 300
    },
    {
      "epoch": 0.09207009207009206,
      "grad_norm": 1.5575037002563477,
      "learning_rate": 2.9447579447579448e-05,
      "loss": 0.0927,
      "step": 310
    },
    {
      "epoch": 0.09504009504009504,
      "grad_norm": 1.6837743520736694,
      "learning_rate": 2.942975942975943e-05,
      "loss": 0.0997,
      "step": 320
    },
    {
      "epoch": 0.09801009801009801,
      "grad_norm": 0.6957836151123047,
      "learning_rate": 2.9411939411939412e-05,
      "loss": 0.07,
      "step": 330
    },
    {
      "epoch": 0.10098010098010098,
      "grad_norm": 1.7029621601104736,
      "learning_rate": 2.9394119394119395e-05,
      "loss": 0.109,
      "step": 340
    },
    {
      "epoch": 0.10395010395010396,
      "grad_norm": 1.519026756286621,
      "learning_rate": 2.9376299376299377e-05,
      "loss": 0.093,
      "step": 350
    },
    {
      "epoch": 0.10692010692010692,
      "grad_norm": 0.8110401034355164,
      "learning_rate": 2.935847935847936e-05,
      "loss": 0.0795,
      "step": 360
    },
    {
      "epoch": 0.10989010989010989,
      "grad_norm": 0.5369464159011841,
      "learning_rate": 2.934065934065934e-05,
      "loss": 0.0877,
      "step": 370
    },
    {
      "epoch": 0.11286011286011285,
      "grad_norm": 2.6125996112823486,
      "learning_rate": 2.9322839322839323e-05,
      "loss": 0.0852,
      "step": 380
    },
    {
      "epoch": 0.11583011583011583,
      "grad_norm": 0.8178906440734863,
      "learning_rate": 2.930501930501931e-05,
      "loss": 0.0708,
      "step": 390
    },
    {
      "epoch": 0.1188001188001188,
      "grad_norm": 1.6640368700027466,
      "learning_rate": 2.9287199287199288e-05,
      "loss": 0.0951,
      "step": 400
    },
    {
      "epoch": 0.12177012177012177,
      "grad_norm": 0.5659627318382263,
      "learning_rate": 2.926937926937927e-05,
      "loss": 0.0927,
      "step": 410
    },
    {
      "epoch": 0.12474012474012475,
      "grad_norm": 1.324707269668579,
      "learning_rate": 2.9251559251559252e-05,
      "loss": 0.0749,
      "step": 420
    },
    {
      "epoch": 0.1277101277101277,
      "grad_norm": 1.9675109386444092,
      "learning_rate": 2.9233739233739234e-05,
      "loss": 0.0827,
      "step": 430
    },
    {
      "epoch": 0.13068013068013068,
      "grad_norm": 1.659543752670288,
      "learning_rate": 2.9215919215919216e-05,
      "loss": 0.0938,
      "step": 440
    },
    {
      "epoch": 0.13365013365013365,
      "grad_norm": 1.0365532636642456,
      "learning_rate": 2.91980991980992e-05,
      "loss": 0.0922,
      "step": 450
    },
    {
      "epoch": 0.1366201366201366,
      "grad_norm": 1.2649776935577393,
      "learning_rate": 2.9180279180279184e-05,
      "loss": 0.0741,
      "step": 460
    },
    {
      "epoch": 0.13959013959013958,
      "grad_norm": 0.6527109146118164,
      "learning_rate": 2.9162459162459163e-05,
      "loss": 0.0865,
      "step": 470
    },
    {
      "epoch": 0.14256014256014257,
      "grad_norm": 1.706261396408081,
      "learning_rate": 2.9144639144639145e-05,
      "loss": 0.0805,
      "step": 480
    },
    {
      "epoch": 0.14553014553014554,
      "grad_norm": 1.399427890777588,
      "learning_rate": 2.9126819126819127e-05,
      "loss": 0.0821,
      "step": 490
    },
    {
      "epoch": 0.1485001485001485,
      "grad_norm": 1.1516929864883423,
      "learning_rate": 2.910899910899911e-05,
      "loss": 0.0877,
      "step": 500
    },
    {
      "epoch": 0.15147015147015147,
      "grad_norm": 0.6054642200469971,
      "learning_rate": 2.909117909117909e-05,
      "loss": 0.0732,
      "step": 510
    },
    {
      "epoch": 0.15444015444015444,
      "grad_norm": 1.518789291381836,
      "learning_rate": 2.9073359073359074e-05,
      "loss": 0.0929,
      "step": 520
    },
    {
      "epoch": 0.1574101574101574,
      "grad_norm": 0.8014304637908936,
      "learning_rate": 2.905553905553906e-05,
      "loss": 0.0948,
      "step": 530
    },
    {
      "epoch": 0.16038016038016037,
      "grad_norm": 1.0207316875457764,
      "learning_rate": 2.9037719037719038e-05,
      "loss": 0.083,
      "step": 540
    },
    {
      "epoch": 0.16335016335016336,
      "grad_norm": 0.8969537019729614,
      "learning_rate": 2.901989901989902e-05,
      "loss": 0.0819,
      "step": 550
    },
    {
      "epoch": 0.16632016632016633,
      "grad_norm": 1.3450151681900024,
      "learning_rate": 2.9002079002079002e-05,
      "loss": 0.0886,
      "step": 560
    },
    {
      "epoch": 0.1692901692901693,
      "grad_norm": 1.7601408958435059,
      "learning_rate": 2.8984258984258984e-05,
      "loss": 0.0898,
      "step": 570
    },
    {
      "epoch": 0.17226017226017226,
      "grad_norm": 1.2938804626464844,
      "learning_rate": 2.8966438966438967e-05,
      "loss": 0.0776,
      "step": 580
    },
    {
      "epoch": 0.17523017523017523,
      "grad_norm": 0.87087082862854,
      "learning_rate": 2.894861894861895e-05,
      "loss": 0.0869,
      "step": 590
    },
    {
      "epoch": 0.1782001782001782,
      "grad_norm": 0.7267048358917236,
      "learning_rate": 2.8930798930798934e-05,
      "loss": 0.0877,
      "step": 600
    },
    {
      "epoch": 0.18117018117018116,
      "grad_norm": 0.9188293218612671,
      "learning_rate": 2.8912978912978913e-05,
      "loss": 0.0701,
      "step": 610
    },
    {
      "epoch": 0.18414018414018413,
      "grad_norm": 1.1492931842803955,
      "learning_rate": 2.8895158895158895e-05,
      "loss": 0.0744,
      "step": 620
    },
    {
      "epoch": 0.18711018711018712,
      "grad_norm": 1.3128528594970703,
      "learning_rate": 2.8877338877338877e-05,
      "loss": 0.0807,
      "step": 630
    },
    {
      "epoch": 0.1900801900801901,
      "grad_norm": 0.9953811168670654,
      "learning_rate": 2.885951885951886e-05,
      "loss": 0.0727,
      "step": 640
    },
    {
      "epoch": 0.19305019305019305,
      "grad_norm": 0.6890339851379395,
      "learning_rate": 2.8841698841698842e-05,
      "loss": 0.0888,
      "step": 650
    },
    {
      "epoch": 0.19602019602019602,
      "grad_norm": 0.5367225408554077,
      "learning_rate": 2.8823878823878824e-05,
      "loss": 0.0912,
      "step": 660
    },
    {
      "epoch": 0.19899019899019899,
      "grad_norm": 0.6967876553535461,
      "learning_rate": 2.880605880605881e-05,
      "loss": 0.0831,
      "step": 670
    },
    {
      "epoch": 0.20196020196020195,
      "grad_norm": 1.0245673656463623,
      "learning_rate": 2.878823878823879e-05,
      "loss": 0.0906,
      "step": 680
    },
    {
      "epoch": 0.20493020493020492,
      "grad_norm": 0.6193961501121521,
      "learning_rate": 2.877041877041877e-05,
      "loss": 0.0798,
      "step": 690
    },
    {
      "epoch": 0.2079002079002079,
      "grad_norm": 1.8188307285308838,
      "learning_rate": 2.8752598752598753e-05,
      "loss": 0.0859,
      "step": 700
    },
    {
      "epoch": 0.21087021087021088,
      "grad_norm": 1.4549452066421509,
      "learning_rate": 2.8734778734778735e-05,
      "loss": 0.0759,
      "step": 710
    },
    {
      "epoch": 0.21384021384021384,
      "grad_norm": 1.0474627017974854,
      "learning_rate": 2.8716958716958717e-05,
      "loss": 0.0809,
      "step": 720
    },
    {
      "epoch": 0.2168102168102168,
      "grad_norm": 0.8879497051239014,
      "learning_rate": 2.86991386991387e-05,
      "loss": 0.0725,
      "step": 730
    },
    {
      "epoch": 0.21978021978021978,
      "grad_norm": 0.6862474679946899,
      "learning_rate": 2.8681318681318685e-05,
      "loss": 0.08,
      "step": 740
    },
    {
      "epoch": 0.22275022275022274,
      "grad_norm": 1.7506073713302612,
      "learning_rate": 2.8663498663498664e-05,
      "loss": 0.0992,
      "step": 750
    },
    {
      "epoch": 0.2257202257202257,
      "grad_norm": 0.9330048561096191,
      "learning_rate": 2.8645678645678646e-05,
      "loss": 0.0936,
      "step": 760
    },
    {
      "epoch": 0.2286902286902287,
      "grad_norm": 1.0609358549118042,
      "learning_rate": 2.8627858627858628e-05,
      "loss": 0.0816,
      "step": 770
    },
    {
      "epoch": 0.23166023166023167,
      "grad_norm": 0.6896756291389465,
      "learning_rate": 2.861003861003861e-05,
      "loss": 0.073,
      "step": 780
    },
    {
      "epoch": 0.23463023463023464,
      "grad_norm": 1.0784275531768799,
      "learning_rate": 2.8592218592218592e-05,
      "loss": 0.0799,
      "step": 790
    },
    {
      "epoch": 0.2376002376002376,
      "grad_norm": 1.0042153596878052,
      "learning_rate": 2.8574398574398574e-05,
      "loss": 0.079,
      "step": 800
    },
    {
      "epoch": 0.24057024057024057,
      "grad_norm": 1.3999987840652466,
      "learning_rate": 2.855657855657856e-05,
      "loss": 0.0841,
      "step": 810
    },
    {
      "epoch": 0.24354024354024353,
      "grad_norm": 0.6107861995697021,
      "learning_rate": 2.853875853875854e-05,
      "loss": 0.067,
      "step": 820
    },
    {
      "epoch": 0.2465102465102465,
      "grad_norm": 0.8890926837921143,
      "learning_rate": 2.852093852093852e-05,
      "loss": 0.0639,
      "step": 830
    },
    {
      "epoch": 0.2494802494802495,
      "grad_norm": 0.7977913618087769,
      "learning_rate": 2.8503118503118503e-05,
      "loss": 0.0745,
      "step": 840
    },
    {
      "epoch": 0.25245025245025243,
      "grad_norm": 1.1050574779510498,
      "learning_rate": 2.8485298485298485e-05,
      "loss": 0.0828,
      "step": 850
    },
    {
      "epoch": 0.2554202554202554,
      "grad_norm": 1.4865517616271973,
      "learning_rate": 2.8467478467478467e-05,
      "loss": 0.0971,
      "step": 860
    },
    {
      "epoch": 0.25839025839025836,
      "grad_norm": 0.8170812726020813,
      "learning_rate": 2.844965844965845e-05,
      "loss": 0.0736,
      "step": 870
    },
    {
      "epoch": 0.26136026136026136,
      "grad_norm": 0.8887490630149841,
      "learning_rate": 2.8431838431838435e-05,
      "loss": 0.0843,
      "step": 880
    },
    {
      "epoch": 0.26433026433026435,
      "grad_norm": 1.2814133167266846,
      "learning_rate": 2.8414018414018414e-05,
      "loss": 0.07,
      "step": 890
    },
    {
      "epoch": 0.2673002673002673,
      "grad_norm": 1.0275837182998657,
      "learning_rate": 2.8396198396198396e-05,
      "loss": 0.1014,
      "step": 900
    },
    {
      "epoch": 0.2702702702702703,
      "grad_norm": 2.31121563911438,
      "learning_rate": 2.8378378378378378e-05,
      "loss": 0.0798,
      "step": 910
    },
    {
      "epoch": 0.2732402732402732,
      "grad_norm": 0.7157280445098877,
      "learning_rate": 2.836055836055836e-05,
      "loss": 0.0856,
      "step": 920
    },
    {
      "epoch": 0.2762102762102762,
      "grad_norm": 1.51650071144104,
      "learning_rate": 2.8342738342738343e-05,
      "loss": 0.0886,
      "step": 930
    },
    {
      "epoch": 0.27918027918027916,
      "grad_norm": 1.0991870164871216,
      "learning_rate": 2.8324918324918325e-05,
      "loss": 0.0832,
      "step": 940
    },
    {
      "epoch": 0.28215028215028215,
      "grad_norm": 1.0901234149932861,
      "learning_rate": 2.830709830709831e-05,
      "loss": 0.0763,
      "step": 950
    },
    {
      "epoch": 0.28512028512028514,
      "grad_norm": 1.1311888694763184,
      "learning_rate": 2.8289278289278293e-05,
      "loss": 0.0766,
      "step": 960
    },
    {
      "epoch": 0.2880902880902881,
      "grad_norm": 0.6374753713607788,
      "learning_rate": 2.827145827145827e-05,
      "loss": 0.0736,
      "step": 970
    },
    {
      "epoch": 0.2910602910602911,
      "grad_norm": 1.2088044881820679,
      "learning_rate": 2.8253638253638253e-05,
      "loss": 0.0681,
      "step": 980
    },
    {
      "epoch": 0.294030294030294,
      "grad_norm": 0.7970484495162964,
      "learning_rate": 2.8235818235818236e-05,
      "loss": 0.0803,
      "step": 990
    },
    {
      "epoch": 0.297000297000297,
      "grad_norm": 1.2192399501800537,
      "learning_rate": 2.8217998217998218e-05,
      "loss": 0.0896,
      "step": 1000
    },
    {
      "epoch": 0.29997029997029995,
      "grad_norm": 1.2083388566970825,
      "learning_rate": 2.82001782001782e-05,
      "loss": 0.0795,
      "step": 1010
    },
    {
      "epoch": 0.30294030294030294,
      "grad_norm": 1.0027815103530884,
      "learning_rate": 2.8182358182358186e-05,
      "loss": 0.0874,
      "step": 1020
    },
    {
      "epoch": 0.30591030591030594,
      "grad_norm": 1.0552337169647217,
      "learning_rate": 2.8164538164538168e-05,
      "loss": 0.0718,
      "step": 1030
    },
    {
      "epoch": 0.3088803088803089,
      "grad_norm": 1.1285655498504639,
      "learning_rate": 2.8146718146718146e-05,
      "loss": 0.0723,
      "step": 1040
    },
    {
      "epoch": 0.31185031185031187,
      "grad_norm": 1.0405571460723877,
      "learning_rate": 2.812889812889813e-05,
      "loss": 0.0929,
      "step": 1050
    },
    {
      "epoch": 0.3148203148203148,
      "grad_norm": 0.9266412258148193,
      "learning_rate": 2.811107811107811e-05,
      "loss": 0.0831,
      "step": 1060
    },
    {
      "epoch": 0.3177903177903178,
      "grad_norm": 1.0532838106155396,
      "learning_rate": 2.8093258093258093e-05,
      "loss": 0.0767,
      "step": 1070
    },
    {
      "epoch": 0.32076032076032074,
      "grad_norm": 0.8444746136665344,
      "learning_rate": 2.8075438075438075e-05,
      "loss": 0.0779,
      "step": 1080
    },
    {
      "epoch": 0.32373032373032373,
      "grad_norm": 0.7850976586341858,
      "learning_rate": 2.805761805761806e-05,
      "loss": 0.0803,
      "step": 1090
    },
    {
      "epoch": 0.3267003267003267,
      "grad_norm": 1.0749289989471436,
      "learning_rate": 2.8039798039798043e-05,
      "loss": 0.0891,
      "step": 1100
    },
    {
      "epoch": 0.32967032967032966,
      "grad_norm": 0.9671429395675659,
      "learning_rate": 2.802197802197802e-05,
      "loss": 0.0734,
      "step": 1110
    },
    {
      "epoch": 0.33264033264033266,
      "grad_norm": 0.665507435798645,
      "learning_rate": 2.8004158004158004e-05,
      "loss": 0.0771,
      "step": 1120
    },
    {
      "epoch": 0.3356103356103356,
      "grad_norm": 0.6605273485183716,
      "learning_rate": 2.7986337986337986e-05,
      "loss": 0.0698,
      "step": 1130
    },
    {
      "epoch": 0.3385803385803386,
      "grad_norm": 0.6776021122932434,
      "learning_rate": 2.7968517968517968e-05,
      "loss": 0.0694,
      "step": 1140
    },
    {
      "epoch": 0.34155034155034153,
      "grad_norm": 0.8488766551017761,
      "learning_rate": 2.795069795069795e-05,
      "loss": 0.0664,
      "step": 1150
    },
    {
      "epoch": 0.3445203445203445,
      "grad_norm": 1.0404934883117676,
      "learning_rate": 2.7932877932877936e-05,
      "loss": 0.0759,
      "step": 1160
    },
    {
      "epoch": 0.3474903474903475,
      "grad_norm": 1.0605899095535278,
      "learning_rate": 2.7915057915057918e-05,
      "loss": 0.0863,
      "step": 1170
    },
    {
      "epoch": 0.35046035046035046,
      "grad_norm": 1.034513235092163,
      "learning_rate": 2.7897237897237897e-05,
      "loss": 0.0815,
      "step": 1180
    },
    {
      "epoch": 0.35343035343035345,
      "grad_norm": 1.6698535680770874,
      "learning_rate": 2.787941787941788e-05,
      "loss": 0.0869,
      "step": 1190
    },
    {
      "epoch": 0.3564003564003564,
      "grad_norm": 0.8041000366210938,
      "learning_rate": 2.786159786159786e-05,
      "loss": 0.0739,
      "step": 1200
    },
    {
      "epoch": 0.3593703593703594,
      "grad_norm": 0.8568106889724731,
      "learning_rate": 2.7843777843777843e-05,
      "loss": 0.0697,
      "step": 1210
    },
    {
      "epoch": 0.3623403623403623,
      "grad_norm": 1.3246437311172485,
      "learning_rate": 2.7825957825957826e-05,
      "loss": 0.0761,
      "step": 1220
    },
    {
      "epoch": 0.3653103653103653,
      "grad_norm": 1.048767328262329,
      "learning_rate": 2.780813780813781e-05,
      "loss": 0.0806,
      "step": 1230
    },
    {
      "epoch": 0.36828036828036825,
      "grad_norm": 0.8982730507850647,
      "learning_rate": 2.7790317790317793e-05,
      "loss": 0.0751,
      "step": 1240
    },
    {
      "epoch": 0.37125037125037125,
      "grad_norm": 0.6694151759147644,
      "learning_rate": 2.7772497772497772e-05,
      "loss": 0.0729,
      "step": 1250
    },
    {
      "epoch": 0.37422037422037424,
      "grad_norm": 0.5974312424659729,
      "learning_rate": 2.7754677754677754e-05,
      "loss": 0.0839,
      "step": 1260
    },
    {
      "epoch": 0.3771903771903772,
      "grad_norm": 1.6992155313491821,
      "learning_rate": 2.7736857736857736e-05,
      "loss": 0.0683,
      "step": 1270
    },
    {
      "epoch": 0.3801603801603802,
      "grad_norm": 1.6791558265686035,
      "learning_rate": 2.771903771903772e-05,
      "loss": 0.0809,
      "step": 1280
    },
    {
      "epoch": 0.3831303831303831,
      "grad_norm": 1.4574382305145264,
      "learning_rate": 2.77012177012177e-05,
      "loss": 0.0723,
      "step": 1290
    },
    {
      "epoch": 0.3861003861003861,
      "grad_norm": 0.6830180287361145,
      "learning_rate": 2.7683397683397686e-05,
      "loss": 0.0619,
      "step": 1300
    },
    {
      "epoch": 0.38907038907038904,
      "grad_norm": 0.7997535467147827,
      "learning_rate": 2.766557766557767e-05,
      "loss": 0.0797,
      "step": 1310
    },
    {
      "epoch": 0.39204039204039204,
      "grad_norm": 0.8323992490768433,
      "learning_rate": 2.7647757647757647e-05,
      "loss": 0.0717,
      "step": 1320
    },
    {
      "epoch": 0.39501039501039503,
      "grad_norm": 0.5575417876243591,
      "learning_rate": 2.762993762993763e-05,
      "loss": 0.072,
      "step": 1330
    },
    {
      "epoch": 0.39798039798039797,
      "grad_norm": 0.9314611554145813,
      "learning_rate": 2.761211761211761e-05,
      "loss": 0.0594,
      "step": 1340
    },
    {
      "epoch": 0.40095040095040096,
      "grad_norm": 1.7011877298355103,
      "learning_rate": 2.7594297594297594e-05,
      "loss": 0.0702,
      "step": 1350
    },
    {
      "epoch": 0.4039204039204039,
      "grad_norm": 0.8633953332901001,
      "learning_rate": 2.7576477576477576e-05,
      "loss": 0.0769,
      "step": 1360
    },
    {
      "epoch": 0.4068904068904069,
      "grad_norm": 0.6947808265686035,
      "learning_rate": 2.755865755865756e-05,
      "loss": 0.069,
      "step": 1370
    },
    {
      "epoch": 0.40986040986040984,
      "grad_norm": 0.5774619579315186,
      "learning_rate": 2.7540837540837544e-05,
      "loss": 0.0642,
      "step": 1380
    },
    {
      "epoch": 0.41283041283041283,
      "grad_norm": 1.0650389194488525,
      "learning_rate": 2.7523017523017522e-05,
      "loss": 0.0603,
      "step": 1390
    },
    {
      "epoch": 0.4158004158004158,
      "grad_norm": 0.9271636605262756,
      "learning_rate": 2.7505197505197505e-05,
      "loss": 0.0695,
      "step": 1400
    },
    {
      "epoch": 0.41877041877041876,
      "grad_norm": 1.1117199659347534,
      "learning_rate": 2.7487377487377487e-05,
      "loss": 0.0691,
      "step": 1410
    },
    {
      "epoch": 0.42174042174042176,
      "grad_norm": 0.7966795563697815,
      "learning_rate": 2.746955746955747e-05,
      "loss": 0.068,
      "step": 1420
    },
    {
      "epoch": 0.4247104247104247,
      "grad_norm": 1.0240360498428345,
      "learning_rate": 2.745173745173745e-05,
      "loss": 0.0656,
      "step": 1430
    },
    {
      "epoch": 0.4276804276804277,
      "grad_norm": 0.9488998651504517,
      "learning_rate": 2.7433917433917437e-05,
      "loss": 0.0666,
      "step": 1440
    },
    {
      "epoch": 0.4306504306504306,
      "grad_norm": 1.2475918531417847,
      "learning_rate": 2.741609741609742e-05,
      "loss": 0.0791,
      "step": 1450
    },
    {
      "epoch": 0.4336204336204336,
      "grad_norm": 1.125158667564392,
      "learning_rate": 2.7398277398277398e-05,
      "loss": 0.0768,
      "step": 1460
    },
    {
      "epoch": 0.4365904365904366,
      "grad_norm": 0.8701415657997131,
      "learning_rate": 2.738045738045738e-05,
      "loss": 0.0869,
      "step": 1470
    },
    {
      "epoch": 0.43956043956043955,
      "grad_norm": 1.3756756782531738,
      "learning_rate": 2.7362637362637362e-05,
      "loss": 0.0814,
      "step": 1480
    },
    {
      "epoch": 0.44253044253044255,
      "grad_norm": 1.0403056144714355,
      "learning_rate": 2.7344817344817344e-05,
      "loss": 0.0678,
      "step": 1490
    },
    {
      "epoch": 0.4455004455004455,
      "grad_norm": 0.45081251859664917,
      "learning_rate": 2.7326997326997326e-05,
      "loss": 0.0598,
      "step": 1500
    },
    {
      "epoch": 0.4484704484704485,
      "grad_norm": 0.4850378930568695,
      "learning_rate": 2.7309177309177312e-05,
      "loss": 0.0733,
      "step": 1510
    },
    {
      "epoch": 0.4514404514404514,
      "grad_norm": 1.3777422904968262,
      "learning_rate": 2.7291357291357294e-05,
      "loss": 0.068,
      "step": 1520
    },
    {
      "epoch": 0.4544104544104544,
      "grad_norm": 1.1350257396697998,
      "learning_rate": 2.7273537273537276e-05,
      "loss": 0.0789,
      "step": 1530
    },
    {
      "epoch": 0.4573804573804574,
      "grad_norm": 0.4942154288291931,
      "learning_rate": 2.7255717255717255e-05,
      "loss": 0.0721,
      "step": 1540
    },
    {
      "epoch": 0.46035046035046034,
      "grad_norm": 0.8224741816520691,
      "learning_rate": 2.7237897237897237e-05,
      "loss": 0.0683,
      "step": 1550
    },
    {
      "epoch": 0.46332046332046334,
      "grad_norm": 0.7868162393569946,
      "learning_rate": 2.722007722007722e-05,
      "loss": 0.0795,
      "step": 1560
    },
    {
      "epoch": 0.4662904662904663,
      "grad_norm": 0.9320521354675293,
      "learning_rate": 2.72022572022572e-05,
      "loss": 0.0809,
      "step": 1570
    },
    {
      "epoch": 0.46926046926046927,
      "grad_norm": 1.2369945049285889,
      "learning_rate": 2.7184437184437187e-05,
      "loss": 0.0606,
      "step": 1580
    },
    {
      "epoch": 0.4722304722304722,
      "grad_norm": 0.934948742389679,
      "learning_rate": 2.716661716661717e-05,
      "loss": 0.0659,
      "step": 1590
    },
    {
      "epoch": 0.4752004752004752,
      "grad_norm": 0.8786292672157288,
      "learning_rate": 2.714879714879715e-05,
      "loss": 0.0979,
      "step": 1600
    },
    {
      "epoch": 0.4781704781704782,
      "grad_norm": 0.5932102799415588,
      "learning_rate": 2.713097713097713e-05,
      "loss": 0.0868,
      "step": 1610
    },
    {
      "epoch": 0.48114048114048114,
      "grad_norm": 0.9076425433158875,
      "learning_rate": 2.7113157113157112e-05,
      "loss": 0.0734,
      "step": 1620
    },
    {
      "epoch": 0.48411048411048413,
      "grad_norm": 0.8655311465263367,
      "learning_rate": 2.7095337095337095e-05,
      "loss": 0.0582,
      "step": 1630
    },
    {
      "epoch": 0.48708048708048707,
      "grad_norm": 0.6988229155540466,
      "learning_rate": 2.7077517077517077e-05,
      "loss": 0.0815,
      "step": 1640
    },
    {
      "epoch": 0.49005049005049006,
      "grad_norm": 0.9337241649627686,
      "learning_rate": 2.7059697059697062e-05,
      "loss": 0.0675,
      "step": 1650
    },
    {
      "epoch": 0.493020493020493,
      "grad_norm": 1.3023432493209839,
      "learning_rate": 2.7041877041877044e-05,
      "loss": 0.0777,
      "step": 1660
    },
    {
      "epoch": 0.495990495990496,
      "grad_norm": 1.0824613571166992,
      "learning_rate": 2.7024057024057027e-05,
      "loss": 0.0707,
      "step": 1670
    },
    {
      "epoch": 0.498960498960499,
      "grad_norm": 0.7362037897109985,
      "learning_rate": 2.7006237006237005e-05,
      "loss": 0.0653,
      "step": 1680
    },
    {
      "epoch": 0.5019305019305019,
      "grad_norm": 0.8872225284576416,
      "learning_rate": 2.6988416988416988e-05,
      "loss": 0.0737,
      "step": 1690
    },
    {
      "epoch": 0.5049005049005049,
      "grad_norm": 1.3285760879516602,
      "learning_rate": 2.697059697059697e-05,
      "loss": 0.074,
      "step": 1700
    },
    {
      "epoch": 0.5078705078705079,
      "grad_norm": 0.8795962929725647,
      "learning_rate": 2.6952776952776952e-05,
      "loss": 0.0825,
      "step": 1710
    },
    {
      "epoch": 0.5108405108405109,
      "grad_norm": 1.0784797668457031,
      "learning_rate": 2.6934956934956937e-05,
      "loss": 0.0792,
      "step": 1720
    },
    {
      "epoch": 0.5138105138105138,
      "grad_norm": 1.0349229574203491,
      "learning_rate": 2.691713691713692e-05,
      "loss": 0.0775,
      "step": 1730
    },
    {
      "epoch": 0.5167805167805167,
      "grad_norm": 0.4279988706111908,
      "learning_rate": 2.6899316899316902e-05,
      "loss": 0.0746,
      "step": 1740
    },
    {
      "epoch": 0.5197505197505198,
      "grad_norm": 0.7496609091758728,
      "learning_rate": 2.688149688149688e-05,
      "loss": 0.0501,
      "step": 1750
    },
    {
      "epoch": 0.5227205227205227,
      "grad_norm": 0.5489044785499573,
      "learning_rate": 2.6863676863676863e-05,
      "loss": 0.0464,
      "step": 1760
    },
    {
      "epoch": 0.5256905256905257,
      "grad_norm": 0.6475281119346619,
      "learning_rate": 2.6845856845856845e-05,
      "loss": 0.0732,
      "step": 1770
    },
    {
      "epoch": 0.5286605286605287,
      "grad_norm": 0.9776890873908997,
      "learning_rate": 2.6828036828036827e-05,
      "loss": 0.0755,
      "step": 1780
    },
    {
      "epoch": 0.5316305316305316,
      "grad_norm": 0.8436118364334106,
      "learning_rate": 2.6810216810216813e-05,
      "loss": 0.0743,
      "step": 1790
    },
    {
      "epoch": 0.5346005346005346,
      "grad_norm": 1.0720716714859009,
      "learning_rate": 2.6792396792396795e-05,
      "loss": 0.0821,
      "step": 1800
    },
    {
      "epoch": 0.5375705375705375,
      "grad_norm": 1.699569582939148,
      "learning_rate": 2.6774576774576777e-05,
      "loss": 0.0664,
      "step": 1810
    },
    {
      "epoch": 0.5405405405405406,
      "grad_norm": 0.975058376789093,
      "learning_rate": 2.6756756756756756e-05,
      "loss": 0.0699,
      "step": 1820
    },
    {
      "epoch": 0.5435105435105435,
      "grad_norm": 0.9048154950141907,
      "learning_rate": 2.6738936738936738e-05,
      "loss": 0.0862,
      "step": 1830
    },
    {
      "epoch": 0.5464805464805464,
      "grad_norm": 1.2828189134597778,
      "learning_rate": 2.672111672111672e-05,
      "loss": 0.0683,
      "step": 1840
    },
    {
      "epoch": 0.5494505494505495,
      "grad_norm": 1.5076489448547363,
      "learning_rate": 2.6703296703296702e-05,
      "loss": 0.0827,
      "step": 1850
    },
    {
      "epoch": 0.5524205524205524,
      "grad_norm": 0.8987582325935364,
      "learning_rate": 2.6685476685476688e-05,
      "loss": 0.0637,
      "step": 1860
    },
    {
      "epoch": 0.5553905553905554,
      "grad_norm": 0.7532503008842468,
      "learning_rate": 2.666765666765667e-05,
      "loss": 0.0847,
      "step": 1870
    },
    {
      "epoch": 0.5583605583605583,
      "grad_norm": 0.8142328858375549,
      "learning_rate": 2.6649836649836652e-05,
      "loss": 0.0732,
      "step": 1880
    },
    {
      "epoch": 0.5613305613305614,
      "grad_norm": 0.6130629181861877,
      "learning_rate": 2.663201663201663e-05,
      "loss": 0.0858,
      "step": 1890
    },
    {
      "epoch": 0.5643005643005643,
      "grad_norm": 1.0010277032852173,
      "learning_rate": 2.6614196614196613e-05,
      "loss": 0.0773,
      "step": 1900
    },
    {
      "epoch": 0.5672705672705672,
      "grad_norm": 0.8659054040908813,
      "learning_rate": 2.6596376596376595e-05,
      "loss": 0.0778,
      "step": 1910
    },
    {
      "epoch": 0.5702405702405703,
      "grad_norm": 0.7538540959358215,
      "learning_rate": 2.6578556578556577e-05,
      "loss": 0.071,
      "step": 1920
    },
    {
      "epoch": 0.5732105732105732,
      "grad_norm": 0.9332623481750488,
      "learning_rate": 2.6560736560736563e-05,
      "loss": 0.0795,
      "step": 1930
    },
    {
      "epoch": 0.5761805761805762,
      "grad_norm": 1.3811964988708496,
      "learning_rate": 2.6542916542916545e-05,
      "loss": 0.0782,
      "step": 1940
    },
    {
      "epoch": 0.5791505791505791,
      "grad_norm": 0.8431365489959717,
      "learning_rate": 2.6525096525096527e-05,
      "loss": 0.0727,
      "step": 1950
    },
    {
      "epoch": 0.5821205821205822,
      "grad_norm": 0.6309846639633179,
      "learning_rate": 2.6507276507276506e-05,
      "loss": 0.074,
      "step": 1960
    },
    {
      "epoch": 0.5850905850905851,
      "grad_norm": 1.0400676727294922,
      "learning_rate": 2.648945648945649e-05,
      "loss": 0.0683,
      "step": 1970
    },
    {
      "epoch": 0.588060588060588,
      "grad_norm": 1.1572082042694092,
      "learning_rate": 2.647163647163647e-05,
      "loss": 0.0718,
      "step": 1980
    },
    {
      "epoch": 0.5910305910305911,
      "grad_norm": 0.6798633337020874,
      "learning_rate": 2.6453816453816453e-05,
      "loss": 0.0695,
      "step": 1990
    },
    {
      "epoch": 0.594000594000594,
      "grad_norm": 0.49746111035346985,
      "learning_rate": 2.6435996435996438e-05,
      "loss": 0.0726,
      "step": 2000
    },
    {
      "epoch": 0.596970596970597,
      "grad_norm": 0.7728568315505981,
      "learning_rate": 2.641817641817642e-05,
      "loss": 0.0681,
      "step": 2010
    },
    {
      "epoch": 0.5999405999405999,
      "grad_norm": 1.0050711631774902,
      "learning_rate": 2.6400356400356403e-05,
      "loss": 0.0783,
      "step": 2020
    },
    {
      "epoch": 0.6029106029106029,
      "grad_norm": 0.6406471729278564,
      "learning_rate": 2.638253638253638e-05,
      "loss": 0.0743,
      "step": 2030
    },
    {
      "epoch": 0.6058806058806059,
      "grad_norm": 0.5787233710289001,
      "learning_rate": 2.6364716364716364e-05,
      "loss": 0.0692,
      "step": 2040
    },
    {
      "epoch": 0.6088506088506088,
      "grad_norm": 1.646081566810608,
      "learning_rate": 2.6346896346896346e-05,
      "loss": 0.0712,
      "step": 2050
    },
    {
      "epoch": 0.6118206118206119,
      "grad_norm": 1.0009373426437378,
      "learning_rate": 2.6329076329076328e-05,
      "loss": 0.0775,
      "step": 2060
    },
    {
      "epoch": 0.6147906147906148,
      "grad_norm": 0.7711787819862366,
      "learning_rate": 2.6311256311256313e-05,
      "loss": 0.0649,
      "step": 2070
    },
    {
      "epoch": 0.6177606177606177,
      "grad_norm": 0.6170060038566589,
      "learning_rate": 2.6293436293436296e-05,
      "loss": 0.0722,
      "step": 2080
    },
    {
      "epoch": 0.6207306207306207,
      "grad_norm": 0.5135714411735535,
      "learning_rate": 2.6275616275616278e-05,
      "loss": 0.0676,
      "step": 2090
    },
    {
      "epoch": 0.6237006237006237,
      "grad_norm": 0.6718676090240479,
      "learning_rate": 2.625779625779626e-05,
      "loss": 0.0748,
      "step": 2100
    },
    {
      "epoch": 0.6266706266706267,
      "grad_norm": 0.47525379061698914,
      "learning_rate": 2.623997623997624e-05,
      "loss": 0.0685,
      "step": 2110
    },
    {
      "epoch": 0.6296406296406296,
      "grad_norm": 1.864790916442871,
      "learning_rate": 2.622215622215622e-05,
      "loss": 0.0934,
      "step": 2120
    },
    {
      "epoch": 0.6326106326106327,
      "grad_norm": 1.857503890991211,
      "learning_rate": 2.6204336204336203e-05,
      "loss": 0.0736,
      "step": 2130
    },
    {
      "epoch": 0.6355806355806356,
      "grad_norm": 0.8403912782669067,
      "learning_rate": 2.618651618651619e-05,
      "loss": 0.0686,
      "step": 2140
    },
    {
      "epoch": 0.6385506385506385,
      "grad_norm": 1.2553832530975342,
      "learning_rate": 2.616869616869617e-05,
      "loss": 0.075,
      "step": 2150
    },
    {
      "epoch": 0.6415206415206415,
      "grad_norm": 0.75575190782547,
      "learning_rate": 2.6150876150876153e-05,
      "loss": 0.072,
      "step": 2160
    },
    {
      "epoch": 0.6444906444906445,
      "grad_norm": 0.7265666723251343,
      "learning_rate": 2.6133056133056135e-05,
      "loss": 0.081,
      "step": 2170
    },
    {
      "epoch": 0.6474606474606475,
      "grad_norm": 0.4849323332309723,
      "learning_rate": 2.6115236115236114e-05,
      "loss": 0.0801,
      "step": 2180
    },
    {
      "epoch": 0.6504306504306504,
      "grad_norm": 1.0659674406051636,
      "learning_rate": 2.6097416097416096e-05,
      "loss": 0.07,
      "step": 2190
    },
    {
      "epoch": 0.6534006534006535,
      "grad_norm": 1.1422368288040161,
      "learning_rate": 2.6079596079596078e-05,
      "loss": 0.0689,
      "step": 2200
    },
    {
      "epoch": 0.6563706563706564,
      "grad_norm": 0.6629343032836914,
      "learning_rate": 2.6061776061776064e-05,
      "loss": 0.0647,
      "step": 2210
    },
    {
      "epoch": 0.6593406593406593,
      "grad_norm": 1.0721797943115234,
      "learning_rate": 2.6043956043956046e-05,
      "loss": 0.065,
      "step": 2220
    },
    {
      "epoch": 0.6623106623106623,
      "grad_norm": 1.3788233995437622,
      "learning_rate": 2.6026136026136028e-05,
      "loss": 0.0743,
      "step": 2230
    },
    {
      "epoch": 0.6652806652806653,
      "grad_norm": 1.3036271333694458,
      "learning_rate": 2.600831600831601e-05,
      "loss": 0.0883,
      "step": 2240
    },
    {
      "epoch": 0.6682506682506683,
      "grad_norm": 0.8231455683708191,
      "learning_rate": 2.599049599049599e-05,
      "loss": 0.0684,
      "step": 2250
    },
    {
      "epoch": 0.6712206712206712,
      "grad_norm": 1.1133084297180176,
      "learning_rate": 2.597267597267597e-05,
      "loss": 0.0658,
      "step": 2260
    },
    {
      "epoch": 0.6741906741906742,
      "grad_norm": 1.1839704513549805,
      "learning_rate": 2.5954855954855953e-05,
      "loss": 0.0613,
      "step": 2270
    },
    {
      "epoch": 0.6771606771606772,
      "grad_norm": 0.9162219762802124,
      "learning_rate": 2.593703593703594e-05,
      "loss": 0.0611,
      "step": 2280
    },
    {
      "epoch": 0.6801306801306801,
      "grad_norm": 0.9693069458007812,
      "learning_rate": 2.591921591921592e-05,
      "loss": 0.0738,
      "step": 2290
    },
    {
      "epoch": 0.6831006831006831,
      "grad_norm": 1.2180196046829224,
      "learning_rate": 2.5901395901395903e-05,
      "loss": 0.0706,
      "step": 2300
    },
    {
      "epoch": 0.6860706860706861,
      "grad_norm": 1.0316444635391235,
      "learning_rate": 2.5883575883575886e-05,
      "loss": 0.0678,
      "step": 2310
    },
    {
      "epoch": 0.689040689040689,
      "grad_norm": 0.9395328164100647,
      "learning_rate": 2.5865755865755864e-05,
      "loss": 0.0745,
      "step": 2320
    },
    {
      "epoch": 0.692010692010692,
      "grad_norm": 0.9199443459510803,
      "learning_rate": 2.5847935847935846e-05,
      "loss": 0.0736,
      "step": 2330
    },
    {
      "epoch": 0.694980694980695,
      "grad_norm": 0.6675525307655334,
      "learning_rate": 2.583011583011583e-05,
      "loss": 0.07,
      "step": 2340
    },
    {
      "epoch": 0.697950697950698,
      "grad_norm": 1.0296847820281982,
      "learning_rate": 2.5812295812295814e-05,
      "loss": 0.0667,
      "step": 2350
    },
    {
      "epoch": 0.7009207009207009,
      "grad_norm": 0.7614896893501282,
      "learning_rate": 2.5794475794475796e-05,
      "loss": 0.0598,
      "step": 2360
    },
    {
      "epoch": 0.7038907038907039,
      "grad_norm": 0.3479407727718353,
      "learning_rate": 2.577665577665578e-05,
      "loss": 0.073,
      "step": 2370
    },
    {
      "epoch": 0.7068607068607069,
      "grad_norm": 0.906836986541748,
      "learning_rate": 2.575883575883576e-05,
      "loss": 0.066,
      "step": 2380
    },
    {
      "epoch": 0.7098307098307098,
      "grad_norm": 1.1523551940917969,
      "learning_rate": 2.574101574101574e-05,
      "loss": 0.0752,
      "step": 2390
    },
    {
      "epoch": 0.7128007128007128,
      "grad_norm": 1.5044126510620117,
      "learning_rate": 2.572319572319572e-05,
      "loss": 0.0629,
      "step": 2400
    },
    {
      "epoch": 0.7157707157707157,
      "grad_norm": 0.7415446639060974,
      "learning_rate": 2.5705375705375707e-05,
      "loss": 0.0851,
      "step": 2410
    },
    {
      "epoch": 0.7187407187407188,
      "grad_norm": 0.9454631805419922,
      "learning_rate": 2.568755568755569e-05,
      "loss": 0.0818,
      "step": 2420
    },
    {
      "epoch": 0.7217107217107217,
      "grad_norm": 1.3333286046981812,
      "learning_rate": 2.566973566973567e-05,
      "loss": 0.0781,
      "step": 2430
    },
    {
      "epoch": 0.7246807246807246,
      "grad_norm": 1.030158519744873,
      "learning_rate": 2.5651915651915654e-05,
      "loss": 0.0752,
      "step": 2440
    },
    {
      "epoch": 0.7276507276507277,
      "grad_norm": 0.9422637820243835,
      "learning_rate": 2.5634095634095636e-05,
      "loss": 0.0779,
      "step": 2450
    },
    {
      "epoch": 0.7306207306207306,
      "grad_norm": 0.8229959011077881,
      "learning_rate": 2.5616275616275615e-05,
      "loss": 0.0644,
      "step": 2460
    },
    {
      "epoch": 0.7335907335907336,
      "grad_norm": 0.7680384516716003,
      "learning_rate": 2.5598455598455597e-05,
      "loss": 0.0717,
      "step": 2470
    },
    {
      "epoch": 0.7365607365607365,
      "grad_norm": 1.118505835533142,
      "learning_rate": 2.5580635580635582e-05,
      "loss": 0.0777,
      "step": 2480
    },
    {
      "epoch": 0.7395307395307396,
      "grad_norm": 0.7873930335044861,
      "learning_rate": 2.5562815562815565e-05,
      "loss": 0.0843,
      "step": 2490
    },
    {
      "epoch": 0.7425007425007425,
      "grad_norm": 0.688714861869812,
      "learning_rate": 2.5544995544995547e-05,
      "loss": 0.0664,
      "step": 2500
    },
    {
      "epoch": 0.7454707454707454,
      "grad_norm": 0.6850858330726624,
      "learning_rate": 2.552717552717553e-05,
      "loss": 0.0715,
      "step": 2510
    },
    {
      "epoch": 0.7484407484407485,
      "grad_norm": 0.71670001745224,
      "learning_rate": 2.550935550935551e-05,
      "loss": 0.0579,
      "step": 2520
    },
    {
      "epoch": 0.7514107514107514,
      "grad_norm": 0.8807306885719299,
      "learning_rate": 2.549153549153549e-05,
      "loss": 0.0568,
      "step": 2530
    },
    {
      "epoch": 0.7543807543807544,
      "grad_norm": 1.108122706413269,
      "learning_rate": 2.5473715473715472e-05,
      "loss": 0.0726,
      "step": 2540
    },
    {
      "epoch": 0.7573507573507573,
      "grad_norm": 0.91880863904953,
      "learning_rate": 2.5455895455895458e-05,
      "loss": 0.0697,
      "step": 2550
    },
    {
      "epoch": 0.7603207603207603,
      "grad_norm": 1.2890892028808594,
      "learning_rate": 2.543807543807544e-05,
      "loss": 0.0742,
      "step": 2560
    },
    {
      "epoch": 0.7632907632907633,
      "grad_norm": 1.3207858800888062,
      "learning_rate": 2.5420255420255422e-05,
      "loss": 0.0602,
      "step": 2570
    },
    {
      "epoch": 0.7662607662607662,
      "grad_norm": 1.0641006231307983,
      "learning_rate": 2.5402435402435404e-05,
      "loss": 0.0783,
      "step": 2580
    },
    {
      "epoch": 0.7692307692307693,
      "grad_norm": 0.7104412317276001,
      "learning_rate": 2.5384615384615386e-05,
      "loss": 0.0622,
      "step": 2590
    },
    {
      "epoch": 0.7722007722007722,
      "grad_norm": 1.2291561365127563,
      "learning_rate": 2.5366795366795365e-05,
      "loss": 0.0789,
      "step": 2600
    },
    {
      "epoch": 0.7751707751707752,
      "grad_norm": 0.9593638181686401,
      "learning_rate": 2.5348975348975347e-05,
      "loss": 0.0604,
      "step": 2610
    },
    {
      "epoch": 0.7781407781407781,
      "grad_norm": 0.7314882278442383,
      "learning_rate": 2.5331155331155333e-05,
      "loss": 0.0679,
      "step": 2620
    },
    {
      "epoch": 0.7811107811107811,
      "grad_norm": 1.0600396394729614,
      "learning_rate": 2.5313335313335315e-05,
      "loss": 0.0723,
      "step": 2630
    },
    {
      "epoch": 0.7840807840807841,
      "grad_norm": 0.8725117444992065,
      "learning_rate": 2.5295515295515297e-05,
      "loss": 0.0627,
      "step": 2640
    },
    {
      "epoch": 0.787050787050787,
      "grad_norm": 0.7169741988182068,
      "learning_rate": 2.527769527769528e-05,
      "loss": 0.0674,
      "step": 2650
    },
    {
      "epoch": 0.7900207900207901,
      "grad_norm": 0.8746957182884216,
      "learning_rate": 2.525987525987526e-05,
      "loss": 0.0729,
      "step": 2660
    },
    {
      "epoch": 0.792990792990793,
      "grad_norm": 0.5545559525489807,
      "learning_rate": 2.524205524205524e-05,
      "loss": 0.0829,
      "step": 2670
    },
    {
      "epoch": 0.7959607959607959,
      "grad_norm": 1.095009207725525,
      "learning_rate": 2.5224235224235222e-05,
      "loss": 0.0705,
      "step": 2680
    },
    {
      "epoch": 0.7989307989307989,
      "grad_norm": 1.138685941696167,
      "learning_rate": 2.5206415206415208e-05,
      "loss": 0.0683,
      "step": 2690
    },
    {
      "epoch": 0.8019008019008019,
      "grad_norm": 0.8513908386230469,
      "learning_rate": 2.518859518859519e-05,
      "loss": 0.0721,
      "step": 2700
    },
    {
      "epoch": 0.8048708048708049,
      "grad_norm": 0.6944445967674255,
      "learning_rate": 2.5170775170775172e-05,
      "loss": 0.0666,
      "step": 2710
    },
    {
      "epoch": 0.8078408078408078,
      "grad_norm": 0.7438215613365173,
      "learning_rate": 2.5152955152955155e-05,
      "loss": 0.0587,
      "step": 2720
    },
    {
      "epoch": 0.8108108108108109,
      "grad_norm": 1.1805411577224731,
      "learning_rate": 2.5135135135135137e-05,
      "loss": 0.0701,
      "step": 2730
    },
    {
      "epoch": 0.8137808137808138,
      "grad_norm": 0.8039736151695251,
      "learning_rate": 2.511731511731512e-05,
      "loss": 0.0611,
      "step": 2740
    },
    {
      "epoch": 0.8167508167508167,
      "grad_norm": 0.9000430107116699,
      "learning_rate": 2.5099495099495098e-05,
      "loss": 0.076,
      "step": 2750
    },
    {
      "epoch": 0.8197208197208197,
      "grad_norm": 1.0777641534805298,
      "learning_rate": 2.5081675081675083e-05,
      "loss": 0.0683,
      "step": 2760
    },
    {
      "epoch": 0.8226908226908227,
      "grad_norm": 1.0703097581863403,
      "learning_rate": 2.5063855063855065e-05,
      "loss": 0.0684,
      "step": 2770
    },
    {
      "epoch": 0.8256608256608257,
      "grad_norm": 1.3522082567214966,
      "learning_rate": 2.5046035046035048e-05,
      "loss": 0.0687,
      "step": 2780
    },
    {
      "epoch": 0.8286308286308286,
      "grad_norm": 0.916243314743042,
      "learning_rate": 2.502821502821503e-05,
      "loss": 0.071,
      "step": 2790
    },
    {
      "epoch": 0.8316008316008316,
      "grad_norm": 0.844256579875946,
      "learning_rate": 2.5010395010395012e-05,
      "loss": 0.0819,
      "step": 2800
    },
    {
      "epoch": 0.8345708345708346,
      "grad_norm": 0.5879322290420532,
      "learning_rate": 2.4992574992574994e-05,
      "loss": 0.0723,
      "step": 2810
    },
    {
      "epoch": 0.8375408375408375,
      "grad_norm": 1.0017356872558594,
      "learning_rate": 2.4974754974754973e-05,
      "loss": 0.0723,
      "step": 2820
    },
    {
      "epoch": 0.8405108405108405,
      "grad_norm": 1.0885043144226074,
      "learning_rate": 2.495693495693496e-05,
      "loss": 0.0607,
      "step": 2830
    },
    {
      "epoch": 0.8434808434808435,
      "grad_norm": 0.7798848748207092,
      "learning_rate": 2.493911493911494e-05,
      "loss": 0.057,
      "step": 2840
    },
    {
      "epoch": 0.8464508464508465,
      "grad_norm": 1.2266145944595337,
      "learning_rate": 2.4921294921294923e-05,
      "loss": 0.0785,
      "step": 2850
    },
    {
      "epoch": 0.8494208494208494,
      "grad_norm": 1.3794151544570923,
      "learning_rate": 2.4903474903474905e-05,
      "loss": 0.0711,
      "step": 2860
    },
    {
      "epoch": 0.8523908523908524,
      "grad_norm": 1.2152888774871826,
      "learning_rate": 2.4885654885654887e-05,
      "loss": 0.0783,
      "step": 2870
    },
    {
      "epoch": 0.8553608553608554,
      "grad_norm": 0.5685151219367981,
      "learning_rate": 2.486783486783487e-05,
      "loss": 0.0818,
      "step": 2880
    },
    {
      "epoch": 0.8583308583308583,
      "grad_norm": 1.0047805309295654,
      "learning_rate": 2.4850014850014848e-05,
      "loss": 0.0768,
      "step": 2890
    },
    {
      "epoch": 0.8613008613008613,
      "grad_norm": 0.4588945806026459,
      "learning_rate": 2.4832194832194834e-05,
      "loss": 0.0616,
      "step": 2900
    },
    {
      "epoch": 0.8642708642708643,
      "grad_norm": 0.815482497215271,
      "learning_rate": 2.4814374814374816e-05,
      "loss": 0.0734,
      "step": 2910
    },
    {
      "epoch": 0.8672408672408672,
      "grad_norm": 1.1462020874023438,
      "learning_rate": 2.4796554796554798e-05,
      "loss": 0.0716,
      "step": 2920
    },
    {
      "epoch": 0.8702108702108702,
      "grad_norm": 1.8708317279815674,
      "learning_rate": 2.477873477873478e-05,
      "loss": 0.0653,
      "step": 2930
    },
    {
      "epoch": 0.8731808731808732,
      "grad_norm": 0.5544032454490662,
      "learning_rate": 2.4760914760914762e-05,
      "loss": 0.07,
      "step": 2940
    },
    {
      "epoch": 0.8761508761508762,
      "grad_norm": 0.7917482256889343,
      "learning_rate": 2.4743094743094744e-05,
      "loss": 0.0656,
      "step": 2950
    },
    {
      "epoch": 0.8791208791208791,
      "grad_norm": 1.3552825450897217,
      "learning_rate": 2.4725274725274723e-05,
      "loss": 0.064,
      "step": 2960
    },
    {
      "epoch": 0.882090882090882,
      "grad_norm": 0.7009034752845764,
      "learning_rate": 2.470745470745471e-05,
      "loss": 0.0707,
      "step": 2970
    },
    {
      "epoch": 0.8850608850608851,
      "grad_norm": 0.835366427898407,
      "learning_rate": 2.468963468963469e-05,
      "loss": 0.0724,
      "step": 2980
    },
    {
      "epoch": 0.888030888030888,
      "grad_norm": 0.8953189253807068,
      "learning_rate": 2.4671814671814673e-05,
      "loss": 0.0692,
      "step": 2990
    },
    {
      "epoch": 0.891000891000891,
      "grad_norm": 1.1693567037582397,
      "learning_rate": 2.4653994653994655e-05,
      "loss": 0.0737,
      "step": 3000
    },
    {
      "epoch": 0.893970893970894,
      "grad_norm": 0.7871369123458862,
      "learning_rate": 2.4636174636174637e-05,
      "loss": 0.0812,
      "step": 3010
    },
    {
      "epoch": 0.896940896940897,
      "grad_norm": 0.7140387296676636,
      "learning_rate": 2.461835461835462e-05,
      "loss": 0.0703,
      "step": 3020
    },
    {
      "epoch": 0.8999108999108999,
      "grad_norm": 0.7188717126846313,
      "learning_rate": 2.46005346005346e-05,
      "loss": 0.062,
      "step": 3030
    },
    {
      "epoch": 0.9028809028809028,
      "grad_norm": 0.5643990635871887,
      "learning_rate": 2.4582714582714584e-05,
      "loss": 0.066,
      "step": 3040
    },
    {
      "epoch": 0.9058509058509059,
      "grad_norm": 0.8118051886558533,
      "learning_rate": 2.4564894564894566e-05,
      "loss": 0.0777,
      "step": 3050
    },
    {
      "epoch": 0.9088209088209088,
      "grad_norm": 0.5903987884521484,
      "learning_rate": 2.454707454707455e-05,
      "loss": 0.0691,
      "step": 3060
    },
    {
      "epoch": 0.9117909117909118,
      "grad_norm": 0.4959113299846649,
      "learning_rate": 2.452925452925453e-05,
      "loss": 0.0706,
      "step": 3070
    },
    {
      "epoch": 0.9147609147609148,
      "grad_norm": 0.6881820559501648,
      "learning_rate": 2.4511434511434513e-05,
      "loss": 0.0793,
      "step": 3080
    },
    {
      "epoch": 0.9177309177309178,
      "grad_norm": 1.0045626163482666,
      "learning_rate": 2.4493614493614495e-05,
      "loss": 0.0669,
      "step": 3090
    },
    {
      "epoch": 0.9207009207009207,
      "grad_norm": 0.6501134037971497,
      "learning_rate": 2.4475794475794474e-05,
      "loss": 0.0618,
      "step": 3100
    },
    {
      "epoch": 0.9236709236709236,
      "grad_norm": 0.7777647972106934,
      "learning_rate": 2.445797445797446e-05,
      "loss": 0.0802,
      "step": 3110
    },
    {
      "epoch": 0.9266409266409267,
      "grad_norm": 0.9002485275268555,
      "learning_rate": 2.444015444015444e-05,
      "loss": 0.0686,
      "step": 3120
    },
    {
      "epoch": 0.9296109296109296,
      "grad_norm": 0.6177936792373657,
      "learning_rate": 2.4422334422334424e-05,
      "loss": 0.0688,
      "step": 3130
    },
    {
      "epoch": 0.9325809325809326,
      "grad_norm": 0.6512916088104248,
      "learning_rate": 2.4404514404514406e-05,
      "loss": 0.068,
      "step": 3140
    },
    {
      "epoch": 0.9355509355509356,
      "grad_norm": 0.670874297618866,
      "learning_rate": 2.4386694386694388e-05,
      "loss": 0.0804,
      "step": 3150
    },
    {
      "epoch": 0.9385209385209385,
      "grad_norm": 0.9932950139045715,
      "learning_rate": 2.436887436887437e-05,
      "loss": 0.0639,
      "step": 3160
    },
    {
      "epoch": 0.9414909414909415,
      "grad_norm": 0.8478681445121765,
      "learning_rate": 2.435105435105435e-05,
      "loss": 0.0617,
      "step": 3170
    },
    {
      "epoch": 0.9444609444609444,
      "grad_norm": 0.6210806369781494,
      "learning_rate": 2.4333234333234334e-05,
      "loss": 0.0651,
      "step": 3180
    },
    {
      "epoch": 0.9474309474309475,
      "grad_norm": 0.8721588850021362,
      "learning_rate": 2.4315414315414317e-05,
      "loss": 0.0849,
      "step": 3190
    },
    {
      "epoch": 0.9504009504009504,
      "grad_norm": 0.5963863730430603,
      "learning_rate": 2.42975942975943e-05,
      "loss": 0.063,
      "step": 3200
    },
    {
      "epoch": 0.9533709533709533,
      "grad_norm": 1.1241910457611084,
      "learning_rate": 2.427977427977428e-05,
      "loss": 0.0669,
      "step": 3210
    },
    {
      "epoch": 0.9563409563409564,
      "grad_norm": 1.101955771446228,
      "learning_rate": 2.4261954261954263e-05,
      "loss": 0.067,
      "step": 3220
    },
    {
      "epoch": 0.9593109593109593,
      "grad_norm": 1.1508077383041382,
      "learning_rate": 2.4244134244134245e-05,
      "loss": 0.0666,
      "step": 3230
    },
    {
      "epoch": 0.9622809622809623,
      "grad_norm": 0.8571543097496033,
      "learning_rate": 2.4226314226314224e-05,
      "loss": 0.0811,
      "step": 3240
    },
    {
      "epoch": 0.9652509652509652,
      "grad_norm": 0.7995705008506775,
      "learning_rate": 2.420849420849421e-05,
      "loss": 0.0602,
      "step": 3250
    },
    {
      "epoch": 0.9682209682209683,
      "grad_norm": 0.6367846727371216,
      "learning_rate": 2.4190674190674192e-05,
      "loss": 0.0635,
      "step": 3260
    },
    {
      "epoch": 0.9711909711909712,
      "grad_norm": 0.6590407490730286,
      "learning_rate": 2.4172854172854174e-05,
      "loss": 0.0599,
      "step": 3270
    },
    {
      "epoch": 0.9741609741609741,
      "grad_norm": 1.3518911600112915,
      "learning_rate": 2.4155034155034156e-05,
      "loss": 0.0816,
      "step": 3280
    },
    {
      "epoch": 0.9771309771309772,
      "grad_norm": 0.7235366106033325,
      "learning_rate": 2.4137214137214138e-05,
      "loss": 0.0585,
      "step": 3290
    },
    {
      "epoch": 0.9801009801009801,
      "grad_norm": 0.7149487733840942,
      "learning_rate": 2.411939411939412e-05,
      "loss": 0.0611,
      "step": 3300
    },
    {
      "epoch": 0.9830709830709831,
      "grad_norm": 0.5758341550827026,
      "learning_rate": 2.4101574101574103e-05,
      "loss": 0.0575,
      "step": 3310
    },
    {
      "epoch": 0.986040986040986,
      "grad_norm": 0.6312692761421204,
      "learning_rate": 2.4083754083754085e-05,
      "loss": 0.0696,
      "step": 3320
    },
    {
      "epoch": 0.989010989010989,
      "grad_norm": 0.6071924567222595,
      "learning_rate": 2.4065934065934067e-05,
      "loss": 0.0736,
      "step": 3330
    },
    {
      "epoch": 0.991980991980992,
      "grad_norm": 1.0844005346298218,
      "learning_rate": 2.404811404811405e-05,
      "loss": 0.0608,
      "step": 3340
    },
    {
      "epoch": 0.9949509949509949,
      "grad_norm": 0.7785583734512329,
      "learning_rate": 2.403029403029403e-05,
      "loss": 0.0737,
      "step": 3350
    },
    {
      "epoch": 0.997920997920998,
      "grad_norm": 0.7778565287590027,
      "learning_rate": 2.4012474012474013e-05,
      "loss": 0.0584,
      "step": 3360
    },
    {
      "epoch": 1.0,
      "eval_f1": 0.33031292965957215,
      "eval_loss": 0.05296385660767555,
      "eval_runtime": 760.7693,
      "eval_samples_per_second": 49.974,
      "eval_steps_per_second": 0.782,
      "step": 3367
    },
    {
      "epoch": 1.0008910008910008,
      "grad_norm": 0.5195960402488708,
      "learning_rate": 2.3994653994653996e-05,
      "loss": 0.0643,
      "step": 3370
    },
    {
      "epoch": 1.0038610038610039,
      "grad_norm": 1.0091626644134521,
      "learning_rate": 2.3976833976833978e-05,
      "loss": 0.0659,
      "step": 3380
    },
    {
      "epoch": 1.006831006831007,
      "grad_norm": 0.6962494254112244,
      "learning_rate": 2.395901395901396e-05,
      "loss": 0.0741,
      "step": 3390
    },
    {
      "epoch": 1.0098010098010097,
      "grad_norm": 0.5195145010948181,
      "learning_rate": 2.3941193941193942e-05,
      "loss": 0.0661,
      "step": 3400
    },
    {
      "epoch": 1.0127710127710128,
      "grad_norm": 0.40490421652793884,
      "learning_rate": 2.3923373923373924e-05,
      "loss": 0.0667,
      "step": 3410
    },
    {
      "epoch": 1.0157410157410158,
      "grad_norm": 0.8673891425132751,
      "learning_rate": 2.3905553905553906e-05,
      "loss": 0.0648,
      "step": 3420
    },
    {
      "epoch": 1.0187110187110187,
      "grad_norm": 0.6482574343681335,
      "learning_rate": 2.388773388773389e-05,
      "loss": 0.0683,
      "step": 3430
    },
    {
      "epoch": 1.0216810216810217,
      "grad_norm": 1.0671459436416626,
      "learning_rate": 2.386991386991387e-05,
      "loss": 0.0731,
      "step": 3440
    },
    {
      "epoch": 1.0246510246510248,
      "grad_norm": 1.1997050046920776,
      "learning_rate": 2.3852093852093853e-05,
      "loss": 0.0824,
      "step": 3450
    },
    {
      "epoch": 1.0276210276210276,
      "grad_norm": 1.1399210691452026,
      "learning_rate": 2.3834273834273835e-05,
      "loss": 0.0631,
      "step": 3460
    },
    {
      "epoch": 1.0305910305910306,
      "grad_norm": 1.4414496421813965,
      "learning_rate": 2.3816453816453817e-05,
      "loss": 0.0644,
      "step": 3470
    },
    {
      "epoch": 1.0335610335610335,
      "grad_norm": 0.7612940669059753,
      "learning_rate": 2.37986337986338e-05,
      "loss": 0.0608,
      "step": 3480
    },
    {
      "epoch": 1.0365310365310365,
      "grad_norm": 1.8347103595733643,
      "learning_rate": 2.378081378081378e-05,
      "loss": 0.0853,
      "step": 3490
    },
    {
      "epoch": 1.0395010395010396,
      "grad_norm": 0.7422579526901245,
      "learning_rate": 2.3762993762993764e-05,
      "loss": 0.065,
      "step": 3500
    },
    {
      "epoch": 1.0424710424710424,
      "grad_norm": 0.4676852524280548,
      "learning_rate": 2.3745173745173746e-05,
      "loss": 0.0686,
      "step": 3510
    },
    {
      "epoch": 1.0454410454410454,
      "grad_norm": 0.7909456491470337,
      "learning_rate": 2.3727353727353728e-05,
      "loss": 0.0808,
      "step": 3520
    },
    {
      "epoch": 1.0484110484110485,
      "grad_norm": 1.0909233093261719,
      "learning_rate": 2.370953370953371e-05,
      "loss": 0.0786,
      "step": 3530
    },
    {
      "epoch": 1.0513810513810513,
      "grad_norm": 0.9881356358528137,
      "learning_rate": 2.3691713691713692e-05,
      "loss": 0.0745,
      "step": 3540
    },
    {
      "epoch": 1.0543510543510544,
      "grad_norm": 0.6421000361442566,
      "learning_rate": 2.3673893673893675e-05,
      "loss": 0.0681,
      "step": 3550
    },
    {
      "epoch": 1.0573210573210574,
      "grad_norm": 1.2213356494903564,
      "learning_rate": 2.3656073656073657e-05,
      "loss": 0.0691,
      "step": 3560
    },
    {
      "epoch": 1.0602910602910602,
      "grad_norm": 1.1838406324386597,
      "learning_rate": 2.363825363825364e-05,
      "loss": 0.0727,
      "step": 3570
    },
    {
      "epoch": 1.0632610632610633,
      "grad_norm": 0.8688098192214966,
      "learning_rate": 2.362043362043362e-05,
      "loss": 0.059,
      "step": 3580
    },
    {
      "epoch": 1.0662310662310661,
      "grad_norm": 0.891708493232727,
      "learning_rate": 2.3602613602613603e-05,
      "loss": 0.0652,
      "step": 3590
    },
    {
      "epoch": 1.0692010692010692,
      "grad_norm": 0.8974260091781616,
      "learning_rate": 2.3584793584793586e-05,
      "loss": 0.0627,
      "step": 3600
    },
    {
      "epoch": 1.0721710721710722,
      "grad_norm": 0.9412424564361572,
      "learning_rate": 2.3566973566973568e-05,
      "loss": 0.0773,
      "step": 3610
    },
    {
      "epoch": 1.075141075141075,
      "grad_norm": 0.8957776427268982,
      "learning_rate": 2.354915354915355e-05,
      "loss": 0.0771,
      "step": 3620
    },
    {
      "epoch": 1.078111078111078,
      "grad_norm": 0.9389250874519348,
      "learning_rate": 2.3531333531333532e-05,
      "loss": 0.0631,
      "step": 3630
    },
    {
      "epoch": 1.0810810810810811,
      "grad_norm": 0.716650664806366,
      "learning_rate": 2.3513513513513514e-05,
      "loss": 0.0525,
      "step": 3640
    },
    {
      "epoch": 1.084051084051084,
      "grad_norm": 0.779859721660614,
      "learning_rate": 2.3495693495693496e-05,
      "loss": 0.0824,
      "step": 3650
    },
    {
      "epoch": 1.087021087021087,
      "grad_norm": 0.5671702027320862,
      "learning_rate": 2.347787347787348e-05,
      "loss": 0.0783,
      "step": 3660
    },
    {
      "epoch": 1.08999108999109,
      "grad_norm": 0.8227300643920898,
      "learning_rate": 2.346005346005346e-05,
      "loss": 0.0463,
      "step": 3670
    },
    {
      "epoch": 1.092961092961093,
      "grad_norm": 0.6845151782035828,
      "learning_rate": 2.3442233442233443e-05,
      "loss": 0.0614,
      "step": 3680
    },
    {
      "epoch": 1.095931095931096,
      "grad_norm": 1.5416568517684937,
      "learning_rate": 2.3424413424413425e-05,
      "loss": 0.0752,
      "step": 3690
    },
    {
      "epoch": 1.098901098901099,
      "grad_norm": 0.8240406513214111,
      "learning_rate": 2.3406593406593407e-05,
      "loss": 0.0693,
      "step": 3700
    },
    {
      "epoch": 1.1018711018711018,
      "grad_norm": 0.713246762752533,
      "learning_rate": 2.338877338877339e-05,
      "loss": 0.051,
      "step": 3710
    },
    {
      "epoch": 1.1048411048411049,
      "grad_norm": 0.9789752960205078,
      "learning_rate": 2.337095337095337e-05,
      "loss": 0.0681,
      "step": 3720
    },
    {
      "epoch": 1.107811107811108,
      "grad_norm": 1.0538547039031982,
      "learning_rate": 2.3353133353133354e-05,
      "loss": 0.0622,
      "step": 3730
    },
    {
      "epoch": 1.1107811107811107,
      "grad_norm": 0.9167300462722778,
      "learning_rate": 2.3335313335313336e-05,
      "loss": 0.0695,
      "step": 3740
    },
    {
      "epoch": 1.1137511137511138,
      "grad_norm": 0.740674614906311,
      "learning_rate": 2.3317493317493318e-05,
      "loss": 0.0666,
      "step": 3750
    },
    {
      "epoch": 1.1167211167211166,
      "grad_norm": 0.6684421300888062,
      "learning_rate": 2.32996732996733e-05,
      "loss": 0.0711,
      "step": 3760
    },
    {
      "epoch": 1.1196911196911197,
      "grad_norm": 0.6935126781463623,
      "learning_rate": 2.3281853281853282e-05,
      "loss": 0.0653,
      "step": 3770
    },
    {
      "epoch": 1.1226611226611227,
      "grad_norm": 1.0922201871871948,
      "learning_rate": 2.3264033264033265e-05,
      "loss": 0.0653,
      "step": 3780
    },
    {
      "epoch": 1.1256311256311256,
      "grad_norm": 0.747627854347229,
      "learning_rate": 2.3246213246213247e-05,
      "loss": 0.0672,
      "step": 3790
    },
    {
      "epoch": 1.1286011286011286,
      "grad_norm": 1.3024978637695312,
      "learning_rate": 2.322839322839323e-05,
      "loss": 0.0725,
      "step": 3800
    },
    {
      "epoch": 1.1315711315711316,
      "grad_norm": 0.9191585779190063,
      "learning_rate": 2.321057321057321e-05,
      "loss": 0.0662,
      "step": 3810
    },
    {
      "epoch": 1.1345411345411345,
      "grad_norm": 1.0653009414672852,
      "learning_rate": 2.3192753192753193e-05,
      "loss": 0.0685,
      "step": 3820
    },
    {
      "epoch": 1.1375111375111375,
      "grad_norm": 1.1122196912765503,
      "learning_rate": 2.3174933174933175e-05,
      "loss": 0.073,
      "step": 3830
    },
    {
      "epoch": 1.1404811404811406,
      "grad_norm": 0.8093435764312744,
      "learning_rate": 2.3157113157113158e-05,
      "loss": 0.0709,
      "step": 3840
    },
    {
      "epoch": 1.1434511434511434,
      "grad_norm": 0.4847230613231659,
      "learning_rate": 2.313929313929314e-05,
      "loss": 0.0625,
      "step": 3850
    },
    {
      "epoch": 1.1464211464211465,
      "grad_norm": 0.8691696524620056,
      "learning_rate": 2.3121473121473122e-05,
      "loss": 0.0558,
      "step": 3860
    },
    {
      "epoch": 1.1493911493911493,
      "grad_norm": 1.140037178993225,
      "learning_rate": 2.3103653103653104e-05,
      "loss": 0.0704,
      "step": 3870
    },
    {
      "epoch": 1.1523611523611523,
      "grad_norm": 0.8088730573654175,
      "learning_rate": 2.3085833085833086e-05,
      "loss": 0.0666,
      "step": 3880
    },
    {
      "epoch": 1.1553311553311554,
      "grad_norm": 0.6530190110206604,
      "learning_rate": 2.306801306801307e-05,
      "loss": 0.0658,
      "step": 3890
    },
    {
      "epoch": 1.1583011583011582,
      "grad_norm": 0.7550728917121887,
      "learning_rate": 2.305019305019305e-05,
      "loss": 0.0715,
      "step": 3900
    },
    {
      "epoch": 1.1612711612711613,
      "grad_norm": 0.6546505689620972,
      "learning_rate": 2.3032373032373033e-05,
      "loss": 0.0682,
      "step": 3910
    },
    {
      "epoch": 1.1642411642411643,
      "grad_norm": 1.3673691749572754,
      "learning_rate": 2.3014553014553015e-05,
      "loss": 0.0628,
      "step": 3920
    },
    {
      "epoch": 1.1672111672111671,
      "grad_norm": 1.3295186758041382,
      "learning_rate": 2.2996732996732997e-05,
      "loss": 0.0632,
      "step": 3930
    },
    {
      "epoch": 1.1701811701811702,
      "grad_norm": 0.8255389332771301,
      "learning_rate": 2.297891297891298e-05,
      "loss": 0.0598,
      "step": 3940
    },
    {
      "epoch": 1.1731511731511732,
      "grad_norm": 0.7321466207504272,
      "learning_rate": 2.2961092961092965e-05,
      "loss": 0.0521,
      "step": 3950
    },
    {
      "epoch": 1.176121176121176,
      "grad_norm": 1.0865906476974487,
      "learning_rate": 2.2943272943272944e-05,
      "loss": 0.0705,
      "step": 3960
    },
    {
      "epoch": 1.179091179091179,
      "grad_norm": 0.691181480884552,
      "learning_rate": 2.2925452925452926e-05,
      "loss": 0.0805,
      "step": 3970
    },
    {
      "epoch": 1.1820611820611822,
      "grad_norm": 0.6251980662345886,
      "learning_rate": 2.2907632907632908e-05,
      "loss": 0.061,
      "step": 3980
    },
    {
      "epoch": 1.185031185031185,
      "grad_norm": 0.7137186527252197,
      "learning_rate": 2.288981288981289e-05,
      "loss": 0.0657,
      "step": 3990
    },
    {
      "epoch": 1.188001188001188,
      "grad_norm": 0.5930067896842957,
      "learning_rate": 2.2871992871992872e-05,
      "loss": 0.065,
      "step": 4000
    },
    {
      "epoch": 1.190971190971191,
      "grad_norm": 0.8055261373519897,
      "learning_rate": 2.2854172854172855e-05,
      "loss": 0.0605,
      "step": 4010
    },
    {
      "epoch": 1.193941193941194,
      "grad_norm": 0.8491489887237549,
      "learning_rate": 2.283635283635284e-05,
      "loss": 0.0768,
      "step": 4020
    },
    {
      "epoch": 1.196911196911197,
      "grad_norm": 0.5159242749214172,
      "learning_rate": 2.281853281853282e-05,
      "loss": 0.0577,
      "step": 4030
    },
    {
      "epoch": 1.1998811998811998,
      "grad_norm": 1.0682775974273682,
      "learning_rate": 2.28007128007128e-05,
      "loss": 0.0669,
      "step": 4040
    },
    {
      "epoch": 1.2028512028512028,
      "grad_norm": 0.7418850660324097,
      "learning_rate": 2.2782892782892783e-05,
      "loss": 0.064,
      "step": 4050
    },
    {
      "epoch": 1.2058212058212059,
      "grad_norm": 1.3865784406661987,
      "learning_rate": 2.2765072765072765e-05,
      "loss": 0.056,
      "step": 4060
    },
    {
      "epoch": 1.2087912087912087,
      "grad_norm": 1.0742945671081543,
      "learning_rate": 2.2747252747252748e-05,
      "loss": 0.0495,
      "step": 4070
    },
    {
      "epoch": 1.2117612117612118,
      "grad_norm": 1.8992480039596558,
      "learning_rate": 2.272943272943273e-05,
      "loss": 0.0603,
      "step": 4080
    },
    {
      "epoch": 1.2147312147312148,
      "grad_norm": 0.701547384262085,
      "learning_rate": 2.2711612711612715e-05,
      "loss": 0.0733,
      "step": 4090
    },
    {
      "epoch": 1.2177012177012176,
      "grad_norm": 0.7614421844482422,
      "learning_rate": 2.2693792693792694e-05,
      "loss": 0.0634,
      "step": 4100
    },
    {
      "epoch": 1.2206712206712207,
      "grad_norm": 0.6779104471206665,
      "learning_rate": 2.2675972675972676e-05,
      "loss": 0.0631,
      "step": 4110
    },
    {
      "epoch": 1.2236412236412235,
      "grad_norm": 0.7199757695198059,
      "learning_rate": 2.265815265815266e-05,
      "loss": 0.0628,
      "step": 4120
    },
    {
      "epoch": 1.2266112266112266,
      "grad_norm": 0.8466992378234863,
      "learning_rate": 2.264033264033264e-05,
      "loss": 0.0784,
      "step": 4130
    },
    {
      "epoch": 1.2295812295812296,
      "grad_norm": 0.5136345028877258,
      "learning_rate": 2.2622512622512623e-05,
      "loss": 0.0866,
      "step": 4140
    },
    {
      "epoch": 1.2325512325512324,
      "grad_norm": 0.7699841260910034,
      "learning_rate": 2.2604692604692605e-05,
      "loss": 0.0607,
      "step": 4150
    },
    {
      "epoch": 1.2355212355212355,
      "grad_norm": 0.61440509557724,
      "learning_rate": 2.258687258687259e-05,
      "loss": 0.0721,
      "step": 4160
    },
    {
      "epoch": 1.2384912384912385,
      "grad_norm": 0.5147203803062439,
      "learning_rate": 2.256905256905257e-05,
      "loss": 0.0615,
      "step": 4170
    },
    {
      "epoch": 1.2414612414612414,
      "grad_norm": 1.0179754495620728,
      "learning_rate": 2.255123255123255e-05,
      "loss": 0.0669,
      "step": 4180
    },
    {
      "epoch": 1.2444312444312444,
      "grad_norm": 1.2487351894378662,
      "learning_rate": 2.2533412533412534e-05,
      "loss": 0.0705,
      "step": 4190
    },
    {
      "epoch": 1.2474012474012475,
      "grad_norm": 0.8838121294975281,
      "learning_rate": 2.2515592515592516e-05,
      "loss": 0.0582,
      "step": 4200
    },
    {
      "epoch": 1.2503712503712503,
      "grad_norm": 0.5740695595741272,
      "learning_rate": 2.2497772497772498e-05,
      "loss": 0.0704,
      "step": 4210
    },
    {
      "epoch": 1.2533412533412533,
      "grad_norm": 1.0215144157409668,
      "learning_rate": 2.247995247995248e-05,
      "loss": 0.0672,
      "step": 4220
    },
    {
      "epoch": 1.2563112563112564,
      "grad_norm": 0.5340459942817688,
      "learning_rate": 2.2462132462132466e-05,
      "loss": 0.0637,
      "step": 4230
    },
    {
      "epoch": 1.2592812592812592,
      "grad_norm": 1.070460557937622,
      "learning_rate": 2.2444312444312444e-05,
      "loss": 0.0545,
      "step": 4240
    },
    {
      "epoch": 1.2622512622512623,
      "grad_norm": 0.8050958514213562,
      "learning_rate": 2.2426492426492427e-05,
      "loss": 0.0732,
      "step": 4250
    },
    {
      "epoch": 1.2652212652212653,
      "grad_norm": 1.1751881837844849,
      "learning_rate": 2.240867240867241e-05,
      "loss": 0.0643,
      "step": 4260
    },
    {
      "epoch": 1.2681912681912682,
      "grad_norm": 0.7636763453483582,
      "learning_rate": 2.239085239085239e-05,
      "loss": 0.0653,
      "step": 4270
    },
    {
      "epoch": 1.2711612711612712,
      "grad_norm": 0.9791406393051147,
      "learning_rate": 2.2373032373032373e-05,
      "loss": 0.0584,
      "step": 4280
    },
    {
      "epoch": 1.2741312741312742,
      "grad_norm": 0.7061544060707092,
      "learning_rate": 2.2355212355212355e-05,
      "loss": 0.0702,
      "step": 4290
    },
    {
      "epoch": 1.277101277101277,
      "grad_norm": 1.0918234586715698,
      "learning_rate": 2.233739233739234e-05,
      "loss": 0.0597,
      "step": 4300
    },
    {
      "epoch": 1.2800712800712801,
      "grad_norm": 1.0463823080062866,
      "learning_rate": 2.231957231957232e-05,
      "loss": 0.0571,
      "step": 4310
    },
    {
      "epoch": 1.2830412830412832,
      "grad_norm": 0.8148181438446045,
      "learning_rate": 2.2301752301752302e-05,
      "loss": 0.0825,
      "step": 4320
    },
    {
      "epoch": 1.286011286011286,
      "grad_norm": 0.9247533082962036,
      "learning_rate": 2.2283932283932284e-05,
      "loss": 0.0643,
      "step": 4330
    },
    {
      "epoch": 1.288981288981289,
      "grad_norm": 0.8927013278007507,
      "learning_rate": 2.2266112266112266e-05,
      "loss": 0.0695,
      "step": 4340
    },
    {
      "epoch": 1.2919512919512919,
      "grad_norm": 1.1298048496246338,
      "learning_rate": 2.2248292248292248e-05,
      "loss": 0.0601,
      "step": 4350
    },
    {
      "epoch": 1.294921294921295,
      "grad_norm": 0.40580281615257263,
      "learning_rate": 2.223047223047223e-05,
      "loss": 0.0742,
      "step": 4360
    },
    {
      "epoch": 1.2978912978912978,
      "grad_norm": 0.471123069524765,
      "learning_rate": 2.2212652212652216e-05,
      "loss": 0.0638,
      "step": 4370
    },
    {
      "epoch": 1.3008613008613008,
      "grad_norm": 1.53095543384552,
      "learning_rate": 2.2194832194832195e-05,
      "loss": 0.0695,
      "step": 4380
    },
    {
      "epoch": 1.3038313038313039,
      "grad_norm": 0.8293925523757935,
      "learning_rate": 2.2177012177012177e-05,
      "loss": 0.0526,
      "step": 4390
    },
    {
      "epoch": 1.3068013068013067,
      "grad_norm": 0.66612708568573,
      "learning_rate": 2.215919215919216e-05,
      "loss": 0.0776,
      "step": 4400
    },
    {
      "epoch": 1.3097713097713097,
      "grad_norm": 0.9548154473304749,
      "learning_rate": 2.214137214137214e-05,
      "loss": 0.0729,
      "step": 4410
    },
    {
      "epoch": 1.3127413127413128,
      "grad_norm": 0.5580644011497498,
      "learning_rate": 2.2123552123552123e-05,
      "loss": 0.0831,
      "step": 4420
    },
    {
      "epoch": 1.3157113157113156,
      "grad_norm": 0.5851168036460876,
      "learning_rate": 2.2105732105732106e-05,
      "loss": 0.0569,
      "step": 4430
    },
    {
      "epoch": 1.3186813186813187,
      "grad_norm": 0.4882819652557373,
      "learning_rate": 2.208791208791209e-05,
      "loss": 0.0692,
      "step": 4440
    },
    {
      "epoch": 1.3216513216513217,
      "grad_norm": 0.44905969500541687,
      "learning_rate": 2.207009207009207e-05,
      "loss": 0.0577,
      "step": 4450
    },
    {
      "epoch": 1.3246213246213245,
      "grad_norm": 0.8258479237556458,
      "learning_rate": 2.2052272052272052e-05,
      "loss": 0.0516,
      "step": 4460
    },
    {
      "epoch": 1.3275913275913276,
      "grad_norm": 0.8680855631828308,
      "learning_rate": 2.2034452034452034e-05,
      "loss": 0.0445,
      "step": 4470
    },
    {
      "epoch": 1.3305613305613306,
      "grad_norm": 0.6597207188606262,
      "learning_rate": 2.2016632016632017e-05,
      "loss": 0.0558,
      "step": 4480
    },
    {
      "epoch": 1.3335313335313335,
      "grad_norm": 1.0117040872573853,
      "learning_rate": 2.1998811998812e-05,
      "loss": 0.0676,
      "step": 4490
    },
    {
      "epoch": 1.3365013365013365,
      "grad_norm": 0.6703979969024658,
      "learning_rate": 2.198099198099198e-05,
      "loss": 0.0517,
      "step": 4500
    },
    {
      "epoch": 1.3394713394713396,
      "grad_norm": 1.2186622619628906,
      "learning_rate": 2.1963171963171966e-05,
      "loss": 0.0653,
      "step": 4510
    },
    {
      "epoch": 1.3424413424413424,
      "grad_norm": 0.6504166722297668,
      "learning_rate": 2.1945351945351945e-05,
      "loss": 0.0737,
      "step": 4520
    },
    {
      "epoch": 1.3454113454113454,
      "grad_norm": 0.8014166355133057,
      "learning_rate": 2.1927531927531927e-05,
      "loss": 0.0667,
      "step": 4530
    },
    {
      "epoch": 1.3483813483813485,
      "grad_norm": 0.6877809166908264,
      "learning_rate": 2.190971190971191e-05,
      "loss": 0.0683,
      "step": 4540
    },
    {
      "epoch": 1.3513513513513513,
      "grad_norm": 1.1891028881072998,
      "learning_rate": 2.1891891891891892e-05,
      "loss": 0.0659,
      "step": 4550
    },
    {
      "epoch": 1.3543213543213544,
      "grad_norm": 1.254156470298767,
      "learning_rate": 2.1874071874071874e-05,
      "loss": 0.065,
      "step": 4560
    },
    {
      "epoch": 1.3572913572913574,
      "grad_norm": 1.4208029508590698,
      "learning_rate": 2.1856251856251856e-05,
      "loss": 0.0733,
      "step": 4570
    },
    {
      "epoch": 1.3602613602613602,
      "grad_norm": 1.2454571723937988,
      "learning_rate": 2.183843183843184e-05,
      "loss": 0.0639,
      "step": 4580
    },
    {
      "epoch": 1.3632313632313633,
      "grad_norm": 0.7917842864990234,
      "learning_rate": 2.1820611820611824e-05,
      "loss": 0.0719,
      "step": 4590
    },
    {
      "epoch": 1.3662013662013661,
      "grad_norm": 0.7804074287414551,
      "learning_rate": 2.1802791802791803e-05,
      "loss": 0.0667,
      "step": 4600
    },
    {
      "epoch": 1.3691713691713692,
      "grad_norm": 0.9394906163215637,
      "learning_rate": 2.1784971784971785e-05,
      "loss": 0.0772,
      "step": 4610
    },
    {
      "epoch": 1.3721413721413722,
      "grad_norm": 0.9124456644058228,
      "learning_rate": 2.1767151767151767e-05,
      "loss": 0.0641,
      "step": 4620
    },
    {
      "epoch": 1.375111375111375,
      "grad_norm": 0.8035851716995239,
      "learning_rate": 2.174933174933175e-05,
      "loss": 0.0537,
      "step": 4630
    },
    {
      "epoch": 1.378081378081378,
      "grad_norm": 0.603728711605072,
      "learning_rate": 2.173151173151173e-05,
      "loss": 0.0671,
      "step": 4640
    },
    {
      "epoch": 1.381051381051381,
      "grad_norm": 0.5484776496887207,
      "learning_rate": 2.1713691713691717e-05,
      "loss": 0.0641,
      "step": 4650
    },
    {
      "epoch": 1.384021384021384,
      "grad_norm": 1.0918580293655396,
      "learning_rate": 2.16958716958717e-05,
      "loss": 0.0701,
      "step": 4660
    },
    {
      "epoch": 1.386991386991387,
      "grad_norm": 0.7597218751907349,
      "learning_rate": 2.1678051678051678e-05,
      "loss": 0.0612,
      "step": 4670
    },
    {
      "epoch": 1.3899613899613898,
      "grad_norm": 1.020501971244812,
      "learning_rate": 2.166023166023166e-05,
      "loss": 0.0683,
      "step": 4680
    },
    {
      "epoch": 1.392931392931393,
      "grad_norm": 0.3588350713253021,
      "learning_rate": 2.1642411642411642e-05,
      "loss": 0.0578,
      "step": 4690
    },
    {
      "epoch": 1.395901395901396,
      "grad_norm": 0.6665700674057007,
      "learning_rate": 2.1624591624591624e-05,
      "loss": 0.0586,
      "step": 4700
    },
    {
      "epoch": 1.3988713988713988,
      "grad_norm": 0.5543581247329712,
      "learning_rate": 2.1606771606771606e-05,
      "loss": 0.0572,
      "step": 4710
    },
    {
      "epoch": 1.4018414018414018,
      "grad_norm": 0.8856662511825562,
      "learning_rate": 2.1588951588951592e-05,
      "loss": 0.0647,
      "step": 4720
    },
    {
      "epoch": 1.4048114048114049,
      "grad_norm": 0.7327395081520081,
      "learning_rate": 2.1571131571131574e-05,
      "loss": 0.055,
      "step": 4730
    },
    {
      "epoch": 1.4077814077814077,
      "grad_norm": 1.0437580347061157,
      "learning_rate": 2.1553311553311553e-05,
      "loss": 0.052,
      "step": 4740
    },
    {
      "epoch": 1.4107514107514108,
      "grad_norm": 0.6243680119514465,
      "learning_rate": 2.1535491535491535e-05,
      "loss": 0.0606,
      "step": 4750
    },
    {
      "epoch": 1.4137214137214138,
      "grad_norm": 0.8211525678634644,
      "learning_rate": 2.1517671517671517e-05,
      "loss": 0.0586,
      "step": 4760
    },
    {
      "epoch": 1.4166914166914166,
      "grad_norm": 0.8621878027915955,
      "learning_rate": 2.14998514998515e-05,
      "loss": 0.0732,
      "step": 4770
    },
    {
      "epoch": 1.4196614196614197,
      "grad_norm": 0.9054310321807861,
      "learning_rate": 2.148203148203148e-05,
      "loss": 0.0548,
      "step": 4780
    },
    {
      "epoch": 1.4226314226314227,
      "grad_norm": 1.122381567955017,
      "learning_rate": 2.1464211464211467e-05,
      "loss": 0.0588,
      "step": 4790
    },
    {
      "epoch": 1.4256014256014256,
      "grad_norm": 1.5067578554153442,
      "learning_rate": 2.144639144639145e-05,
      "loss": 0.0592,
      "step": 4800
    },
    {
      "epoch": 1.4285714285714286,
      "grad_norm": 0.9946634769439697,
      "learning_rate": 2.1428571428571428e-05,
      "loss": 0.0588,
      "step": 4810
    },
    {
      "epoch": 1.4315414315414317,
      "grad_norm": 0.7225719094276428,
      "learning_rate": 2.141075141075141e-05,
      "loss": 0.0669,
      "step": 4820
    },
    {
      "epoch": 1.4345114345114345,
      "grad_norm": 0.7336903810501099,
      "learning_rate": 2.1392931392931392e-05,
      "loss": 0.0707,
      "step": 4830
    },
    {
      "epoch": 1.4374814374814375,
      "grad_norm": 1.2703900337219238,
      "learning_rate": 2.1375111375111375e-05,
      "loss": 0.0666,
      "step": 4840
    },
    {
      "epoch": 1.4404514404514406,
      "grad_norm": 1.0433528423309326,
      "learning_rate": 2.1357291357291357e-05,
      "loss": 0.0754,
      "step": 4850
    },
    {
      "epoch": 1.4434214434214434,
      "grad_norm": 0.6614531874656677,
      "learning_rate": 2.1339471339471342e-05,
      "loss": 0.0685,
      "step": 4860
    },
    {
      "epoch": 1.4463914463914465,
      "grad_norm": 0.7256604433059692,
      "learning_rate": 2.1321651321651325e-05,
      "loss": 0.0488,
      "step": 4870
    },
    {
      "epoch": 1.4493614493614493,
      "grad_norm": 0.781690776348114,
      "learning_rate": 2.1303831303831303e-05,
      "loss": 0.0599,
      "step": 4880
    },
    {
      "epoch": 1.4523314523314523,
      "grad_norm": 0.5681362748146057,
      "learning_rate": 2.1286011286011286e-05,
      "loss": 0.0654,
      "step": 4890
    },
    {
      "epoch": 1.4553014553014554,
      "grad_norm": 0.6243995428085327,
      "learning_rate": 2.1268191268191268e-05,
      "loss": 0.0695,
      "step": 4900
    },
    {
      "epoch": 1.4582714582714582,
      "grad_norm": 0.7592119574546814,
      "learning_rate": 2.125037125037125e-05,
      "loss": 0.0562,
      "step": 4910
    },
    {
      "epoch": 1.4612414612414613,
      "grad_norm": 0.740118682384491,
      "learning_rate": 2.1232551232551232e-05,
      "loss": 0.0617,
      "step": 4920
    },
    {
      "epoch": 1.464211464211464,
      "grad_norm": 0.8962658643722534,
      "learning_rate": 2.1214731214731218e-05,
      "loss": 0.0657,
      "step": 4930
    },
    {
      "epoch": 1.4671814671814671,
      "grad_norm": 0.6510963439941406,
      "learning_rate": 2.11969111969112e-05,
      "loss": 0.0639,
      "step": 4940
    },
    {
      "epoch": 1.4701514701514702,
      "grad_norm": 1.0592265129089355,
      "learning_rate": 2.117909117909118e-05,
      "loss": 0.0644,
      "step": 4950
    },
    {
      "epoch": 1.473121473121473,
      "grad_norm": 0.9842652678489685,
      "learning_rate": 2.116127116127116e-05,
      "loss": 0.0573,
      "step": 4960
    },
    {
      "epoch": 1.476091476091476,
      "grad_norm": 0.7804544568061829,
      "learning_rate": 2.1143451143451143e-05,
      "loss": 0.0605,
      "step": 4970
    },
    {
      "epoch": 1.4790614790614791,
      "grad_norm": 0.7872318625450134,
      "learning_rate": 2.1125631125631125e-05,
      "loss": 0.0561,
      "step": 4980
    },
    {
      "epoch": 1.482031482031482,
      "grad_norm": 0.9400390982627869,
      "learning_rate": 2.1107811107811107e-05,
      "loss": 0.0684,
      "step": 4990
    },
    {
      "epoch": 1.485001485001485,
      "grad_norm": 1.317718505859375,
      "learning_rate": 2.1089991089991093e-05,
      "loss": 0.0586,
      "step": 5000
    },
    {
      "epoch": 1.487971487971488,
      "grad_norm": 0.41417181491851807,
      "learning_rate": 2.1072171072171075e-05,
      "loss": 0.0671,
      "step": 5010
    },
    {
      "epoch": 1.4909414909414909,
      "grad_norm": 0.775679886341095,
      "learning_rate": 2.1054351054351054e-05,
      "loss": 0.0753,
      "step": 5020
    },
    {
      "epoch": 1.493911493911494,
      "grad_norm": 1.1018742322921753,
      "learning_rate": 2.1036531036531036e-05,
      "loss": 0.0684,
      "step": 5030
    },
    {
      "epoch": 1.496881496881497,
      "grad_norm": 0.877463161945343,
      "learning_rate": 2.1018711018711018e-05,
      "loss": 0.0622,
      "step": 5040
    },
    {
      "epoch": 1.4998514998514998,
      "grad_norm": 0.44593024253845215,
      "learning_rate": 2.1000891000891e-05,
      "loss": 0.0829,
      "step": 5050
    },
    {
      "epoch": 1.5028215028215028,
      "grad_norm": 0.7530653476715088,
      "learning_rate": 2.0983070983070982e-05,
      "loss": 0.0573,
      "step": 5060
    },
    {
      "epoch": 1.505791505791506,
      "grad_norm": 0.6737465262413025,
      "learning_rate": 2.0965250965250968e-05,
      "loss": 0.0657,
      "step": 5070
    },
    {
      "epoch": 1.5087615087615087,
      "grad_norm": 1.2247141599655151,
      "learning_rate": 2.094743094743095e-05,
      "loss": 0.0715,
      "step": 5080
    },
    {
      "epoch": 1.5117315117315118,
      "grad_norm": 0.5894985795021057,
      "learning_rate": 2.092961092961093e-05,
      "loss": 0.0635,
      "step": 5090
    },
    {
      "epoch": 1.5147015147015148,
      "grad_norm": 0.6736055016517639,
      "learning_rate": 2.091179091179091e-05,
      "loss": 0.0699,
      "step": 5100
    },
    {
      "epoch": 1.5176715176715176,
      "grad_norm": 1.1154353618621826,
      "learning_rate": 2.0893970893970893e-05,
      "loss": 0.0662,
      "step": 5110
    },
    {
      "epoch": 1.5206415206415207,
      "grad_norm": 1.1602790355682373,
      "learning_rate": 2.0876150876150875e-05,
      "loss": 0.0611,
      "step": 5120
    },
    {
      "epoch": 1.5236115236115237,
      "grad_norm": 0.6141194701194763,
      "learning_rate": 2.0858330858330858e-05,
      "loss": 0.0655,
      "step": 5130
    },
    {
      "epoch": 1.5265815265815266,
      "grad_norm": 0.7376190423965454,
      "learning_rate": 2.0840510840510843e-05,
      "loss": 0.055,
      "step": 5140
    },
    {
      "epoch": 1.5295515295515294,
      "grad_norm": 1.0901671648025513,
      "learning_rate": 2.0822690822690825e-05,
      "loss": 0.0678,
      "step": 5150
    },
    {
      "epoch": 1.5325215325215327,
      "grad_norm": 0.7825741171836853,
      "learning_rate": 2.0804870804870808e-05,
      "loss": 0.0722,
      "step": 5160
    },
    {
      "epoch": 1.5354915354915355,
      "grad_norm": 1.3154313564300537,
      "learning_rate": 2.0787050787050786e-05,
      "loss": 0.0651,
      "step": 5170
    },
    {
      "epoch": 1.5384615384615383,
      "grad_norm": 0.9979254603385925,
      "learning_rate": 2.076923076923077e-05,
      "loss": 0.0623,
      "step": 5180
    },
    {
      "epoch": 1.5414315414315416,
      "grad_norm": 0.6638979911804199,
      "learning_rate": 2.075141075141075e-05,
      "loss": 0.0724,
      "step": 5190
    },
    {
      "epoch": 1.5444015444015444,
      "grad_norm": 0.9916415214538574,
      "learning_rate": 2.0733590733590733e-05,
      "loss": 0.0665,
      "step": 5200
    },
    {
      "epoch": 1.5473715473715473,
      "grad_norm": 1.4179552793502808,
      "learning_rate": 2.071577071577072e-05,
      "loss": 0.062,
      "step": 5210
    },
    {
      "epoch": 1.5503415503415503,
      "grad_norm": 0.7684439420700073,
      "learning_rate": 2.06979506979507e-05,
      "loss": 0.0575,
      "step": 5220
    },
    {
      "epoch": 1.5533115533115534,
      "grad_norm": 0.7808251976966858,
      "learning_rate": 2.0680130680130683e-05,
      "loss": 0.065,
      "step": 5230
    },
    {
      "epoch": 1.5562815562815562,
      "grad_norm": 0.6247586011886597,
      "learning_rate": 2.066231066231066e-05,
      "loss": 0.0611,
      "step": 5240
    },
    {
      "epoch": 1.5592515592515592,
      "grad_norm": 0.9080334305763245,
      "learning_rate": 2.0644490644490644e-05,
      "loss": 0.0591,
      "step": 5250
    },
    {
      "epoch": 1.5622215622215623,
      "grad_norm": 1.0106734037399292,
      "learning_rate": 2.0626670626670626e-05,
      "loss": 0.055,
      "step": 5260
    },
    {
      "epoch": 1.565191565191565,
      "grad_norm": 0.8516258001327515,
      "learning_rate": 2.0608850608850608e-05,
      "loss": 0.0609,
      "step": 5270
    },
    {
      "epoch": 1.5681615681615682,
      "grad_norm": 0.7109993100166321,
      "learning_rate": 2.0591030591030594e-05,
      "loss": 0.0644,
      "step": 5280
    },
    {
      "epoch": 1.5711315711315712,
      "grad_norm": 1.2328643798828125,
      "learning_rate": 2.0573210573210576e-05,
      "loss": 0.0654,
      "step": 5290
    },
    {
      "epoch": 1.574101574101574,
      "grad_norm": 0.8040093183517456,
      "learning_rate": 2.0555390555390558e-05,
      "loss": 0.0558,
      "step": 5300
    },
    {
      "epoch": 1.577071577071577,
      "grad_norm": 0.4768785834312439,
      "learning_rate": 2.0537570537570537e-05,
      "loss": 0.0486,
      "step": 5310
    },
    {
      "epoch": 1.5800415800415801,
      "grad_norm": 0.8780914545059204,
      "learning_rate": 2.051975051975052e-05,
      "loss": 0.0597,
      "step": 5320
    },
    {
      "epoch": 1.583011583011583,
      "grad_norm": 0.5771424770355225,
      "learning_rate": 2.05019305019305e-05,
      "loss": 0.0696,
      "step": 5330
    },
    {
      "epoch": 1.585981585981586,
      "grad_norm": 0.8761149644851685,
      "learning_rate": 2.0484110484110483e-05,
      "loss": 0.0617,
      "step": 5340
    },
    {
      "epoch": 1.588951588951589,
      "grad_norm": 0.8866630792617798,
      "learning_rate": 2.046629046629047e-05,
      "loss": 0.0619,
      "step": 5350
    },
    {
      "epoch": 1.5919215919215919,
      "grad_norm": 0.8091713190078735,
      "learning_rate": 2.044847044847045e-05,
      "loss": 0.0705,
      "step": 5360
    },
    {
      "epoch": 1.594891594891595,
      "grad_norm": 1.079195261001587,
      "learning_rate": 2.0430650430650433e-05,
      "loss": 0.0555,
      "step": 5370
    },
    {
      "epoch": 1.597861597861598,
      "grad_norm": 0.5570653676986694,
      "learning_rate": 2.0412830412830412e-05,
      "loss": 0.0759,
      "step": 5380
    },
    {
      "epoch": 1.6008316008316008,
      "grad_norm": 0.8151302933692932,
      "learning_rate": 2.0395010395010394e-05,
      "loss": 0.0644,
      "step": 5390
    },
    {
      "epoch": 1.6038016038016036,
      "grad_norm": 0.9399856925010681,
      "learning_rate": 2.0377190377190376e-05,
      "loss": 0.0507,
      "step": 5400
    },
    {
      "epoch": 1.606771606771607,
      "grad_norm": 0.7044445872306824,
      "learning_rate": 2.035937035937036e-05,
      "loss": 0.0638,
      "step": 5410
    },
    {
      "epoch": 1.6097416097416097,
      "grad_norm": 0.9745432734489441,
      "learning_rate": 2.0341550341550344e-05,
      "loss": 0.0588,
      "step": 5420
    },
    {
      "epoch": 1.6127116127116126,
      "grad_norm": 0.8628720045089722,
      "learning_rate": 2.0323730323730326e-05,
      "loss": 0.0627,
      "step": 5430
    },
    {
      "epoch": 1.6156816156816158,
      "grad_norm": 0.8440065979957581,
      "learning_rate": 2.0305910305910308e-05,
      "loss": 0.056,
      "step": 5440
    },
    {
      "epoch": 1.6186516186516187,
      "grad_norm": 0.7349764108657837,
      "learning_rate": 2.0288090288090287e-05,
      "loss": 0.0524,
      "step": 5450
    },
    {
      "epoch": 1.6216216216216215,
      "grad_norm": 0.4457996189594269,
      "learning_rate": 2.027027027027027e-05,
      "loss": 0.0678,
      "step": 5460
    },
    {
      "epoch": 1.6245916245916245,
      "grad_norm": 1.1030367612838745,
      "learning_rate": 2.025245025245025e-05,
      "loss": 0.075,
      "step": 5470
    },
    {
      "epoch": 1.6275616275616276,
      "grad_norm": 0.6578485369682312,
      "learning_rate": 2.0234630234630234e-05,
      "loss": 0.0586,
      "step": 5480
    },
    {
      "epoch": 1.6305316305316304,
      "grad_norm": 0.8512071967124939,
      "learning_rate": 2.021681021681022e-05,
      "loss": 0.0619,
      "step": 5490
    },
    {
      "epoch": 1.6335016335016335,
      "grad_norm": 0.3718603551387787,
      "learning_rate": 2.01989901989902e-05,
      "loss": 0.077,
      "step": 5500
    },
    {
      "epoch": 1.6364716364716365,
      "grad_norm": 0.6400578618049622,
      "learning_rate": 2.0181170181170183e-05,
      "loss": 0.0443,
      "step": 5510
    },
    {
      "epoch": 1.6394416394416393,
      "grad_norm": 0.6334373354911804,
      "learning_rate": 2.0163350163350162e-05,
      "loss": 0.0749,
      "step": 5520
    },
    {
      "epoch": 1.6424116424116424,
      "grad_norm": 1.0908116102218628,
      "learning_rate": 2.0145530145530144e-05,
      "loss": 0.0597,
      "step": 5530
    },
    {
      "epoch": 1.6453816453816454,
      "grad_norm": 0.8702746033668518,
      "learning_rate": 2.0127710127710127e-05,
      "loss": 0.0622,
      "step": 5540
    },
    {
      "epoch": 1.6483516483516483,
      "grad_norm": 0.816605269908905,
      "learning_rate": 2.010989010989011e-05,
      "loss": 0.0555,
      "step": 5550
    },
    {
      "epoch": 1.6513216513216513,
      "grad_norm": 1.7561945915222168,
      "learning_rate": 2.0092070092070094e-05,
      "loss": 0.0609,
      "step": 5560
    },
    {
      "epoch": 1.6542916542916544,
      "grad_norm": 1.0770184993743896,
      "learning_rate": 2.0074250074250076e-05,
      "loss": 0.0737,
      "step": 5570
    },
    {
      "epoch": 1.6572616572616572,
      "grad_norm": 0.8192417025566101,
      "learning_rate": 2.005643005643006e-05,
      "loss": 0.0607,
      "step": 5580
    },
    {
      "epoch": 1.6602316602316602,
      "grad_norm": 0.49099376797676086,
      "learning_rate": 2.0038610038610037e-05,
      "loss": 0.0621,
      "step": 5590
    },
    {
      "epoch": 1.6632016632016633,
      "grad_norm": 0.8861784338951111,
      "learning_rate": 2.002079002079002e-05,
      "loss": 0.0548,
      "step": 5600
    },
    {
      "epoch": 1.6661716661716661,
      "grad_norm": 0.6367491483688354,
      "learning_rate": 2.0002970002970002e-05,
      "loss": 0.0685,
      "step": 5610
    },
    {
      "epoch": 1.6691416691416692,
      "grad_norm": 0.639358401298523,
      "learning_rate": 1.9985149985149984e-05,
      "loss": 0.0572,
      "step": 5620
    },
    {
      "epoch": 1.6721116721116722,
      "grad_norm": 0.8413973450660706,
      "learning_rate": 1.996732996732997e-05,
      "loss": 0.063,
      "step": 5630
    },
    {
      "epoch": 1.675081675081675,
      "grad_norm": 1.1640899181365967,
      "learning_rate": 1.994950994950995e-05,
      "loss": 0.0683,
      "step": 5640
    },
    {
      "epoch": 1.678051678051678,
      "grad_norm": 1.299309253692627,
      "learning_rate": 1.9931689931689934e-05,
      "loss": 0.0676,
      "step": 5650
    },
    {
      "epoch": 1.6810216810216811,
      "grad_norm": 1.032827377319336,
      "learning_rate": 1.9913869913869913e-05,
      "loss": 0.0658,
      "step": 5660
    },
    {
      "epoch": 1.683991683991684,
      "grad_norm": 0.36365097761154175,
      "learning_rate": 1.9896049896049895e-05,
      "loss": 0.0605,
      "step": 5670
    },
    {
      "epoch": 1.6869616869616868,
      "grad_norm": 0.8830529451370239,
      "learning_rate": 1.9878229878229877e-05,
      "loss": 0.0676,
      "step": 5680
    },
    {
      "epoch": 1.68993168993169,
      "grad_norm": 0.5479172468185425,
      "learning_rate": 1.986040986040986e-05,
      "loss": 0.0798,
      "step": 5690
    },
    {
      "epoch": 1.692901692901693,
      "grad_norm": 1.1123652458190918,
      "learning_rate": 1.9842589842589845e-05,
      "loss": 0.0528,
      "step": 5700
    },
    {
      "epoch": 1.6958716958716957,
      "grad_norm": 0.9681738018989563,
      "learning_rate": 1.9824769824769827e-05,
      "loss": 0.0641,
      "step": 5710
    },
    {
      "epoch": 1.698841698841699,
      "grad_norm": 0.7309338450431824,
      "learning_rate": 1.980694980694981e-05,
      "loss": 0.0764,
      "step": 5720
    },
    {
      "epoch": 1.7018117018117018,
      "grad_norm": 0.8220120668411255,
      "learning_rate": 1.978912978912979e-05,
      "loss": 0.0679,
      "step": 5730
    },
    {
      "epoch": 1.7047817047817047,
      "grad_norm": 0.9567219018936157,
      "learning_rate": 1.977130977130977e-05,
      "loss": 0.0637,
      "step": 5740
    },
    {
      "epoch": 1.7077517077517077,
      "grad_norm": 0.643904447555542,
      "learning_rate": 1.9753489753489752e-05,
      "loss": 0.0672,
      "step": 5750
    },
    {
      "epoch": 1.7107217107217108,
      "grad_norm": 1.4598885774612427,
      "learning_rate": 1.9735669735669734e-05,
      "loss": 0.067,
      "step": 5760
    },
    {
      "epoch": 1.7136917136917136,
      "grad_norm": 0.8001982569694519,
      "learning_rate": 1.971784971784972e-05,
      "loss": 0.0579,
      "step": 5770
    },
    {
      "epoch": 1.7166617166617166,
      "grad_norm": 0.858055591583252,
      "learning_rate": 1.9700029700029702e-05,
      "loss": 0.0548,
      "step": 5780
    },
    {
      "epoch": 1.7196317196317197,
      "grad_norm": 1.0781798362731934,
      "learning_rate": 1.9682209682209684e-05,
      "loss": 0.067,
      "step": 5790
    },
    {
      "epoch": 1.7226017226017225,
      "grad_norm": 0.9981959462165833,
      "learning_rate": 1.9664389664389666e-05,
      "loss": 0.0689,
      "step": 5800
    },
    {
      "epoch": 1.7255717255717256,
      "grad_norm": 0.9247766137123108,
      "learning_rate": 1.9646569646569645e-05,
      "loss": 0.0595,
      "step": 5810
    },
    {
      "epoch": 1.7285417285417286,
      "grad_norm": 0.8209492564201355,
      "learning_rate": 1.9628749628749627e-05,
      "loss": 0.0627,
      "step": 5820
    },
    {
      "epoch": 1.7315117315117314,
      "grad_norm": 1.1569684743881226,
      "learning_rate": 1.961092961092961e-05,
      "loss": 0.0727,
      "step": 5830
    },
    {
      "epoch": 1.7344817344817345,
      "grad_norm": 0.6315358281135559,
      "learning_rate": 1.9593109593109595e-05,
      "loss": 0.0673,
      "step": 5840
    },
    {
      "epoch": 1.7374517374517375,
      "grad_norm": 0.7351091504096985,
      "learning_rate": 1.9575289575289577e-05,
      "loss": 0.0673,
      "step": 5850
    },
    {
      "epoch": 1.7404217404217404,
      "grad_norm": 0.8699774742126465,
      "learning_rate": 1.955746955746956e-05,
      "loss": 0.0732,
      "step": 5860
    },
    {
      "epoch": 1.7433917433917434,
      "grad_norm": 0.992885172367096,
      "learning_rate": 1.953964953964954e-05,
      "loss": 0.0674,
      "step": 5870
    },
    {
      "epoch": 1.7463617463617465,
      "grad_norm": 0.8643231391906738,
      "learning_rate": 1.952182952182952e-05,
      "loss": 0.0755,
      "step": 5880
    },
    {
      "epoch": 1.7493317493317493,
      "grad_norm": 0.8017232418060303,
      "learning_rate": 1.9504009504009503e-05,
      "loss": 0.0671,
      "step": 5890
    },
    {
      "epoch": 1.7523017523017523,
      "grad_norm": 0.5877105593681335,
      "learning_rate": 1.9486189486189485e-05,
      "loss": 0.0658,
      "step": 5900
    },
    {
      "epoch": 1.7552717552717554,
      "grad_norm": 1.359218716621399,
      "learning_rate": 1.946836946836947e-05,
      "loss": 0.0626,
      "step": 5910
    },
    {
      "epoch": 1.7582417582417582,
      "grad_norm": 0.6869223117828369,
      "learning_rate": 1.9450549450549452e-05,
      "loss": 0.0775,
      "step": 5920
    },
    {
      "epoch": 1.7612117612117613,
      "grad_norm": 0.7375755906105042,
      "learning_rate": 1.9432729432729435e-05,
      "loss": 0.0599,
      "step": 5930
    },
    {
      "epoch": 1.7641817641817643,
      "grad_norm": 0.9355472326278687,
      "learning_rate": 1.9414909414909417e-05,
      "loss": 0.0632,
      "step": 5940
    },
    {
      "epoch": 1.7671517671517671,
      "grad_norm": 1.0422033071517944,
      "learning_rate": 1.9397089397089396e-05,
      "loss": 0.0644,
      "step": 5950
    },
    {
      "epoch": 1.77012177012177,
      "grad_norm": 0.7586382627487183,
      "learning_rate": 1.9379269379269378e-05,
      "loss": 0.0615,
      "step": 5960
    },
    {
      "epoch": 1.7730917730917732,
      "grad_norm": 1.0248056650161743,
      "learning_rate": 1.936144936144936e-05,
      "loss": 0.0708,
      "step": 5970
    },
    {
      "epoch": 1.776061776061776,
      "grad_norm": 0.9860632419586182,
      "learning_rate": 1.9343629343629345e-05,
      "loss": 0.0735,
      "step": 5980
    },
    {
      "epoch": 1.779031779031779,
      "grad_norm": 0.8044900894165039,
      "learning_rate": 1.9325809325809328e-05,
      "loss": 0.051,
      "step": 5990
    },
    {
      "epoch": 1.7820017820017822,
      "grad_norm": 0.7025960087776184,
      "learning_rate": 1.930798930798931e-05,
      "loss": 0.0535,
      "step": 6000
    },
    {
      "epoch": 1.784971784971785,
      "grad_norm": 0.9908897876739502,
      "learning_rate": 1.9290169290169292e-05,
      "loss": 0.0578,
      "step": 6010
    },
    {
      "epoch": 1.7879417879417878,
      "grad_norm": 1.2506797313690186,
      "learning_rate": 1.927234927234927e-05,
      "loss": 0.065,
      "step": 6020
    },
    {
      "epoch": 1.7909117909117909,
      "grad_norm": 1.023766040802002,
      "learning_rate": 1.9254529254529253e-05,
      "loss": 0.0631,
      "step": 6030
    },
    {
      "epoch": 1.793881793881794,
      "grad_norm": 0.8149070739746094,
      "learning_rate": 1.9236709236709235e-05,
      "loss": 0.0621,
      "step": 6040
    },
    {
      "epoch": 1.7968517968517967,
      "grad_norm": 0.6476550698280334,
      "learning_rate": 1.921888921888922e-05,
      "loss": 0.066,
      "step": 6050
    },
    {
      "epoch": 1.7998217998217998,
      "grad_norm": 0.65140700340271,
      "learning_rate": 1.9201069201069203e-05,
      "loss": 0.0611,
      "step": 6060
    },
    {
      "epoch": 1.8027918027918028,
      "grad_norm": 0.9502988457679749,
      "learning_rate": 1.9183249183249185e-05,
      "loss": 0.0615,
      "step": 6070
    },
    {
      "epoch": 1.8057618057618057,
      "grad_norm": 0.916491687297821,
      "learning_rate": 1.9165429165429167e-05,
      "loss": 0.0643,
      "step": 6080
    },
    {
      "epoch": 1.8087318087318087,
      "grad_norm": 1.169666051864624,
      "learning_rate": 1.9147609147609146e-05,
      "loss": 0.0715,
      "step": 6090
    },
    {
      "epoch": 1.8117018117018118,
      "grad_norm": 1.0344257354736328,
      "learning_rate": 1.9129789129789128e-05,
      "loss": 0.0605,
      "step": 6100
    },
    {
      "epoch": 1.8146718146718146,
      "grad_norm": 0.49185994267463684,
      "learning_rate": 1.911196911196911e-05,
      "loss": 0.0621,
      "step": 6110
    },
    {
      "epoch": 1.8176418176418176,
      "grad_norm": 0.9639203548431396,
      "learning_rate": 1.9094149094149096e-05,
      "loss": 0.0655,
      "step": 6120
    },
    {
      "epoch": 1.8206118206118207,
      "grad_norm": 0.6532425284385681,
      "learning_rate": 1.9076329076329078e-05,
      "loss": 0.0542,
      "step": 6130
    },
    {
      "epoch": 1.8235818235818235,
      "grad_norm": 0.455677330493927,
      "learning_rate": 1.905850905850906e-05,
      "loss": 0.0791,
      "step": 6140
    },
    {
      "epoch": 1.8265518265518266,
      "grad_norm": 0.5976959466934204,
      "learning_rate": 1.9040689040689042e-05,
      "loss": 0.0586,
      "step": 6150
    },
    {
      "epoch": 1.8295218295218296,
      "grad_norm": 0.6233227849006653,
      "learning_rate": 1.902286902286902e-05,
      "loss": 0.0765,
      "step": 6160
    },
    {
      "epoch": 1.8324918324918325,
      "grad_norm": 1.4189893007278442,
      "learning_rate": 1.9005049005049003e-05,
      "loss": 0.0683,
      "step": 6170
    },
    {
      "epoch": 1.8354618354618355,
      "grad_norm": 0.5655115246772766,
      "learning_rate": 1.8987228987228986e-05,
      "loss": 0.067,
      "step": 6180
    },
    {
      "epoch": 1.8384318384318385,
      "grad_norm": 0.7144678235054016,
      "learning_rate": 1.896940896940897e-05,
      "loss": 0.0537,
      "step": 6190
    },
    {
      "epoch": 1.8414018414018414,
      "grad_norm": 1.3380944728851318,
      "learning_rate": 1.8951588951588953e-05,
      "loss": 0.0697,
      "step": 6200
    },
    {
      "epoch": 1.8443718443718444,
      "grad_norm": 0.6233878135681152,
      "learning_rate": 1.8933768933768935e-05,
      "loss": 0.0702,
      "step": 6210
    },
    {
      "epoch": 1.8473418473418475,
      "grad_norm": 0.7405056357383728,
      "learning_rate": 1.8915948915948918e-05,
      "loss": 0.0654,
      "step": 6220
    },
    {
      "epoch": 1.8503118503118503,
      "grad_norm": 1.0644950866699219,
      "learning_rate": 1.8898128898128896e-05,
      "loss": 0.0654,
      "step": 6230
    },
    {
      "epoch": 1.8532818532818531,
      "grad_norm": 1.2101881504058838,
      "learning_rate": 1.888030888030888e-05,
      "loss": 0.0612,
      "step": 6240
    },
    {
      "epoch": 1.8562518562518564,
      "grad_norm": 0.7648733258247375,
      "learning_rate": 1.886248886248886e-05,
      "loss": 0.0492,
      "step": 6250
    },
    {
      "epoch": 1.8592218592218592,
      "grad_norm": 0.764968991279602,
      "learning_rate": 1.8844668844668846e-05,
      "loss": 0.0582,
      "step": 6260
    },
    {
      "epoch": 1.862191862191862,
      "grad_norm": 0.4919446110725403,
      "learning_rate": 1.882684882684883e-05,
      "loss": 0.0593,
      "step": 6270
    },
    {
      "epoch": 1.865161865161865,
      "grad_norm": 0.784898579120636,
      "learning_rate": 1.880902880902881e-05,
      "loss": 0.0651,
      "step": 6280
    },
    {
      "epoch": 1.8681318681318682,
      "grad_norm": 0.4835127890110016,
      "learning_rate": 1.8791208791208793e-05,
      "loss": 0.0545,
      "step": 6290
    },
    {
      "epoch": 1.871101871101871,
      "grad_norm": 0.7325404286384583,
      "learning_rate": 1.8773388773388775e-05,
      "loss": 0.0622,
      "step": 6300
    },
    {
      "epoch": 1.874071874071874,
      "grad_norm": 0.9660061001777649,
      "learning_rate": 1.8755568755568754e-05,
      "loss": 0.0767,
      "step": 6310
    },
    {
      "epoch": 1.877041877041877,
      "grad_norm": 1.0770442485809326,
      "learning_rate": 1.8737748737748736e-05,
      "loss": 0.0657,
      "step": 6320
    },
    {
      "epoch": 1.88001188001188,
      "grad_norm": 1.3359770774841309,
      "learning_rate": 1.871992871992872e-05,
      "loss": 0.0686,
      "step": 6330
    },
    {
      "epoch": 1.882981882981883,
      "grad_norm": 0.6175670027732849,
      "learning_rate": 1.8702108702108704e-05,
      "loss": 0.0665,
      "step": 6340
    },
    {
      "epoch": 1.885951885951886,
      "grad_norm": 0.7021560072898865,
      "learning_rate": 1.8684288684288686e-05,
      "loss": 0.0621,
      "step": 6350
    },
    {
      "epoch": 1.8889218889218888,
      "grad_norm": 0.5976589918136597,
      "learning_rate": 1.8666468666468668e-05,
      "loss": 0.0612,
      "step": 6360
    },
    {
      "epoch": 1.8918918918918919,
      "grad_norm": 0.6338089108467102,
      "learning_rate": 1.864864864864865e-05,
      "loss": 0.0636,
      "step": 6370
    },
    {
      "epoch": 1.894861894861895,
      "grad_norm": 0.9274225234985352,
      "learning_rate": 1.863082863082863e-05,
      "loss": 0.057,
      "step": 6380
    },
    {
      "epoch": 1.8978318978318978,
      "grad_norm": 0.42748093605041504,
      "learning_rate": 1.861300861300861e-05,
      "loss": 0.0583,
      "step": 6390
    },
    {
      "epoch": 1.9008019008019008,
      "grad_norm": 0.7819433808326721,
      "learning_rate": 1.8595188595188597e-05,
      "loss": 0.0581,
      "step": 6400
    },
    {
      "epoch": 1.9037719037719039,
      "grad_norm": 0.5644189119338989,
      "learning_rate": 1.857736857736858e-05,
      "loss": 0.0554,
      "step": 6410
    },
    {
      "epoch": 1.9067419067419067,
      "grad_norm": 0.6357256770133972,
      "learning_rate": 1.855954855954856e-05,
      "loss": 0.0755,
      "step": 6420
    },
    {
      "epoch": 1.9097119097119097,
      "grad_norm": 0.737325131893158,
      "learning_rate": 1.8541728541728543e-05,
      "loss": 0.0742,
      "step": 6430
    },
    {
      "epoch": 1.9126819126819128,
      "grad_norm": 0.6249455809593201,
      "learning_rate": 1.8523908523908525e-05,
      "loss": 0.0624,
      "step": 6440
    },
    {
      "epoch": 1.9156519156519156,
      "grad_norm": 0.8265432119369507,
      "learning_rate": 1.8506088506088504e-05,
      "loss": 0.0535,
      "step": 6450
    },
    {
      "epoch": 1.9186219186219187,
      "grad_norm": 0.5451765656471252,
      "learning_rate": 1.8488268488268486e-05,
      "loss": 0.0646,
      "step": 6460
    },
    {
      "epoch": 1.9215919215919217,
      "grad_norm": 0.9091644883155823,
      "learning_rate": 1.8470448470448472e-05,
      "loss": 0.0579,
      "step": 6470
    },
    {
      "epoch": 1.9245619245619245,
      "grad_norm": 0.6501501798629761,
      "learning_rate": 1.8452628452628454e-05,
      "loss": 0.0518,
      "step": 6480
    },
    {
      "epoch": 1.9275319275319274,
      "grad_norm": 0.6668860912322998,
      "learning_rate": 1.8434808434808436e-05,
      "loss": 0.0484,
      "step": 6490
    },
    {
      "epoch": 1.9305019305019306,
      "grad_norm": 0.6235078573226929,
      "learning_rate": 1.841698841698842e-05,
      "loss": 0.0571,
      "step": 6500
    },
    {
      "epoch": 1.9334719334719335,
      "grad_norm": 0.5391401648521423,
      "learning_rate": 1.83991683991684e-05,
      "loss": 0.042,
      "step": 6510
    },
    {
      "epoch": 1.9364419364419363,
      "grad_norm": 0.4596608877182007,
      "learning_rate": 1.838134838134838e-05,
      "loss": 0.0666,
      "step": 6520
    },
    {
      "epoch": 1.9394119394119396,
      "grad_norm": 0.9191550612449646,
      "learning_rate": 1.836352836352836e-05,
      "loss": 0.0731,
      "step": 6530
    },
    {
      "epoch": 1.9423819423819424,
      "grad_norm": 0.8240092992782593,
      "learning_rate": 1.8345708345708347e-05,
      "loss": 0.0635,
      "step": 6540
    },
    {
      "epoch": 1.9453519453519452,
      "grad_norm": 0.8021003007888794,
      "learning_rate": 1.832788832788833e-05,
      "loss": 0.0675,
      "step": 6550
    },
    {
      "epoch": 1.9483219483219483,
      "grad_norm": 1.1347585916519165,
      "learning_rate": 1.831006831006831e-05,
      "loss": 0.0621,
      "step": 6560
    },
    {
      "epoch": 1.9512919512919513,
      "grad_norm": 0.6139947175979614,
      "learning_rate": 1.8292248292248294e-05,
      "loss": 0.0543,
      "step": 6570
    },
    {
      "epoch": 1.9542619542619541,
      "grad_norm": 0.7635927796363831,
      "learning_rate": 1.8274428274428276e-05,
      "loss": 0.0592,
      "step": 6580
    },
    {
      "epoch": 1.9572319572319572,
      "grad_norm": 0.43468376994132996,
      "learning_rate": 1.8256608256608254e-05,
      "loss": 0.0612,
      "step": 6590
    },
    {
      "epoch": 1.9602019602019602,
      "grad_norm": 0.7748926877975464,
      "learning_rate": 1.8238788238788237e-05,
      "loss": 0.0604,
      "step": 6600
    },
    {
      "epoch": 1.963171963171963,
      "grad_norm": 0.8012065887451172,
      "learning_rate": 1.8220968220968222e-05,
      "loss": 0.0611,
      "step": 6610
    },
    {
      "epoch": 1.9661419661419661,
      "grad_norm": 0.9480249285697937,
      "learning_rate": 1.8203148203148204e-05,
      "loss": 0.0557,
      "step": 6620
    },
    {
      "epoch": 1.9691119691119692,
      "grad_norm": 0.8358305096626282,
      "learning_rate": 1.8185328185328187e-05,
      "loss": 0.0598,
      "step": 6630
    },
    {
      "epoch": 1.972081972081972,
      "grad_norm": 0.5806179642677307,
      "learning_rate": 1.816750816750817e-05,
      "loss": 0.0646,
      "step": 6640
    },
    {
      "epoch": 1.975051975051975,
      "grad_norm": 1.0674052238464355,
      "learning_rate": 1.814968814968815e-05,
      "loss": 0.075,
      "step": 6650
    },
    {
      "epoch": 1.978021978021978,
      "grad_norm": 0.7299931645393372,
      "learning_rate": 1.813186813186813e-05,
      "loss": 0.0526,
      "step": 6660
    },
    {
      "epoch": 1.980991980991981,
      "grad_norm": 0.7215370535850525,
      "learning_rate": 1.8114048114048112e-05,
      "loss": 0.063,
      "step": 6670
    },
    {
      "epoch": 1.983961983961984,
      "grad_norm": 0.6642675995826721,
      "learning_rate": 1.8096228096228097e-05,
      "loss": 0.0534,
      "step": 6680
    },
    {
      "epoch": 1.986931986931987,
      "grad_norm": 1.6818110942840576,
      "learning_rate": 1.807840807840808e-05,
      "loss": 0.0663,
      "step": 6690
    },
    {
      "epoch": 1.9899019899019899,
      "grad_norm": 0.8374834060668945,
      "learning_rate": 1.8060588060588062e-05,
      "loss": 0.057,
      "step": 6700
    },
    {
      "epoch": 1.992871992871993,
      "grad_norm": 0.5147525668144226,
      "learning_rate": 1.8042768042768044e-05,
      "loss": 0.0629,
      "step": 6710
    },
    {
      "epoch": 1.995841995841996,
      "grad_norm": 0.8637810945510864,
      "learning_rate": 1.8024948024948026e-05,
      "loss": 0.068,
      "step": 6720
    },
    {
      "epoch": 1.9988119988119988,
      "grad_norm": 1.2453491687774658,
      "learning_rate": 1.8007128007128005e-05,
      "loss": 0.0679,
      "step": 6730
    },
    {
      "epoch": 2.0,
      "eval_f1": 0.33031292965957215,
      "eval_loss": 0.04602367803454399,
      "eval_runtime": 175.7476,
      "eval_samples_per_second": 216.327,
      "eval_steps_per_second": 3.386,
      "step": 6734
    },
    {
      "epoch": 2.0017820017820016,
      "grad_norm": 1.147544264793396,
      "learning_rate": 1.7989307989307987e-05,
      "loss": 0.0478,
      "step": 6740
    },
    {
      "epoch": 2.004752004752005,
      "grad_norm": 0.7158094644546509,
      "learning_rate": 1.7971487971487973e-05,
      "loss": 0.0597,
      "step": 6750
    },
    {
      "epoch": 2.0077220077220077,
      "grad_norm": 0.5224217176437378,
      "learning_rate": 1.7953667953667955e-05,
      "loss": 0.0655,
      "step": 6760
    },
    {
      "epoch": 2.0106920106920105,
      "grad_norm": 1.072407603263855,
      "learning_rate": 1.7935847935847937e-05,
      "loss": 0.058,
      "step": 6770
    },
    {
      "epoch": 2.013662013662014,
      "grad_norm": 1.3633906841278076,
      "learning_rate": 1.791802791802792e-05,
      "loss": 0.0601,
      "step": 6780
    },
    {
      "epoch": 2.0166320166320166,
      "grad_norm": 1.6582151651382446,
      "learning_rate": 1.79002079002079e-05,
      "loss": 0.0629,
      "step": 6790
    },
    {
      "epoch": 2.0196020196020195,
      "grad_norm": 0.9400825500488281,
      "learning_rate": 1.788238788238788e-05,
      "loss": 0.06,
      "step": 6800
    },
    {
      "epoch": 2.0225720225720227,
      "grad_norm": 0.9094600677490234,
      "learning_rate": 1.7864567864567862e-05,
      "loss": 0.0583,
      "step": 6810
    },
    {
      "epoch": 2.0255420255420256,
      "grad_norm": 0.7568134069442749,
      "learning_rate": 1.7846747846747848e-05,
      "loss": 0.0584,
      "step": 6820
    },
    {
      "epoch": 2.0285120285120284,
      "grad_norm": 0.4175792634487152,
      "learning_rate": 1.782892782892783e-05,
      "loss": 0.056,
      "step": 6830
    },
    {
      "epoch": 2.0314820314820317,
      "grad_norm": 0.7946698665618896,
      "learning_rate": 1.7811107811107812e-05,
      "loss": 0.0611,
      "step": 6840
    },
    {
      "epoch": 2.0344520344520345,
      "grad_norm": 0.5856500267982483,
      "learning_rate": 1.7793287793287794e-05,
      "loss": 0.056,
      "step": 6850
    },
    {
      "epoch": 2.0374220374220373,
      "grad_norm": 0.7179090976715088,
      "learning_rate": 1.7775467775467776e-05,
      "loss": 0.0602,
      "step": 6860
    },
    {
      "epoch": 2.0403920403920406,
      "grad_norm": 1.6203516721725464,
      "learning_rate": 1.7757647757647755e-05,
      "loss": 0.059,
      "step": 6870
    },
    {
      "epoch": 2.0433620433620434,
      "grad_norm": 0.522436797618866,
      "learning_rate": 1.7739827739827737e-05,
      "loss": 0.0616,
      "step": 6880
    },
    {
      "epoch": 2.0463320463320462,
      "grad_norm": 1.3147087097167969,
      "learning_rate": 1.7722007722007723e-05,
      "loss": 0.0658,
      "step": 6890
    },
    {
      "epoch": 2.0493020493020495,
      "grad_norm": 0.7292523980140686,
      "learning_rate": 1.7704187704187705e-05,
      "loss": 0.0559,
      "step": 6900
    },
    {
      "epoch": 2.0522720522720523,
      "grad_norm": 0.7819983959197998,
      "learning_rate": 1.7686367686367687e-05,
      "loss": 0.0581,
      "step": 6910
    },
    {
      "epoch": 2.055242055242055,
      "grad_norm": 0.7037408947944641,
      "learning_rate": 1.766854766854767e-05,
      "loss": 0.0653,
      "step": 6920
    },
    {
      "epoch": 2.0582120582120584,
      "grad_norm": 0.940587043762207,
      "learning_rate": 1.765072765072765e-05,
      "loss": 0.063,
      "step": 6930
    },
    {
      "epoch": 2.0611820611820613,
      "grad_norm": 0.7256175875663757,
      "learning_rate": 1.7632907632907634e-05,
      "loss": 0.0611,
      "step": 6940
    },
    {
      "epoch": 2.064152064152064,
      "grad_norm": 0.7367241978645325,
      "learning_rate": 1.7615087615087613e-05,
      "loss": 0.0548,
      "step": 6950
    },
    {
      "epoch": 2.067122067122067,
      "grad_norm": 0.9881260991096497,
      "learning_rate": 1.7597267597267598e-05,
      "loss": 0.054,
      "step": 6960
    },
    {
      "epoch": 2.07009207009207,
      "grad_norm": 0.8852285742759705,
      "learning_rate": 1.757944757944758e-05,
      "loss": 0.065,
      "step": 6970
    },
    {
      "epoch": 2.073062073062073,
      "grad_norm": 0.8763765692710876,
      "learning_rate": 1.7561627561627563e-05,
      "loss": 0.0714,
      "step": 6980
    },
    {
      "epoch": 2.076032076032076,
      "grad_norm": 0.9801186919212341,
      "learning_rate": 1.7543807543807545e-05,
      "loss": 0.0644,
      "step": 6990
    },
    {
      "epoch": 2.079002079002079,
      "grad_norm": 0.8674045205116272,
      "learning_rate": 1.7525987525987527e-05,
      "loss": 0.0597,
      "step": 7000
    },
    {
      "epoch": 2.081972081972082,
      "grad_norm": 0.8817034959793091,
      "learning_rate": 1.750816750816751e-05,
      "loss": 0.0558,
      "step": 7010
    },
    {
      "epoch": 2.0849420849420848,
      "grad_norm": 0.8658657073974609,
      "learning_rate": 1.7490347490347488e-05,
      "loss": 0.0673,
      "step": 7020
    },
    {
      "epoch": 2.087912087912088,
      "grad_norm": 0.5216271877288818,
      "learning_rate": 1.7472527472527473e-05,
      "loss": 0.057,
      "step": 7030
    },
    {
      "epoch": 2.090882090882091,
      "grad_norm": 1.3478199243545532,
      "learning_rate": 1.7454707454707456e-05,
      "loss": 0.0623,
      "step": 7040
    },
    {
      "epoch": 2.0938520938520937,
      "grad_norm": 1.1730366945266724,
      "learning_rate": 1.7436887436887438e-05,
      "loss": 0.0663,
      "step": 7050
    },
    {
      "epoch": 2.096822096822097,
      "grad_norm": 0.7896762490272522,
      "learning_rate": 1.741906741906742e-05,
      "loss": 0.0677,
      "step": 7060
    },
    {
      "epoch": 2.0997920997921,
      "grad_norm": 0.6003497838973999,
      "learning_rate": 1.7401247401247402e-05,
      "loss": 0.062,
      "step": 7070
    },
    {
      "epoch": 2.1027621027621026,
      "grad_norm": 1.0001591444015503,
      "learning_rate": 1.7383427383427384e-05,
      "loss": 0.0653,
      "step": 7080
    },
    {
      "epoch": 2.105732105732106,
      "grad_norm": 1.3104040622711182,
      "learning_rate": 1.7365607365607363e-05,
      "loss": 0.0653,
      "step": 7090
    },
    {
      "epoch": 2.1087021087021087,
      "grad_norm": 0.6923945546150208,
      "learning_rate": 1.734778734778735e-05,
      "loss": 0.0474,
      "step": 7100
    },
    {
      "epoch": 2.1116721116721116,
      "grad_norm": 0.6234051585197449,
      "learning_rate": 1.732996732996733e-05,
      "loss": 0.0617,
      "step": 7110
    },
    {
      "epoch": 2.114642114642115,
      "grad_norm": 0.9373548626899719,
      "learning_rate": 1.7312147312147313e-05,
      "loss": 0.055,
      "step": 7120
    },
    {
      "epoch": 2.1176121176121177,
      "grad_norm": 0.9294105768203735,
      "learning_rate": 1.7294327294327295e-05,
      "loss": 0.0573,
      "step": 7130
    },
    {
      "epoch": 2.1205821205821205,
      "grad_norm": 0.7299237847328186,
      "learning_rate": 1.7276507276507277e-05,
      "loss": 0.0528,
      "step": 7140
    },
    {
      "epoch": 2.1235521235521237,
      "grad_norm": 0.5744438767433167,
      "learning_rate": 1.725868725868726e-05,
      "loss": 0.0606,
      "step": 7150
    },
    {
      "epoch": 2.1265221265221266,
      "grad_norm": 0.7229251861572266,
      "learning_rate": 1.7240867240867238e-05,
      "loss": 0.0543,
      "step": 7160
    },
    {
      "epoch": 2.1294921294921294,
      "grad_norm": 1.3666146993637085,
      "learning_rate": 1.7223047223047224e-05,
      "loss": 0.0617,
      "step": 7170
    },
    {
      "epoch": 2.1324621324621322,
      "grad_norm": 0.6104298830032349,
      "learning_rate": 1.7205227205227206e-05,
      "loss": 0.0507,
      "step": 7180
    },
    {
      "epoch": 2.1354321354321355,
      "grad_norm": 0.5680792331695557,
      "learning_rate": 1.7187407187407188e-05,
      "loss": 0.052,
      "step": 7190
    },
    {
      "epoch": 2.1384021384021383,
      "grad_norm": 0.5435447692871094,
      "learning_rate": 1.716958716958717e-05,
      "loss": 0.0578,
      "step": 7200
    },
    {
      "epoch": 2.141372141372141,
      "grad_norm": 0.6102536916732788,
      "learning_rate": 1.7151767151767152e-05,
      "loss": 0.0717,
      "step": 7210
    },
    {
      "epoch": 2.1443421443421444,
      "grad_norm": 0.9239901304244995,
      "learning_rate": 1.7133947133947135e-05,
      "loss": 0.0595,
      "step": 7220
    },
    {
      "epoch": 2.1473121473121473,
      "grad_norm": 0.7742301225662231,
      "learning_rate": 1.7116127116127117e-05,
      "loss": 0.0613,
      "step": 7230
    },
    {
      "epoch": 2.15028215028215,
      "grad_norm": 0.7156841158866882,
      "learning_rate": 1.70983070983071e-05,
      "loss": 0.0673,
      "step": 7240
    },
    {
      "epoch": 2.1532521532521534,
      "grad_norm": 0.8358856439590454,
      "learning_rate": 1.708048708048708e-05,
      "loss": 0.0662,
      "step": 7250
    },
    {
      "epoch": 2.156222156222156,
      "grad_norm": 1.2078478336334229,
      "learning_rate": 1.7062667062667063e-05,
      "loss": 0.0612,
      "step": 7260
    },
    {
      "epoch": 2.159192159192159,
      "grad_norm": 0.7166781425476074,
      "learning_rate": 1.7044847044847045e-05,
      "loss": 0.0758,
      "step": 7270
    },
    {
      "epoch": 2.1621621621621623,
      "grad_norm": 0.5501638650894165,
      "learning_rate": 1.7027027027027028e-05,
      "loss": 0.0544,
      "step": 7280
    },
    {
      "epoch": 2.165132165132165,
      "grad_norm": 0.6720577478408813,
      "learning_rate": 1.700920700920701e-05,
      "loss": 0.0682,
      "step": 7290
    },
    {
      "epoch": 2.168102168102168,
      "grad_norm": 1.2910141944885254,
      "learning_rate": 1.6991386991386992e-05,
      "loss": 0.0563,
      "step": 7300
    },
    {
      "epoch": 2.171072171072171,
      "grad_norm": 0.834705114364624,
      "learning_rate": 1.6973566973566974e-05,
      "loss": 0.0562,
      "step": 7310
    },
    {
      "epoch": 2.174042174042174,
      "grad_norm": 0.9644896984100342,
      "learning_rate": 1.6955746955746956e-05,
      "loss": 0.0582,
      "step": 7320
    },
    {
      "epoch": 2.177012177012177,
      "grad_norm": 0.7104148864746094,
      "learning_rate": 1.693792693792694e-05,
      "loss": 0.0624,
      "step": 7330
    },
    {
      "epoch": 2.17998217998218,
      "grad_norm": 1.1333881616592407,
      "learning_rate": 1.692010692010692e-05,
      "loss": 0.0681,
      "step": 7340
    },
    {
      "epoch": 2.182952182952183,
      "grad_norm": 0.5454867482185364,
      "learning_rate": 1.6902286902286903e-05,
      "loss": 0.0504,
      "step": 7350
    },
    {
      "epoch": 2.185922185922186,
      "grad_norm": 0.7545517086982727,
      "learning_rate": 1.6884466884466885e-05,
      "loss": 0.0546,
      "step": 7360
    },
    {
      "epoch": 2.188892188892189,
      "grad_norm": 0.8815241456031799,
      "learning_rate": 1.6866646866646867e-05,
      "loss": 0.0645,
      "step": 7370
    },
    {
      "epoch": 2.191862191862192,
      "grad_norm": 0.8765430450439453,
      "learning_rate": 1.684882684882685e-05,
      "loss": 0.0658,
      "step": 7380
    },
    {
      "epoch": 2.1948321948321947,
      "grad_norm": 0.8555752038955688,
      "learning_rate": 1.683100683100683e-05,
      "loss": 0.06,
      "step": 7390
    },
    {
      "epoch": 2.197802197802198,
      "grad_norm": 0.7630652785301208,
      "learning_rate": 1.6813186813186814e-05,
      "loss": 0.0701,
      "step": 7400
    },
    {
      "epoch": 2.200772200772201,
      "grad_norm": 0.6194639205932617,
      "learning_rate": 1.6795366795366796e-05,
      "loss": 0.057,
      "step": 7410
    },
    {
      "epoch": 2.2037422037422036,
      "grad_norm": 0.4202905297279358,
      "learning_rate": 1.6777546777546778e-05,
      "loss": 0.0486,
      "step": 7420
    },
    {
      "epoch": 2.206712206712207,
      "grad_norm": 0.6983931660652161,
      "learning_rate": 1.675972675972676e-05,
      "loss": 0.0638,
      "step": 7430
    },
    {
      "epoch": 2.2096822096822097,
      "grad_norm": 0.521522045135498,
      "learning_rate": 1.6741906741906742e-05,
      "loss": 0.0659,
      "step": 7440
    },
    {
      "epoch": 2.2126522126522126,
      "grad_norm": 0.4995703101158142,
      "learning_rate": 1.6724086724086725e-05,
      "loss": 0.0631,
      "step": 7450
    },
    {
      "epoch": 2.215622215622216,
      "grad_norm": 1.18669593334198,
      "learning_rate": 1.6706266706266707e-05,
      "loss": 0.0589,
      "step": 7460
    },
    {
      "epoch": 2.2185922185922187,
      "grad_norm": 1.027181625366211,
      "learning_rate": 1.668844668844669e-05,
      "loss": 0.0664,
      "step": 7470
    },
    {
      "epoch": 2.2215622215622215,
      "grad_norm": 0.9086653590202332,
      "learning_rate": 1.667062667062667e-05,
      "loss": 0.0589,
      "step": 7480
    },
    {
      "epoch": 2.2245322245322248,
      "grad_norm": 0.6986064314842224,
      "learning_rate": 1.6652806652806653e-05,
      "loss": 0.061,
      "step": 7490
    },
    {
      "epoch": 2.2275022275022276,
      "grad_norm": 1.728353500366211,
      "learning_rate": 1.6634986634986635e-05,
      "loss": 0.0496,
      "step": 7500
    },
    {
      "epoch": 2.2304722304722304,
      "grad_norm": 0.635563313961029,
      "learning_rate": 1.6617166617166618e-05,
      "loss": 0.0711,
      "step": 7510
    },
    {
      "epoch": 2.2334422334422332,
      "grad_norm": 0.9441576600074768,
      "learning_rate": 1.65993465993466e-05,
      "loss": 0.0554,
      "step": 7520
    },
    {
      "epoch": 2.2364122364122365,
      "grad_norm": 0.984162449836731,
      "learning_rate": 1.6581526581526582e-05,
      "loss": 0.059,
      "step": 7530
    },
    {
      "epoch": 2.2393822393822393,
      "grad_norm": 0.7663610577583313,
      "learning_rate": 1.6563706563706564e-05,
      "loss": 0.0654,
      "step": 7540
    },
    {
      "epoch": 2.242352242352242,
      "grad_norm": 0.8404085636138916,
      "learning_rate": 1.6545886545886546e-05,
      "loss": 0.0723,
      "step": 7550
    },
    {
      "epoch": 2.2453222453222454,
      "grad_norm": 0.9056734442710876,
      "learning_rate": 1.652806652806653e-05,
      "loss": 0.0632,
      "step": 7560
    },
    {
      "epoch": 2.2482922482922483,
      "grad_norm": 1.249064326286316,
      "learning_rate": 1.651024651024651e-05,
      "loss": 0.0762,
      "step": 7570
    },
    {
      "epoch": 2.251262251262251,
      "grad_norm": 0.7174326181411743,
      "learning_rate": 1.6492426492426496e-05,
      "loss": 0.054,
      "step": 7580
    },
    {
      "epoch": 2.2542322542322544,
      "grad_norm": 0.9042089581489563,
      "learning_rate": 1.6474606474606475e-05,
      "loss": 0.0668,
      "step": 7590
    },
    {
      "epoch": 2.257202257202257,
      "grad_norm": 0.6631519794464111,
      "learning_rate": 1.6456786456786457e-05,
      "loss": 0.0599,
      "step": 7600
    },
    {
      "epoch": 2.26017226017226,
      "grad_norm": 0.6724472045898438,
      "learning_rate": 1.643896643896644e-05,
      "loss": 0.0675,
      "step": 7610
    },
    {
      "epoch": 2.2631422631422633,
      "grad_norm": 0.48432090878486633,
      "learning_rate": 1.642114642114642e-05,
      "loss": 0.0565,
      "step": 7620
    },
    {
      "epoch": 2.266112266112266,
      "grad_norm": 0.5798602104187012,
      "learning_rate": 1.6403326403326404e-05,
      "loss": 0.0584,
      "step": 7630
    },
    {
      "epoch": 2.269082269082269,
      "grad_norm": 0.676134467124939,
      "learning_rate": 1.6385506385506386e-05,
      "loss": 0.0651,
      "step": 7640
    },
    {
      "epoch": 2.2720522720522722,
      "grad_norm": 0.6854906678199768,
      "learning_rate": 1.636768636768637e-05,
      "loss": 0.0711,
      "step": 7650
    },
    {
      "epoch": 2.275022275022275,
      "grad_norm": 1.0705246925354004,
      "learning_rate": 1.634986634986635e-05,
      "loss": 0.0635,
      "step": 7660
    },
    {
      "epoch": 2.277992277992278,
      "grad_norm": 0.6841071248054504,
      "learning_rate": 1.6332046332046332e-05,
      "loss": 0.0534,
      "step": 7670
    },
    {
      "epoch": 2.280962280962281,
      "grad_norm": 1.0041640996932983,
      "learning_rate": 1.6314226314226314e-05,
      "loss": 0.053,
      "step": 7680
    },
    {
      "epoch": 2.283932283932284,
      "grad_norm": 0.7719654440879822,
      "learning_rate": 1.6296406296406297e-05,
      "loss": 0.0715,
      "step": 7690
    },
    {
      "epoch": 2.286902286902287,
      "grad_norm": 0.5428666472434998,
      "learning_rate": 1.627858627858628e-05,
      "loss": 0.0483,
      "step": 7700
    },
    {
      "epoch": 2.2898722898722896,
      "grad_norm": 0.9091588854789734,
      "learning_rate": 1.626076626076626e-05,
      "loss": 0.0525,
      "step": 7710
    },
    {
      "epoch": 2.292842292842293,
      "grad_norm": 0.8281648755073547,
      "learning_rate": 1.6242946242946247e-05,
      "loss": 0.0662,
      "step": 7720
    },
    {
      "epoch": 2.2958122958122957,
      "grad_norm": 1.2040926218032837,
      "learning_rate": 1.6225126225126225e-05,
      "loss": 0.0561,
      "step": 7730
    },
    {
      "epoch": 2.2987822987822986,
      "grad_norm": 1.1443513631820679,
      "learning_rate": 1.6207306207306207e-05,
      "loss": 0.0462,
      "step": 7740
    },
    {
      "epoch": 2.301752301752302,
      "grad_norm": 0.990245521068573,
      "learning_rate": 1.618948618948619e-05,
      "loss": 0.0647,
      "step": 7750
    },
    {
      "epoch": 2.3047223047223047,
      "grad_norm": 0.6413756608963013,
      "learning_rate": 1.6171666171666172e-05,
      "loss": 0.0713,
      "step": 7760
    },
    {
      "epoch": 2.3076923076923075,
      "grad_norm": 0.5494916439056396,
      "learning_rate": 1.6153846153846154e-05,
      "loss": 0.0594,
      "step": 7770
    },
    {
      "epoch": 2.3106623106623108,
      "grad_norm": 0.6467922329902649,
      "learning_rate": 1.6136026136026136e-05,
      "loss": 0.0586,
      "step": 7780
    },
    {
      "epoch": 2.3136323136323136,
      "grad_norm": 0.5509008169174194,
      "learning_rate": 1.6118206118206122e-05,
      "loss": 0.0574,
      "step": 7790
    },
    {
      "epoch": 2.3166023166023164,
      "grad_norm": 0.7247257232666016,
      "learning_rate": 1.61003861003861e-05,
      "loss": 0.0676,
      "step": 7800
    },
    {
      "epoch": 2.3195723195723197,
      "grad_norm": 0.7993832230567932,
      "learning_rate": 1.6082566082566083e-05,
      "loss": 0.0666,
      "step": 7810
    },
    {
      "epoch": 2.3225423225423225,
      "grad_norm": 0.5373691916465759,
      "learning_rate": 1.6064746064746065e-05,
      "loss": 0.0551,
      "step": 7820
    },
    {
      "epoch": 2.3255123255123253,
      "grad_norm": 0.8234823942184448,
      "learning_rate": 1.6046926046926047e-05,
      "loss": 0.0696,
      "step": 7830
    },
    {
      "epoch": 2.3284823284823286,
      "grad_norm": 0.6193700432777405,
      "learning_rate": 1.602910602910603e-05,
      "loss": 0.0742,
      "step": 7840
    },
    {
      "epoch": 2.3314523314523314,
      "grad_norm": 0.9918487071990967,
      "learning_rate": 1.601128601128601e-05,
      "loss": 0.0636,
      "step": 7850
    },
    {
      "epoch": 2.3344223344223343,
      "grad_norm": 0.47731947898864746,
      "learning_rate": 1.5993465993465997e-05,
      "loss": 0.0605,
      "step": 7860
    },
    {
      "epoch": 2.3373923373923375,
      "grad_norm": 1.2060436010360718,
      "learning_rate": 1.5975645975645976e-05,
      "loss": 0.0532,
      "step": 7870
    },
    {
      "epoch": 2.3403623403623404,
      "grad_norm": 0.9285519123077393,
      "learning_rate": 1.5957825957825958e-05,
      "loss": 0.0638,
      "step": 7880
    },
    {
      "epoch": 2.343332343332343,
      "grad_norm": 1.4281196594238281,
      "learning_rate": 1.594000594000594e-05,
      "loss": 0.0678,
      "step": 7890
    },
    {
      "epoch": 2.3463023463023465,
      "grad_norm": 1.6847813129425049,
      "learning_rate": 1.5922185922185922e-05,
      "loss": 0.0629,
      "step": 7900
    },
    {
      "epoch": 2.3492723492723493,
      "grad_norm": 0.5582916140556335,
      "learning_rate": 1.5904365904365904e-05,
      "loss": 0.0614,
      "step": 7910
    },
    {
      "epoch": 2.352242352242352,
      "grad_norm": 0.5601425170898438,
      "learning_rate": 1.5886545886545887e-05,
      "loss": 0.0549,
      "step": 7920
    },
    {
      "epoch": 2.3552123552123554,
      "grad_norm": 0.8730356097221375,
      "learning_rate": 1.5868725868725872e-05,
      "loss": 0.051,
      "step": 7930
    },
    {
      "epoch": 2.358182358182358,
      "grad_norm": 0.9987317323684692,
      "learning_rate": 1.585090585090585e-05,
      "loss": 0.0598,
      "step": 7940
    },
    {
      "epoch": 2.361152361152361,
      "grad_norm": 0.689275860786438,
      "learning_rate": 1.5833085833085833e-05,
      "loss": 0.0477,
      "step": 7950
    },
    {
      "epoch": 2.3641223641223643,
      "grad_norm": 0.8642667531967163,
      "learning_rate": 1.5815265815265815e-05,
      "loss": 0.0619,
      "step": 7960
    },
    {
      "epoch": 2.367092367092367,
      "grad_norm": 0.675285279750824,
      "learning_rate": 1.5797445797445797e-05,
      "loss": 0.0671,
      "step": 7970
    },
    {
      "epoch": 2.37006237006237,
      "grad_norm": 0.9872873425483704,
      "learning_rate": 1.577962577962578e-05,
      "loss": 0.0523,
      "step": 7980
    },
    {
      "epoch": 2.3730323730323732,
      "grad_norm": 0.7550463676452637,
      "learning_rate": 1.5761805761805762e-05,
      "loss": 0.0616,
      "step": 7990
    },
    {
      "epoch": 2.376002376002376,
      "grad_norm": 1.0801265239715576,
      "learning_rate": 1.5743985743985747e-05,
      "loss": 0.0565,
      "step": 8000
    },
    {
      "epoch": 2.378972378972379,
      "grad_norm": 1.093015432357788,
      "learning_rate": 1.5726165726165726e-05,
      "loss": 0.0666,
      "step": 8010
    },
    {
      "epoch": 2.381942381942382,
      "grad_norm": 0.882056713104248,
      "learning_rate": 1.5708345708345708e-05,
      "loss": 0.0609,
      "step": 8020
    },
    {
      "epoch": 2.384912384912385,
      "grad_norm": 0.7684606909751892,
      "learning_rate": 1.569052569052569e-05,
      "loss": 0.063,
      "step": 8030
    },
    {
      "epoch": 2.387882387882388,
      "grad_norm": 0.9712556600570679,
      "learning_rate": 1.5672705672705673e-05,
      "loss": 0.0598,
      "step": 8040
    },
    {
      "epoch": 2.390852390852391,
      "grad_norm": 0.7231224179267883,
      "learning_rate": 1.5654885654885655e-05,
      "loss": 0.0671,
      "step": 8050
    },
    {
      "epoch": 2.393822393822394,
      "grad_norm": 0.6880158185958862,
      "learning_rate": 1.5637065637065637e-05,
      "loss": 0.0638,
      "step": 8060
    },
    {
      "epoch": 2.3967923967923968,
      "grad_norm": 0.8751171231269836,
      "learning_rate": 1.5619245619245622e-05,
      "loss": 0.0607,
      "step": 8070
    },
    {
      "epoch": 2.3997623997623996,
      "grad_norm": 1.0133845806121826,
      "learning_rate": 1.56014256014256e-05,
      "loss": 0.0663,
      "step": 8080
    },
    {
      "epoch": 2.402732402732403,
      "grad_norm": 0.4937121272087097,
      "learning_rate": 1.5583605583605583e-05,
      "loss": 0.0586,
      "step": 8090
    },
    {
      "epoch": 2.4057024057024057,
      "grad_norm": 0.6717655062675476,
      "learning_rate": 1.5565785565785566e-05,
      "loss": 0.0709,
      "step": 8100
    },
    {
      "epoch": 2.4086724086724085,
      "grad_norm": 0.9288782477378845,
      "learning_rate": 1.5547965547965548e-05,
      "loss": 0.0728,
      "step": 8110
    },
    {
      "epoch": 2.4116424116424118,
      "grad_norm": 0.9981745481491089,
      "learning_rate": 1.553014553014553e-05,
      "loss": 0.0518,
      "step": 8120
    },
    {
      "epoch": 2.4146124146124146,
      "grad_norm": 0.6858815550804138,
      "learning_rate": 1.5512325512325512e-05,
      "loss": 0.059,
      "step": 8130
    },
    {
      "epoch": 2.4175824175824174,
      "grad_norm": 1.0234894752502441,
      "learning_rate": 1.5494505494505498e-05,
      "loss": 0.0656,
      "step": 8140
    },
    {
      "epoch": 2.4205524205524207,
      "grad_norm": 1.0168789625167847,
      "learning_rate": 1.547668547668548e-05,
      "loss": 0.0661,
      "step": 8150
    },
    {
      "epoch": 2.4235224235224235,
      "grad_norm": 0.5430660247802734,
      "learning_rate": 1.545886545886546e-05,
      "loss": 0.0498,
      "step": 8160
    },
    {
      "epoch": 2.4264924264924264,
      "grad_norm": 0.6519795656204224,
      "learning_rate": 1.544104544104544e-05,
      "loss": 0.0565,
      "step": 8170
    },
    {
      "epoch": 2.4294624294624296,
      "grad_norm": 0.4860493540763855,
      "learning_rate": 1.5423225423225423e-05,
      "loss": 0.0502,
      "step": 8180
    },
    {
      "epoch": 2.4324324324324325,
      "grad_norm": 1.287544846534729,
      "learning_rate": 1.5405405405405405e-05,
      "loss": 0.0619,
      "step": 8190
    },
    {
      "epoch": 2.4354024354024353,
      "grad_norm": 0.7595537900924683,
      "learning_rate": 1.5387585387585387e-05,
      "loss": 0.062,
      "step": 8200
    },
    {
      "epoch": 2.4383724383724386,
      "grad_norm": 1.2238869667053223,
      "learning_rate": 1.5369765369765373e-05,
      "loss": 0.0602,
      "step": 8210
    },
    {
      "epoch": 2.4413424413424414,
      "grad_norm": 0.6044248342514038,
      "learning_rate": 1.5351945351945355e-05,
      "loss": 0.064,
      "step": 8220
    },
    {
      "epoch": 2.444312444312444,
      "grad_norm": 1.0515096187591553,
      "learning_rate": 1.5334125334125334e-05,
      "loss": 0.0623,
      "step": 8230
    },
    {
      "epoch": 2.447282447282447,
      "grad_norm": 0.3388879895210266,
      "learning_rate": 1.5316305316305316e-05,
      "loss": 0.0508,
      "step": 8240
    },
    {
      "epoch": 2.4502524502524503,
      "grad_norm": 1.4969494342803955,
      "learning_rate": 1.5298485298485298e-05,
      "loss": 0.0562,
      "step": 8250
    },
    {
      "epoch": 2.453222453222453,
      "grad_norm": 1.1669758558273315,
      "learning_rate": 1.528066528066528e-05,
      "loss": 0.0694,
      "step": 8260
    },
    {
      "epoch": 2.456192456192456,
      "grad_norm": 0.8401341438293457,
      "learning_rate": 1.5262845262845263e-05,
      "loss": 0.0596,
      "step": 8270
    },
    {
      "epoch": 2.4591624591624592,
      "grad_norm": 0.8268706798553467,
      "learning_rate": 1.5245025245025246e-05,
      "loss": 0.0595,
      "step": 8280
    },
    {
      "epoch": 2.462132462132462,
      "grad_norm": 1.1533724069595337,
      "learning_rate": 1.5227205227205229e-05,
      "loss": 0.0586,
      "step": 8290
    },
    {
      "epoch": 2.465102465102465,
      "grad_norm": 0.972823441028595,
      "learning_rate": 1.520938520938521e-05,
      "loss": 0.0431,
      "step": 8300
    },
    {
      "epoch": 2.468072468072468,
      "grad_norm": 0.6986684799194336,
      "learning_rate": 1.5191565191565193e-05,
      "loss": 0.0591,
      "step": 8310
    },
    {
      "epoch": 2.471042471042471,
      "grad_norm": 0.6188581585884094,
      "learning_rate": 1.5173745173745173e-05,
      "loss": 0.0557,
      "step": 8320
    },
    {
      "epoch": 2.474012474012474,
      "grad_norm": 0.7166512608528137,
      "learning_rate": 1.5155925155925156e-05,
      "loss": 0.0562,
      "step": 8330
    },
    {
      "epoch": 2.476982476982477,
      "grad_norm": 0.6496429443359375,
      "learning_rate": 1.5138105138105138e-05,
      "loss": 0.0555,
      "step": 8340
    },
    {
      "epoch": 2.47995247995248,
      "grad_norm": 0.9905614852905273,
      "learning_rate": 1.5120285120285122e-05,
      "loss": 0.0574,
      "step": 8350
    },
    {
      "epoch": 2.4829224829224827,
      "grad_norm": 0.5973777770996094,
      "learning_rate": 1.5102465102465104e-05,
      "loss": 0.0636,
      "step": 8360
    },
    {
      "epoch": 2.485892485892486,
      "grad_norm": 0.4405939280986786,
      "learning_rate": 1.5084645084645086e-05,
      "loss": 0.0795,
      "step": 8370
    },
    {
      "epoch": 2.488862488862489,
      "grad_norm": 0.6318250894546509,
      "learning_rate": 1.5066825066825068e-05,
      "loss": 0.0599,
      "step": 8380
    },
    {
      "epoch": 2.4918324918324917,
      "grad_norm": 0.5132951140403748,
      "learning_rate": 1.5049005049005049e-05,
      "loss": 0.0632,
      "step": 8390
    },
    {
      "epoch": 2.494802494802495,
      "grad_norm": 1.1621384620666504,
      "learning_rate": 1.503118503118503e-05,
      "loss": 0.0716,
      "step": 8400
    },
    {
      "epoch": 2.4977724977724978,
      "grad_norm": 0.7687697410583496,
      "learning_rate": 1.5013365013365013e-05,
      "loss": 0.0543,
      "step": 8410
    },
    {
      "epoch": 2.5007425007425006,
      "grad_norm": 1.1648590564727783,
      "learning_rate": 1.4995544995544995e-05,
      "loss": 0.0525,
      "step": 8420
    },
    {
      "epoch": 2.503712503712504,
      "grad_norm": 1.093809723854065,
      "learning_rate": 1.4977724977724977e-05,
      "loss": 0.0577,
      "step": 8430
    },
    {
      "epoch": 2.5066825066825067,
      "grad_norm": 0.5859802961349487,
      "learning_rate": 1.4959904959904961e-05,
      "loss": 0.0505,
      "step": 8440
    },
    {
      "epoch": 2.5096525096525095,
      "grad_norm": 0.818012535572052,
      "learning_rate": 1.4942084942084943e-05,
      "loss": 0.06,
      "step": 8450
    },
    {
      "epoch": 2.512622512622513,
      "grad_norm": 0.4179311990737915,
      "learning_rate": 1.4924264924264924e-05,
      "loss": 0.0538,
      "step": 8460
    },
    {
      "epoch": 2.5155925155925156,
      "grad_norm": 1.3955974578857422,
      "learning_rate": 1.4906444906444908e-05,
      "loss": 0.0721,
      "step": 8470
    },
    {
      "epoch": 2.5185625185625184,
      "grad_norm": 0.8818016052246094,
      "learning_rate": 1.488862488862489e-05,
      "loss": 0.0578,
      "step": 8480
    },
    {
      "epoch": 2.5215325215325217,
      "grad_norm": 0.7490390539169312,
      "learning_rate": 1.487080487080487e-05,
      "loss": 0.0528,
      "step": 8490
    },
    {
      "epoch": 2.5245025245025245,
      "grad_norm": 0.7568694949150085,
      "learning_rate": 1.4852984852984852e-05,
      "loss": 0.0649,
      "step": 8500
    },
    {
      "epoch": 2.5274725274725274,
      "grad_norm": 0.7171285152435303,
      "learning_rate": 1.4835164835164836e-05,
      "loss": 0.0474,
      "step": 8510
    },
    {
      "epoch": 2.5304425304425306,
      "grad_norm": 0.5041258931159973,
      "learning_rate": 1.4817344817344818e-05,
      "loss": 0.0562,
      "step": 8520
    },
    {
      "epoch": 2.5334125334125335,
      "grad_norm": 0.5090057253837585,
      "learning_rate": 1.4799524799524799e-05,
      "loss": 0.0709,
      "step": 8530
    },
    {
      "epoch": 2.5363825363825363,
      "grad_norm": 0.5413128137588501,
      "learning_rate": 1.4781704781704783e-05,
      "loss": 0.0619,
      "step": 8540
    },
    {
      "epoch": 2.5393525393525396,
      "grad_norm": 0.8878335952758789,
      "learning_rate": 1.4763884763884765e-05,
      "loss": 0.0625,
      "step": 8550
    },
    {
      "epoch": 2.5423225423225424,
      "grad_norm": 0.7137174606323242,
      "learning_rate": 1.4746064746064745e-05,
      "loss": 0.057,
      "step": 8560
    },
    {
      "epoch": 2.5452925452925452,
      "grad_norm": 0.7620881795883179,
      "learning_rate": 1.4728244728244728e-05,
      "loss": 0.0681,
      "step": 8570
    },
    {
      "epoch": 2.5482625482625485,
      "grad_norm": 1.1073065996170044,
      "learning_rate": 1.4710424710424711e-05,
      "loss": 0.0708,
      "step": 8580
    },
    {
      "epoch": 2.5512325512325513,
      "grad_norm": 0.7626820802688599,
      "learning_rate": 1.4692604692604694e-05,
      "loss": 0.0616,
      "step": 8590
    },
    {
      "epoch": 2.554202554202554,
      "grad_norm": 0.849026083946228,
      "learning_rate": 1.4674784674784674e-05,
      "loss": 0.0511,
      "step": 8600
    },
    {
      "epoch": 2.5571725571725574,
      "grad_norm": 0.7794767618179321,
      "learning_rate": 1.4656964656964658e-05,
      "loss": 0.0527,
      "step": 8610
    },
    {
      "epoch": 2.5601425601425603,
      "grad_norm": 0.6230559349060059,
      "learning_rate": 1.463914463914464e-05,
      "loss": 0.0615,
      "step": 8620
    },
    {
      "epoch": 2.563112563112563,
      "grad_norm": 0.517436146736145,
      "learning_rate": 1.4621324621324622e-05,
      "loss": 0.0562,
      "step": 8630
    },
    {
      "epoch": 2.5660825660825664,
      "grad_norm": 0.835220217704773,
      "learning_rate": 1.4603504603504603e-05,
      "loss": 0.0548,
      "step": 8640
    },
    {
      "epoch": 2.569052569052569,
      "grad_norm": 0.8735977411270142,
      "learning_rate": 1.4585684585684587e-05,
      "loss": 0.0728,
      "step": 8650
    },
    {
      "epoch": 2.572022572022572,
      "grad_norm": 0.27027377486228943,
      "learning_rate": 1.4567864567864569e-05,
      "loss": 0.0512,
      "step": 8660
    },
    {
      "epoch": 2.574992574992575,
      "grad_norm": 0.804161548614502,
      "learning_rate": 1.455004455004455e-05,
      "loss": 0.0598,
      "step": 8670
    },
    {
      "epoch": 2.577962577962578,
      "grad_norm": 0.9159836173057556,
      "learning_rate": 1.4532224532224533e-05,
      "loss": 0.0755,
      "step": 8680
    },
    {
      "epoch": 2.580932580932581,
      "grad_norm": 1.0621222257614136,
      "learning_rate": 1.4514404514404515e-05,
      "loss": 0.0599,
      "step": 8690
    },
    {
      "epoch": 2.5839025839025838,
      "grad_norm": 0.61686110496521,
      "learning_rate": 1.4496584496584498e-05,
      "loss": 0.0606,
      "step": 8700
    },
    {
      "epoch": 2.586872586872587,
      "grad_norm": 0.3363722860813141,
      "learning_rate": 1.4478764478764478e-05,
      "loss": 0.0451,
      "step": 8710
    },
    {
      "epoch": 2.58984258984259,
      "grad_norm": 0.5933698415756226,
      "learning_rate": 1.4460944460944462e-05,
      "loss": 0.0635,
      "step": 8720
    },
    {
      "epoch": 2.5928125928125927,
      "grad_norm": 0.6424260139465332,
      "learning_rate": 1.4443124443124444e-05,
      "loss": 0.0537,
      "step": 8730
    },
    {
      "epoch": 2.5957825957825955,
      "grad_norm": 0.981282651424408,
      "learning_rate": 1.4425304425304425e-05,
      "loss": 0.0635,
      "step": 8740
    },
    {
      "epoch": 2.598752598752599,
      "grad_norm": 1.4847664833068848,
      "learning_rate": 1.4407484407484408e-05,
      "loss": 0.0664,
      "step": 8750
    },
    {
      "epoch": 2.6017226017226016,
      "grad_norm": 0.5225690007209778,
      "learning_rate": 1.438966438966439e-05,
      "loss": 0.061,
      "step": 8760
    },
    {
      "epoch": 2.6046926046926044,
      "grad_norm": 0.6397286057472229,
      "learning_rate": 1.4371844371844373e-05,
      "loss": 0.0674,
      "step": 8770
    },
    {
      "epoch": 2.6076626076626077,
      "grad_norm": 0.8102043867111206,
      "learning_rate": 1.4354024354024353e-05,
      "loss": 0.067,
      "step": 8780
    },
    {
      "epoch": 2.6106326106326105,
      "grad_norm": 0.9729601144790649,
      "learning_rate": 1.4336204336204337e-05,
      "loss": 0.0569,
      "step": 8790
    },
    {
      "epoch": 2.6136026136026134,
      "grad_norm": 0.6307176947593689,
      "learning_rate": 1.431838431838432e-05,
      "loss": 0.056,
      "step": 8800
    },
    {
      "epoch": 2.6165726165726166,
      "grad_norm": 0.8650787472724915,
      "learning_rate": 1.43005643005643e-05,
      "loss": 0.0674,
      "step": 8810
    },
    {
      "epoch": 2.6195426195426195,
      "grad_norm": 0.9029502868652344,
      "learning_rate": 1.4282744282744284e-05,
      "loss": 0.0533,
      "step": 8820
    },
    {
      "epoch": 2.6225126225126223,
      "grad_norm": 0.817317008972168,
      "learning_rate": 1.4264924264924266e-05,
      "loss": 0.0596,
      "step": 8830
    },
    {
      "epoch": 2.6254826254826256,
      "grad_norm": 0.9638757705688477,
      "learning_rate": 1.4247104247104248e-05,
      "loss": 0.0603,
      "step": 8840
    },
    {
      "epoch": 2.6284526284526284,
      "grad_norm": 0.9621570110321045,
      "learning_rate": 1.4229284229284228e-05,
      "loss": 0.0612,
      "step": 8850
    },
    {
      "epoch": 2.631422631422631,
      "grad_norm": 0.504012942314148,
      "learning_rate": 1.4211464211464212e-05,
      "loss": 0.0747,
      "step": 8860
    },
    {
      "epoch": 2.6343926343926345,
      "grad_norm": 1.0219043493270874,
      "learning_rate": 1.4193644193644194e-05,
      "loss": 0.052,
      "step": 8870
    },
    {
      "epoch": 2.6373626373626373,
      "grad_norm": 0.7350105047225952,
      "learning_rate": 1.4175824175824177e-05,
      "loss": 0.0604,
      "step": 8880
    },
    {
      "epoch": 2.64033264033264,
      "grad_norm": 0.5058907270431519,
      "learning_rate": 1.4158004158004159e-05,
      "loss": 0.0558,
      "step": 8890
    },
    {
      "epoch": 2.6433026433026434,
      "grad_norm": 0.6319795846939087,
      "learning_rate": 1.4140184140184141e-05,
      "loss": 0.0505,
      "step": 8900
    },
    {
      "epoch": 2.6462726462726462,
      "grad_norm": 0.6916008591651917,
      "learning_rate": 1.4122364122364123e-05,
      "loss": 0.0686,
      "step": 8910
    },
    {
      "epoch": 2.649242649242649,
      "grad_norm": 1.57939612865448,
      "learning_rate": 1.4104544104544104e-05,
      "loss": 0.0534,
      "step": 8920
    },
    {
      "epoch": 2.6522126522126523,
      "grad_norm": 1.0057226419448853,
      "learning_rate": 1.4086724086724087e-05,
      "loss": 0.0573,
      "step": 8930
    },
    {
      "epoch": 2.655182655182655,
      "grad_norm": 0.6452613472938538,
      "learning_rate": 1.406890406890407e-05,
      "loss": 0.0678,
      "step": 8940
    },
    {
      "epoch": 2.658152658152658,
      "grad_norm": 0.45321300625801086,
      "learning_rate": 1.4051084051084052e-05,
      "loss": 0.0453,
      "step": 8950
    },
    {
      "epoch": 2.6611226611226613,
      "grad_norm": 0.46493178606033325,
      "learning_rate": 1.4033264033264034e-05,
      "loss": 0.0579,
      "step": 8960
    },
    {
      "epoch": 2.664092664092664,
      "grad_norm": 0.6382163763046265,
      "learning_rate": 1.4015444015444016e-05,
      "loss": 0.0537,
      "step": 8970
    },
    {
      "epoch": 2.667062667062667,
      "grad_norm": 0.5830327272415161,
      "learning_rate": 1.3997623997623998e-05,
      "loss": 0.0529,
      "step": 8980
    },
    {
      "epoch": 2.67003267003267,
      "grad_norm": 0.8241320252418518,
      "learning_rate": 1.3979803979803979e-05,
      "loss": 0.0501,
      "step": 8990
    },
    {
      "epoch": 2.673002673002673,
      "grad_norm": 1.3200924396514893,
      "learning_rate": 1.3961983961983963e-05,
      "loss": 0.0637,
      "step": 9000
    },
    {
      "epoch": 2.675972675972676,
      "grad_norm": 0.5963950157165527,
      "learning_rate": 1.3944163944163945e-05,
      "loss": 0.0581,
      "step": 9010
    },
    {
      "epoch": 2.678942678942679,
      "grad_norm": 0.5137681365013123,
      "learning_rate": 1.3926343926343927e-05,
      "loss": 0.0504,
      "step": 9020
    },
    {
      "epoch": 2.681912681912682,
      "grad_norm": 0.8717916011810303,
      "learning_rate": 1.390852390852391e-05,
      "loss": 0.0681,
      "step": 9030
    },
    {
      "epoch": 2.684882684882685,
      "grad_norm": 0.5380828976631165,
      "learning_rate": 1.3890703890703891e-05,
      "loss": 0.0568,
      "step": 9040
    },
    {
      "epoch": 2.687852687852688,
      "grad_norm": 0.8956130743026733,
      "learning_rate": 1.3872883872883874e-05,
      "loss": 0.0623,
      "step": 9050
    },
    {
      "epoch": 2.690822690822691,
      "grad_norm": 0.6086248159408569,
      "learning_rate": 1.3855063855063854e-05,
      "loss": 0.0643,
      "step": 9060
    },
    {
      "epoch": 2.6937926937926937,
      "grad_norm": 0.8992329835891724,
      "learning_rate": 1.3837243837243838e-05,
      "loss": 0.0649,
      "step": 9070
    },
    {
      "epoch": 2.696762696762697,
      "grad_norm": 0.3477851450443268,
      "learning_rate": 1.381942381942382e-05,
      "loss": 0.0479,
      "step": 9080
    },
    {
      "epoch": 2.6997326997327,
      "grad_norm": 1.4136226177215576,
      "learning_rate": 1.3801603801603802e-05,
      "loss": 0.0529,
      "step": 9090
    },
    {
      "epoch": 2.7027027027027026,
      "grad_norm": 1.031639814376831,
      "learning_rate": 1.3783783783783784e-05,
      "loss": 0.0607,
      "step": 9100
    },
    {
      "epoch": 2.705672705672706,
      "grad_norm": 0.9110945463180542,
      "learning_rate": 1.3765963765963767e-05,
      "loss": 0.054,
      "step": 9110
    },
    {
      "epoch": 2.7086427086427087,
      "grad_norm": 1.111244797706604,
      "learning_rate": 1.3748143748143749e-05,
      "loss": 0.065,
      "step": 9120
    },
    {
      "epoch": 2.7116127116127116,
      "grad_norm": 0.507455587387085,
      "learning_rate": 1.373032373032373e-05,
      "loss": 0.0605,
      "step": 9130
    },
    {
      "epoch": 2.714582714582715,
      "grad_norm": 1.2011653184890747,
      "learning_rate": 1.3712503712503713e-05,
      "loss": 0.0535,
      "step": 9140
    },
    {
      "epoch": 2.7175527175527177,
      "grad_norm": 0.48414701223373413,
      "learning_rate": 1.3694683694683695e-05,
      "loss": 0.0488,
      "step": 9150
    },
    {
      "epoch": 2.7205227205227205,
      "grad_norm": 1.0523313283920288,
      "learning_rate": 1.3676863676863677e-05,
      "loss": 0.0657,
      "step": 9160
    },
    {
      "epoch": 2.7234927234927238,
      "grad_norm": 0.8370203971862793,
      "learning_rate": 1.365904365904366e-05,
      "loss": 0.0563,
      "step": 9170
    },
    {
      "epoch": 2.7264627264627266,
      "grad_norm": 0.6961177587509155,
      "learning_rate": 1.3641223641223642e-05,
      "loss": 0.0587,
      "step": 9180
    },
    {
      "epoch": 2.7294327294327294,
      "grad_norm": 1.089735507965088,
      "learning_rate": 1.3623403623403624e-05,
      "loss": 0.0606,
      "step": 9190
    },
    {
      "epoch": 2.7324027324027322,
      "grad_norm": 0.8496239185333252,
      "learning_rate": 1.3605583605583606e-05,
      "loss": 0.0572,
      "step": 9200
    },
    {
      "epoch": 2.7353727353727355,
      "grad_norm": 0.8729379773139954,
      "learning_rate": 1.3587763587763588e-05,
      "loss": 0.0604,
      "step": 9210
    },
    {
      "epoch": 2.7383427383427383,
      "grad_norm": 0.8139004111289978,
      "learning_rate": 1.356994356994357e-05,
      "loss": 0.052,
      "step": 9220
    },
    {
      "epoch": 2.741312741312741,
      "grad_norm": 1.0032668113708496,
      "learning_rate": 1.3552123552123553e-05,
      "loss": 0.0578,
      "step": 9230
    },
    {
      "epoch": 2.7442827442827444,
      "grad_norm": 0.5807051062583923,
      "learning_rate": 1.3534303534303535e-05,
      "loss": 0.0581,
      "step": 9240
    },
    {
      "epoch": 2.7472527472527473,
      "grad_norm": 0.5372444987297058,
      "learning_rate": 1.3516483516483517e-05,
      "loss": 0.0606,
      "step": 9250
    },
    {
      "epoch": 2.75022275022275,
      "grad_norm": 2.1565425395965576,
      "learning_rate": 1.3498663498663499e-05,
      "loss": 0.0545,
      "step": 9260
    },
    {
      "epoch": 2.753192753192753,
      "grad_norm": 0.6508318185806274,
      "learning_rate": 1.3480843480843481e-05,
      "loss": 0.0591,
      "step": 9270
    },
    {
      "epoch": 2.756162756162756,
      "grad_norm": 0.36386728286743164,
      "learning_rate": 1.3463023463023463e-05,
      "loss": 0.049,
      "step": 9280
    },
    {
      "epoch": 2.759132759132759,
      "grad_norm": 1.4432202577590942,
      "learning_rate": 1.3445203445203446e-05,
      "loss": 0.067,
      "step": 9290
    },
    {
      "epoch": 2.762102762102762,
      "grad_norm": 0.7495296001434326,
      "learning_rate": 1.3427383427383428e-05,
      "loss": 0.0551,
      "step": 9300
    },
    {
      "epoch": 2.765072765072765,
      "grad_norm": 0.5541219711303711,
      "learning_rate": 1.340956340956341e-05,
      "loss": 0.0594,
      "step": 9310
    },
    {
      "epoch": 2.768042768042768,
      "grad_norm": 0.8535263538360596,
      "learning_rate": 1.3391743391743392e-05,
      "loss": 0.0583,
      "step": 9320
    },
    {
      "epoch": 2.7710127710127708,
      "grad_norm": 0.784618616104126,
      "learning_rate": 1.3373923373923374e-05,
      "loss": 0.0705,
      "step": 9330
    },
    {
      "epoch": 2.773982773982774,
      "grad_norm": 0.6920385360717773,
      "learning_rate": 1.3356103356103356e-05,
      "loss": 0.0686,
      "step": 9340
    },
    {
      "epoch": 2.776952776952777,
      "grad_norm": 0.7097575068473816,
      "learning_rate": 1.3338283338283339e-05,
      "loss": 0.0546,
      "step": 9350
    },
    {
      "epoch": 2.7799227799227797,
      "grad_norm": 0.9163897633552551,
      "learning_rate": 1.332046332046332e-05,
      "loss": 0.0604,
      "step": 9360
    },
    {
      "epoch": 2.782892782892783,
      "grad_norm": 0.5482726097106934,
      "learning_rate": 1.3302643302643303e-05,
      "loss": 0.0561,
      "step": 9370
    },
    {
      "epoch": 2.785862785862786,
      "grad_norm": 0.5546408891677856,
      "learning_rate": 1.3284823284823285e-05,
      "loss": 0.0664,
      "step": 9380
    },
    {
      "epoch": 2.7888327888327886,
      "grad_norm": 0.9671948552131653,
      "learning_rate": 1.3267003267003267e-05,
      "loss": 0.049,
      "step": 9390
    },
    {
      "epoch": 2.791802791802792,
      "grad_norm": 0.8189311027526855,
      "learning_rate": 1.324918324918325e-05,
      "loss": 0.0464,
      "step": 9400
    },
    {
      "epoch": 2.7947727947727947,
      "grad_norm": 0.5405160188674927,
      "learning_rate": 1.3231363231363232e-05,
      "loss": 0.0673,
      "step": 9410
    },
    {
      "epoch": 2.7977427977427975,
      "grad_norm": 0.8173415660858154,
      "learning_rate": 1.3213543213543214e-05,
      "loss": 0.0589,
      "step": 9420
    },
    {
      "epoch": 2.800712800712801,
      "grad_norm": 0.6421013474464417,
      "learning_rate": 1.3195723195723196e-05,
      "loss": 0.0591,
      "step": 9430
    },
    {
      "epoch": 2.8036828036828036,
      "grad_norm": 1.540049433708191,
      "learning_rate": 1.3177903177903178e-05,
      "loss": 0.072,
      "step": 9440
    },
    {
      "epoch": 2.8066528066528065,
      "grad_norm": 0.6752909421920776,
      "learning_rate": 1.316008316008316e-05,
      "loss": 0.049,
      "step": 9450
    },
    {
      "epoch": 2.8096228096228097,
      "grad_norm": 0.9367174506187439,
      "learning_rate": 1.3142263142263142e-05,
      "loss": 0.0738,
      "step": 9460
    },
    {
      "epoch": 2.8125928125928126,
      "grad_norm": 1.2231054306030273,
      "learning_rate": 1.3124443124443125e-05,
      "loss": 0.0625,
      "step": 9470
    },
    {
      "epoch": 2.8155628155628154,
      "grad_norm": 0.4123198091983795,
      "learning_rate": 1.3106623106623107e-05,
      "loss": 0.0518,
      "step": 9480
    },
    {
      "epoch": 2.8185328185328187,
      "grad_norm": 0.6436600089073181,
      "learning_rate": 1.3088803088803089e-05,
      "loss": 0.0725,
      "step": 9490
    },
    {
      "epoch": 2.8215028215028215,
      "grad_norm": 0.6609872579574585,
      "learning_rate": 1.3070983070983071e-05,
      "loss": 0.0594,
      "step": 9500
    },
    {
      "epoch": 2.8244728244728243,
      "grad_norm": 0.4559807777404785,
      "learning_rate": 1.3053163053163053e-05,
      "loss": 0.063,
      "step": 9510
    },
    {
      "epoch": 2.8274428274428276,
      "grad_norm": 1.0290307998657227,
      "learning_rate": 1.3035343035343037e-05,
      "loss": 0.0618,
      "step": 9520
    },
    {
      "epoch": 2.8304128304128304,
      "grad_norm": 0.7586894035339355,
      "learning_rate": 1.3017523017523018e-05,
      "loss": 0.0584,
      "step": 9530
    },
    {
      "epoch": 2.8333828333828333,
      "grad_norm": 0.714316725730896,
      "learning_rate": 1.2999702999703e-05,
      "loss": 0.0448,
      "step": 9540
    },
    {
      "epoch": 2.8363528363528365,
      "grad_norm": 0.6114319562911987,
      "learning_rate": 1.2981882981882982e-05,
      "loss": 0.0689,
      "step": 9550
    },
    {
      "epoch": 2.8393228393228394,
      "grad_norm": 0.7381054162979126,
      "learning_rate": 1.2964062964062964e-05,
      "loss": 0.0577,
      "step": 9560
    },
    {
      "epoch": 2.842292842292842,
      "grad_norm": 0.7892597317695618,
      "learning_rate": 1.2946242946242946e-05,
      "loss": 0.0697,
      "step": 9570
    },
    {
      "epoch": 2.8452628452628455,
      "grad_norm": 0.6848814487457275,
      "learning_rate": 1.2928422928422929e-05,
      "loss": 0.0521,
      "step": 9580
    },
    {
      "epoch": 2.8482328482328483,
      "grad_norm": 0.5784212946891785,
      "learning_rate": 1.2910602910602912e-05,
      "loss": 0.0647,
      "step": 9590
    },
    {
      "epoch": 2.851202851202851,
      "grad_norm": 0.6273770332336426,
      "learning_rate": 1.2892782892782893e-05,
      "loss": 0.0677,
      "step": 9600
    },
    {
      "epoch": 2.8541728541728544,
      "grad_norm": 0.7045747637748718,
      "learning_rate": 1.2874962874962875e-05,
      "loss": 0.0686,
      "step": 9610
    },
    {
      "epoch": 2.857142857142857,
      "grad_norm": 0.8756235241889954,
      "learning_rate": 1.2857142857142857e-05,
      "loss": 0.0551,
      "step": 9620
    },
    {
      "epoch": 2.86011286011286,
      "grad_norm": 1.0615090131759644,
      "learning_rate": 1.283932283932284e-05,
      "loss": 0.0676,
      "step": 9630
    },
    {
      "epoch": 2.8630828630828633,
      "grad_norm": 0.42012715339660645,
      "learning_rate": 1.2821502821502822e-05,
      "loss": 0.0492,
      "step": 9640
    },
    {
      "epoch": 2.866052866052866,
      "grad_norm": 0.8934495449066162,
      "learning_rate": 1.2803682803682804e-05,
      "loss": 0.0523,
      "step": 9650
    },
    {
      "epoch": 2.869022869022869,
      "grad_norm": 1.0009864568710327,
      "learning_rate": 1.2785862785862788e-05,
      "loss": 0.0662,
      "step": 9660
    },
    {
      "epoch": 2.8719928719928722,
      "grad_norm": 0.8075212836265564,
      "learning_rate": 1.2768042768042768e-05,
      "loss": 0.0553,
      "step": 9670
    },
    {
      "epoch": 2.874962874962875,
      "grad_norm": 0.7965303063392639,
      "learning_rate": 1.275022275022275e-05,
      "loss": 0.0573,
      "step": 9680
    },
    {
      "epoch": 2.877932877932878,
      "grad_norm": 0.6948648691177368,
      "learning_rate": 1.2732402732402732e-05,
      "loss": 0.0557,
      "step": 9690
    },
    {
      "epoch": 2.880902880902881,
      "grad_norm": 0.7285399436950684,
      "learning_rate": 1.2714582714582715e-05,
      "loss": 0.0624,
      "step": 9700
    },
    {
      "epoch": 2.883872883872884,
      "grad_norm": 0.37577903270721436,
      "learning_rate": 1.2696762696762697e-05,
      "loss": 0.0517,
      "step": 9710
    },
    {
      "epoch": 2.886842886842887,
      "grad_norm": 0.36712825298309326,
      "learning_rate": 1.2678942678942679e-05,
      "loss": 0.0567,
      "step": 9720
    },
    {
      "epoch": 2.88981288981289,
      "grad_norm": 1.1397478580474854,
      "learning_rate": 1.2661122661122663e-05,
      "loss": 0.0636,
      "step": 9730
    },
    {
      "epoch": 2.892782892782893,
      "grad_norm": 0.5818475484848022,
      "learning_rate": 1.2643302643302643e-05,
      "loss": 0.0535,
      "step": 9740
    },
    {
      "epoch": 2.8957528957528957,
      "grad_norm": 0.7430572509765625,
      "learning_rate": 1.2625482625482625e-05,
      "loss": 0.0484,
      "step": 9750
    },
    {
      "epoch": 2.8987228987228986,
      "grad_norm": 1.0265908241271973,
      "learning_rate": 1.2607662607662608e-05,
      "loss": 0.0566,
      "step": 9760
    },
    {
      "epoch": 2.901692901692902,
      "grad_norm": 1.191502332687378,
      "learning_rate": 1.258984258984259e-05,
      "loss": 0.0574,
      "step": 9770
    },
    {
      "epoch": 2.9046629046629047,
      "grad_norm": 0.7020824551582336,
      "learning_rate": 1.2572022572022572e-05,
      "loss": 0.0498,
      "step": 9780
    },
    {
      "epoch": 2.9076329076329075,
      "grad_norm": 1.4909850358963013,
      "learning_rate": 1.2554202554202554e-05,
      "loss": 0.0648,
      "step": 9790
    },
    {
      "epoch": 2.9106029106029108,
      "grad_norm": 0.9916106462478638,
      "learning_rate": 1.2536382536382538e-05,
      "loss": 0.068,
      "step": 9800
    },
    {
      "epoch": 2.9135729135729136,
      "grad_norm": 0.7778554558753967,
      "learning_rate": 1.2518562518562518e-05,
      "loss": 0.0518,
      "step": 9810
    },
    {
      "epoch": 2.9165429165429164,
      "grad_norm": 1.0887690782546997,
      "learning_rate": 1.25007425007425e-05,
      "loss": 0.0668,
      "step": 9820
    },
    {
      "epoch": 2.9195129195129192,
      "grad_norm": 0.5052692294120789,
      "learning_rate": 1.2482922482922483e-05,
      "loss": 0.0489,
      "step": 9830
    },
    {
      "epoch": 2.9224829224829225,
      "grad_norm": 0.9965596199035645,
      "learning_rate": 1.2465102465102467e-05,
      "loss": 0.0596,
      "step": 9840
    },
    {
      "epoch": 2.9254529254529253,
      "grad_norm": 0.6010634303092957,
      "learning_rate": 1.2447282447282447e-05,
      "loss": 0.0556,
      "step": 9850
    },
    {
      "epoch": 2.928422928422928,
      "grad_norm": 0.7271102070808411,
      "learning_rate": 1.242946242946243e-05,
      "loss": 0.0539,
      "step": 9860
    },
    {
      "epoch": 2.9313929313929314,
      "grad_norm": 0.5833537578582764,
      "learning_rate": 1.2411642411642413e-05,
      "loss": 0.0551,
      "step": 9870
    },
    {
      "epoch": 2.9343629343629343,
      "grad_norm": 1.5592166185379028,
      "learning_rate": 1.2393822393822394e-05,
      "loss": 0.0649,
      "step": 9880
    },
    {
      "epoch": 2.937332937332937,
      "grad_norm": 0.6563842296600342,
      "learning_rate": 1.2376002376002376e-05,
      "loss": 0.047,
      "step": 9890
    },
    {
      "epoch": 2.9403029403029404,
      "grad_norm": 0.8599936366081238,
      "learning_rate": 1.2358182358182358e-05,
      "loss": 0.0665,
      "step": 9900
    },
    {
      "epoch": 2.943272943272943,
      "grad_norm": 0.6941415667533875,
      "learning_rate": 1.2340362340362342e-05,
      "loss": 0.0603,
      "step": 9910
    },
    {
      "epoch": 2.946242946242946,
      "grad_norm": 0.6341265439987183,
      "learning_rate": 1.2322542322542322e-05,
      "loss": 0.0701,
      "step": 9920
    },
    {
      "epoch": 2.9492129492129493,
      "grad_norm": 0.8768404722213745,
      "learning_rate": 1.2304722304722305e-05,
      "loss": 0.0635,
      "step": 9930
    },
    {
      "epoch": 2.952182952182952,
      "grad_norm": 0.7118542194366455,
      "learning_rate": 1.2286902286902288e-05,
      "loss": 0.0592,
      "step": 9940
    },
    {
      "epoch": 2.955152955152955,
      "grad_norm": 1.1415071487426758,
      "learning_rate": 1.2269082269082269e-05,
      "loss": 0.0649,
      "step": 9950
    },
    {
      "epoch": 2.9581229581229582,
      "grad_norm": 1.042884349822998,
      "learning_rate": 1.2251262251262251e-05,
      "loss": 0.0522,
      "step": 9960
    },
    {
      "epoch": 2.961092961092961,
      "grad_norm": 0.5634980201721191,
      "learning_rate": 1.2233442233442233e-05,
      "loss": 0.0532,
      "step": 9970
    },
    {
      "epoch": 2.964062964062964,
      "grad_norm": 0.8798786401748657,
      "learning_rate": 1.2215622215622217e-05,
      "loss": 0.0575,
      "step": 9980
    },
    {
      "epoch": 2.967032967032967,
      "grad_norm": 0.651884913444519,
      "learning_rate": 1.2197802197802198e-05,
      "loss": 0.0469,
      "step": 9990
    },
    {
      "epoch": 2.97000297000297,
      "grad_norm": 0.9131767749786377,
      "learning_rate": 1.217998217998218e-05,
      "loss": 0.0576,
      "step": 10000
    },
    {
      "epoch": 2.972972972972973,
      "grad_norm": 1.2109713554382324,
      "learning_rate": 1.2162162162162164e-05,
      "loss": 0.0663,
      "step": 10010
    },
    {
      "epoch": 2.975942975942976,
      "grad_norm": 0.9525397419929504,
      "learning_rate": 1.2144342144342144e-05,
      "loss": 0.0531,
      "step": 10020
    },
    {
      "epoch": 2.978912978912979,
      "grad_norm": 0.7746742963790894,
      "learning_rate": 1.2126522126522126e-05,
      "loss": 0.0619,
      "step": 10030
    },
    {
      "epoch": 2.9818829818829817,
      "grad_norm": 0.526714026927948,
      "learning_rate": 1.2108702108702108e-05,
      "loss": 0.0639,
      "step": 10040
    },
    {
      "epoch": 2.984852984852985,
      "grad_norm": 0.42681199312210083,
      "learning_rate": 1.2090882090882092e-05,
      "loss": 0.059,
      "step": 10050
    },
    {
      "epoch": 2.987822987822988,
      "grad_norm": 1.26163911819458,
      "learning_rate": 1.2073062073062073e-05,
      "loss": 0.0534,
      "step": 10060
    },
    {
      "epoch": 2.9907929907929907,
      "grad_norm": 0.6416770815849304,
      "learning_rate": 1.2055242055242055e-05,
      "loss": 0.0668,
      "step": 10070
    },
    {
      "epoch": 2.993762993762994,
      "grad_norm": 0.7979917526245117,
      "learning_rate": 1.2037422037422039e-05,
      "loss": 0.0553,
      "step": 10080
    },
    {
      "epoch": 2.9967329967329968,
      "grad_norm": 0.6158313751220703,
      "learning_rate": 1.2019602019602021e-05,
      "loss": 0.0388,
      "step": 10090
    },
    {
      "epoch": 2.9997029997029996,
      "grad_norm": 0.5872673392295837,
      "learning_rate": 1.2001782001782001e-05,
      "loss": 0.0508,
      "step": 10100
    },
    {
      "epoch": 3.0,
      "eval_f1": 0.33031292965957215,
      "eval_loss": 0.04517492279410362,
      "eval_runtime": 166.1304,
      "eval_samples_per_second": 228.85,
      "eval_steps_per_second": 3.582,
      "step": 10101
    },
    {
      "epoch": 3.002673002673003,
      "grad_norm": 0.6985778212547302,
      "learning_rate": 1.1983961983961984e-05,
      "loss": 0.056,
      "step": 10110
    },
    {
      "epoch": 3.0056430056430057,
      "grad_norm": 0.88740074634552,
      "learning_rate": 1.1966141966141967e-05,
      "loss": 0.0639,
      "step": 10120
    },
    {
      "epoch": 3.0086130086130085,
      "grad_norm": 1.2202911376953125,
      "learning_rate": 1.1948321948321948e-05,
      "loss": 0.0697,
      "step": 10130
    },
    {
      "epoch": 3.011583011583012,
      "grad_norm": 0.9488741755485535,
      "learning_rate": 1.193050193050193e-05,
      "loss": 0.0473,
      "step": 10140
    },
    {
      "epoch": 3.0145530145530146,
      "grad_norm": 0.6430271863937378,
      "learning_rate": 1.1912681912681914e-05,
      "loss": 0.0654,
      "step": 10150
    },
    {
      "epoch": 3.0175230175230174,
      "grad_norm": 0.4088257849216461,
      "learning_rate": 1.1894861894861896e-05,
      "loss": 0.0608,
      "step": 10160
    },
    {
      "epoch": 3.0204930204930207,
      "grad_norm": 0.8268032670021057,
      "learning_rate": 1.1877041877041877e-05,
      "loss": 0.0543,
      "step": 10170
    },
    {
      "epoch": 3.0234630234630235,
      "grad_norm": 0.7411820888519287,
      "learning_rate": 1.1859221859221859e-05,
      "loss": 0.045,
      "step": 10180
    },
    {
      "epoch": 3.0264330264330264,
      "grad_norm": 1.4683622121810913,
      "learning_rate": 1.1841401841401843e-05,
      "loss": 0.0657,
      "step": 10190
    },
    {
      "epoch": 3.029403029403029,
      "grad_norm": 0.926177978515625,
      "learning_rate": 1.1823581823581823e-05,
      "loss": 0.0535,
      "step": 10200
    },
    {
      "epoch": 3.0323730323730325,
      "grad_norm": 0.816768229007721,
      "learning_rate": 1.1805761805761805e-05,
      "loss": 0.0591,
      "step": 10210
    },
    {
      "epoch": 3.0353430353430353,
      "grad_norm": 0.4738346040248871,
      "learning_rate": 1.1787941787941789e-05,
      "loss": 0.0697,
      "step": 10220
    },
    {
      "epoch": 3.038313038313038,
      "grad_norm": 0.748884379863739,
      "learning_rate": 1.1770121770121771e-05,
      "loss": 0.0489,
      "step": 10230
    },
    {
      "epoch": 3.0412830412830414,
      "grad_norm": 0.4384136199951172,
      "learning_rate": 1.1752301752301752e-05,
      "loss": 0.058,
      "step": 10240
    },
    {
      "epoch": 3.044253044253044,
      "grad_norm": 0.8452009558677673,
      "learning_rate": 1.1734481734481734e-05,
      "loss": 0.0579,
      "step": 10250
    },
    {
      "epoch": 3.047223047223047,
      "grad_norm": 1.2820180654525757,
      "learning_rate": 1.1716661716661718e-05,
      "loss": 0.0506,
      "step": 10260
    },
    {
      "epoch": 3.0501930501930503,
      "grad_norm": 0.6100145578384399,
      "learning_rate": 1.1698841698841698e-05,
      "loss": 0.0493,
      "step": 10270
    },
    {
      "epoch": 3.053163053163053,
      "grad_norm": 0.8073909282684326,
      "learning_rate": 1.168102168102168e-05,
      "loss": 0.059,
      "step": 10280
    },
    {
      "epoch": 3.056133056133056,
      "grad_norm": 0.6318356394767761,
      "learning_rate": 1.1663201663201664e-05,
      "loss": 0.0595,
      "step": 10290
    },
    {
      "epoch": 3.0591030591030592,
      "grad_norm": 0.7712961435317993,
      "learning_rate": 1.1645381645381647e-05,
      "loss": 0.0636,
      "step": 10300
    },
    {
      "epoch": 3.062073062073062,
      "grad_norm": 0.9575150012969971,
      "learning_rate": 1.1627561627561627e-05,
      "loss": 0.067,
      "step": 10310
    },
    {
      "epoch": 3.065043065043065,
      "grad_norm": 0.9202025532722473,
      "learning_rate": 1.160974160974161e-05,
      "loss": 0.0711,
      "step": 10320
    },
    {
      "epoch": 3.068013068013068,
      "grad_norm": 1.038837194442749,
      "learning_rate": 1.1591921591921593e-05,
      "loss": 0.0487,
      "step": 10330
    },
    {
      "epoch": 3.070983070983071,
      "grad_norm": 0.6200097799301147,
      "learning_rate": 1.1574101574101574e-05,
      "loss": 0.0578,
      "step": 10340
    },
    {
      "epoch": 3.073953073953074,
      "grad_norm": 0.6585675477981567,
      "learning_rate": 1.1556281556281556e-05,
      "loss": 0.0648,
      "step": 10350
    },
    {
      "epoch": 3.076923076923077,
      "grad_norm": 0.8432527184486389,
      "learning_rate": 1.153846153846154e-05,
      "loss": 0.0574,
      "step": 10360
    },
    {
      "epoch": 3.07989307989308,
      "grad_norm": 0.8519158959388733,
      "learning_rate": 1.1520641520641522e-05,
      "loss": 0.0435,
      "step": 10370
    },
    {
      "epoch": 3.0828630828630827,
      "grad_norm": 0.5639305710792542,
      "learning_rate": 1.1502821502821502e-05,
      "loss": 0.0548,
      "step": 10380
    },
    {
      "epoch": 3.085833085833086,
      "grad_norm": 1.1483186483383179,
      "learning_rate": 1.1485001485001484e-05,
      "loss": 0.067,
      "step": 10390
    },
    {
      "epoch": 3.088803088803089,
      "grad_norm": 0.7661743760108948,
      "learning_rate": 1.1467181467181468e-05,
      "loss": 0.0409,
      "step": 10400
    },
    {
      "epoch": 3.0917730917730917,
      "grad_norm": 0.42964890599250793,
      "learning_rate": 1.144936144936145e-05,
      "loss": 0.0656,
      "step": 10410
    },
    {
      "epoch": 3.094743094743095,
      "grad_norm": 0.4453743100166321,
      "learning_rate": 1.1431541431541431e-05,
      "loss": 0.0638,
      "step": 10420
    },
    {
      "epoch": 3.0977130977130978,
      "grad_norm": 1.3230763673782349,
      "learning_rate": 1.1413721413721415e-05,
      "loss": 0.0644,
      "step": 10430
    },
    {
      "epoch": 3.1006831006831006,
      "grad_norm": 0.6661444306373596,
      "learning_rate": 1.1395901395901397e-05,
      "loss": 0.0585,
      "step": 10440
    },
    {
      "epoch": 3.1036531036531034,
      "grad_norm": 0.6051294207572937,
      "learning_rate": 1.1378081378081377e-05,
      "loss": 0.0583,
      "step": 10450
    },
    {
      "epoch": 3.1066231066231067,
      "grad_norm": 0.9865986704826355,
      "learning_rate": 1.136026136026136e-05,
      "loss": 0.0547,
      "step": 10460
    },
    {
      "epoch": 3.1095931095931095,
      "grad_norm": 0.6704586148262024,
      "learning_rate": 1.1342441342441343e-05,
      "loss": 0.0582,
      "step": 10470
    },
    {
      "epoch": 3.1125631125631124,
      "grad_norm": 0.8656442761421204,
      "learning_rate": 1.1324621324621326e-05,
      "loss": 0.0574,
      "step": 10480
    },
    {
      "epoch": 3.1155331155331156,
      "grad_norm": 0.704118549823761,
      "learning_rate": 1.1306801306801306e-05,
      "loss": 0.0509,
      "step": 10490
    },
    {
      "epoch": 3.1185031185031185,
      "grad_norm": 0.5805778503417969,
      "learning_rate": 1.128898128898129e-05,
      "loss": 0.0587,
      "step": 10500
    },
    {
      "epoch": 3.1214731214731213,
      "grad_norm": 1.2092708349227905,
      "learning_rate": 1.1271161271161272e-05,
      "loss": 0.069,
      "step": 10510
    },
    {
      "epoch": 3.1244431244431246,
      "grad_norm": 0.7527390718460083,
      "learning_rate": 1.1253341253341253e-05,
      "loss": 0.059,
      "step": 10520
    },
    {
      "epoch": 3.1274131274131274,
      "grad_norm": 0.5099059343338013,
      "learning_rate": 1.1235521235521235e-05,
      "loss": 0.0616,
      "step": 10530
    },
    {
      "epoch": 3.13038313038313,
      "grad_norm": 0.6300451755523682,
      "learning_rate": 1.1217701217701219e-05,
      "loss": 0.0625,
      "step": 10540
    },
    {
      "epoch": 3.1333531333531335,
      "grad_norm": 0.9511438608169556,
      "learning_rate": 1.11998811998812e-05,
      "loss": 0.0734,
      "step": 10550
    },
    {
      "epoch": 3.1363231363231363,
      "grad_norm": 0.6538392901420593,
      "learning_rate": 1.1182061182061181e-05,
      "loss": 0.0429,
      "step": 10560
    },
    {
      "epoch": 3.139293139293139,
      "grad_norm": 0.983573317527771,
      "learning_rate": 1.1164241164241165e-05,
      "loss": 0.049,
      "step": 10570
    },
    {
      "epoch": 3.1422631422631424,
      "grad_norm": 0.8676197528839111,
      "learning_rate": 1.1146421146421147e-05,
      "loss": 0.0639,
      "step": 10580
    },
    {
      "epoch": 3.1452331452331452,
      "grad_norm": 0.4240647256374359,
      "learning_rate": 1.1128601128601128e-05,
      "loss": 0.063,
      "step": 10590
    },
    {
      "epoch": 3.148203148203148,
      "grad_norm": 0.9053698182106018,
      "learning_rate": 1.111078111078111e-05,
      "loss": 0.0409,
      "step": 10600
    },
    {
      "epoch": 3.1511731511731513,
      "grad_norm": 0.6372582316398621,
      "learning_rate": 1.1092961092961094e-05,
      "loss": 0.0648,
      "step": 10610
    },
    {
      "epoch": 3.154143154143154,
      "grad_norm": 0.7005014419555664,
      "learning_rate": 1.1075141075141076e-05,
      "loss": 0.0563,
      "step": 10620
    },
    {
      "epoch": 3.157113157113157,
      "grad_norm": 0.8473065495491028,
      "learning_rate": 1.1057321057321056e-05,
      "loss": 0.0729,
      "step": 10630
    },
    {
      "epoch": 3.1600831600831603,
      "grad_norm": 0.3668000102043152,
      "learning_rate": 1.103950103950104e-05,
      "loss": 0.0609,
      "step": 10640
    },
    {
      "epoch": 3.163053163053163,
      "grad_norm": 0.3873116374015808,
      "learning_rate": 1.1021681021681022e-05,
      "loss": 0.0427,
      "step": 10650
    },
    {
      "epoch": 3.166023166023166,
      "grad_norm": 0.6118089556694031,
      "learning_rate": 1.1003861003861003e-05,
      "loss": 0.0426,
      "step": 10660
    },
    {
      "epoch": 3.168993168993169,
      "grad_norm": 0.8138056993484497,
      "learning_rate": 1.0986040986040985e-05,
      "loss": 0.0626,
      "step": 10670
    },
    {
      "epoch": 3.171963171963172,
      "grad_norm": 0.45243504643440247,
      "learning_rate": 1.0968220968220969e-05,
      "loss": 0.0643,
      "step": 10680
    },
    {
      "epoch": 3.174933174933175,
      "grad_norm": 0.8468220233917236,
      "learning_rate": 1.0950400950400951e-05,
      "loss": 0.0628,
      "step": 10690
    },
    {
      "epoch": 3.177903177903178,
      "grad_norm": 0.3471866846084595,
      "learning_rate": 1.0932580932580932e-05,
      "loss": 0.056,
      "step": 10700
    },
    {
      "epoch": 3.180873180873181,
      "grad_norm": 0.6115337014198303,
      "learning_rate": 1.0914760914760916e-05,
      "loss": 0.0571,
      "step": 10710
    },
    {
      "epoch": 3.1838431838431838,
      "grad_norm": 0.7158623933792114,
      "learning_rate": 1.0896940896940898e-05,
      "loss": 0.0541,
      "step": 10720
    },
    {
      "epoch": 3.186813186813187,
      "grad_norm": 0.49794185161590576,
      "learning_rate": 1.087912087912088e-05,
      "loss": 0.0491,
      "step": 10730
    },
    {
      "epoch": 3.18978318978319,
      "grad_norm": 0.786274254322052,
      "learning_rate": 1.086130086130086e-05,
      "loss": 0.0675,
      "step": 10740
    },
    {
      "epoch": 3.1927531927531927,
      "grad_norm": 1.0238100290298462,
      "learning_rate": 1.0843480843480844e-05,
      "loss": 0.0413,
      "step": 10750
    },
    {
      "epoch": 3.1957231957231955,
      "grad_norm": 1.032472848892212,
      "learning_rate": 1.0825660825660826e-05,
      "loss": 0.0607,
      "step": 10760
    },
    {
      "epoch": 3.198693198693199,
      "grad_norm": 0.7296070456504822,
      "learning_rate": 1.0807840807840807e-05,
      "loss": 0.0532,
      "step": 10770
    },
    {
      "epoch": 3.2016632016632016,
      "grad_norm": 1.2393643856048584,
      "learning_rate": 1.079002079002079e-05,
      "loss": 0.0602,
      "step": 10780
    },
    {
      "epoch": 3.2046332046332044,
      "grad_norm": 0.6754175424575806,
      "learning_rate": 1.0772200772200773e-05,
      "loss": 0.0592,
      "step": 10790
    },
    {
      "epoch": 3.2076032076032077,
      "grad_norm": 0.5455211997032166,
      "learning_rate": 1.0754380754380755e-05,
      "loss": 0.0622,
      "step": 10800
    },
    {
      "epoch": 3.2105732105732105,
      "grad_norm": 0.449032187461853,
      "learning_rate": 1.0736560736560736e-05,
      "loss": 0.0497,
      "step": 10810
    },
    {
      "epoch": 3.2135432135432134,
      "grad_norm": 0.3489013612270355,
      "learning_rate": 1.071874071874072e-05,
      "loss": 0.0669,
      "step": 10820
    },
    {
      "epoch": 3.2165132165132166,
      "grad_norm": 1.101258397102356,
      "learning_rate": 1.0700920700920702e-05,
      "loss": 0.0718,
      "step": 10830
    },
    {
      "epoch": 3.2194832194832195,
      "grad_norm": 0.637738049030304,
      "learning_rate": 1.0683100683100682e-05,
      "loss": 0.0455,
      "step": 10840
    },
    {
      "epoch": 3.2224532224532223,
      "grad_norm": 0.8752096891403198,
      "learning_rate": 1.0665280665280666e-05,
      "loss": 0.0671,
      "step": 10850
    },
    {
      "epoch": 3.2254232254232256,
      "grad_norm": 0.5893465876579285,
      "learning_rate": 1.0647460647460648e-05,
      "loss": 0.0627,
      "step": 10860
    },
    {
      "epoch": 3.2283932283932284,
      "grad_norm": 1.188922643661499,
      "learning_rate": 1.062964062964063e-05,
      "loss": 0.066,
      "step": 10870
    },
    {
      "epoch": 3.2313632313632312,
      "grad_norm": 0.6569589376449585,
      "learning_rate": 1.0611820611820612e-05,
      "loss": 0.0481,
      "step": 10880
    },
    {
      "epoch": 3.2343332343332345,
      "grad_norm": 0.7973625063896179,
      "learning_rate": 1.0594000594000595e-05,
      "loss": 0.0532,
      "step": 10890
    },
    {
      "epoch": 3.2373032373032373,
      "grad_norm": 1.0945541858673096,
      "learning_rate": 1.0576180576180577e-05,
      "loss": 0.0557,
      "step": 10900
    },
    {
      "epoch": 3.24027324027324,
      "grad_norm": 0.6560423374176025,
      "learning_rate": 1.0558360558360557e-05,
      "loss": 0.048,
      "step": 10910
    },
    {
      "epoch": 3.2432432432432434,
      "grad_norm": 0.7343136072158813,
      "learning_rate": 1.0540540540540541e-05,
      "loss": 0.0588,
      "step": 10920
    },
    {
      "epoch": 3.2462132462132463,
      "grad_norm": 0.6441863775253296,
      "learning_rate": 1.0522720522720523e-05,
      "loss": 0.0501,
      "step": 10930
    },
    {
      "epoch": 3.249183249183249,
      "grad_norm": 0.5556747913360596,
      "learning_rate": 1.0504900504900505e-05,
      "loss": 0.0521,
      "step": 10940
    },
    {
      "epoch": 3.252153252153252,
      "grad_norm": 0.636785089969635,
      "learning_rate": 1.0487080487080488e-05,
      "loss": 0.0593,
      "step": 10950
    },
    {
      "epoch": 3.255123255123255,
      "grad_norm": 1.0058456659317017,
      "learning_rate": 1.046926046926047e-05,
      "loss": 0.0683,
      "step": 10960
    },
    {
      "epoch": 3.258093258093258,
      "grad_norm": 0.6078860759735107,
      "learning_rate": 1.0451440451440452e-05,
      "loss": 0.0525,
      "step": 10970
    },
    {
      "epoch": 3.261063261063261,
      "grad_norm": 0.675046980381012,
      "learning_rate": 1.0433620433620434e-05,
      "loss": 0.0538,
      "step": 10980
    },
    {
      "epoch": 3.264033264033264,
      "grad_norm": 0.4468725323677063,
      "learning_rate": 1.0415800415800416e-05,
      "loss": 0.0669,
      "step": 10990
    },
    {
      "epoch": 3.267003267003267,
      "grad_norm": 0.7147549986839294,
      "learning_rate": 1.0397980397980398e-05,
      "loss": 0.0454,
      "step": 11000
    },
    {
      "epoch": 3.2699732699732698,
      "grad_norm": 0.8929649591445923,
      "learning_rate": 1.038016038016038e-05,
      "loss": 0.0399,
      "step": 11010
    },
    {
      "epoch": 3.272943272943273,
      "grad_norm": 0.3554942011833191,
      "learning_rate": 1.0362340362340363e-05,
      "loss": 0.0501,
      "step": 11020
    },
    {
      "epoch": 3.275913275913276,
      "grad_norm": 0.4639175236225128,
      "learning_rate": 1.0344520344520345e-05,
      "loss": 0.0614,
      "step": 11030
    },
    {
      "epoch": 3.2788832788832787,
      "grad_norm": 0.8245081901550293,
      "learning_rate": 1.0326700326700327e-05,
      "loss": 0.0581,
      "step": 11040
    },
    {
      "epoch": 3.281853281853282,
      "grad_norm": 0.9380563497543335,
      "learning_rate": 1.030888030888031e-05,
      "loss": 0.0677,
      "step": 11050
    },
    {
      "epoch": 3.284823284823285,
      "grad_norm": 1.2146899700164795,
      "learning_rate": 1.0291060291060291e-05,
      "loss": 0.0511,
      "step": 11060
    },
    {
      "epoch": 3.2877932877932876,
      "grad_norm": 0.7667972445487976,
      "learning_rate": 1.0273240273240274e-05,
      "loss": 0.0475,
      "step": 11070
    },
    {
      "epoch": 3.290763290763291,
      "grad_norm": 0.6132957339286804,
      "learning_rate": 1.0255420255420256e-05,
      "loss": 0.0539,
      "step": 11080
    },
    {
      "epoch": 3.2937332937332937,
      "grad_norm": 1.2011501789093018,
      "learning_rate": 1.0237600237600238e-05,
      "loss": 0.0475,
      "step": 11090
    },
    {
      "epoch": 3.2967032967032965,
      "grad_norm": 0.6205746531486511,
      "learning_rate": 1.021978021978022e-05,
      "loss": 0.0593,
      "step": 11100
    },
    {
      "epoch": 3.2996732996733,
      "grad_norm": 1.1285995244979858,
      "learning_rate": 1.0201960201960202e-05,
      "loss": 0.0693,
      "step": 11110
    },
    {
      "epoch": 3.3026433026433026,
      "grad_norm": 1.0713976621627808,
      "learning_rate": 1.0184140184140184e-05,
      "loss": 0.062,
      "step": 11120
    },
    {
      "epoch": 3.3056133056133055,
      "grad_norm": 0.6657689213752747,
      "learning_rate": 1.0166320166320167e-05,
      "loss": 0.0544,
      "step": 11130
    },
    {
      "epoch": 3.3085833085833087,
      "grad_norm": 0.9671308994293213,
      "learning_rate": 1.0148500148500149e-05,
      "loss": 0.0586,
      "step": 11140
    },
    {
      "epoch": 3.3115533115533116,
      "grad_norm": 0.6092202663421631,
      "learning_rate": 1.0130680130680131e-05,
      "loss": 0.0646,
      "step": 11150
    },
    {
      "epoch": 3.3145233145233144,
      "grad_norm": 0.8114497065544128,
      "learning_rate": 1.0112860112860113e-05,
      "loss": 0.0723,
      "step": 11160
    },
    {
      "epoch": 3.3174933174933177,
      "grad_norm": 1.0398253202438354,
      "learning_rate": 1.0095040095040095e-05,
      "loss": 0.0516,
      "step": 11170
    },
    {
      "epoch": 3.3204633204633205,
      "grad_norm": 0.773544430732727,
      "learning_rate": 1.0077220077220078e-05,
      "loss": 0.0553,
      "step": 11180
    },
    {
      "epoch": 3.3234333234333233,
      "grad_norm": 1.1099355220794678,
      "learning_rate": 1.005940005940006e-05,
      "loss": 0.0662,
      "step": 11190
    },
    {
      "epoch": 3.3264033264033266,
      "grad_norm": 0.7854142189025879,
      "learning_rate": 1.0041580041580042e-05,
      "loss": 0.0578,
      "step": 11200
    },
    {
      "epoch": 3.3293733293733294,
      "grad_norm": 0.47787368297576904,
      "learning_rate": 1.0023760023760024e-05,
      "loss": 0.0634,
      "step": 11210
    },
    {
      "epoch": 3.3323433323433322,
      "grad_norm": 0.5865157246589661,
      "learning_rate": 1.0005940005940006e-05,
      "loss": 0.0425,
      "step": 11220
    },
    {
      "epoch": 3.3353133353133355,
      "grad_norm": 0.7458857297897339,
      "learning_rate": 9.988119988119988e-06,
      "loss": 0.0659,
      "step": 11230
    },
    {
      "epoch": 3.3382833382833383,
      "grad_norm": 0.73867267370224,
      "learning_rate": 9.97029997029997e-06,
      "loss": 0.0615,
      "step": 11240
    },
    {
      "epoch": 3.341253341253341,
      "grad_norm": 0.9785704016685486,
      "learning_rate": 9.952479952479953e-06,
      "loss": 0.0452,
      "step": 11250
    },
    {
      "epoch": 3.3442233442233444,
      "grad_norm": 0.7227851152420044,
      "learning_rate": 9.934659934659935e-06,
      "loss": 0.0572,
      "step": 11260
    },
    {
      "epoch": 3.3471933471933473,
      "grad_norm": 0.6161572933197021,
      "learning_rate": 9.916839916839917e-06,
      "loss": 0.0457,
      "step": 11270
    },
    {
      "epoch": 3.35016335016335,
      "grad_norm": 0.6703232526779175,
      "learning_rate": 9.8990198990199e-06,
      "loss": 0.0682,
      "step": 11280
    },
    {
      "epoch": 3.3531333531333534,
      "grad_norm": 0.6447526216506958,
      "learning_rate": 9.881199881199881e-06,
      "loss": 0.0596,
      "step": 11290
    },
    {
      "epoch": 3.356103356103356,
      "grad_norm": 0.46125978231430054,
      "learning_rate": 9.863379863379865e-06,
      "loss": 0.0659,
      "step": 11300
    },
    {
      "epoch": 3.359073359073359,
      "grad_norm": 1.2563014030456543,
      "learning_rate": 9.845559845559846e-06,
      "loss": 0.057,
      "step": 11310
    },
    {
      "epoch": 3.362043362043362,
      "grad_norm": 0.8137726187705994,
      "learning_rate": 9.827739827739828e-06,
      "loss": 0.0461,
      "step": 11320
    },
    {
      "epoch": 3.365013365013365,
      "grad_norm": 0.5662651658058167,
      "learning_rate": 9.80991980991981e-06,
      "loss": 0.0484,
      "step": 11330
    },
    {
      "epoch": 3.367983367983368,
      "grad_norm": 0.5154465436935425,
      "learning_rate": 9.792099792099792e-06,
      "loss": 0.0579,
      "step": 11340
    },
    {
      "epoch": 3.3709533709533708,
      "grad_norm": 0.7611321806907654,
      "learning_rate": 9.774279774279774e-06,
      "loss": 0.0613,
      "step": 11350
    },
    {
      "epoch": 3.373923373923374,
      "grad_norm": 0.8335089087486267,
      "learning_rate": 9.756459756459757e-06,
      "loss": 0.0493,
      "step": 11360
    },
    {
      "epoch": 3.376893376893377,
      "grad_norm": 1.004220724105835,
      "learning_rate": 9.73863973863974e-06,
      "loss": 0.0511,
      "step": 11370
    },
    {
      "epoch": 3.3798633798633797,
      "grad_norm": 0.5322721600532532,
      "learning_rate": 9.720819720819721e-06,
      "loss": 0.0551,
      "step": 11380
    },
    {
      "epoch": 3.382833382833383,
      "grad_norm": 0.7684707641601562,
      "learning_rate": 9.702999702999703e-06,
      "loss": 0.0658,
      "step": 11390
    },
    {
      "epoch": 3.385803385803386,
      "grad_norm": 0.7341310381889343,
      "learning_rate": 9.685179685179685e-06,
      "loss": 0.0626,
      "step": 11400
    },
    {
      "epoch": 3.3887733887733886,
      "grad_norm": 0.7289576530456543,
      "learning_rate": 9.667359667359667e-06,
      "loss": 0.0508,
      "step": 11410
    },
    {
      "epoch": 3.391743391743392,
      "grad_norm": 0.9594446420669556,
      "learning_rate": 9.64953964953965e-06,
      "loss": 0.0559,
      "step": 11420
    },
    {
      "epoch": 3.3947133947133947,
      "grad_norm": 0.7913696765899658,
      "learning_rate": 9.631719631719632e-06,
      "loss": 0.0648,
      "step": 11430
    },
    {
      "epoch": 3.3976833976833976,
      "grad_norm": 0.8502475619316101,
      "learning_rate": 9.613899613899616e-06,
      "loss": 0.0531,
      "step": 11440
    },
    {
      "epoch": 3.400653400653401,
      "grad_norm": 0.8771364092826843,
      "learning_rate": 9.596079596079596e-06,
      "loss": 0.0577,
      "step": 11450
    },
    {
      "epoch": 3.4036234036234037,
      "grad_norm": 0.6660274267196655,
      "learning_rate": 9.578259578259578e-06,
      "loss": 0.0496,
      "step": 11460
    },
    {
      "epoch": 3.4065934065934065,
      "grad_norm": 0.7120699286460876,
      "learning_rate": 9.56043956043956e-06,
      "loss": 0.0533,
      "step": 11470
    },
    {
      "epoch": 3.4095634095634098,
      "grad_norm": 0.8846875429153442,
      "learning_rate": 9.542619542619543e-06,
      "loss": 0.0557,
      "step": 11480
    },
    {
      "epoch": 3.4125334125334126,
      "grad_norm": 0.5555750131607056,
      "learning_rate": 9.524799524799525e-06,
      "loss": 0.049,
      "step": 11490
    },
    {
      "epoch": 3.4155034155034154,
      "grad_norm": 0.6334190964698792,
      "learning_rate": 9.506979506979507e-06,
      "loss": 0.0679,
      "step": 11500
    },
    {
      "epoch": 3.4184734184734182,
      "grad_norm": 0.462439626455307,
      "learning_rate": 9.48915948915949e-06,
      "loss": 0.0572,
      "step": 11510
    },
    {
      "epoch": 3.4214434214434215,
      "grad_norm": 0.6081441640853882,
      "learning_rate": 9.471339471339471e-06,
      "loss": 0.0487,
      "step": 11520
    },
    {
      "epoch": 3.4244134244134243,
      "grad_norm": 0.7960311770439148,
      "learning_rate": 9.453519453519453e-06,
      "loss": 0.0628,
      "step": 11530
    },
    {
      "epoch": 3.427383427383427,
      "grad_norm": 0.5583157539367676,
      "learning_rate": 9.435699435699436e-06,
      "loss": 0.0666,
      "step": 11540
    },
    {
      "epoch": 3.4303534303534304,
      "grad_norm": 0.8321641683578491,
      "learning_rate": 9.417879417879418e-06,
      "loss": 0.0527,
      "step": 11550
    },
    {
      "epoch": 3.4333234333234333,
      "grad_norm": 0.6398223042488098,
      "learning_rate": 9.4000594000594e-06,
      "loss": 0.0421,
      "step": 11560
    },
    {
      "epoch": 3.436293436293436,
      "grad_norm": 0.8610657453536987,
      "learning_rate": 9.382239382239382e-06,
      "loss": 0.0512,
      "step": 11570
    },
    {
      "epoch": 3.4392634392634394,
      "grad_norm": 0.8731980919837952,
      "learning_rate": 9.364419364419366e-06,
      "loss": 0.0529,
      "step": 11580
    },
    {
      "epoch": 3.442233442233442,
      "grad_norm": 0.6579751372337341,
      "learning_rate": 9.346599346599347e-06,
      "loss": 0.0437,
      "step": 11590
    },
    {
      "epoch": 3.445203445203445,
      "grad_norm": 1.1427984237670898,
      "learning_rate": 9.328779328779329e-06,
      "loss": 0.0565,
      "step": 11600
    },
    {
      "epoch": 3.4481734481734483,
      "grad_norm": 0.7703096866607666,
      "learning_rate": 9.31095931095931e-06,
      "loss": 0.0473,
      "step": 11610
    },
    {
      "epoch": 3.451143451143451,
      "grad_norm": 1.2040685415267944,
      "learning_rate": 9.293139293139295e-06,
      "loss": 0.0579,
      "step": 11620
    },
    {
      "epoch": 3.454113454113454,
      "grad_norm": 0.7224343419075012,
      "learning_rate": 9.275319275319275e-06,
      "loss": 0.0523,
      "step": 11630
    },
    {
      "epoch": 3.457083457083457,
      "grad_norm": 1.0151997804641724,
      "learning_rate": 9.257499257499257e-06,
      "loss": 0.0559,
      "step": 11640
    },
    {
      "epoch": 3.46005346005346,
      "grad_norm": 0.5172938704490662,
      "learning_rate": 9.239679239679241e-06,
      "loss": 0.0518,
      "step": 11650
    },
    {
      "epoch": 3.463023463023463,
      "grad_norm": 0.864036500453949,
      "learning_rate": 9.221859221859222e-06,
      "loss": 0.0407,
      "step": 11660
    },
    {
      "epoch": 3.465993465993466,
      "grad_norm": 0.603539228439331,
      "learning_rate": 9.204039204039204e-06,
      "loss": 0.0589,
      "step": 11670
    },
    {
      "epoch": 3.468963468963469,
      "grad_norm": 0.6137470006942749,
      "learning_rate": 9.186219186219186e-06,
      "loss": 0.0657,
      "step": 11680
    },
    {
      "epoch": 3.471933471933472,
      "grad_norm": 0.8344042897224426,
      "learning_rate": 9.16839916839917e-06,
      "loss": 0.0562,
      "step": 11690
    },
    {
      "epoch": 3.474903474903475,
      "grad_norm": 0.9684216976165771,
      "learning_rate": 9.15057915057915e-06,
      "loss": 0.0482,
      "step": 11700
    },
    {
      "epoch": 3.477873477873478,
      "grad_norm": 0.5925269722938538,
      "learning_rate": 9.132759132759133e-06,
      "loss": 0.0563,
      "step": 11710
    },
    {
      "epoch": 3.4808434808434807,
      "grad_norm": 1.1107386350631714,
      "learning_rate": 9.114939114939116e-06,
      "loss": 0.0585,
      "step": 11720
    },
    {
      "epoch": 3.483813483813484,
      "grad_norm": 0.7802149653434753,
      "learning_rate": 9.097119097119097e-06,
      "loss": 0.0639,
      "step": 11730
    },
    {
      "epoch": 3.486783486783487,
      "grad_norm": 0.9649807214736938,
      "learning_rate": 9.079299079299079e-06,
      "loss": 0.0687,
      "step": 11740
    },
    {
      "epoch": 3.4897534897534896,
      "grad_norm": 0.5262308716773987,
      "learning_rate": 9.061479061479061e-06,
      "loss": 0.0432,
      "step": 11750
    },
    {
      "epoch": 3.492723492723493,
      "grad_norm": 0.8198150396347046,
      "learning_rate": 9.043659043659045e-06,
      "loss": 0.0583,
      "step": 11760
    },
    {
      "epoch": 3.4956934956934957,
      "grad_norm": 1.3013529777526855,
      "learning_rate": 9.025839025839026e-06,
      "loss": 0.0562,
      "step": 11770
    },
    {
      "epoch": 3.4986634986634986,
      "grad_norm": 0.875752329826355,
      "learning_rate": 9.008019008019008e-06,
      "loss": 0.0499,
      "step": 11780
    },
    {
      "epoch": 3.501633501633502,
      "grad_norm": 0.8648601174354553,
      "learning_rate": 8.990198990198992e-06,
      "loss": 0.0502,
      "step": 11790
    },
    {
      "epoch": 3.5046035046035047,
      "grad_norm": 0.9301609992980957,
      "learning_rate": 8.972378972378972e-06,
      "loss": 0.0478,
      "step": 11800
    },
    {
      "epoch": 3.5075735075735075,
      "grad_norm": 0.9498510360717773,
      "learning_rate": 8.954558954558954e-06,
      "loss": 0.0503,
      "step": 11810
    },
    {
      "epoch": 3.5105435105435108,
      "grad_norm": 0.5462694764137268,
      "learning_rate": 8.936738936738936e-06,
      "loss": 0.0575,
      "step": 11820
    },
    {
      "epoch": 3.5135135135135136,
      "grad_norm": 0.8553217649459839,
      "learning_rate": 8.91891891891892e-06,
      "loss": 0.0655,
      "step": 11830
    },
    {
      "epoch": 3.5164835164835164,
      "grad_norm": 0.9228159189224243,
      "learning_rate": 8.9010989010989e-06,
      "loss": 0.0611,
      "step": 11840
    },
    {
      "epoch": 3.5194535194535197,
      "grad_norm": 0.6614230275154114,
      "learning_rate": 8.883278883278883e-06,
      "loss": 0.0465,
      "step": 11850
    },
    {
      "epoch": 3.5224235224235225,
      "grad_norm": 0.9171364307403564,
      "learning_rate": 8.865458865458867e-06,
      "loss": 0.0524,
      "step": 11860
    },
    {
      "epoch": 3.5253935253935254,
      "grad_norm": 0.4737289249897003,
      "learning_rate": 8.847638847638847e-06,
      "loss": 0.0688,
      "step": 11870
    },
    {
      "epoch": 3.5283635283635286,
      "grad_norm": 0.79871666431427,
      "learning_rate": 8.82981882981883e-06,
      "loss": 0.0556,
      "step": 11880
    },
    {
      "epoch": 3.5313335313335315,
      "grad_norm": 0.9557964205741882,
      "learning_rate": 8.811998811998812e-06,
      "loss": 0.0592,
      "step": 11890
    },
    {
      "epoch": 3.5343035343035343,
      "grad_norm": 0.9244213104248047,
      "learning_rate": 8.794178794178795e-06,
      "loss": 0.0513,
      "step": 11900
    },
    {
      "epoch": 3.5372735372735375,
      "grad_norm": 0.5867542624473572,
      "learning_rate": 8.776358776358776e-06,
      "loss": 0.0504,
      "step": 11910
    },
    {
      "epoch": 3.5402435402435404,
      "grad_norm": 0.47742247581481934,
      "learning_rate": 8.758538758538758e-06,
      "loss": 0.0529,
      "step": 11920
    },
    {
      "epoch": 3.543213543213543,
      "grad_norm": 1.0986418724060059,
      "learning_rate": 8.740718740718742e-06,
      "loss": 0.0546,
      "step": 11930
    },
    {
      "epoch": 3.546183546183546,
      "grad_norm": 0.7719411253929138,
      "learning_rate": 8.722898722898724e-06,
      "loss": 0.0564,
      "step": 11940
    },
    {
      "epoch": 3.5491535491535493,
      "grad_norm": 0.9818161129951477,
      "learning_rate": 8.705078705078705e-06,
      "loss": 0.0691,
      "step": 11950
    },
    {
      "epoch": 3.552123552123552,
      "grad_norm": 0.6486049294471741,
      "learning_rate": 8.687258687258687e-06,
      "loss": 0.0662,
      "step": 11960
    },
    {
      "epoch": 3.555093555093555,
      "grad_norm": 0.6224350929260254,
      "learning_rate": 8.66943866943867e-06,
      "loss": 0.0565,
      "step": 11970
    },
    {
      "epoch": 3.5580635580635582,
      "grad_norm": 1.0095922946929932,
      "learning_rate": 8.651618651618651e-06,
      "loss": 0.0609,
      "step": 11980
    },
    {
      "epoch": 3.561033561033561,
      "grad_norm": 0.5099046230316162,
      "learning_rate": 8.633798633798633e-06,
      "loss": 0.0578,
      "step": 11990
    },
    {
      "epoch": 3.564003564003564,
      "grad_norm": 0.9583487510681152,
      "learning_rate": 8.615978615978617e-06,
      "loss": 0.0751,
      "step": 12000
    },
    {
      "epoch": 3.5669735669735667,
      "grad_norm": 0.8922184109687805,
      "learning_rate": 8.5981585981586e-06,
      "loss": 0.0469,
      "step": 12010
    },
    {
      "epoch": 3.56994356994357,
      "grad_norm": 0.6828520894050598,
      "learning_rate": 8.58033858033858e-06,
      "loss": 0.0482,
      "step": 12020
    },
    {
      "epoch": 3.572913572913573,
      "grad_norm": 1.372278094291687,
      "learning_rate": 8.562518562518562e-06,
      "loss": 0.0606,
      "step": 12030
    },
    {
      "epoch": 3.5758835758835756,
      "grad_norm": 0.5054883360862732,
      "learning_rate": 8.544698544698546e-06,
      "loss": 0.0465,
      "step": 12040
    },
    {
      "epoch": 3.578853578853579,
      "grad_norm": 0.564597487449646,
      "learning_rate": 8.526878526878526e-06,
      "loss": 0.0516,
      "step": 12050
    },
    {
      "epoch": 3.5818235818235817,
      "grad_norm": 0.6574399471282959,
      "learning_rate": 8.509058509058509e-06,
      "loss": 0.0511,
      "step": 12060
    },
    {
      "epoch": 3.5847935847935846,
      "grad_norm": 0.4628748893737793,
      "learning_rate": 8.491238491238492e-06,
      "loss": 0.0494,
      "step": 12070
    },
    {
      "epoch": 3.587763587763588,
      "grad_norm": 0.957274854183197,
      "learning_rate": 8.473418473418475e-06,
      "loss": 0.0558,
      "step": 12080
    },
    {
      "epoch": 3.5907335907335907,
      "grad_norm": 0.9441186785697937,
      "learning_rate": 8.455598455598455e-06,
      "loss": 0.0572,
      "step": 12090
    },
    {
      "epoch": 3.5937035937035935,
      "grad_norm": 0.7726499438285828,
      "learning_rate": 8.437778437778437e-06,
      "loss": 0.058,
      "step": 12100
    },
    {
      "epoch": 3.5966735966735968,
      "grad_norm": 0.7733856439590454,
      "learning_rate": 8.419958419958421e-06,
      "loss": 0.0383,
      "step": 12110
    },
    {
      "epoch": 3.5996435996435996,
      "grad_norm": 0.8538611531257629,
      "learning_rate": 8.402138402138402e-06,
      "loss": 0.0647,
      "step": 12120
    },
    {
      "epoch": 3.6026136026136024,
      "grad_norm": 0.826570451259613,
      "learning_rate": 8.384318384318384e-06,
      "loss": 0.0521,
      "step": 12130
    },
    {
      "epoch": 3.6055836055836057,
      "grad_norm": 1.6991403102874756,
      "learning_rate": 8.366498366498368e-06,
      "loss": 0.058,
      "step": 12140
    },
    {
      "epoch": 3.6085536085536085,
      "grad_norm": 0.7238519191741943,
      "learning_rate": 8.34867834867835e-06,
      "loss": 0.0655,
      "step": 12150
    },
    {
      "epoch": 3.6115236115236113,
      "grad_norm": 0.7092881798744202,
      "learning_rate": 8.33085833085833e-06,
      "loss": 0.0623,
      "step": 12160
    },
    {
      "epoch": 3.6144936144936146,
      "grad_norm": 0.5333400964736938,
      "learning_rate": 8.313038313038312e-06,
      "loss": 0.0503,
      "step": 12170
    },
    {
      "epoch": 3.6174636174636174,
      "grad_norm": 0.9490695595741272,
      "learning_rate": 8.295218295218296e-06,
      "loss": 0.0495,
      "step": 12180
    },
    {
      "epoch": 3.6204336204336203,
      "grad_norm": 0.5925624370574951,
      "learning_rate": 8.277398277398278e-06,
      "loss": 0.0575,
      "step": 12190
    },
    {
      "epoch": 3.6234036234036235,
      "grad_norm": 0.8142616748809814,
      "learning_rate": 8.259578259578259e-06,
      "loss": 0.0561,
      "step": 12200
    },
    {
      "epoch": 3.6263736263736264,
      "grad_norm": 0.8163090348243713,
      "learning_rate": 8.241758241758243e-06,
      "loss": 0.0576,
      "step": 12210
    },
    {
      "epoch": 3.629343629343629,
      "grad_norm": 0.8602144122123718,
      "learning_rate": 8.223938223938225e-06,
      "loss": 0.0543,
      "step": 12220
    },
    {
      "epoch": 3.6323136323136325,
      "grad_norm": 0.5057087540626526,
      "learning_rate": 8.206118206118205e-06,
      "loss": 0.0658,
      "step": 12230
    },
    {
      "epoch": 3.6352836352836353,
      "grad_norm": 1.0046048164367676,
      "learning_rate": 8.188298188298188e-06,
      "loss": 0.0546,
      "step": 12240
    },
    {
      "epoch": 3.638253638253638,
      "grad_norm": 0.5388000011444092,
      "learning_rate": 8.170478170478171e-06,
      "loss": 0.0535,
      "step": 12250
    },
    {
      "epoch": 3.6412236412236414,
      "grad_norm": 0.31448933482170105,
      "learning_rate": 8.152658152658154e-06,
      "loss": 0.0561,
      "step": 12260
    },
    {
      "epoch": 3.644193644193644,
      "grad_norm": 0.7060285210609436,
      "learning_rate": 8.134838134838134e-06,
      "loss": 0.0628,
      "step": 12270
    },
    {
      "epoch": 3.647163647163647,
      "grad_norm": 0.44246843457221985,
      "learning_rate": 8.117018117018118e-06,
      "loss": 0.0549,
      "step": 12280
    },
    {
      "epoch": 3.6501336501336503,
      "grad_norm": 0.8844314813613892,
      "learning_rate": 8.0991980991981e-06,
      "loss": 0.0525,
      "step": 12290
    },
    {
      "epoch": 3.653103653103653,
      "grad_norm": 1.1188846826553345,
      "learning_rate": 8.08137808137808e-06,
      "loss": 0.0563,
      "step": 12300
    },
    {
      "epoch": 3.656073656073656,
      "grad_norm": 0.7898098230361938,
      "learning_rate": 8.063558063558063e-06,
      "loss": 0.0571,
      "step": 12310
    },
    {
      "epoch": 3.6590436590436592,
      "grad_norm": 0.5534541606903076,
      "learning_rate": 8.045738045738047e-06,
      "loss": 0.0648,
      "step": 12320
    },
    {
      "epoch": 3.662013662013662,
      "grad_norm": 0.5332062840461731,
      "learning_rate": 8.027918027918029e-06,
      "loss": 0.0674,
      "step": 12330
    },
    {
      "epoch": 3.664983664983665,
      "grad_norm": 1.169073224067688,
      "learning_rate": 8.01009801009801e-06,
      "loss": 0.0532,
      "step": 12340
    },
    {
      "epoch": 3.667953667953668,
      "grad_norm": 0.5271252393722534,
      "learning_rate": 7.992277992277993e-06,
      "loss": 0.0593,
      "step": 12350
    },
    {
      "epoch": 3.670923670923671,
      "grad_norm": 0.8392048478126526,
      "learning_rate": 7.974457974457975e-06,
      "loss": 0.0468,
      "step": 12360
    },
    {
      "epoch": 3.673893673893674,
      "grad_norm": 0.7559081315994263,
      "learning_rate": 7.956637956637956e-06,
      "loss": 0.0476,
      "step": 12370
    },
    {
      "epoch": 3.676863676863677,
      "grad_norm": 0.6734028458595276,
      "learning_rate": 7.938817938817938e-06,
      "loss": 0.0532,
      "step": 12380
    },
    {
      "epoch": 3.67983367983368,
      "grad_norm": 0.7153018712997437,
      "learning_rate": 7.920997920997922e-06,
      "loss": 0.0597,
      "step": 12390
    },
    {
      "epoch": 3.6828036828036828,
      "grad_norm": 0.7017358541488647,
      "learning_rate": 7.903177903177904e-06,
      "loss": 0.0642,
      "step": 12400
    },
    {
      "epoch": 3.685773685773686,
      "grad_norm": 0.8669632077217102,
      "learning_rate": 7.885357885357884e-06,
      "loss": 0.0521,
      "step": 12410
    },
    {
      "epoch": 3.688743688743689,
      "grad_norm": 0.5131626725196838,
      "learning_rate": 7.867537867537868e-06,
      "loss": 0.0578,
      "step": 12420
    },
    {
      "epoch": 3.6917136917136917,
      "grad_norm": 1.0312336683273315,
      "learning_rate": 7.84971784971785e-06,
      "loss": 0.0641,
      "step": 12430
    },
    {
      "epoch": 3.694683694683695,
      "grad_norm": 0.620993435382843,
      "learning_rate": 7.831897831897831e-06,
      "loss": 0.0577,
      "step": 12440
    },
    {
      "epoch": 3.697653697653698,
      "grad_norm": 0.9431599974632263,
      "learning_rate": 7.814077814077813e-06,
      "loss": 0.054,
      "step": 12450
    },
    {
      "epoch": 3.7006237006237006,
      "grad_norm": 0.5963833928108215,
      "learning_rate": 7.796257796257797e-06,
      "loss": 0.0617,
      "step": 12460
    },
    {
      "epoch": 3.7035937035937034,
      "grad_norm": 0.8309038877487183,
      "learning_rate": 7.77843777843778e-06,
      "loss": 0.064,
      "step": 12470
    },
    {
      "epoch": 3.7065637065637067,
      "grad_norm": 0.38153156638145447,
      "learning_rate": 7.76061776061776e-06,
      "loss": 0.0574,
      "step": 12480
    },
    {
      "epoch": 3.7095337095337095,
      "grad_norm": 0.6947796940803528,
      "learning_rate": 7.742797742797744e-06,
      "loss": 0.0582,
      "step": 12490
    },
    {
      "epoch": 3.7125037125037124,
      "grad_norm": 0.29392537474632263,
      "learning_rate": 7.724977724977726e-06,
      "loss": 0.0599,
      "step": 12500
    },
    {
      "epoch": 3.7154737154737156,
      "grad_norm": 0.7719232439994812,
      "learning_rate": 7.707157707157708e-06,
      "loss": 0.0454,
      "step": 12510
    },
    {
      "epoch": 3.7184437184437185,
      "grad_norm": 0.8208956718444824,
      "learning_rate": 7.689337689337688e-06,
      "loss": 0.0627,
      "step": 12520
    },
    {
      "epoch": 3.7214137214137213,
      "grad_norm": 0.7686058878898621,
      "learning_rate": 7.671517671517672e-06,
      "loss": 0.055,
      "step": 12530
    },
    {
      "epoch": 3.724383724383724,
      "grad_norm": 0.7790077328681946,
      "learning_rate": 7.653697653697654e-06,
      "loss": 0.0595,
      "step": 12540
    },
    {
      "epoch": 3.7273537273537274,
      "grad_norm": 0.5429967045783997,
      "learning_rate": 7.635877635877635e-06,
      "loss": 0.0596,
      "step": 12550
    },
    {
      "epoch": 3.73032373032373,
      "grad_norm": 0.4277282655239105,
      "learning_rate": 7.618057618057619e-06,
      "loss": 0.057,
      "step": 12560
    },
    {
      "epoch": 3.733293733293733,
      "grad_norm": 0.9140810966491699,
      "learning_rate": 7.600237600237601e-06,
      "loss": 0.0572,
      "step": 12570
    },
    {
      "epoch": 3.7362637362637363,
      "grad_norm": 0.9258958697319031,
      "learning_rate": 7.582417582417582e-06,
      "loss": 0.0467,
      "step": 12580
    },
    {
      "epoch": 3.739233739233739,
      "grad_norm": 0.4420478045940399,
      "learning_rate": 7.564597564597564e-06,
      "loss": 0.0633,
      "step": 12590
    },
    {
      "epoch": 3.742203742203742,
      "grad_norm": 0.642549455165863,
      "learning_rate": 7.546777546777547e-06,
      "loss": 0.059,
      "step": 12600
    },
    {
      "epoch": 3.7451737451737452,
      "grad_norm": 0.9195041060447693,
      "learning_rate": 7.528957528957529e-06,
      "loss": 0.059,
      "step": 12610
    },
    {
      "epoch": 3.748143748143748,
      "grad_norm": 0.7634923458099365,
      "learning_rate": 7.511137511137511e-06,
      "loss": 0.0476,
      "step": 12620
    },
    {
      "epoch": 3.751113751113751,
      "grad_norm": 0.7769595980644226,
      "learning_rate": 7.493317493317493e-06,
      "loss": 0.0573,
      "step": 12630
    },
    {
      "epoch": 3.754083754083754,
      "grad_norm": 0.8632429242134094,
      "learning_rate": 7.475497475497476e-06,
      "loss": 0.0769,
      "step": 12640
    },
    {
      "epoch": 3.757053757053757,
      "grad_norm": 0.7088262438774109,
      "learning_rate": 7.457677457677457e-06,
      "loss": 0.0542,
      "step": 12650
    },
    {
      "epoch": 3.76002376002376,
      "grad_norm": 0.7802964448928833,
      "learning_rate": 7.4398574398574404e-06,
      "loss": 0.0541,
      "step": 12660
    },
    {
      "epoch": 3.762993762993763,
      "grad_norm": 0.8181250095367432,
      "learning_rate": 7.422037422037423e-06,
      "loss": 0.0555,
      "step": 12670
    },
    {
      "epoch": 3.765963765963766,
      "grad_norm": 0.48934659361839294,
      "learning_rate": 7.404217404217404e-06,
      "loss": 0.0552,
      "step": 12680
    },
    {
      "epoch": 3.7689337689337687,
      "grad_norm": 0.7606164216995239,
      "learning_rate": 7.386397386397387e-06,
      "loss": 0.052,
      "step": 12690
    },
    {
      "epoch": 3.771903771903772,
      "grad_norm": 0.5713620185852051,
      "learning_rate": 7.368577368577368e-06,
      "loss": 0.0497,
      "step": 12700
    },
    {
      "epoch": 3.774873774873775,
      "grad_norm": 0.4329434037208557,
      "learning_rate": 7.350757350757351e-06,
      "loss": 0.0508,
      "step": 12710
    },
    {
      "epoch": 3.7778437778437777,
      "grad_norm": 0.5861251950263977,
      "learning_rate": 7.332937332937333e-06,
      "loss": 0.0574,
      "step": 12720
    },
    {
      "epoch": 3.780813780813781,
      "grad_norm": 0.9137320518493652,
      "learning_rate": 7.315117315117316e-06,
      "loss": 0.0473,
      "step": 12730
    },
    {
      "epoch": 3.7837837837837838,
      "grad_norm": 0.4197285771369934,
      "learning_rate": 7.297297297297298e-06,
      "loss": 0.0557,
      "step": 12740
    },
    {
      "epoch": 3.7867537867537866,
      "grad_norm": 0.8672818541526794,
      "learning_rate": 7.27947727947728e-06,
      "loss": 0.0547,
      "step": 12750
    },
    {
      "epoch": 3.78972378972379,
      "grad_norm": 1.1099815368652344,
      "learning_rate": 7.261657261657262e-06,
      "loss": 0.0547,
      "step": 12760
    },
    {
      "epoch": 3.7926937926937927,
      "grad_norm": 0.6481065154075623,
      "learning_rate": 7.2438372438372435e-06,
      "loss": 0.0552,
      "step": 12770
    },
    {
      "epoch": 3.7956637956637955,
      "grad_norm": 0.6549276113510132,
      "learning_rate": 7.2260172260172265e-06,
      "loss": 0.052,
      "step": 12780
    },
    {
      "epoch": 3.798633798633799,
      "grad_norm": 1.2700341939926147,
      "learning_rate": 7.208197208197208e-06,
      "loss": 0.052,
      "step": 12790
    },
    {
      "epoch": 3.8016038016038016,
      "grad_norm": 0.7553129196166992,
      "learning_rate": 7.190377190377191e-06,
      "loss": 0.0506,
      "step": 12800
    },
    {
      "epoch": 3.8045738045738045,
      "grad_norm": 0.6122118830680847,
      "learning_rate": 7.172557172557173e-06,
      "loss": 0.0588,
      "step": 12810
    },
    {
      "epoch": 3.8075438075438077,
      "grad_norm": 1.0293030738830566,
      "learning_rate": 7.154737154737155e-06,
      "loss": 0.0626,
      "step": 12820
    },
    {
      "epoch": 3.8105138105138106,
      "grad_norm": 0.8084139227867126,
      "learning_rate": 7.136917136917137e-06,
      "loss": 0.0523,
      "step": 12830
    },
    {
      "epoch": 3.8134838134838134,
      "grad_norm": 0.83390212059021,
      "learning_rate": 7.119097119097119e-06,
      "loss": 0.0561,
      "step": 12840
    },
    {
      "epoch": 3.8164538164538166,
      "grad_norm": 0.8345874547958374,
      "learning_rate": 7.101277101277102e-06,
      "loss": 0.0578,
      "step": 12850
    },
    {
      "epoch": 3.8194238194238195,
      "grad_norm": 0.9062905311584473,
      "learning_rate": 7.083457083457083e-06,
      "loss": 0.0488,
      "step": 12860
    },
    {
      "epoch": 3.8223938223938223,
      "grad_norm": 0.855656623840332,
      "learning_rate": 7.065637065637066e-06,
      "loss": 0.0469,
      "step": 12870
    },
    {
      "epoch": 3.8253638253638256,
      "grad_norm": 0.7304750084877014,
      "learning_rate": 7.047817047817048e-06,
      "loss": 0.0612,
      "step": 12880
    },
    {
      "epoch": 3.8283338283338284,
      "grad_norm": 1.3436956405639648,
      "learning_rate": 7.02999702999703e-06,
      "loss": 0.0692,
      "step": 12890
    },
    {
      "epoch": 3.8313038313038312,
      "grad_norm": 0.6037256121635437,
      "learning_rate": 7.0121770121770125e-06,
      "loss": 0.0527,
      "step": 12900
    },
    {
      "epoch": 3.8342738342738345,
      "grad_norm": 0.9017062187194824,
      "learning_rate": 6.994356994356995e-06,
      "loss": 0.0544,
      "step": 12910
    },
    {
      "epoch": 3.8372438372438373,
      "grad_norm": 0.5321645736694336,
      "learning_rate": 6.976536976536977e-06,
      "loss": 0.0616,
      "step": 12920
    },
    {
      "epoch": 3.84021384021384,
      "grad_norm": 0.7365061044692993,
      "learning_rate": 6.958716958716958e-06,
      "loss": 0.0544,
      "step": 12930
    },
    {
      "epoch": 3.8431838431838434,
      "grad_norm": 0.6113480925559998,
      "learning_rate": 6.940896940896941e-06,
      "loss": 0.0547,
      "step": 12940
    },
    {
      "epoch": 3.8461538461538463,
      "grad_norm": 0.6009132266044617,
      "learning_rate": 6.923076923076923e-06,
      "loss": 0.0585,
      "step": 12950
    },
    {
      "epoch": 3.849123849123849,
      "grad_norm": 0.9174481630325317,
      "learning_rate": 6.9052569052569056e-06,
      "loss": 0.0622,
      "step": 12960
    },
    {
      "epoch": 3.8520938520938524,
      "grad_norm": 0.3774741590023041,
      "learning_rate": 6.887436887436888e-06,
      "loss": 0.0545,
      "step": 12970
    },
    {
      "epoch": 3.855063855063855,
      "grad_norm": 1.134770154953003,
      "learning_rate": 6.86961686961687e-06,
      "loss": 0.0471,
      "step": 12980
    },
    {
      "epoch": 3.858033858033858,
      "grad_norm": 0.8583316802978516,
      "learning_rate": 6.851796851796852e-06,
      "loss": 0.0535,
      "step": 12990
    },
    {
      "epoch": 3.861003861003861,
      "grad_norm": 0.6052166223526001,
      "learning_rate": 6.833976833976834e-06,
      "loss": 0.0557,
      "step": 13000
    },
    {
      "epoch": 3.863973863973864,
      "grad_norm": 1.0991127490997314,
      "learning_rate": 6.816156816156816e-06,
      "loss": 0.0528,
      "step": 13010
    },
    {
      "epoch": 3.866943866943867,
      "grad_norm": 0.7694152593612671,
      "learning_rate": 6.7983367983367986e-06,
      "loss": 0.0689,
      "step": 13020
    },
    {
      "epoch": 3.8699138699138698,
      "grad_norm": 0.9624953866004944,
      "learning_rate": 6.780516780516781e-06,
      "loss": 0.0585,
      "step": 13030
    },
    {
      "epoch": 3.872883872883873,
      "grad_norm": 0.613135814666748,
      "learning_rate": 6.762696762696763e-06,
      "loss": 0.0636,
      "step": 13040
    },
    {
      "epoch": 3.875853875853876,
      "grad_norm": 0.6408237814903259,
      "learning_rate": 6.744876744876745e-06,
      "loss": 0.0525,
      "step": 13050
    },
    {
      "epoch": 3.8788238788238787,
      "grad_norm": 0.5397794842720032,
      "learning_rate": 6.727056727056727e-06,
      "loss": 0.0538,
      "step": 13060
    },
    {
      "epoch": 3.8817938817938815,
      "grad_norm": 0.569254457950592,
      "learning_rate": 6.7092367092367094e-06,
      "loss": 0.0494,
      "step": 13070
    },
    {
      "epoch": 3.884763884763885,
      "grad_norm": 0.7283187508583069,
      "learning_rate": 6.691416691416692e-06,
      "loss": 0.0428,
      "step": 13080
    },
    {
      "epoch": 3.8877338877338876,
      "grad_norm": 0.5344458818435669,
      "learning_rate": 6.673596673596674e-06,
      "loss": 0.0487,
      "step": 13090
    },
    {
      "epoch": 3.8907038907038904,
      "grad_norm": 0.8869258165359497,
      "learning_rate": 6.655776655776656e-06,
      "loss": 0.0517,
      "step": 13100
    },
    {
      "epoch": 3.8936738936738937,
      "grad_norm": 0.6164690256118774,
      "learning_rate": 6.637956637956638e-06,
      "loss": 0.056,
      "step": 13110
    },
    {
      "epoch": 3.8966438966438965,
      "grad_norm": 1.2430282831192017,
      "learning_rate": 6.62013662013662e-06,
      "loss": 0.0546,
      "step": 13120
    },
    {
      "epoch": 3.8996138996138994,
      "grad_norm": 0.5532137751579285,
      "learning_rate": 6.6023166023166025e-06,
      "loss": 0.0545,
      "step": 13130
    },
    {
      "epoch": 3.9025839025839026,
      "grad_norm": 0.9797311425209045,
      "learning_rate": 6.584496584496585e-06,
      "loss": 0.0532,
      "step": 13140
    },
    {
      "epoch": 3.9055539055539055,
      "grad_norm": 1.118308186531067,
      "learning_rate": 6.566676566676567e-06,
      "loss": 0.0609,
      "step": 13150
    },
    {
      "epoch": 3.9085239085239083,
      "grad_norm": 0.8075883388519287,
      "learning_rate": 6.548856548856549e-06,
      "loss": 0.0623,
      "step": 13160
    },
    {
      "epoch": 3.9114939114939116,
      "grad_norm": 0.5950072407722473,
      "learning_rate": 6.531036531036531e-06,
      "loss": 0.0556,
      "step": 13170
    },
    {
      "epoch": 3.9144639144639144,
      "grad_norm": 0.7014681696891785,
      "learning_rate": 6.513216513216513e-06,
      "loss": 0.0685,
      "step": 13180
    },
    {
      "epoch": 3.9174339174339172,
      "grad_norm": 0.8518680334091187,
      "learning_rate": 6.4953964953964955e-06,
      "loss": 0.0561,
      "step": 13190
    },
    {
      "epoch": 3.9204039204039205,
      "grad_norm": 0.61479651927948,
      "learning_rate": 6.477576477576478e-06,
      "loss": 0.0512,
      "step": 13200
    },
    {
      "epoch": 3.9233739233739233,
      "grad_norm": 0.8155126571655273,
      "learning_rate": 6.45975645975646e-06,
      "loss": 0.0469,
      "step": 13210
    },
    {
      "epoch": 3.926343926343926,
      "grad_norm": 0.8962894082069397,
      "learning_rate": 6.441936441936442e-06,
      "loss": 0.0517,
      "step": 13220
    },
    {
      "epoch": 3.9293139293139294,
      "grad_norm": 0.8040223717689514,
      "learning_rate": 6.424116424116425e-06,
      "loss": 0.0546,
      "step": 13230
    },
    {
      "epoch": 3.9322839322839322,
      "grad_norm": 0.6404465436935425,
      "learning_rate": 6.406296406296406e-06,
      "loss": 0.057,
      "step": 13240
    },
    {
      "epoch": 3.935253935253935,
      "grad_norm": 0.6395756602287292,
      "learning_rate": 6.3884763884763885e-06,
      "loss": 0.0552,
      "step": 13250
    },
    {
      "epoch": 3.9382239382239383,
      "grad_norm": 0.8880889415740967,
      "learning_rate": 6.370656370656371e-06,
      "loss": 0.0529,
      "step": 13260
    },
    {
      "epoch": 3.941193941193941,
      "grad_norm": 0.8830543160438538,
      "learning_rate": 6.352836352836353e-06,
      "loss": 0.0532,
      "step": 13270
    },
    {
      "epoch": 3.944163944163944,
      "grad_norm": 1.0157254934310913,
      "learning_rate": 6.335016335016335e-06,
      "loss": 0.0504,
      "step": 13280
    },
    {
      "epoch": 3.9471339471339473,
      "grad_norm": 0.43136101961135864,
      "learning_rate": 6.317196317196317e-06,
      "loss": 0.0422,
      "step": 13290
    },
    {
      "epoch": 3.95010395010395,
      "grad_norm": 0.990959107875824,
      "learning_rate": 6.2993762993763e-06,
      "loss": 0.0685,
      "step": 13300
    },
    {
      "epoch": 3.953073953073953,
      "grad_norm": 0.7046756744384766,
      "learning_rate": 6.2815562815562815e-06,
      "loss": 0.0482,
      "step": 13310
    },
    {
      "epoch": 3.956043956043956,
      "grad_norm": 0.8372275829315186,
      "learning_rate": 6.2637362637362645e-06,
      "loss": 0.0458,
      "step": 13320
    },
    {
      "epoch": 3.959013959013959,
      "grad_norm": 0.7267201542854309,
      "learning_rate": 6.245916245916246e-06,
      "loss": 0.0461,
      "step": 13330
    },
    {
      "epoch": 3.961983961983962,
      "grad_norm": 1.0572090148925781,
      "learning_rate": 6.228096228096228e-06,
      "loss": 0.0502,
      "step": 13340
    },
    {
      "epoch": 3.964953964953965,
      "grad_norm": 1.4934335947036743,
      "learning_rate": 6.21027621027621e-06,
      "loss": 0.0622,
      "step": 13350
    },
    {
      "epoch": 3.967923967923968,
      "grad_norm": 0.9395650029182434,
      "learning_rate": 6.192456192456192e-06,
      "loss": 0.0438,
      "step": 13360
    },
    {
      "epoch": 3.970893970893971,
      "grad_norm": 0.7308318018913269,
      "learning_rate": 6.174636174636175e-06,
      "loss": 0.0547,
      "step": 13370
    },
    {
      "epoch": 3.973863973863974,
      "grad_norm": 1.219499111175537,
      "learning_rate": 6.156816156816157e-06,
      "loss": 0.0698,
      "step": 13380
    },
    {
      "epoch": 3.976833976833977,
      "grad_norm": 1.007973074913025,
      "learning_rate": 6.13899613899614e-06,
      "loss": 0.0598,
      "step": 13390
    },
    {
      "epoch": 3.9798039798039797,
      "grad_norm": 0.569019615650177,
      "learning_rate": 6.121176121176121e-06,
      "loss": 0.0567,
      "step": 13400
    },
    {
      "epoch": 3.982773982773983,
      "grad_norm": 1.0321439504623413,
      "learning_rate": 6.103356103356103e-06,
      "loss": 0.0512,
      "step": 13410
    },
    {
      "epoch": 3.985743985743986,
      "grad_norm": 0.6399171352386475,
      "learning_rate": 6.085536085536085e-06,
      "loss": 0.0551,
      "step": 13420
    },
    {
      "epoch": 3.9887139887139886,
      "grad_norm": 0.7629494071006775,
      "learning_rate": 6.0677160677160676e-06,
      "loss": 0.0577,
      "step": 13430
    },
    {
      "epoch": 3.991683991683992,
      "grad_norm": 0.564344584941864,
      "learning_rate": 6.049896049896051e-06,
      "loss": 0.0512,
      "step": 13440
    },
    {
      "epoch": 3.9946539946539947,
      "grad_norm": 0.5358871817588806,
      "learning_rate": 6.032076032076032e-06,
      "loss": 0.0552,
      "step": 13450
    },
    {
      "epoch": 3.9976239976239976,
      "grad_norm": 0.9634397625923157,
      "learning_rate": 6.014256014256015e-06,
      "loss": 0.0539,
      "step": 13460
    },
    {
      "epoch": 4.0,
      "eval_f1": 0.33031292965957215,
      "eval_loss": 0.043538980185985565,
      "eval_runtime": 164.5887,
      "eval_samples_per_second": 230.994,
      "eval_steps_per_second": 3.615,
      "step": 13468
    },
    {
      "epoch": 4.000594000594001,
      "grad_norm": 1.27556574344635,
      "learning_rate": 5.996435996435996e-06,
      "loss": 0.0485,
      "step": 13470
    },
    {
      "epoch": 4.003564003564003,
      "grad_norm": 1.009068489074707,
      "learning_rate": 5.978615978615979e-06,
      "loss": 0.0567,
      "step": 13480
    },
    {
      "epoch": 4.0065340065340065,
      "grad_norm": 0.7276560068130493,
      "learning_rate": 5.960795960795961e-06,
      "loss": 0.0475,
      "step": 13490
    },
    {
      "epoch": 4.00950400950401,
      "grad_norm": 1.1445473432540894,
      "learning_rate": 5.942975942975943e-06,
      "loss": 0.0555,
      "step": 13500
    },
    {
      "epoch": 4.012474012474012,
      "grad_norm": 0.8062208294868469,
      "learning_rate": 5.925155925155926e-06,
      "loss": 0.0482,
      "step": 13510
    },
    {
      "epoch": 4.015444015444015,
      "grad_norm": 0.8470934629440308,
      "learning_rate": 5.907335907335907e-06,
      "loss": 0.0553,
      "step": 13520
    },
    {
      "epoch": 4.018414018414019,
      "grad_norm": 0.5474221706390381,
      "learning_rate": 5.88951588951589e-06,
      "loss": 0.0534,
      "step": 13530
    },
    {
      "epoch": 4.021384021384021,
      "grad_norm": 0.6959888339042664,
      "learning_rate": 5.8716958716958714e-06,
      "loss": 0.0538,
      "step": 13540
    },
    {
      "epoch": 4.024354024354024,
      "grad_norm": 0.9099267721176147,
      "learning_rate": 5.8538758538758545e-06,
      "loss": 0.055,
      "step": 13550
    },
    {
      "epoch": 4.027324027324028,
      "grad_norm": 1.2985259294509888,
      "learning_rate": 5.836055836055836e-06,
      "loss": 0.0594,
      "step": 13560
    },
    {
      "epoch": 4.03029403029403,
      "grad_norm": 0.9521999359130859,
      "learning_rate": 5.818235818235818e-06,
      "loss": 0.0588,
      "step": 13570
    },
    {
      "epoch": 4.033264033264033,
      "grad_norm": 0.5641045570373535,
      "learning_rate": 5.800415800415801e-06,
      "loss": 0.0638,
      "step": 13580
    },
    {
      "epoch": 4.0362340362340365,
      "grad_norm": 1.3589891195297241,
      "learning_rate": 5.782595782595782e-06,
      "loss": 0.0558,
      "step": 13590
    },
    {
      "epoch": 4.039204039204039,
      "grad_norm": 1.1346710920333862,
      "learning_rate": 5.764775764775765e-06,
      "loss": 0.0548,
      "step": 13600
    },
    {
      "epoch": 4.042174042174042,
      "grad_norm": 1.2047168016433716,
      "learning_rate": 5.746955746955747e-06,
      "loss": 0.0485,
      "step": 13610
    },
    {
      "epoch": 4.0451440451440455,
      "grad_norm": 0.673353910446167,
      "learning_rate": 5.72913572913573e-06,
      "loss": 0.0518,
      "step": 13620
    },
    {
      "epoch": 4.048114048114048,
      "grad_norm": 0.6137884855270386,
      "learning_rate": 5.711315711315711e-06,
      "loss": 0.0594,
      "step": 13630
    },
    {
      "epoch": 4.051084051084051,
      "grad_norm": 0.7243009805679321,
      "learning_rate": 5.693495693495694e-06,
      "loss": 0.0528,
      "step": 13640
    },
    {
      "epoch": 4.054054054054054,
      "grad_norm": 0.5329965353012085,
      "learning_rate": 5.675675675675676e-06,
      "loss": 0.049,
      "step": 13650
    },
    {
      "epoch": 4.057024057024057,
      "grad_norm": 0.4849202334880829,
      "learning_rate": 5.6578556578556575e-06,
      "loss": 0.0541,
      "step": 13660
    },
    {
      "epoch": 4.05999405999406,
      "grad_norm": 0.6682773232460022,
      "learning_rate": 5.6400356400356405e-06,
      "loss": 0.0571,
      "step": 13670
    },
    {
      "epoch": 4.062964062964063,
      "grad_norm": 0.7048366069793701,
      "learning_rate": 5.622215622215622e-06,
      "loss": 0.0582,
      "step": 13680
    },
    {
      "epoch": 4.065934065934066,
      "grad_norm": 0.7056940793991089,
      "learning_rate": 5.604395604395605e-06,
      "loss": 0.0437,
      "step": 13690
    },
    {
      "epoch": 4.068904068904069,
      "grad_norm": 0.6416358351707458,
      "learning_rate": 5.586575586575586e-06,
      "loss": 0.0458,
      "step": 13700
    },
    {
      "epoch": 4.071874071874072,
      "grad_norm": 0.7514241933822632,
      "learning_rate": 5.568755568755569e-06,
      "loss": 0.0589,
      "step": 13710
    },
    {
      "epoch": 4.074844074844075,
      "grad_norm": 0.6604739427566528,
      "learning_rate": 5.550935550935551e-06,
      "loss": 0.0562,
      "step": 13720
    },
    {
      "epoch": 4.077814077814078,
      "grad_norm": 0.4404093325138092,
      "learning_rate": 5.533115533115533e-06,
      "loss": 0.0561,
      "step": 13730
    },
    {
      "epoch": 4.080784080784081,
      "grad_norm": 0.5778560042381287,
      "learning_rate": 5.515295515295516e-06,
      "loss": 0.0664,
      "step": 13740
    },
    {
      "epoch": 4.0837540837540836,
      "grad_norm": 0.8316758275032043,
      "learning_rate": 5.497475497475497e-06,
      "loss": 0.0623,
      "step": 13750
    },
    {
      "epoch": 4.086724086724087,
      "grad_norm": 1.186763882637024,
      "learning_rate": 5.47965547965548e-06,
      "loss": 0.0524,
      "step": 13760
    },
    {
      "epoch": 4.08969408969409,
      "grad_norm": 0.4923029839992523,
      "learning_rate": 5.461835461835461e-06,
      "loss": 0.0601,
      "step": 13770
    },
    {
      "epoch": 4.0926640926640925,
      "grad_norm": 0.8885074257850647,
      "learning_rate": 5.444015444015444e-06,
      "loss": 0.0549,
      "step": 13780
    },
    {
      "epoch": 4.095634095634096,
      "grad_norm": 1.2533655166625977,
      "learning_rate": 5.4261954261954265e-06,
      "loss": 0.0662,
      "step": 13790
    },
    {
      "epoch": 4.098604098604099,
      "grad_norm": 0.7636746168136597,
      "learning_rate": 5.408375408375409e-06,
      "loss": 0.056,
      "step": 13800
    },
    {
      "epoch": 4.101574101574101,
      "grad_norm": 0.5362575054168701,
      "learning_rate": 5.390555390555391e-06,
      "loss": 0.0492,
      "step": 13810
    },
    {
      "epoch": 4.104544104544105,
      "grad_norm": 0.8624958992004395,
      "learning_rate": 5.372735372735372e-06,
      "loss": 0.0472,
      "step": 13820
    },
    {
      "epoch": 4.107514107514108,
      "grad_norm": 0.8159658908843994,
      "learning_rate": 5.354915354915355e-06,
      "loss": 0.052,
      "step": 13830
    },
    {
      "epoch": 4.11048411048411,
      "grad_norm": 1.0133655071258545,
      "learning_rate": 5.337095337095337e-06,
      "loss": 0.0675,
      "step": 13840
    },
    {
      "epoch": 4.113454113454114,
      "grad_norm": 0.7075998783111572,
      "learning_rate": 5.3192753192753196e-06,
      "loss": 0.0521,
      "step": 13850
    },
    {
      "epoch": 4.116424116424117,
      "grad_norm": 0.7996563911437988,
      "learning_rate": 5.301455301455302e-06,
      "loss": 0.0569,
      "step": 13860
    },
    {
      "epoch": 4.119394119394119,
      "grad_norm": 0.7214605808258057,
      "learning_rate": 5.283635283635284e-06,
      "loss": 0.0444,
      "step": 13870
    },
    {
      "epoch": 4.1223641223641225,
      "grad_norm": 0.5179925560951233,
      "learning_rate": 5.265815265815266e-06,
      "loss": 0.0473,
      "step": 13880
    },
    {
      "epoch": 4.125334125334125,
      "grad_norm": 0.7396313548088074,
      "learning_rate": 5.247995247995247e-06,
      "loss": 0.0549,
      "step": 13890
    },
    {
      "epoch": 4.128304128304128,
      "grad_norm": 0.8080036640167236,
      "learning_rate": 5.2301752301752304e-06,
      "loss": 0.0415,
      "step": 13900
    },
    {
      "epoch": 4.1312741312741315,
      "grad_norm": 0.6321327090263367,
      "learning_rate": 5.212355212355213e-06,
      "loss": 0.0503,
      "step": 13910
    },
    {
      "epoch": 4.134244134244134,
      "grad_norm": 0.8221378922462463,
      "learning_rate": 5.194535194535195e-06,
      "loss": 0.0456,
      "step": 13920
    },
    {
      "epoch": 4.137214137214137,
      "grad_norm": 1.2352089881896973,
      "learning_rate": 5.176715176715177e-06,
      "loss": 0.0631,
      "step": 13930
    },
    {
      "epoch": 4.14018414018414,
      "grad_norm": 0.7337819933891296,
      "learning_rate": 5.158895158895159e-06,
      "loss": 0.0568,
      "step": 13940
    },
    {
      "epoch": 4.143154143154143,
      "grad_norm": 0.5924921035766602,
      "learning_rate": 5.141075141075141e-06,
      "loss": 0.0591,
      "step": 13950
    },
    {
      "epoch": 4.146124146124146,
      "grad_norm": 0.6949560642242432,
      "learning_rate": 5.1232551232551234e-06,
      "loss": 0.0448,
      "step": 13960
    },
    {
      "epoch": 4.149094149094149,
      "grad_norm": 0.686150074005127,
      "learning_rate": 5.105435105435106e-06,
      "loss": 0.0603,
      "step": 13970
    },
    {
      "epoch": 4.152064152064152,
      "grad_norm": 0.9512919783592224,
      "learning_rate": 5.087615087615088e-06,
      "loss": 0.0757,
      "step": 13980
    },
    {
      "epoch": 4.155034155034155,
      "grad_norm": 1.0221737623214722,
      "learning_rate": 5.06979506979507e-06,
      "loss": 0.0431,
      "step": 13990
    },
    {
      "epoch": 4.158004158004158,
      "grad_norm": 0.5151529312133789,
      "learning_rate": 5.051975051975052e-06,
      "loss": 0.0446,
      "step": 14000
    },
    {
      "epoch": 4.160974160974161,
      "grad_norm": 0.5607339143753052,
      "learning_rate": 5.034155034155034e-06,
      "loss": 0.0582,
      "step": 14010
    },
    {
      "epoch": 4.163944163944164,
      "grad_norm": 0.5247116088867188,
      "learning_rate": 5.0163350163350165e-06,
      "loss": 0.0601,
      "step": 14020
    },
    {
      "epoch": 4.166914166914167,
      "grad_norm": 0.5927850604057312,
      "learning_rate": 4.998514998514999e-06,
      "loss": 0.0519,
      "step": 14030
    },
    {
      "epoch": 4.1698841698841695,
      "grad_norm": 1.483964443206787,
      "learning_rate": 4.980694980694981e-06,
      "loss": 0.0551,
      "step": 14040
    },
    {
      "epoch": 4.172854172854173,
      "grad_norm": 0.7660327553749084,
      "learning_rate": 4.962874962874963e-06,
      "loss": 0.0511,
      "step": 14050
    },
    {
      "epoch": 4.175824175824176,
      "grad_norm": 0.7715831995010376,
      "learning_rate": 4.945054945054945e-06,
      "loss": 0.069,
      "step": 14060
    },
    {
      "epoch": 4.1787941787941785,
      "grad_norm": 0.8233699798583984,
      "learning_rate": 4.927234927234927e-06,
      "loss": 0.0532,
      "step": 14070
    },
    {
      "epoch": 4.181764181764182,
      "grad_norm": 1.1859806776046753,
      "learning_rate": 4.9094149094149095e-06,
      "loss": 0.0636,
      "step": 14080
    },
    {
      "epoch": 4.184734184734185,
      "grad_norm": 0.6633349061012268,
      "learning_rate": 4.891594891594892e-06,
      "loss": 0.0535,
      "step": 14090
    },
    {
      "epoch": 4.187704187704187,
      "grad_norm": 0.6353622674942017,
      "learning_rate": 4.873774873774874e-06,
      "loss": 0.0511,
      "step": 14100
    },
    {
      "epoch": 4.190674190674191,
      "grad_norm": 0.7265152335166931,
      "learning_rate": 4.855954855954856e-06,
      "loss": 0.0562,
      "step": 14110
    },
    {
      "epoch": 4.193644193644194,
      "grad_norm": 0.8203964233398438,
      "learning_rate": 4.838134838134839e-06,
      "loss": 0.062,
      "step": 14120
    },
    {
      "epoch": 4.196614196614196,
      "grad_norm": 0.6999956369400024,
      "learning_rate": 4.82031482031482e-06,
      "loss": 0.0468,
      "step": 14130
    },
    {
      "epoch": 4.1995841995842,
      "grad_norm": 0.7191495299339294,
      "learning_rate": 4.8024948024948025e-06,
      "loss": 0.0521,
      "step": 14140
    },
    {
      "epoch": 4.202554202554203,
      "grad_norm": 1.0970999002456665,
      "learning_rate": 4.784674784674785e-06,
      "loss": 0.0538,
      "step": 14150
    },
    {
      "epoch": 4.205524205524205,
      "grad_norm": 0.8598058223724365,
      "learning_rate": 4.766854766854767e-06,
      "loss": 0.0607,
      "step": 14160
    },
    {
      "epoch": 4.2084942084942085,
      "grad_norm": 0.6349924206733704,
      "learning_rate": 4.749034749034749e-06,
      "loss": 0.0684,
      "step": 14170
    },
    {
      "epoch": 4.211464211464212,
      "grad_norm": 0.836054265499115,
      "learning_rate": 4.731214731214731e-06,
      "loss": 0.0429,
      "step": 14180
    },
    {
      "epoch": 4.214434214434214,
      "grad_norm": 1.0408257246017456,
      "learning_rate": 4.713394713394714e-06,
      "loss": 0.0579,
      "step": 14190
    },
    {
      "epoch": 4.2174042174042174,
      "grad_norm": 0.69316166639328,
      "learning_rate": 4.6955746955746955e-06,
      "loss": 0.0506,
      "step": 14200
    },
    {
      "epoch": 4.220374220374221,
      "grad_norm": 0.9772447347640991,
      "learning_rate": 4.677754677754678e-06,
      "loss": 0.0597,
      "step": 14210
    },
    {
      "epoch": 4.223344223344223,
      "grad_norm": 0.7558261156082153,
      "learning_rate": 4.65993465993466e-06,
      "loss": 0.0596,
      "step": 14220
    },
    {
      "epoch": 4.226314226314226,
      "grad_norm": 1.0452693700790405,
      "learning_rate": 4.642114642114642e-06,
      "loss": 0.0593,
      "step": 14230
    },
    {
      "epoch": 4.22928422928423,
      "grad_norm": 0.8529985547065735,
      "learning_rate": 4.624294624294624e-06,
      "loss": 0.0565,
      "step": 14240
    },
    {
      "epoch": 4.232254232254232,
      "grad_norm": 0.7685134410858154,
      "learning_rate": 4.606474606474606e-06,
      "loss": 0.0399,
      "step": 14250
    },
    {
      "epoch": 4.235224235224235,
      "grad_norm": 0.6068404912948608,
      "learning_rate": 4.588654588654589e-06,
      "loss": 0.0574,
      "step": 14260
    },
    {
      "epoch": 4.238194238194239,
      "grad_norm": 0.9283270835876465,
      "learning_rate": 4.570834570834571e-06,
      "loss": 0.0578,
      "step": 14270
    },
    {
      "epoch": 4.241164241164241,
      "grad_norm": 1.1383968591690063,
      "learning_rate": 4.553014553014554e-06,
      "loss": 0.0581,
      "step": 14280
    },
    {
      "epoch": 4.244134244134244,
      "grad_norm": 1.4468823671340942,
      "learning_rate": 4.535194535194535e-06,
      "loss": 0.0472,
      "step": 14290
    },
    {
      "epoch": 4.2471042471042475,
      "grad_norm": 0.7596200108528137,
      "learning_rate": 4.517374517374517e-06,
      "loss": 0.0529,
      "step": 14300
    },
    {
      "epoch": 4.25007425007425,
      "grad_norm": 0.31641915440559387,
      "learning_rate": 4.499554499554499e-06,
      "loss": 0.0515,
      "step": 14310
    },
    {
      "epoch": 4.253044253044253,
      "grad_norm": 0.7532891035079956,
      "learning_rate": 4.481734481734482e-06,
      "loss": 0.0618,
      "step": 14320
    },
    {
      "epoch": 4.256014256014256,
      "grad_norm": 0.5696304440498352,
      "learning_rate": 4.463914463914465e-06,
      "loss": 0.0508,
      "step": 14330
    },
    {
      "epoch": 4.258984258984259,
      "grad_norm": 0.7461338639259338,
      "learning_rate": 4.446094446094446e-06,
      "loss": 0.0619,
      "step": 14340
    },
    {
      "epoch": 4.261954261954262,
      "grad_norm": 0.6547114849090576,
      "learning_rate": 4.428274428274429e-06,
      "loss": 0.054,
      "step": 14350
    },
    {
      "epoch": 4.2649242649242645,
      "grad_norm": 0.47888076305389404,
      "learning_rate": 4.41045441045441e-06,
      "loss": 0.0528,
      "step": 14360
    },
    {
      "epoch": 4.267894267894268,
      "grad_norm": 0.500554084777832,
      "learning_rate": 4.392634392634393e-06,
      "loss": 0.0508,
      "step": 14370
    },
    {
      "epoch": 4.270864270864271,
      "grad_norm": 0.8352609276771545,
      "learning_rate": 4.374814374814375e-06,
      "loss": 0.0517,
      "step": 14380
    },
    {
      "epoch": 4.273834273834273,
      "grad_norm": 0.8386610150337219,
      "learning_rate": 4.356994356994357e-06,
      "loss": 0.0537,
      "step": 14390
    },
    {
      "epoch": 4.276804276804277,
      "grad_norm": 0.6090155243873596,
      "learning_rate": 4.33917433917434e-06,
      "loss": 0.0624,
      "step": 14400
    },
    {
      "epoch": 4.27977427977428,
      "grad_norm": 0.5954572558403015,
      "learning_rate": 4.321354321354321e-06,
      "loss": 0.0562,
      "step": 14410
    },
    {
      "epoch": 4.282744282744282,
      "grad_norm": 1.196431279182434,
      "learning_rate": 4.303534303534304e-06,
      "loss": 0.0625,
      "step": 14420
    },
    {
      "epoch": 4.285714285714286,
      "grad_norm": 0.8351851105690002,
      "learning_rate": 4.2857142857142855e-06,
      "loss": 0.0473,
      "step": 14430
    },
    {
      "epoch": 4.288684288684289,
      "grad_norm": 1.5514023303985596,
      "learning_rate": 4.2678942678942685e-06,
      "loss": 0.0521,
      "step": 14440
    },
    {
      "epoch": 4.291654291654291,
      "grad_norm": 0.5265604257583618,
      "learning_rate": 4.25007425007425e-06,
      "loss": 0.0527,
      "step": 14450
    },
    {
      "epoch": 4.2946242946242945,
      "grad_norm": 0.746711015701294,
      "learning_rate": 4.232254232254232e-06,
      "loss": 0.0539,
      "step": 14460
    },
    {
      "epoch": 4.297594297594298,
      "grad_norm": 1.2712050676345825,
      "learning_rate": 4.214434214434215e-06,
      "loss": 0.061,
      "step": 14470
    },
    {
      "epoch": 4.3005643005643,
      "grad_norm": 0.7480892539024353,
      "learning_rate": 4.196614196614196e-06,
      "loss": 0.0546,
      "step": 14480
    },
    {
      "epoch": 4.303534303534303,
      "grad_norm": 0.9045110940933228,
      "learning_rate": 4.178794178794179e-06,
      "loss": 0.0528,
      "step": 14490
    },
    {
      "epoch": 4.306504306504307,
      "grad_norm": 0.7714174389839172,
      "learning_rate": 4.160974160974161e-06,
      "loss": 0.0498,
      "step": 14500
    },
    {
      "epoch": 4.309474309474309,
      "grad_norm": 0.9806069731712341,
      "learning_rate": 4.143154143154144e-06,
      "loss": 0.0642,
      "step": 14510
    },
    {
      "epoch": 4.312444312444312,
      "grad_norm": 0.8302409052848816,
      "learning_rate": 4.125334125334125e-06,
      "loss": 0.0682,
      "step": 14520
    },
    {
      "epoch": 4.315414315414316,
      "grad_norm": 1.0357773303985596,
      "learning_rate": 4.107514107514108e-06,
      "loss": 0.0543,
      "step": 14530
    },
    {
      "epoch": 4.318384318384318,
      "grad_norm": 0.711145281791687,
      "learning_rate": 4.08969408969409e-06,
      "loss": 0.0546,
      "step": 14540
    },
    {
      "epoch": 4.321354321354321,
      "grad_norm": 0.4952409863471985,
      "learning_rate": 4.0718740718740715e-06,
      "loss": 0.0483,
      "step": 14550
    },
    {
      "epoch": 4.324324324324325,
      "grad_norm": 0.7028197050094604,
      "learning_rate": 4.0540540540540545e-06,
      "loss": 0.0449,
      "step": 14560
    },
    {
      "epoch": 4.327294327294327,
      "grad_norm": 0.7618030905723572,
      "learning_rate": 4.036234036234036e-06,
      "loss": 0.0533,
      "step": 14570
    },
    {
      "epoch": 4.33026433026433,
      "grad_norm": 0.4399206340312958,
      "learning_rate": 4.018414018414019e-06,
      "loss": 0.0504,
      "step": 14580
    },
    {
      "epoch": 4.3332343332343335,
      "grad_norm": 0.4202236533164978,
      "learning_rate": 4.000594000594e-06,
      "loss": 0.0553,
      "step": 14590
    },
    {
      "epoch": 4.336204336204336,
      "grad_norm": 0.4028279185295105,
      "learning_rate": 3.982773982773983e-06,
      "loss": 0.0548,
      "step": 14600
    },
    {
      "epoch": 4.339174339174339,
      "grad_norm": 0.7991787195205688,
      "learning_rate": 3.964953964953965e-06,
      "loss": 0.0462,
      "step": 14610
    },
    {
      "epoch": 4.342144342144342,
      "grad_norm": 0.8941323757171631,
      "learning_rate": 3.947133947133947e-06,
      "loss": 0.0396,
      "step": 14620
    },
    {
      "epoch": 4.345114345114345,
      "grad_norm": 0.6208845973014832,
      "learning_rate": 3.92931392931393e-06,
      "loss": 0.0491,
      "step": 14630
    },
    {
      "epoch": 4.348084348084348,
      "grad_norm": 0.7600147724151611,
      "learning_rate": 3.911493911493911e-06,
      "loss": 0.0625,
      "step": 14640
    },
    {
      "epoch": 4.351054351054351,
      "grad_norm": 0.4878835678100586,
      "learning_rate": 3.893673893673894e-06,
      "loss": 0.0517,
      "step": 14650
    },
    {
      "epoch": 4.354024354024354,
      "grad_norm": 0.7598239183425903,
      "learning_rate": 3.875853875853875e-06,
      "loss": 0.0534,
      "step": 14660
    },
    {
      "epoch": 4.356994356994357,
      "grad_norm": 1.0485094785690308,
      "learning_rate": 3.858033858033858e-06,
      "loss": 0.0429,
      "step": 14670
    },
    {
      "epoch": 4.35996435996436,
      "grad_norm": 1.0490139722824097,
      "learning_rate": 3.8402138402138406e-06,
      "loss": 0.0608,
      "step": 14680
    },
    {
      "epoch": 4.362934362934363,
      "grad_norm": 0.6799785494804382,
      "learning_rate": 3.822393822393823e-06,
      "loss": 0.0582,
      "step": 14690
    },
    {
      "epoch": 4.365904365904366,
      "grad_norm": 0.6751995086669922,
      "learning_rate": 3.804573804573805e-06,
      "loss": 0.0436,
      "step": 14700
    },
    {
      "epoch": 4.368874368874369,
      "grad_norm": 0.8852332830429077,
      "learning_rate": 3.7867537867537867e-06,
      "loss": 0.0505,
      "step": 14710
    },
    {
      "epoch": 4.371844371844372,
      "grad_norm": 1.6373037099838257,
      "learning_rate": 3.7689337689337693e-06,
      "loss": 0.0592,
      "step": 14720
    },
    {
      "epoch": 4.374814374814375,
      "grad_norm": 0.8025867342948914,
      "learning_rate": 3.751113751113751e-06,
      "loss": 0.0429,
      "step": 14730
    },
    {
      "epoch": 4.377784377784378,
      "grad_norm": 1.4527711868286133,
      "learning_rate": 3.733293733293733e-06,
      "loss": 0.0479,
      "step": 14740
    },
    {
      "epoch": 4.3807543807543805,
      "grad_norm": 0.726428210735321,
      "learning_rate": 3.7154737154737153e-06,
      "loss": 0.0496,
      "step": 14750
    },
    {
      "epoch": 4.383724383724384,
      "grad_norm": 0.6071370840072632,
      "learning_rate": 3.6976536976536975e-06,
      "loss": 0.0461,
      "step": 14760
    },
    {
      "epoch": 4.386694386694387,
      "grad_norm": 0.7492642998695374,
      "learning_rate": 3.67983367983368e-06,
      "loss": 0.0476,
      "step": 14770
    },
    {
      "epoch": 4.389664389664389,
      "grad_norm": 0.6537764072418213,
      "learning_rate": 3.6620136620136623e-06,
      "loss": 0.0514,
      "step": 14780
    },
    {
      "epoch": 4.392634392634393,
      "grad_norm": 0.9229190349578857,
      "learning_rate": 3.6441936441936444e-06,
      "loss": 0.0522,
      "step": 14790
    },
    {
      "epoch": 4.395604395604396,
      "grad_norm": 0.7735339999198914,
      "learning_rate": 3.6263736263736266e-06,
      "loss": 0.0492,
      "step": 14800
    },
    {
      "epoch": 4.398574398574398,
      "grad_norm": 0.7553290128707886,
      "learning_rate": 3.6085536085536088e-06,
      "loss": 0.0399,
      "step": 14810
    },
    {
      "epoch": 4.401544401544402,
      "grad_norm": 0.988293468952179,
      "learning_rate": 3.5907335907335905e-06,
      "loss": 0.0616,
      "step": 14820
    },
    {
      "epoch": 4.404514404514405,
      "grad_norm": 0.6976864337921143,
      "learning_rate": 3.5729135729135727e-06,
      "loss": 0.045,
      "step": 14830
    },
    {
      "epoch": 4.407484407484407,
      "grad_norm": 1.1107743978500366,
      "learning_rate": 3.5550935550935553e-06,
      "loss": 0.0592,
      "step": 14840
    },
    {
      "epoch": 4.410454410454411,
      "grad_norm": 0.5012500286102295,
      "learning_rate": 3.5372735372735375e-06,
      "loss": 0.0691,
      "step": 14850
    },
    {
      "epoch": 4.413424413424414,
      "grad_norm": 0.6141711473464966,
      "learning_rate": 3.5194535194535196e-06,
      "loss": 0.0504,
      "step": 14860
    },
    {
      "epoch": 4.416394416394416,
      "grad_norm": 0.9581981897354126,
      "learning_rate": 3.501633501633502e-06,
      "loss": 0.0506,
      "step": 14870
    },
    {
      "epoch": 4.4193644193644195,
      "grad_norm": 0.5101330280303955,
      "learning_rate": 3.483813483813484e-06,
      "loss": 0.051,
      "step": 14880
    },
    {
      "epoch": 4.422334422334423,
      "grad_norm": 1.3231449127197266,
      "learning_rate": 3.465993465993466e-06,
      "loss": 0.0683,
      "step": 14890
    },
    {
      "epoch": 4.425304425304425,
      "grad_norm": 0.7032347917556763,
      "learning_rate": 3.448173448173448e-06,
      "loss": 0.0544,
      "step": 14900
    },
    {
      "epoch": 4.428274428274428,
      "grad_norm": 1.0103317499160767,
      "learning_rate": 3.4303534303534305e-06,
      "loss": 0.0686,
      "step": 14910
    },
    {
      "epoch": 4.431244431244432,
      "grad_norm": 0.8349617719650269,
      "learning_rate": 3.4125334125334127e-06,
      "loss": 0.0532,
      "step": 14920
    },
    {
      "epoch": 4.434214434214434,
      "grad_norm": 0.7502866983413696,
      "learning_rate": 3.394713394713395e-06,
      "loss": 0.0585,
      "step": 14930
    },
    {
      "epoch": 4.437184437184437,
      "grad_norm": 0.6169337630271912,
      "learning_rate": 3.376893376893377e-06,
      "loss": 0.0643,
      "step": 14940
    },
    {
      "epoch": 4.440154440154441,
      "grad_norm": 1.1160919666290283,
      "learning_rate": 3.359073359073359e-06,
      "loss": 0.0545,
      "step": 14950
    },
    {
      "epoch": 4.443124443124443,
      "grad_norm": 1.2266151905059814,
      "learning_rate": 3.3412533412533413e-06,
      "loss": 0.0514,
      "step": 14960
    },
    {
      "epoch": 4.446094446094446,
      "grad_norm": 0.4717876613140106,
      "learning_rate": 3.3234333234333235e-06,
      "loss": 0.0492,
      "step": 14970
    },
    {
      "epoch": 4.4490644490644495,
      "grad_norm": 0.7887519598007202,
      "learning_rate": 3.3056133056133057e-06,
      "loss": 0.053,
      "step": 14980
    },
    {
      "epoch": 4.452034452034452,
      "grad_norm": 0.6707944273948669,
      "learning_rate": 3.287793287793288e-06,
      "loss": 0.0636,
      "step": 14990
    },
    {
      "epoch": 4.455004455004455,
      "grad_norm": 0.632908821105957,
      "learning_rate": 3.26997326997327e-06,
      "loss": 0.0563,
      "step": 15000
    },
    {
      "epoch": 4.457974457974458,
      "grad_norm": 0.597366213798523,
      "learning_rate": 3.252153252153252e-06,
      "loss": 0.0605,
      "step": 15010
    },
    {
      "epoch": 4.460944460944461,
      "grad_norm": 0.9385949969291687,
      "learning_rate": 3.2343332343332344e-06,
      "loss": 0.0562,
      "step": 15020
    },
    {
      "epoch": 4.463914463914464,
      "grad_norm": 0.7558770179748535,
      "learning_rate": 3.2165132165132165e-06,
      "loss": 0.0629,
      "step": 15030
    },
    {
      "epoch": 4.4668844668844665,
      "grad_norm": 0.5903274416923523,
      "learning_rate": 3.1986931986931987e-06,
      "loss": 0.0642,
      "step": 15040
    },
    {
      "epoch": 4.46985446985447,
      "grad_norm": 0.564150869846344,
      "learning_rate": 3.1808731808731813e-06,
      "loss": 0.0541,
      "step": 15050
    },
    {
      "epoch": 4.472824472824473,
      "grad_norm": 0.41265928745269775,
      "learning_rate": 3.163053163053163e-06,
      "loss": 0.0532,
      "step": 15060
    },
    {
      "epoch": 4.475794475794475,
      "grad_norm": 0.43676692247390747,
      "learning_rate": 3.1452331452331452e-06,
      "loss": 0.0448,
      "step": 15070
    },
    {
      "epoch": 4.478764478764479,
      "grad_norm": 0.9410478472709656,
      "learning_rate": 3.1274131274131274e-06,
      "loss": 0.0521,
      "step": 15080
    },
    {
      "epoch": 4.481734481734482,
      "grad_norm": 0.7335745096206665,
      "learning_rate": 3.1095931095931096e-06,
      "loss": 0.0431,
      "step": 15090
    },
    {
      "epoch": 4.484704484704484,
      "grad_norm": 0.8039221167564392,
      "learning_rate": 3.0917730917730917e-06,
      "loss": 0.0561,
      "step": 15100
    },
    {
      "epoch": 4.487674487674488,
      "grad_norm": 0.7728500366210938,
      "learning_rate": 3.073953073953074e-06,
      "loss": 0.0487,
      "step": 15110
    },
    {
      "epoch": 4.490644490644491,
      "grad_norm": 0.8309100270271301,
      "learning_rate": 3.0561330561330565e-06,
      "loss": 0.0506,
      "step": 15120
    },
    {
      "epoch": 4.493614493614493,
      "grad_norm": 0.8935027718544006,
      "learning_rate": 3.0383130383130387e-06,
      "loss": 0.0479,
      "step": 15130
    },
    {
      "epoch": 4.4965844965844965,
      "grad_norm": 1.1319783926010132,
      "learning_rate": 3.0204930204930204e-06,
      "loss": 0.0542,
      "step": 15140
    },
    {
      "epoch": 4.4995544995545,
      "grad_norm": 0.41596898436546326,
      "learning_rate": 3.0026730026730026e-06,
      "loss": 0.0375,
      "step": 15150
    },
    {
      "epoch": 4.502524502524502,
      "grad_norm": 1.0437852144241333,
      "learning_rate": 2.9848529848529848e-06,
      "loss": 0.0566,
      "step": 15160
    },
    {
      "epoch": 4.5054945054945055,
      "grad_norm": 0.41360408067703247,
      "learning_rate": 2.967032967032967e-06,
      "loss": 0.0503,
      "step": 15170
    },
    {
      "epoch": 4.508464508464509,
      "grad_norm": 0.5282860994338989,
      "learning_rate": 2.949212949212949e-06,
      "loss": 0.0495,
      "step": 15180
    },
    {
      "epoch": 4.511434511434511,
      "grad_norm": 0.764944851398468,
      "learning_rate": 2.9313929313929317e-06,
      "loss": 0.0546,
      "step": 15190
    },
    {
      "epoch": 4.514404514404514,
      "grad_norm": 0.7760726809501648,
      "learning_rate": 2.913572913572914e-06,
      "loss": 0.0552,
      "step": 15200
    },
    {
      "epoch": 4.517374517374518,
      "grad_norm": 0.6407870650291443,
      "learning_rate": 2.895752895752896e-06,
      "loss": 0.0624,
      "step": 15210
    },
    {
      "epoch": 4.52034452034452,
      "grad_norm": 0.7598323822021484,
      "learning_rate": 2.877932877932878e-06,
      "loss": 0.0547,
      "step": 15220
    },
    {
      "epoch": 4.523314523314523,
      "grad_norm": 0.6085183024406433,
      "learning_rate": 2.86011286011286e-06,
      "loss": 0.047,
      "step": 15230
    },
    {
      "epoch": 4.526284526284527,
      "grad_norm": 0.5744462609291077,
      "learning_rate": 2.842292842292842e-06,
      "loss": 0.0476,
      "step": 15240
    },
    {
      "epoch": 4.529254529254529,
      "grad_norm": 0.6744916439056396,
      "learning_rate": 2.8244728244728243e-06,
      "loss": 0.0503,
      "step": 15250
    },
    {
      "epoch": 4.532224532224532,
      "grad_norm": 0.939346194267273,
      "learning_rate": 2.806652806652807e-06,
      "loss": 0.0602,
      "step": 15260
    },
    {
      "epoch": 4.5351945351945355,
      "grad_norm": 1.4114429950714111,
      "learning_rate": 2.788832788832789e-06,
      "loss": 0.0659,
      "step": 15270
    },
    {
      "epoch": 4.538164538164538,
      "grad_norm": 0.6457445025444031,
      "learning_rate": 2.7710127710127712e-06,
      "loss": 0.0554,
      "step": 15280
    },
    {
      "epoch": 4.541134541134541,
      "grad_norm": 0.6651527881622314,
      "learning_rate": 2.7531927531927534e-06,
      "loss": 0.0468,
      "step": 15290
    },
    {
      "epoch": 4.5441045441045445,
      "grad_norm": 0.45064395666122437,
      "learning_rate": 2.7353727353727356e-06,
      "loss": 0.0649,
      "step": 15300
    },
    {
      "epoch": 4.547074547074547,
      "grad_norm": 0.4725130498409271,
      "learning_rate": 2.7175527175527173e-06,
      "loss": 0.051,
      "step": 15310
    },
    {
      "epoch": 4.55004455004455,
      "grad_norm": 0.7358287572860718,
      "learning_rate": 2.6997326997326995e-06,
      "loss": 0.0481,
      "step": 15320
    },
    {
      "epoch": 4.553014553014553,
      "grad_norm": 0.6313060522079468,
      "learning_rate": 2.681912681912682e-06,
      "loss": 0.0463,
      "step": 15330
    },
    {
      "epoch": 4.555984555984556,
      "grad_norm": 1.0503184795379639,
      "learning_rate": 2.6640926640926642e-06,
      "loss": 0.0597,
      "step": 15340
    },
    {
      "epoch": 4.558954558954559,
      "grad_norm": 1.3355445861816406,
      "learning_rate": 2.6462726462726464e-06,
      "loss": 0.0545,
      "step": 15350
    },
    {
      "epoch": 4.561924561924562,
      "grad_norm": 0.9509602785110474,
      "learning_rate": 2.6284526284526286e-06,
      "loss": 0.0629,
      "step": 15360
    },
    {
      "epoch": 4.564894564894565,
      "grad_norm": 1.1099562644958496,
      "learning_rate": 2.6106326106326108e-06,
      "loss": 0.0707,
      "step": 15370
    },
    {
      "epoch": 4.567864567864568,
      "grad_norm": 1.1157077550888062,
      "learning_rate": 2.592812592812593e-06,
      "loss": 0.0609,
      "step": 15380
    },
    {
      "epoch": 4.57083457083457,
      "grad_norm": 0.5642300844192505,
      "learning_rate": 2.574992574992575e-06,
      "loss": 0.0572,
      "step": 15390
    },
    {
      "epoch": 4.573804573804574,
      "grad_norm": 1.1523653268814087,
      "learning_rate": 2.5571725571725573e-06,
      "loss": 0.0585,
      "step": 15400
    },
    {
      "epoch": 4.576774576774577,
      "grad_norm": 0.6620275378227234,
      "learning_rate": 2.5393525393525394e-06,
      "loss": 0.0428,
      "step": 15410
    },
    {
      "epoch": 4.579744579744579,
      "grad_norm": 0.8893596529960632,
      "learning_rate": 2.5215325215325216e-06,
      "loss": 0.0529,
      "step": 15420
    },
    {
      "epoch": 4.5827145827145825,
      "grad_norm": 0.3965216875076294,
      "learning_rate": 2.5037125037125038e-06,
      "loss": 0.0501,
      "step": 15430
    },
    {
      "epoch": 4.585684585684586,
      "grad_norm": 0.888308584690094,
      "learning_rate": 2.485892485892486e-06,
      "loss": 0.0411,
      "step": 15440
    },
    {
      "epoch": 4.588654588654588,
      "grad_norm": 0.7801947593688965,
      "learning_rate": 2.468072468072468e-06,
      "loss": 0.0465,
      "step": 15450
    },
    {
      "epoch": 4.5916245916245915,
      "grad_norm": 0.5786198377609253,
      "learning_rate": 2.4502524502524507e-06,
      "loss": 0.0546,
      "step": 15460
    },
    {
      "epoch": 4.594594594594595,
      "grad_norm": 0.537132978439331,
      "learning_rate": 2.4324324324324325e-06,
      "loss": 0.0602,
      "step": 15470
    },
    {
      "epoch": 4.597564597564597,
      "grad_norm": 0.9715235829353333,
      "learning_rate": 2.4146124146124146e-06,
      "loss": 0.058,
      "step": 15480
    },
    {
      "epoch": 4.6005346005346,
      "grad_norm": 1.2125133275985718,
      "learning_rate": 2.396792396792397e-06,
      "loss": 0.0663,
      "step": 15490
    },
    {
      "epoch": 4.603504603504604,
      "grad_norm": 0.955084502696991,
      "learning_rate": 2.378972378972379e-06,
      "loss": 0.058,
      "step": 15500
    },
    {
      "epoch": 4.606474606474606,
      "grad_norm": 1.2165902853012085,
      "learning_rate": 2.361152361152361e-06,
      "loss": 0.0537,
      "step": 15510
    },
    {
      "epoch": 4.609444609444609,
      "grad_norm": 0.5308700203895569,
      "learning_rate": 2.3433323433323433e-06,
      "loss": 0.0489,
      "step": 15520
    },
    {
      "epoch": 4.612414612414613,
      "grad_norm": 1.492642879486084,
      "learning_rate": 2.325512325512326e-06,
      "loss": 0.0535,
      "step": 15530
    },
    {
      "epoch": 4.615384615384615,
      "grad_norm": 0.8670569062232971,
      "learning_rate": 2.307692307692308e-06,
      "loss": 0.0635,
      "step": 15540
    },
    {
      "epoch": 4.618354618354618,
      "grad_norm": 0.5621850490570068,
      "learning_rate": 2.28987228987229e-06,
      "loss": 0.0563,
      "step": 15550
    },
    {
      "epoch": 4.6213246213246215,
      "grad_norm": 0.7570291757583618,
      "learning_rate": 2.272052272052272e-06,
      "loss": 0.0554,
      "step": 15560
    },
    {
      "epoch": 4.624294624294624,
      "grad_norm": 0.7704951763153076,
      "learning_rate": 2.254232254232254e-06,
      "loss": 0.0504,
      "step": 15570
    },
    {
      "epoch": 4.627264627264627,
      "grad_norm": 1.1347087621688843,
      "learning_rate": 2.2364122364122363e-06,
      "loss": 0.0586,
      "step": 15580
    },
    {
      "epoch": 4.63023463023463,
      "grad_norm": 0.9112380743026733,
      "learning_rate": 2.2185922185922185e-06,
      "loss": 0.0489,
      "step": 15590
    },
    {
      "epoch": 4.633204633204633,
      "grad_norm": 0.9339880347251892,
      "learning_rate": 2.200772200772201e-06,
      "loss": 0.0487,
      "step": 15600
    },
    {
      "epoch": 4.636174636174636,
      "grad_norm": 0.5124456286430359,
      "learning_rate": 2.1829521829521833e-06,
      "loss": 0.0682,
      "step": 15610
    },
    {
      "epoch": 4.639144639144639,
      "grad_norm": 0.742302417755127,
      "learning_rate": 2.1651321651321654e-06,
      "loss": 0.0579,
      "step": 15620
    },
    {
      "epoch": 4.642114642114642,
      "grad_norm": 0.7987191677093506,
      "learning_rate": 2.147312147312147e-06,
      "loss": 0.0501,
      "step": 15630
    },
    {
      "epoch": 4.645084645084645,
      "grad_norm": 1.0972774028778076,
      "learning_rate": 2.1294921294921294e-06,
      "loss": 0.0657,
      "step": 15640
    },
    {
      "epoch": 4.648054648054648,
      "grad_norm": 0.855499804019928,
      "learning_rate": 2.1116721116721115e-06,
      "loss": 0.0623,
      "step": 15650
    },
    {
      "epoch": 4.651024651024651,
      "grad_norm": 0.6510183215141296,
      "learning_rate": 2.0938520938520937e-06,
      "loss": 0.0556,
      "step": 15660
    },
    {
      "epoch": 4.653994653994654,
      "grad_norm": 0.7300544381141663,
      "learning_rate": 2.0760320760320763e-06,
      "loss": 0.047,
      "step": 15670
    },
    {
      "epoch": 4.656964656964657,
      "grad_norm": 1.2278099060058594,
      "learning_rate": 2.0582120582120585e-06,
      "loss": 0.0551,
      "step": 15680
    },
    {
      "epoch": 4.65993465993466,
      "grad_norm": 0.751706063747406,
      "learning_rate": 2.0403920403920406e-06,
      "loss": 0.0583,
      "step": 15690
    },
    {
      "epoch": 4.662904662904663,
      "grad_norm": 0.32578542828559875,
      "learning_rate": 2.022572022572023e-06,
      "loss": 0.0566,
      "step": 15700
    },
    {
      "epoch": 4.665874665874666,
      "grad_norm": 0.8979377746582031,
      "learning_rate": 2.0047520047520046e-06,
      "loss": 0.0574,
      "step": 15710
    },
    {
      "epoch": 4.6688446688446685,
      "grad_norm": 0.816284716129303,
      "learning_rate": 1.9869319869319867e-06,
      "loss": 0.0534,
      "step": 15720
    },
    {
      "epoch": 4.671814671814672,
      "grad_norm": 0.9114980101585388,
      "learning_rate": 1.969111969111969e-06,
      "loss": 0.0599,
      "step": 15730
    },
    {
      "epoch": 4.674784674784675,
      "grad_norm": 0.8003864884376526,
      "learning_rate": 1.9512919512919515e-06,
      "loss": 0.0571,
      "step": 15740
    },
    {
      "epoch": 4.6777546777546775,
      "grad_norm": 0.7641117572784424,
      "learning_rate": 1.9334719334719337e-06,
      "loss": 0.0523,
      "step": 15750
    },
    {
      "epoch": 4.680724680724681,
      "grad_norm": 0.795187771320343,
      "learning_rate": 1.915651915651916e-06,
      "loss": 0.0652,
      "step": 15760
    },
    {
      "epoch": 4.683694683694684,
      "grad_norm": 0.7843266129493713,
      "learning_rate": 1.8978318978318978e-06,
      "loss": 0.0541,
      "step": 15770
    },
    {
      "epoch": 4.686664686664686,
      "grad_norm": 1.0619968175888062,
      "learning_rate": 1.88001188001188e-06,
      "loss": 0.061,
      "step": 15780
    },
    {
      "epoch": 4.68963468963469,
      "grad_norm": 0.6766983270645142,
      "learning_rate": 1.8621918621918623e-06,
      "loss": 0.0479,
      "step": 15790
    },
    {
      "epoch": 4.692604692604693,
      "grad_norm": 1.0215040445327759,
      "learning_rate": 1.8443718443718445e-06,
      "loss": 0.0479,
      "step": 15800
    },
    {
      "epoch": 4.695574695574695,
      "grad_norm": 1.0211207866668701,
      "learning_rate": 1.8265518265518265e-06,
      "loss": 0.0595,
      "step": 15810
    },
    {
      "epoch": 4.698544698544699,
      "grad_norm": 0.677951455116272,
      "learning_rate": 1.8087318087318088e-06,
      "loss": 0.0389,
      "step": 15820
    },
    {
      "epoch": 4.701514701514702,
      "grad_norm": 0.9057896137237549,
      "learning_rate": 1.790911790911791e-06,
      "loss": 0.0694,
      "step": 15830
    },
    {
      "epoch": 4.704484704484704,
      "grad_norm": 0.49817752838134766,
      "learning_rate": 1.7730917730917732e-06,
      "loss": 0.0572,
      "step": 15840
    },
    {
      "epoch": 4.7074547074547075,
      "grad_norm": 0.554590106010437,
      "learning_rate": 1.7552717552717551e-06,
      "loss": 0.0381,
      "step": 15850
    },
    {
      "epoch": 4.710424710424711,
      "grad_norm": 0.6778869032859802,
      "learning_rate": 1.7374517374517375e-06,
      "loss": 0.0488,
      "step": 15860
    },
    {
      "epoch": 4.713394713394713,
      "grad_norm": 0.38889896869659424,
      "learning_rate": 1.7196317196317197e-06,
      "loss": 0.0454,
      "step": 15870
    },
    {
      "epoch": 4.716364716364716,
      "grad_norm": 0.6715492010116577,
      "learning_rate": 1.7018117018117019e-06,
      "loss": 0.0532,
      "step": 15880
    },
    {
      "epoch": 4.71933471933472,
      "grad_norm": 0.9232339262962341,
      "learning_rate": 1.683991683991684e-06,
      "loss": 0.0407,
      "step": 15890
    },
    {
      "epoch": 4.722304722304722,
      "grad_norm": 0.9645439386367798,
      "learning_rate": 1.6661716661716662e-06,
      "loss": 0.0553,
      "step": 15900
    },
    {
      "epoch": 4.725274725274725,
      "grad_norm": 0.6320471167564392,
      "learning_rate": 1.6483516483516484e-06,
      "loss": 0.0413,
      "step": 15910
    },
    {
      "epoch": 4.728244728244729,
      "grad_norm": 0.7938999533653259,
      "learning_rate": 1.6305316305316306e-06,
      "loss": 0.0497,
      "step": 15920
    },
    {
      "epoch": 4.731214731214731,
      "grad_norm": 0.5799199938774109,
      "learning_rate": 1.6127116127116127e-06,
      "loss": 0.0608,
      "step": 15930
    },
    {
      "epoch": 4.734184734184734,
      "grad_norm": 0.6402101516723633,
      "learning_rate": 1.594891594891595e-06,
      "loss": 0.0524,
      "step": 15940
    },
    {
      "epoch": 4.737154737154738,
      "grad_norm": 0.6040619015693665,
      "learning_rate": 1.577071577071577e-06,
      "loss": 0.0486,
      "step": 15950
    },
    {
      "epoch": 4.74012474012474,
      "grad_norm": 0.4295766055583954,
      "learning_rate": 1.5592515592515594e-06,
      "loss": 0.0481,
      "step": 15960
    },
    {
      "epoch": 4.743094743094743,
      "grad_norm": 0.9524640440940857,
      "learning_rate": 1.5414315414315414e-06,
      "loss": 0.0501,
      "step": 15970
    },
    {
      "epoch": 4.7460647460647465,
      "grad_norm": 0.6644571423530579,
      "learning_rate": 1.5236115236115236e-06,
      "loss": 0.069,
      "step": 15980
    },
    {
      "epoch": 4.749034749034749,
      "grad_norm": 0.9465572237968445,
      "learning_rate": 1.5057915057915057e-06,
      "loss": 0.0499,
      "step": 15990
    },
    {
      "epoch": 4.752004752004752,
      "grad_norm": 0.6358723640441895,
      "learning_rate": 1.4879714879714881e-06,
      "loss": 0.0516,
      "step": 16000
    },
    {
      "epoch": 4.754974754974755,
      "grad_norm": 0.6010512709617615,
      "learning_rate": 1.47015147015147e-06,
      "loss": 0.0378,
      "step": 16010
    },
    {
      "epoch": 4.757944757944758,
      "grad_norm": 0.5566834211349487,
      "learning_rate": 1.4523314523314523e-06,
      "loss": 0.046,
      "step": 16020
    },
    {
      "epoch": 4.760914760914761,
      "grad_norm": 0.8823090195655823,
      "learning_rate": 1.4345114345114346e-06,
      "loss": 0.0516,
      "step": 16030
    },
    {
      "epoch": 4.763884763884764,
      "grad_norm": 0.9315813183784485,
      "learning_rate": 1.4166914166914168e-06,
      "loss": 0.0585,
      "step": 16040
    },
    {
      "epoch": 4.766854766854767,
      "grad_norm": 0.848888635635376,
      "learning_rate": 1.3988713988713988e-06,
      "loss": 0.0544,
      "step": 16050
    },
    {
      "epoch": 4.76982476982477,
      "grad_norm": 0.7777525782585144,
      "learning_rate": 1.381051381051381e-06,
      "loss": 0.0606,
      "step": 16060
    },
    {
      "epoch": 4.772794772794773,
      "grad_norm": 0.7137773036956787,
      "learning_rate": 1.3632313632313633e-06,
      "loss": 0.0561,
      "step": 16070
    },
    {
      "epoch": 4.775764775764776,
      "grad_norm": 0.7126787900924683,
      "learning_rate": 1.3454113454113455e-06,
      "loss": 0.0597,
      "step": 16080
    },
    {
      "epoch": 4.778734778734779,
      "grad_norm": 1.0160871744155884,
      "learning_rate": 1.3275913275913275e-06,
      "loss": 0.0566,
      "step": 16090
    },
    {
      "epoch": 4.781704781704782,
      "grad_norm": 1.4480925798416138,
      "learning_rate": 1.3097713097713098e-06,
      "loss": 0.072,
      "step": 16100
    },
    {
      "epoch": 4.784674784674785,
      "grad_norm": 0.7537574172019958,
      "learning_rate": 1.291951291951292e-06,
      "loss": 0.0528,
      "step": 16110
    },
    {
      "epoch": 4.787644787644788,
      "grad_norm": 1.0016658306121826,
      "learning_rate": 1.2741312741312742e-06,
      "loss": 0.0501,
      "step": 16120
    },
    {
      "epoch": 4.79061479061479,
      "grad_norm": 0.7509961128234863,
      "learning_rate": 1.2563112563112563e-06,
      "loss": 0.0573,
      "step": 16130
    },
    {
      "epoch": 4.7935847935847935,
      "grad_norm": 0.7173300981521606,
      "learning_rate": 1.2384912384912385e-06,
      "loss": 0.0545,
      "step": 16140
    },
    {
      "epoch": 4.796554796554797,
      "grad_norm": 0.6532518267631531,
      "learning_rate": 1.2206712206712207e-06,
      "loss": 0.0711,
      "step": 16150
    },
    {
      "epoch": 4.799524799524799,
      "grad_norm": 0.7350220084190369,
      "learning_rate": 1.2028512028512029e-06,
      "loss": 0.0511,
      "step": 16160
    },
    {
      "epoch": 4.802494802494802,
      "grad_norm": 0.8470587730407715,
      "learning_rate": 1.185031185031185e-06,
      "loss": 0.045,
      "step": 16170
    },
    {
      "epoch": 4.805464805464806,
      "grad_norm": 0.7634150981903076,
      "learning_rate": 1.1672111672111672e-06,
      "loss": 0.0529,
      "step": 16180
    },
    {
      "epoch": 4.808434808434808,
      "grad_norm": 0.8620032668113708,
      "learning_rate": 1.1493911493911494e-06,
      "loss": 0.0527,
      "step": 16190
    },
    {
      "epoch": 4.811404811404811,
      "grad_norm": 0.770086944103241,
      "learning_rate": 1.1315711315711318e-06,
      "loss": 0.0521,
      "step": 16200
    },
    {
      "epoch": 4.814374814374815,
      "grad_norm": 0.6775233149528503,
      "learning_rate": 1.1137511137511137e-06,
      "loss": 0.0465,
      "step": 16210
    },
    {
      "epoch": 4.817344817344817,
      "grad_norm": 0.7453165054321289,
      "learning_rate": 1.0959310959310959e-06,
      "loss": 0.0547,
      "step": 16220
    },
    {
      "epoch": 4.82031482031482,
      "grad_norm": 1.0375672578811646,
      "learning_rate": 1.078111078111078e-06,
      "loss": 0.0583,
      "step": 16230
    },
    {
      "epoch": 4.8232848232848236,
      "grad_norm": 0.8512599468231201,
      "learning_rate": 1.0602910602910604e-06,
      "loss": 0.0587,
      "step": 16240
    },
    {
      "epoch": 4.826254826254826,
      "grad_norm": 0.7857792377471924,
      "learning_rate": 1.0424710424710424e-06,
      "loss": 0.0596,
      "step": 16250
    },
    {
      "epoch": 4.829224829224829,
      "grad_norm": 0.6470934152603149,
      "learning_rate": 1.0246510246510246e-06,
      "loss": 0.0538,
      "step": 16260
    },
    {
      "epoch": 4.8321948321948325,
      "grad_norm": 0.8165239095687866,
      "learning_rate": 1.006831006831007e-06,
      "loss": 0.0567,
      "step": 16270
    },
    {
      "epoch": 4.835164835164835,
      "grad_norm": 0.5124844908714294,
      "learning_rate": 9.890109890109891e-07,
      "loss": 0.0493,
      "step": 16280
    },
    {
      "epoch": 4.838134838134838,
      "grad_norm": 1.0751707553863525,
      "learning_rate": 9.711909711909713e-07,
      "loss": 0.0559,
      "step": 16290
    },
    {
      "epoch": 4.841104841104841,
      "grad_norm": 0.9530063271522522,
      "learning_rate": 9.533709533709534e-07,
      "loss": 0.0474,
      "step": 16300
    },
    {
      "epoch": 4.844074844074844,
      "grad_norm": 1.0915278196334839,
      "learning_rate": 9.355509355509356e-07,
      "loss": 0.0565,
      "step": 16310
    },
    {
      "epoch": 4.847044847044847,
      "grad_norm": 1.2590452432632446,
      "learning_rate": 9.177309177309178e-07,
      "loss": 0.043,
      "step": 16320
    },
    {
      "epoch": 4.85001485001485,
      "grad_norm": 0.6076927781105042,
      "learning_rate": 8.999108999109e-07,
      "loss": 0.056,
      "step": 16330
    },
    {
      "epoch": 4.852984852984853,
      "grad_norm": 0.8309280276298523,
      "learning_rate": 8.820908820908821e-07,
      "loss": 0.0486,
      "step": 16340
    },
    {
      "epoch": 4.855954855954856,
      "grad_norm": 0.8125685453414917,
      "learning_rate": 8.642708642708643e-07,
      "loss": 0.0499,
      "step": 16350
    },
    {
      "epoch": 4.858924858924859,
      "grad_norm": 0.6055704951286316,
      "learning_rate": 8.464508464508465e-07,
      "loss": 0.0454,
      "step": 16360
    },
    {
      "epoch": 4.861894861894862,
      "grad_norm": 0.6113276481628418,
      "learning_rate": 8.286308286308286e-07,
      "loss": 0.0446,
      "step": 16370
    },
    {
      "epoch": 4.864864864864865,
      "grad_norm": 0.6283566951751709,
      "learning_rate": 8.108108108108109e-07,
      "loss": 0.0643,
      "step": 16380
    },
    {
      "epoch": 4.867834867834868,
      "grad_norm": 0.6934186220169067,
      "learning_rate": 7.92990792990793e-07,
      "loss": 0.0506,
      "step": 16390
    },
    {
      "epoch": 4.870804870804871,
      "grad_norm": 0.7844398021697998,
      "learning_rate": 7.751707751707753e-07,
      "loss": 0.0551,
      "step": 16400
    },
    {
      "epoch": 4.873774873774874,
      "grad_norm": 0.7214897274971008,
      "learning_rate": 7.573507573507573e-07,
      "loss": 0.0648,
      "step": 16410
    },
    {
      "epoch": 4.876744876744877,
      "grad_norm": 1.6125606298446655,
      "learning_rate": 7.395307395307396e-07,
      "loss": 0.0462,
      "step": 16420
    },
    {
      "epoch": 4.8797148797148795,
      "grad_norm": 1.0362389087677002,
      "learning_rate": 7.217107217107217e-07,
      "loss": 0.0579,
      "step": 16430
    },
    {
      "epoch": 4.882684882684883,
      "grad_norm": 1.2571192979812622,
      "learning_rate": 7.03890703890704e-07,
      "loss": 0.0526,
      "step": 16440
    },
    {
      "epoch": 4.885654885654886,
      "grad_norm": 0.7572282552719116,
      "learning_rate": 6.860706860706861e-07,
      "loss": 0.0551,
      "step": 16450
    },
    {
      "epoch": 4.888624888624888,
      "grad_norm": 0.7616530060768127,
      "learning_rate": 6.682506682506683e-07,
      "loss": 0.0557,
      "step": 16460
    },
    {
      "epoch": 4.891594891594892,
      "grad_norm": 1.0284298658370972,
      "learning_rate": 6.504306504306505e-07,
      "loss": 0.0501,
      "step": 16470
    },
    {
      "epoch": 4.894564894564894,
      "grad_norm": 0.9173740148544312,
      "learning_rate": 6.326106326106326e-07,
      "loss": 0.0554,
      "step": 16480
    },
    {
      "epoch": 4.897534897534897,
      "grad_norm": 0.5375432968139648,
      "learning_rate": 6.147906147906148e-07,
      "loss": 0.0534,
      "step": 16490
    },
    {
      "epoch": 4.900504900504901,
      "grad_norm": 0.7436822056770325,
      "learning_rate": 5.969705969705971e-07,
      "loss": 0.0528,
      "step": 16500
    },
    {
      "epoch": 4.903474903474903,
      "grad_norm": 0.3775230348110199,
      "learning_rate": 5.791505791505791e-07,
      "loss": 0.0584,
      "step": 16510
    },
    {
      "epoch": 4.906444906444906,
      "grad_norm": 0.792444109916687,
      "learning_rate": 5.613305613305614e-07,
      "loss": 0.0578,
      "step": 16520
    },
    {
      "epoch": 4.9094149094149095,
      "grad_norm": 0.527422308921814,
      "learning_rate": 5.435105435105435e-07,
      "loss": 0.0581,
      "step": 16530
    },
    {
      "epoch": 4.912384912384912,
      "grad_norm": 0.6384781002998352,
      "learning_rate": 5.256905256905258e-07,
      "loss": 0.0492,
      "step": 16540
    },
    {
      "epoch": 4.915354915354915,
      "grad_norm": 0.8578511476516724,
      "learning_rate": 5.078705078705078e-07,
      "loss": 0.069,
      "step": 16550
    },
    {
      "epoch": 4.9183249183249185,
      "grad_norm": 0.7936479449272156,
      "learning_rate": 4.900504900504901e-07,
      "loss": 0.0533,
      "step": 16560
    },
    {
      "epoch": 4.921294921294921,
      "grad_norm": 0.6151940226554871,
      "learning_rate": 4.7223047223047227e-07,
      "loss": 0.0473,
      "step": 16570
    },
    {
      "epoch": 4.924264924264924,
      "grad_norm": 0.8337469696998596,
      "learning_rate": 4.544104544104544e-07,
      "loss": 0.0434,
      "step": 16580
    },
    {
      "epoch": 4.927234927234927,
      "grad_norm": 0.7319076657295227,
      "learning_rate": 4.365904365904366e-07,
      "loss": 0.057,
      "step": 16590
    },
    {
      "epoch": 4.93020493020493,
      "grad_norm": 0.8618406653404236,
      "learning_rate": 4.187704187704188e-07,
      "loss": 0.0613,
      "step": 16600
    },
    {
      "epoch": 4.933174933174933,
      "grad_norm": 1.1733934879302979,
      "learning_rate": 4.0095040095040095e-07,
      "loss": 0.0491,
      "step": 16610
    },
    {
      "epoch": 4.936144936144936,
      "grad_norm": 0.41288742423057556,
      "learning_rate": 3.831303831303831e-07,
      "loss": 0.0443,
      "step": 16620
    },
    {
      "epoch": 4.939114939114939,
      "grad_norm": 0.854500412940979,
      "learning_rate": 3.653103653103653e-07,
      "loss": 0.0625,
      "step": 16630
    },
    {
      "epoch": 4.942084942084942,
      "grad_norm": 0.610560953617096,
      "learning_rate": 3.4749034749034746e-07,
      "loss": 0.0492,
      "step": 16640
    },
    {
      "epoch": 4.945054945054945,
      "grad_norm": 0.6326998472213745,
      "learning_rate": 3.296703296703297e-07,
      "loss": 0.0608,
      "step": 16650
    },
    {
      "epoch": 4.948024948024948,
      "grad_norm": 0.4535106122493744,
      "learning_rate": 3.1185031185031186e-07,
      "loss": 0.0415,
      "step": 16660
    },
    {
      "epoch": 4.950994950994951,
      "grad_norm": 0.8340437412261963,
      "learning_rate": 2.9403029403029403e-07,
      "loss": 0.0505,
      "step": 16670
    },
    {
      "epoch": 4.953964953964954,
      "grad_norm": 0.5445932149887085,
      "learning_rate": 2.762102762102762e-07,
      "loss": 0.0482,
      "step": 16680
    },
    {
      "epoch": 4.956934956934957,
      "grad_norm": 0.6914054155349731,
      "learning_rate": 2.5839025839025837e-07,
      "loss": 0.0588,
      "step": 16690
    },
    {
      "epoch": 4.95990495990496,
      "grad_norm": 0.9241282343864441,
      "learning_rate": 2.4057024057024054e-07,
      "loss": 0.0549,
      "step": 16700
    },
    {
      "epoch": 4.962874962874963,
      "grad_norm": 0.5624737739562988,
      "learning_rate": 2.2275022275022276e-07,
      "loss": 0.0588,
      "step": 16710
    },
    {
      "epoch": 4.9658449658449655,
      "grad_norm": 1.0242942571640015,
      "learning_rate": 2.0493020493020493e-07,
      "loss": 0.0457,
      "step": 16720
    },
    {
      "epoch": 4.968814968814969,
      "grad_norm": 1.025058388710022,
      "learning_rate": 1.8711018711018713e-07,
      "loss": 0.0524,
      "step": 16730
    },
    {
      "epoch": 4.971784971784972,
      "grad_norm": 1.143563985824585,
      "learning_rate": 1.692901692901693e-07,
      "loss": 0.0391,
      "step": 16740
    },
    {
      "epoch": 4.974754974754974,
      "grad_norm": 0.7179445028305054,
      "learning_rate": 1.5147015147015147e-07,
      "loss": 0.054,
      "step": 16750
    },
    {
      "epoch": 4.977724977724978,
      "grad_norm": 1.321466088294983,
      "learning_rate": 1.3365013365013367e-07,
      "loss": 0.0569,
      "step": 16760
    },
    {
      "epoch": 4.980694980694981,
      "grad_norm": 0.7663488388061523,
      "learning_rate": 1.1583011583011584e-07,
      "loss": 0.0542,
      "step": 16770
    },
    {
      "epoch": 4.983664983664983,
      "grad_norm": 0.7234964966773987,
      "learning_rate": 9.801009801009801e-08,
      "loss": 0.0414,
      "step": 16780
    },
    {
      "epoch": 4.986634986634987,
      "grad_norm": 0.8341914415359497,
      "learning_rate": 8.019008019008019e-08,
      "loss": 0.0647,
      "step": 16790
    },
    {
      "epoch": 4.98960498960499,
      "grad_norm": 0.9533319473266602,
      "learning_rate": 6.237006237006238e-08,
      "loss": 0.0513,
      "step": 16800
    },
    {
      "epoch": 4.992574992574992,
      "grad_norm": 1.162724494934082,
      "learning_rate": 4.4550044550044554e-08,
      "loss": 0.0543,
      "step": 16810
    },
    {
      "epoch": 4.9955449955449955,
      "grad_norm": 1.0774705410003662,
      "learning_rate": 2.673002673002673e-08,
      "loss": 0.0559,
      "step": 16820
    },
    {
      "epoch": 4.998514998514999,
      "grad_norm": 0.7029628157615662,
      "learning_rate": 8.91000891000891e-09,
      "loss": 0.0568,
      "step": 16830
    },
    {
      "epoch": 5.0,
      "eval_f1": 0.33031292965957215,
      "eval_loss": 0.041273970156908035,
      "eval_runtime": 1003.0923,
      "eval_samples_per_second": 37.902,
      "eval_steps_per_second": 0.593,
      "step": 16835
    },
    {
      "epoch": 5.0,
      "step": 16835,
      "total_flos": 8.348297673446728e+19,
      "train_loss": 0.062463992788064436,
      "train_runtime": 18625.883,
      "train_samples_per_second": 57.834,
      "train_steps_per_second": 0.904
    }
  ],
  "logging_steps": 10,
  "max_steps": 16835,
  "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": 8.348297673446728e+19,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}