{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999639262652863,
  "eval_steps": 500,
  "global_step": 13860,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 7.214746942750983e-05,
      "grad_norm": 17.337638809319504,
      "learning_rate": 0.0,
      "loss": 1.8015,
      "step": 1
    },
    {
      "epoch": 0.00014429493885501966,
      "grad_norm": 14.708392927501142,
      "learning_rate": 4.5974687827095554e-07,
      "loss": 1.6941,
      "step": 2
    },
    {
      "epoch": 0.0002164424082825295,
      "grad_norm": 3.3347749135984026,
      "learning_rate": 7.286815618830787e-07,
      "loss": 0.9941,
      "step": 3
    },
    {
      "epoch": 0.0002885898777100393,
      "grad_norm": 14.37798616656918,
      "learning_rate": 9.194937565419111e-07,
      "loss": 1.5809,
      "step": 4
    },
    {
      "epoch": 0.00036073734713754915,
      "grad_norm": 12.049786602995688,
      "learning_rate": 1.0674991931940918e-06,
      "loss": 1.5229,
      "step": 5
    },
    {
      "epoch": 0.000432884816565059,
      "grad_norm": 12.684840031548548,
      "learning_rate": 1.1884284401540344e-06,
      "loss": 1.5679,
      "step": 6
    },
    {
      "epoch": 0.0005050322859925688,
      "grad_norm": 15.397559014610025,
      "learning_rate": 1.290672661614585e-06,
      "loss": 1.5783,
      "step": 7
    },
    {
      "epoch": 0.0005771797554200786,
      "grad_norm": 7.863691900060192,
      "learning_rate": 1.3792406348128665e-06,
      "loss": 1.4325,
      "step": 8
    },
    {
      "epoch": 0.0006493272248475885,
      "grad_norm": 11.608930740995001,
      "learning_rate": 1.4573631237661574e-06,
      "loss": 1.4409,
      "step": 9
    },
    {
      "epoch": 0.0007214746942750983,
      "grad_norm": 10.9058750383251,
      "learning_rate": 1.5272460714650477e-06,
      "loss": 1.3914,
      "step": 10
    },
    {
      "epoch": 0.0007936221637026081,
      "grad_norm": 10.952990565107593,
      "learning_rate": 1.5904628872603362e-06,
      "loss": 1.4029,
      "step": 11
    },
    {
      "epoch": 0.000865769633130118,
      "grad_norm": 11.64369126457106,
      "learning_rate": 1.6481753184249899e-06,
      "loss": 1.4575,
      "step": 12
    },
    {
      "epoch": 0.0009379171025576278,
      "grad_norm": 9.356475024170678,
      "learning_rate": 1.7012656086452217e-06,
      "loss": 1.3841,
      "step": 13
    },
    {
      "epoch": 0.0010100645719851376,
      "grad_norm": 2.4849374002596583,
      "learning_rate": 1.7504195398855406e-06,
      "loss": 1.0073,
      "step": 14
    },
    {
      "epoch": 0.0010822120414126475,
      "grad_norm": 11.925975729252787,
      "learning_rate": 1.7961807550771707e-06,
      "loss": 1.3089,
      "step": 15
    },
    {
      "epoch": 0.0011543595108401573,
      "grad_norm": 5.210389161320135,
      "learning_rate": 1.8389875130838222e-06,
      "loss": 1.3466,
      "step": 16
    },
    {
      "epoch": 0.0012265069802676672,
      "grad_norm": 7.969261242791541,
      "learning_rate": 1.879198281313374e-06,
      "loss": 1.3099,
      "step": 17
    },
    {
      "epoch": 0.001298654449695177,
      "grad_norm": 6.450578755280504,
      "learning_rate": 1.9171100020371126e-06,
      "loss": 1.3445,
      "step": 18
    },
    {
      "epoch": 0.0013708019191226869,
      "grad_norm": 2.2542617491049666,
      "learning_rate": 1.9529714134269907e-06,
      "loss": 0.9106,
      "step": 19
    },
    {
      "epoch": 0.0014429493885501966,
      "grad_norm": 5.492564116543458,
      "learning_rate": 1.986992949736003e-06,
      "loss": 1.2779,
      "step": 20
    },
    {
      "epoch": 0.0015150968579777065,
      "grad_norm": 1.9367441118002386,
      "learning_rate": 2.0193542234976638e-06,
      "loss": 0.9611,
      "step": 21
    },
    {
      "epoch": 0.0015872443274052162,
      "grad_norm": 5.04572635015364,
      "learning_rate": 2.0502097655312917e-06,
      "loss": 1.2228,
      "step": 22
    },
    {
      "epoch": 0.0016593917968327262,
      "grad_norm": 4.726185782056054,
      "learning_rate": 2.079693487962469e-06,
      "loss": 1.1193,
      "step": 23
    },
    {
      "epoch": 0.001731539266260236,
      "grad_norm": 12.018217880998673,
      "learning_rate": 2.1079221966959454e-06,
      "loss": 1.1931,
      "step": 24
    },
    {
      "epoch": 0.0018036867356877458,
      "grad_norm": 3.669000157628403,
      "learning_rate": 2.1349983863881835e-06,
      "loss": 1.17,
      "step": 25
    },
    {
      "epoch": 0.0018758342051152556,
      "grad_norm": 4.31357373125358,
      "learning_rate": 2.1610124869161774e-06,
      "loss": 1.0759,
      "step": 26
    },
    {
      "epoch": 0.0019479816745427655,
      "grad_norm": 3.9406611756275916,
      "learning_rate": 2.1860446856492365e-06,
      "loss": 1.2018,
      "step": 27
    },
    {
      "epoch": 0.0020201291439702752,
      "grad_norm": 12.292474717988352,
      "learning_rate": 2.210166418156496e-06,
      "loss": 1.198,
      "step": 28
    },
    {
      "epoch": 0.002092276613397785,
      "grad_norm": 4.637815916092144,
      "learning_rate": 2.2334415972095315e-06,
      "loss": 1.1668,
      "step": 29
    },
    {
      "epoch": 0.002164424082825295,
      "grad_norm": 3.3529179462025738,
      "learning_rate": 2.255927633348126e-06,
      "loss": 1.1874,
      "step": 30
    },
    {
      "epoch": 0.002236571552252805,
      "grad_norm": 4.370409928218843,
      "learning_rate": 2.2776762880418517e-06,
      "loss": 1.2325,
      "step": 31
    },
    {
      "epoch": 0.0023087190216803145,
      "grad_norm": 4.078727823770592,
      "learning_rate": 2.2987343913547777e-06,
      "loss": 1.1141,
      "step": 32
    },
    {
      "epoch": 0.0023808664911078243,
      "grad_norm": 3.432065656260713,
      "learning_rate": 2.3191444491434147e-06,
      "loss": 1.08,
      "step": 33
    },
    {
      "epoch": 0.0024530139605353344,
      "grad_norm": 5.282743886001657,
      "learning_rate": 2.3389451595843297e-06,
      "loss": 1.1397,
      "step": 34
    },
    {
      "epoch": 0.002525161429962844,
      "grad_norm": 3.178195762784223,
      "learning_rate": 2.358171854808677e-06,
      "loss": 1.2568,
      "step": 35
    },
    {
      "epoch": 0.002597308899390354,
      "grad_norm": 7.827452731627904,
      "learning_rate": 2.3768568803080688e-06,
      "loss": 1.1223,
      "step": 36
    },
    {
      "epoch": 0.0026694563688178636,
      "grad_norm": 3.559823218467558,
      "learning_rate": 2.3950299223460323e-06,
      "loss": 1.1665,
      "step": 37
    },
    {
      "epoch": 0.0027416038382453737,
      "grad_norm": 4.6572457087373795,
      "learning_rate": 2.4127182916979464e-06,
      "loss": 1.2391,
      "step": 38
    },
    {
      "epoch": 0.0028137513076728835,
      "grad_norm": 3.251429610169609,
      "learning_rate": 2.4299471705283004e-06,
      "loss": 1.1168,
      "step": 39
    },
    {
      "epoch": 0.002885898777100393,
      "grad_norm": 3.277235499925609,
      "learning_rate": 2.4467398280069582e-06,
      "loss": 1.1053,
      "step": 40
    },
    {
      "epoch": 0.002958046246527903,
      "grad_norm": 6.013533270670491,
      "learning_rate": 2.463117809297464e-06,
      "loss": 1.1581,
      "step": 41
    },
    {
      "epoch": 0.003030193715955413,
      "grad_norm": 3.4298548907400153,
      "learning_rate": 2.4791011017686195e-06,
      "loss": 1.1789,
      "step": 42
    },
    {
      "epoch": 0.0031023411853829228,
      "grad_norm": 1.15717041909563,
      "learning_rate": 2.494708281646002e-06,
      "loss": 0.8829,
      "step": 43
    },
    {
      "epoch": 0.0031744886548104325,
      "grad_norm": 4.384988771465545,
      "learning_rate": 2.5099566438022474e-06,
      "loss": 1.107,
      "step": 44
    },
    {
      "epoch": 0.003246636124237942,
      "grad_norm": 3.3495549945476375,
      "learning_rate": 2.5248623169602493e-06,
      "loss": 1.1958,
      "step": 45
    },
    {
      "epoch": 0.0033187835936654524,
      "grad_norm": 8.719814440651506,
      "learning_rate": 2.539440366233425e-06,
      "loss": 1.1428,
      "step": 46
    },
    {
      "epoch": 0.003390931063092962,
      "grad_norm": 3.0513196560458202,
      "learning_rate": 2.553704884637446e-06,
      "loss": 1.0806,
      "step": 47
    },
    {
      "epoch": 0.003463078532520472,
      "grad_norm": 4.766816137322283,
      "learning_rate": 2.567669074966901e-06,
      "loss": 1.0535,
      "step": 48
    },
    {
      "epoch": 0.0035352260019479815,
      "grad_norm": 9.331868965075948,
      "learning_rate": 2.58134532322917e-06,
      "loss": 1.0947,
      "step": 49
    },
    {
      "epoch": 0.0036073734713754917,
      "grad_norm": 3.8692365053901208,
      "learning_rate": 2.5947452646591396e-06,
      "loss": 1.1547,
      "step": 50
    },
    {
      "epoch": 0.0036795209408030014,
      "grad_norm": 16.823833743708285,
      "learning_rate": 2.6078798431964527e-06,
      "loss": 1.162,
      "step": 51
    },
    {
      "epoch": 0.003751668410230511,
      "grad_norm": 15.022961653118026,
      "learning_rate": 2.6207593651871327e-06,
      "loss": 1.1851,
      "step": 52
    },
    {
      "epoch": 0.003823815879658021,
      "grad_norm": 3.6336176810566356,
      "learning_rate": 2.6333935479697833e-06,
      "loss": 1.1737,
      "step": 53
    },
    {
      "epoch": 0.003895963349085531,
      "grad_norm": 4.692168289028376,
      "learning_rate": 2.6457915639201917e-06,
      "loss": 1.0942,
      "step": 54
    },
    {
      "epoch": 0.003968110818513041,
      "grad_norm": 4.203159261230599,
      "learning_rate": 2.6579620804544284e-06,
      "loss": 1.1204,
      "step": 55
    },
    {
      "epoch": 0.0040402582879405504,
      "grad_norm": 3.1345780646283976,
      "learning_rate": 2.669913296427452e-06,
      "loss": 1.0679,
      "step": 56
    },
    {
      "epoch": 0.00411240575736806,
      "grad_norm": 5.351746582429324,
      "learning_rate": 2.68165297531007e-06,
      "loss": 1.1128,
      "step": 57
    },
    {
      "epoch": 0.00418455322679557,
      "grad_norm": 5.64700455382393,
      "learning_rate": 2.693188475480487e-06,
      "loss": 1.0783,
      "step": 58
    },
    {
      "epoch": 0.00425670069622308,
      "grad_norm": 6.542627381486624,
      "learning_rate": 2.704526777926441e-06,
      "loss": 1.119,
      "step": 59
    },
    {
      "epoch": 0.00432884816565059,
      "grad_norm": 4.2910700756814935,
      "learning_rate": 2.7156745116190816e-06,
      "loss": 1.0756,
      "step": 60
    },
    {
      "epoch": 0.0044009956350781,
      "grad_norm": 6.536321541822717,
      "learning_rate": 2.7266379767895355e-06,
      "loss": 1.0956,
      "step": 61
    },
    {
      "epoch": 0.00447314310450561,
      "grad_norm": 2.921219532929751,
      "learning_rate": 2.7374231663128074e-06,
      "loss": 1.1173,
      "step": 62
    },
    {
      "epoch": 0.004545290573933119,
      "grad_norm": 4.516474765956679,
      "learning_rate": 2.7480357853807425e-06,
      "loss": 1.1448,
      "step": 63
    },
    {
      "epoch": 0.004617438043360629,
      "grad_norm": 4.044582822432931,
      "learning_rate": 2.758481269625733e-06,
      "loss": 1.1513,
      "step": 64
    },
    {
      "epoch": 0.004689585512788139,
      "grad_norm": 3.0318055577799754,
      "learning_rate": 2.7687648018393137e-06,
      "loss": 1.1876,
      "step": 65
    },
    {
      "epoch": 0.0047617329822156485,
      "grad_norm": 3.481452434687666,
      "learning_rate": 2.7788913274143704e-06,
      "loss": 1.1772,
      "step": 66
    },
    {
      "epoch": 0.004833880451643158,
      "grad_norm": 7.823810660070252,
      "learning_rate": 2.7888655686261485e-06,
      "loss": 0.9992,
      "step": 67
    },
    {
      "epoch": 0.004906027921070669,
      "grad_norm": 3.5068151334752584,
      "learning_rate": 2.798692037855285e-06,
      "loss": 1.1365,
      "step": 68
    },
    {
      "epoch": 0.0049781753904981785,
      "grad_norm": 4.283146793048968,
      "learning_rate": 2.8083750498455478e-06,
      "loss": 1.1458,
      "step": 69
    },
    {
      "epoch": 0.005050322859925688,
      "grad_norm": 3.585863166653655,
      "learning_rate": 2.817918733079633e-06,
      "loss": 1.1126,
      "step": 70
    },
    {
      "epoch": 0.005122470329353198,
      "grad_norm": 4.608792336054533,
      "learning_rate": 2.8273270403480783e-06,
      "loss": 1.1099,
      "step": 71
    },
    {
      "epoch": 0.005194617798780708,
      "grad_norm": 6.57739021746117,
      "learning_rate": 2.836603758579024e-06,
      "loss": 1.0892,
      "step": 72
    },
    {
      "epoch": 0.005266765268208217,
      "grad_norm": 3.2103857601261567,
      "learning_rate": 2.8457525179899824e-06,
      "loss": 1.1452,
      "step": 73
    },
    {
      "epoch": 0.005338912737635727,
      "grad_norm": 3.917592297740529,
      "learning_rate": 2.8547768006169884e-06,
      "loss": 1.1134,
      "step": 74
    },
    {
      "epoch": 0.005411060207063237,
      "grad_norm": 8.24474144006798,
      "learning_rate": 2.863679948271262e-06,
      "loss": 1.1336,
      "step": 75
    },
    {
      "epoch": 0.0054832076764907475,
      "grad_norm": 3.890716061426494,
      "learning_rate": 2.872465169968902e-06,
      "loss": 0.973,
      "step": 76
    },
    {
      "epoch": 0.005555355145918257,
      "grad_norm": 3.466639245976305,
      "learning_rate": 2.881135548874922e-06,
      "loss": 1.0966,
      "step": 77
    },
    {
      "epoch": 0.005627502615345767,
      "grad_norm": 3.012834592100587,
      "learning_rate": 2.889694048799256e-06,
      "loss": 1.22,
      "step": 78
    },
    {
      "epoch": 0.005699650084773277,
      "grad_norm": 2.6466274321344074,
      "learning_rate": 2.8981435202789713e-06,
      "loss": 1.0942,
      "step": 79
    },
    {
      "epoch": 0.005771797554200786,
      "grad_norm": 5.580054197971102,
      "learning_rate": 2.906486706277914e-06,
      "loss": 1.1142,
      "step": 80
    },
    {
      "epoch": 0.005843945023628296,
      "grad_norm": 4.23550028470583,
      "learning_rate": 2.9147262475323147e-06,
      "loss": 1.1032,
      "step": 81
    },
    {
      "epoch": 0.005916092493055806,
      "grad_norm": 6.277382946070068,
      "learning_rate": 2.9228646875684196e-06,
      "loss": 1.1217,
      "step": 82
    },
    {
      "epoch": 0.0059882399624833155,
      "grad_norm": 5.487663304846907,
      "learning_rate": 2.9309044774159963e-06,
      "loss": 1.0683,
      "step": 83
    },
    {
      "epoch": 0.006060387431910826,
      "grad_norm": 1.2599272868201703,
      "learning_rate": 2.9388479800395748e-06,
      "loss": 0.9017,
      "step": 84
    },
    {
      "epoch": 0.006132534901338336,
      "grad_norm": 3.6750464196419936,
      "learning_rate": 2.946697474507466e-06,
      "loss": 1.1352,
      "step": 85
    },
    {
      "epoch": 0.0062046823707658455,
      "grad_norm": 5.782928058301454,
      "learning_rate": 2.954455159916957e-06,
      "loss": 1.0053,
      "step": 86
    },
    {
      "epoch": 0.006276829840193355,
      "grad_norm": 5.097798429530634,
      "learning_rate": 2.96212315909261e-06,
      "loss": 1.1113,
      "step": 87
    },
    {
      "epoch": 0.006348977309620865,
      "grad_norm": 5.126109677645036,
      "learning_rate": 2.969703522073203e-06,
      "loss": 1.1312,
      "step": 88
    },
    {
      "epoch": 0.006421124779048375,
      "grad_norm": 10.44877225973654,
      "learning_rate": 2.9771982294016654e-06,
      "loss": 1.0315,
      "step": 89
    },
    {
      "epoch": 0.006493272248475884,
      "grad_norm": 6.810662257948613,
      "learning_rate": 2.9846091952312046e-06,
      "loss": 1.1084,
      "step": 90
    },
    {
      "epoch": 0.006565419717903394,
      "grad_norm": 2.72852846987738,
      "learning_rate": 2.9919382702598064e-06,
      "loss": 1.1078,
      "step": 91
    },
    {
      "epoch": 0.006637567187330905,
      "grad_norm": 4.995834904602329,
      "learning_rate": 2.99918724450438e-06,
      "loss": 1.1317,
      "step": 92
    },
    {
      "epoch": 0.0067097146567584144,
      "grad_norm": 3.610287241405329,
      "learning_rate": 3.0063578499249308e-06,
      "loss": 1.1698,
      "step": 93
    },
    {
      "epoch": 0.006781862126185924,
      "grad_norm": 3.745819049902739,
      "learning_rate": 3.013451762908401e-06,
      "loss": 1.1094,
      "step": 94
    },
    {
      "epoch": 0.006854009595613434,
      "grad_norm": 5.153900535988571,
      "learning_rate": 3.020470606621083e-06,
      "loss": 1.1245,
      "step": 95
    },
    {
      "epoch": 0.006926157065040944,
      "grad_norm": 3.8277795752558172,
      "learning_rate": 3.0274159532378564e-06,
      "loss": 1.1188,
      "step": 96
    },
    {
      "epoch": 0.006998304534468453,
      "grad_norm": 2.9839064944346525,
      "learning_rate": 3.0342893260559215e-06,
      "loss": 1.0637,
      "step": 97
    },
    {
      "epoch": 0.007070452003895963,
      "grad_norm": 4.1383105449276965,
      "learning_rate": 3.0410922015001263e-06,
      "loss": 1.084,
      "step": 98
    },
    {
      "epoch": 0.007142599473323473,
      "grad_norm": 4.180208148793111,
      "learning_rate": 3.0478260110264934e-06,
      "loss": 1.2422,
      "step": 99
    },
    {
      "epoch": 0.007214746942750983,
      "grad_norm": 4.057755972167755,
      "learning_rate": 3.0544921429300953e-06,
      "loss": 1.1038,
      "step": 100
    },
    {
      "epoch": 0.007286894412178493,
      "grad_norm": 3.020758178477608,
      "learning_rate": 3.061091944062968e-06,
      "loss": 1.114,
      "step": 101
    },
    {
      "epoch": 0.007359041881606003,
      "grad_norm": 3.1413719086191816,
      "learning_rate": 3.067626721467408e-06,
      "loss": 1.0761,
      "step": 102
    },
    {
      "epoch": 0.0074311893510335125,
      "grad_norm": 6.23272514980155,
      "learning_rate": 3.0740977439295834e-06,
      "loss": 1.1422,
      "step": 103
    },
    {
      "epoch": 0.007503336820461022,
      "grad_norm": 5.520300420795941,
      "learning_rate": 3.0805062434580884e-06,
      "loss": 0.9977,
      "step": 104
    },
    {
      "epoch": 0.007575484289888532,
      "grad_norm": 8.058073972876134,
      "learning_rate": 3.0868534166917558e-06,
      "loss": 1.0544,
      "step": 105
    },
    {
      "epoch": 0.007647631759316042,
      "grad_norm": 4.916676515903414,
      "learning_rate": 3.093140426240739e-06,
      "loss": 1.1662,
      "step": 106
    },
    {
      "epoch": 0.007719779228743551,
      "grad_norm": 3.715555951092084,
      "learning_rate": 3.0993684019646337e-06,
      "loss": 1.049,
      "step": 107
    },
    {
      "epoch": 0.007791926698171062,
      "grad_norm": 3.6000702056396996,
      "learning_rate": 3.1055384421911475e-06,
      "loss": 1.12,
      "step": 108
    },
    {
      "epoch": 0.00786407416759857,
      "grad_norm": 6.743765318166395,
      "learning_rate": 3.1116516148786068e-06,
      "loss": 1.0975,
      "step": 109
    },
    {
      "epoch": 0.007936221637026081,
      "grad_norm": 0.8621694387470867,
      "learning_rate": 3.1177089587253837e-06,
      "loss": 0.8973,
      "step": 110
    },
    {
      "epoch": 0.00800836910645359,
      "grad_norm": 3.961041872353608,
      "learning_rate": 3.123711484229111e-06,
      "loss": 1.0636,
      "step": 111
    },
    {
      "epoch": 0.008080516575881101,
      "grad_norm": 3.6473240593041774,
      "learning_rate": 3.129660174698407e-06,
      "loss": 1.1546,
      "step": 112
    },
    {
      "epoch": 0.008152664045308611,
      "grad_norm": 5.191673713952798,
      "learning_rate": 3.1355559872196274e-06,
      "loss": 1.1686,
      "step": 113
    },
    {
      "epoch": 0.00822481151473612,
      "grad_norm": 0.8941036351703334,
      "learning_rate": 3.1413998535810255e-06,
      "loss": 0.8828,
      "step": 114
    },
    {
      "epoch": 0.008296958984163631,
      "grad_norm": 3.6307012514318,
      "learning_rate": 3.147192681156561e-06,
      "loss": 1.017,
      "step": 115
    },
    {
      "epoch": 0.00836910645359114,
      "grad_norm": 10.732338880195476,
      "learning_rate": 3.1529353537514425e-06,
      "loss": 1.063,
      "step": 116
    },
    {
      "epoch": 0.00844125392301865,
      "grad_norm": 3.1738577222177664,
      "learning_rate": 3.158628732411379e-06,
      "loss": 1.0517,
      "step": 117
    },
    {
      "epoch": 0.00851340139244616,
      "grad_norm": 3.531006121020547,
      "learning_rate": 3.164273656197397e-06,
      "loss": 0.9583,
      "step": 118
    },
    {
      "epoch": 0.00858554886187367,
      "grad_norm": 7.136065540371976,
      "learning_rate": 3.169870942927959e-06,
      "loss": 0.9422,
      "step": 119
    },
    {
      "epoch": 0.00865769633130118,
      "grad_norm": 6.541325003414291,
      "learning_rate": 3.175421389890037e-06,
      "loss": 1.0631,
      "step": 120
    },
    {
      "epoch": 0.00872984380072869,
      "grad_norm": 8.282872224659696,
      "learning_rate": 3.1809257745206724e-06,
      "loss": 1.0005,
      "step": 121
    },
    {
      "epoch": 0.0088019912701562,
      "grad_norm": 3.310376476302349,
      "learning_rate": 3.186384855060491e-06,
      "loss": 1.1129,
      "step": 122
    },
    {
      "epoch": 0.008874138739583709,
      "grad_norm": 3.8813865699021903,
      "learning_rate": 3.191799371180542e-06,
      "loss": 1.0682,
      "step": 123
    },
    {
      "epoch": 0.00894628620901122,
      "grad_norm": 27.094250103515968,
      "learning_rate": 3.197170044583763e-06,
      "loss": 1.104,
      "step": 124
    },
    {
      "epoch": 0.009018433678438728,
      "grad_norm": 5.243031554626818,
      "learning_rate": 3.202497579582276e-06,
      "loss": 1.0249,
      "step": 125
    },
    {
      "epoch": 0.009090581147866239,
      "grad_norm": 4.088784270975322,
      "learning_rate": 3.207782663651698e-06,
      "loss": 1.1559,
      "step": 126
    },
    {
      "epoch": 0.009162728617293748,
      "grad_norm": 7.473127925521611,
      "learning_rate": 3.213025967963534e-06,
      "loss": 1.0616,
      "step": 127
    },
    {
      "epoch": 0.009234876086721258,
      "grad_norm": 0.8827114553773715,
      "learning_rate": 3.2182281478966887e-06,
      "loss": 0.9413,
      "step": 128
    },
    {
      "epoch": 0.009307023556148769,
      "grad_norm": 5.418304938988321,
      "learning_rate": 3.22338984352908e-06,
      "loss": 1.1086,
      "step": 129
    },
    {
      "epoch": 0.009379171025576278,
      "grad_norm": 4.238470482886693,
      "learning_rate": 3.228511680110269e-06,
      "loss": 1.1925,
      "step": 130
    },
    {
      "epoch": 0.009451318495003788,
      "grad_norm": 4.612321332246856,
      "learning_rate": 3.2335942685159767e-06,
      "loss": 1.1181,
      "step": 131
    },
    {
      "epoch": 0.009523465964431297,
      "grad_norm": 2.8601505082548724,
      "learning_rate": 3.2386382056853257e-06,
      "loss": 1.1346,
      "step": 132
    },
    {
      "epoch": 0.009595613433858808,
      "grad_norm": 3.5771756307832074,
      "learning_rate": 3.2436440750415763e-06,
      "loss": 1.0994,
      "step": 133
    },
    {
      "epoch": 0.009667760903286316,
      "grad_norm": 5.955254922145999,
      "learning_rate": 3.248612446897104e-06,
      "loss": 1.0452,
      "step": 134
    },
    {
      "epoch": 0.009739908372713827,
      "grad_norm": 12.293958616207851,
      "learning_rate": 3.2535438788433284e-06,
      "loss": 1.133,
      "step": 135
    },
    {
      "epoch": 0.009812055842141338,
      "grad_norm": 3.090248052523138,
      "learning_rate": 3.258438916126241e-06,
      "loss": 1.0268,
      "step": 136
    },
    {
      "epoch": 0.009884203311568847,
      "grad_norm": 4.597513745999388,
      "learning_rate": 3.2632980920081904e-06,
      "loss": 1.141,
      "step": 137
    },
    {
      "epoch": 0.009956350780996357,
      "grad_norm": 5.933627502064143,
      "learning_rate": 3.268121928116504e-06,
      "loss": 1.0441,
      "step": 138
    },
    {
      "epoch": 0.010028498250423866,
      "grad_norm": 4.636600483235287,
      "learning_rate": 3.27291093477952e-06,
      "loss": 1.2137,
      "step": 139
    },
    {
      "epoch": 0.010100645719851377,
      "grad_norm": 11.863056159372663,
      "learning_rate": 3.277665611350588e-06,
      "loss": 1.017,
      "step": 140
    },
    {
      "epoch": 0.010172793189278885,
      "grad_norm": 6.7825237431516605,
      "learning_rate": 3.282386446520524e-06,
      "loss": 1.1108,
      "step": 141
    },
    {
      "epoch": 0.010244940658706396,
      "grad_norm": 6.612222116901563,
      "learning_rate": 3.2870739186190344e-06,
      "loss": 1.0935,
      "step": 142
    },
    {
      "epoch": 0.010317088128133907,
      "grad_norm": 4.82839481130711,
      "learning_rate": 3.2917284959055577e-06,
      "loss": 1.0298,
      "step": 143
    },
    {
      "epoch": 0.010389235597561415,
      "grad_norm": 4.637982267818794,
      "learning_rate": 3.2963506368499798e-06,
      "loss": 1.1607,
      "step": 144
    },
    {
      "epoch": 0.010461383066988926,
      "grad_norm": 4.624111193234066,
      "learning_rate": 3.300940790403623e-06,
      "loss": 1.1781,
      "step": 145
    },
    {
      "epoch": 0.010533530536416435,
      "grad_norm": 3.5752010475660776,
      "learning_rate": 3.3054993962609377e-06,
      "loss": 1.032,
      "step": 146
    },
    {
      "epoch": 0.010605678005843945,
      "grad_norm": 5.018176989308021,
      "learning_rate": 3.310026885112249e-06,
      "loss": 1.0559,
      "step": 147
    },
    {
      "epoch": 0.010677825475271454,
      "grad_norm": 5.175116039209856,
      "learning_rate": 3.3145236788879433e-06,
      "loss": 1.1155,
      "step": 148
    },
    {
      "epoch": 0.010749972944698965,
      "grad_norm": 8.053321049072462,
      "learning_rate": 3.3189901909944316e-06,
      "loss": 1.1128,
      "step": 149
    },
    {
      "epoch": 0.010822120414126474,
      "grad_norm": 3.6206164986872618,
      "learning_rate": 3.323426826542218e-06,
      "loss": 1.0404,
      "step": 150
    },
    {
      "epoch": 0.010894267883553984,
      "grad_norm": 4.055185864348165,
      "learning_rate": 3.327833982566395e-06,
      "loss": 1.1234,
      "step": 151
    },
    {
      "epoch": 0.010966415352981495,
      "grad_norm": 5.091073485605943,
      "learning_rate": 3.3322120482398574e-06,
      "loss": 1.1197,
      "step": 152
    },
    {
      "epoch": 0.011038562822409004,
      "grad_norm": 0.8606846280519923,
      "learning_rate": 3.3365614050795313e-06,
      "loss": 0.9196,
      "step": 153
    },
    {
      "epoch": 0.011110710291836514,
      "grad_norm": 5.29802410073305,
      "learning_rate": 3.3408824271458772e-06,
      "loss": 1.0445,
      "step": 154
    },
    {
      "epoch": 0.011182857761264023,
      "grad_norm": 3.456240116263911,
      "learning_rate": 3.3451754812359436e-06,
      "loss": 1.129,
      "step": 155
    },
    {
      "epoch": 0.011255005230691534,
      "grad_norm": 3.516328054701559,
      "learning_rate": 3.3494409270702114e-06,
      "loss": 1.03,
      "step": 156
    },
    {
      "epoch": 0.011327152700119043,
      "grad_norm": 10.117615821049135,
      "learning_rate": 3.3536791174734653e-06,
      "loss": 1.0492,
      "step": 157
    },
    {
      "epoch": 0.011399300169546553,
      "grad_norm": 3.789375339453303,
      "learning_rate": 3.3578903985499274e-06,
      "loss": 1.1456,
      "step": 158
    },
    {
      "epoch": 0.011471447638974064,
      "grad_norm": 3.861692651609919,
      "learning_rate": 3.362075109852862e-06,
      "loss": 1.1442,
      "step": 159
    },
    {
      "epoch": 0.011543595108401573,
      "grad_norm": 3.3746045198790893,
      "learning_rate": 3.3662335845488692e-06,
      "loss": 1.1071,
      "step": 160
    },
    {
      "epoch": 0.011615742577829083,
      "grad_norm": 0.9428207723819193,
      "learning_rate": 3.370366149577054e-06,
      "loss": 0.9145,
      "step": 161
    },
    {
      "epoch": 0.011687890047256592,
      "grad_norm": 3.577391570105489,
      "learning_rate": 3.37447312580327e-06,
      "loss": 1.1628,
      "step": 162
    },
    {
      "epoch": 0.011760037516684103,
      "grad_norm": 2.8747848761971078,
      "learning_rate": 3.3785548281696194e-06,
      "loss": 1.0818,
      "step": 163
    },
    {
      "epoch": 0.011832184986111612,
      "grad_norm": 6.099055245202588,
      "learning_rate": 3.3826115658393753e-06,
      "loss": 1.0186,
      "step": 164
    },
    {
      "epoch": 0.011904332455539122,
      "grad_norm": 5.228978167577481,
      "learning_rate": 3.3866436423375062e-06,
      "loss": 1.0353,
      "step": 165
    },
    {
      "epoch": 0.011976479924966631,
      "grad_norm": 4.975849190318666,
      "learning_rate": 3.3906513556869524e-06,
      "loss": 1.0702,
      "step": 166
    },
    {
      "epoch": 0.012048627394394142,
      "grad_norm": 1.9804880716859807,
      "learning_rate": 3.3946349985408007e-06,
      "loss": 0.8606,
      "step": 167
    },
    {
      "epoch": 0.012120774863821652,
      "grad_norm": 0.9927173911632436,
      "learning_rate": 3.3985948583105305e-06,
      "loss": 0.9011,
      "step": 168
    },
    {
      "epoch": 0.012192922333249161,
      "grad_norm": 2.402879509459682,
      "learning_rate": 3.4025312172904435e-06,
      "loss": 0.9606,
      "step": 169
    },
    {
      "epoch": 0.012265069802676672,
      "grad_norm": 3.2998330102920344,
      "learning_rate": 3.4064443527784217e-06,
      "loss": 1.1173,
      "step": 170
    },
    {
      "epoch": 0.01233721727210418,
      "grad_norm": 3.28530400836333,
      "learning_rate": 3.4103345371931485e-06,
      "loss": 1.0272,
      "step": 171
    },
    {
      "epoch": 0.012409364741531691,
      "grad_norm": 4.849273682312814,
      "learning_rate": 3.414202038187913e-06,
      "loss": 1.1245,
      "step": 172
    },
    {
      "epoch": 0.0124815122109592,
      "grad_norm": 3.7349167434829806,
      "learning_rate": 3.4180471187611114e-06,
      "loss": 1.011,
      "step": 173
    },
    {
      "epoch": 0.01255365968038671,
      "grad_norm": 3.787126246980854,
      "learning_rate": 3.4218700373635654e-06,
      "loss": 1.0684,
      "step": 174
    },
    {
      "epoch": 0.012625807149814221,
      "grad_norm": 2.71948626841862,
      "learning_rate": 3.4256710480027695e-06,
      "loss": 1.1215,
      "step": 175
    },
    {
      "epoch": 0.01269795461924173,
      "grad_norm": 0.7562041765312071,
      "learning_rate": 3.4294504003441584e-06,
      "loss": 0.8668,
      "step": 176
    },
    {
      "epoch": 0.01277010208866924,
      "grad_norm": 5.577524190223588,
      "learning_rate": 3.4332083398095194e-06,
      "loss": 1.1485,
      "step": 177
    },
    {
      "epoch": 0.01284224955809675,
      "grad_norm": 3.031568058277395,
      "learning_rate": 3.4369451076726216e-06,
      "loss": 1.0207,
      "step": 178
    },
    {
      "epoch": 0.01291439702752426,
      "grad_norm": 4.504248626744345,
      "learning_rate": 3.4406609411521665e-06,
      "loss": 1.1042,
      "step": 179
    },
    {
      "epoch": 0.012986544496951769,
      "grad_norm": 3.8307403972296004,
      "learning_rate": 3.4443560735021607e-06,
      "loss": 1.0309,
      "step": 180
    },
    {
      "epoch": 0.01305869196637928,
      "grad_norm": 3.8817021645582157,
      "learning_rate": 3.4480307340997692e-06,
      "loss": 1.0994,
      "step": 181
    },
    {
      "epoch": 0.013130839435806788,
      "grad_norm": 3.941973975193895,
      "learning_rate": 3.4516851485307625e-06,
      "loss": 1.1142,
      "step": 182
    },
    {
      "epoch": 0.013202986905234299,
      "grad_norm": 8.012809987848657,
      "learning_rate": 3.455319538672614e-06,
      "loss": 1.0409,
      "step": 183
    },
    {
      "epoch": 0.01327513437466181,
      "grad_norm": 4.222727391210969,
      "learning_rate": 3.4589341227753358e-06,
      "loss": 1.1141,
      "step": 184
    },
    {
      "epoch": 0.013347281844089318,
      "grad_norm": 4.830947001876187,
      "learning_rate": 3.462529115540124e-06,
      "loss": 0.9764,
      "step": 185
    },
    {
      "epoch": 0.013419429313516829,
      "grad_norm": 4.874702053257432,
      "learning_rate": 3.466104728195886e-06,
      "loss": 1.1224,
      "step": 186
    },
    {
      "epoch": 0.013491576782944338,
      "grad_norm": 4.309767147421104,
      "learning_rate": 3.4696611685737104e-06,
      "loss": 1.1014,
      "step": 187
    },
    {
      "epoch": 0.013563724252371848,
      "grad_norm": 4.008078523548127,
      "learning_rate": 3.4731986411793564e-06,
      "loss": 1.0768,
      "step": 188
    },
    {
      "epoch": 0.013635871721799357,
      "grad_norm": 4.281009639480527,
      "learning_rate": 3.4767173472638216e-06,
      "loss": 0.912,
      "step": 189
    },
    {
      "epoch": 0.013708019191226868,
      "grad_norm": 7.036667292438421,
      "learning_rate": 3.4802174848920384e-06,
      "loss": 1.0169,
      "step": 190
    },
    {
      "epoch": 0.013780166660654378,
      "grad_norm": 4.627219208988508,
      "learning_rate": 3.4836992490097806e-06,
      "loss": 1.0837,
      "step": 191
    },
    {
      "epoch": 0.013852314130081887,
      "grad_norm": 5.430137121671629,
      "learning_rate": 3.487162831508812e-06,
      "loss": 1.1717,
      "step": 192
    },
    {
      "epoch": 0.013924461599509398,
      "grad_norm": 3.6245043185483694,
      "learning_rate": 3.490608421290348e-06,
      "loss": 1.002,
      "step": 193
    },
    {
      "epoch": 0.013996609068936907,
      "grad_norm": 6.281845140493407,
      "learning_rate": 3.4940362043268776e-06,
      "loss": 1.0621,
      "step": 194
    },
    {
      "epoch": 0.014068756538364417,
      "grad_norm": 4.356762393715987,
      "learning_rate": 3.4974463637223924e-06,
      "loss": 1.18,
      "step": 195
    },
    {
      "epoch": 0.014140904007791926,
      "grad_norm": 3.9156606613203624,
      "learning_rate": 3.500839079771081e-06,
      "loss": 0.9632,
      "step": 196
    },
    {
      "epoch": 0.014213051477219437,
      "grad_norm": 2.5947446639434935,
      "learning_rate": 3.504214530014526e-06,
      "loss": 1.0476,
      "step": 197
    },
    {
      "epoch": 0.014285198946646946,
      "grad_norm": 2.4175138523213824,
      "learning_rate": 3.507572889297449e-06,
      "loss": 1.1357,
      "step": 198
    },
    {
      "epoch": 0.014357346416074456,
      "grad_norm": 4.646502026159201,
      "learning_rate": 3.5109143298220628e-06,
      "loss": 1.0803,
      "step": 199
    },
    {
      "epoch": 0.014429493885501967,
      "grad_norm": 4.885940599577786,
      "learning_rate": 3.5142390212010506e-06,
      "loss": 1.0617,
      "step": 200
    },
    {
      "epoch": 0.014501641354929476,
      "grad_norm": 4.952632458382336,
      "learning_rate": 3.517547130509227e-06,
      "loss": 1.1781,
      "step": 201
    },
    {
      "epoch": 0.014573788824356986,
      "grad_norm": 3.8086001352109937,
      "learning_rate": 3.5208388223339235e-06,
      "loss": 1.0851,
      "step": 202
    },
    {
      "epoch": 0.014645936293784495,
      "grad_norm": 4.5047402316952425,
      "learning_rate": 3.5241142588241166e-06,
      "loss": 1.0673,
      "step": 203
    },
    {
      "epoch": 0.014718083763212006,
      "grad_norm": 2.9303633660507504,
      "learning_rate": 3.5273735997383636e-06,
      "loss": 0.9994,
      "step": 204
    },
    {
      "epoch": 0.014790231232639514,
      "grad_norm": 3.899622673952769,
      "learning_rate": 3.530617002491556e-06,
      "loss": 1.0467,
      "step": 205
    },
    {
      "epoch": 0.014862378702067025,
      "grad_norm": 4.3798866229428794,
      "learning_rate": 3.5338446222005387e-06,
      "loss": 1.0089,
      "step": 206
    },
    {
      "epoch": 0.014934526171494536,
      "grad_norm": 5.075586241888964,
      "learning_rate": 3.5370566117286265e-06,
      "loss": 1.0603,
      "step": 207
    },
    {
      "epoch": 0.015006673640922044,
      "grad_norm": 3.026733809523219,
      "learning_rate": 3.5402531217290437e-06,
      "loss": 1.124,
      "step": 208
    },
    {
      "epoch": 0.015078821110349555,
      "grad_norm": 5.292017507061361,
      "learning_rate": 3.543434300687327e-06,
      "loss": 1.067,
      "step": 209
    },
    {
      "epoch": 0.015150968579777064,
      "grad_norm": 10.23429672560716,
      "learning_rate": 3.546600294962711e-06,
      "loss": 0.9818,
      "step": 210
    },
    {
      "epoch": 0.015223116049204575,
      "grad_norm": 4.039277593024617,
      "learning_rate": 3.549751248828536e-06,
      "loss": 1.0073,
      "step": 211
    },
    {
      "epoch": 0.015295263518632083,
      "grad_norm": 7.301230852010208,
      "learning_rate": 3.5528873045116943e-06,
      "loss": 1.0337,
      "step": 212
    },
    {
      "epoch": 0.015367410988059594,
      "grad_norm": 4.939240933374763,
      "learning_rate": 3.5560086022311574e-06,
      "loss": 1.0953,
      "step": 213
    },
    {
      "epoch": 0.015439558457487103,
      "grad_norm": 7.309148819364882,
      "learning_rate": 3.559115280235589e-06,
      "loss": 1.1759,
      "step": 214
    },
    {
      "epoch": 0.015511705926914613,
      "grad_norm": 4.091092477406421,
      "learning_rate": 3.562207474840094e-06,
      "loss": 1.055,
      "step": 215
    },
    {
      "epoch": 0.015583853396342124,
      "grad_norm": 6.389750451123793,
      "learning_rate": 3.5652853204621027e-06,
      "loss": 1.0122,
      "step": 216
    },
    {
      "epoch": 0.015656000865769633,
      "grad_norm": 0.8447796226110148,
      "learning_rate": 3.568348949656437e-06,
      "loss": 0.8685,
      "step": 217
    },
    {
      "epoch": 0.01572814833519714,
      "grad_norm": 5.316067204816268,
      "learning_rate": 3.5713984931495625e-06,
      "loss": 1.1381,
      "step": 218
    },
    {
      "epoch": 0.015800295804624654,
      "grad_norm": 5.575735055679458,
      "learning_rate": 3.574434079873061e-06,
      "loss": 1.0947,
      "step": 219
    },
    {
      "epoch": 0.015872443274052163,
      "grad_norm": 5.210805362244243,
      "learning_rate": 3.57745583699634e-06,
      "loss": 1.0452,
      "step": 220
    },
    {
      "epoch": 0.01594459074347967,
      "grad_norm": 5.804885963548697,
      "learning_rate": 3.5804638899585957e-06,
      "loss": 1.1113,
      "step": 221
    },
    {
      "epoch": 0.01601673821290718,
      "grad_norm": 4.584224660497677,
      "learning_rate": 3.5834583625000667e-06,
      "loss": 1.1196,
      "step": 222
    },
    {
      "epoch": 0.016088885682334693,
      "grad_norm": 0.7977093461286708,
      "learning_rate": 3.58643937669257e-06,
      "loss": 0.8684,
      "step": 223
    },
    {
      "epoch": 0.016161033151762202,
      "grad_norm": 5.081943050853595,
      "learning_rate": 3.5894070529693628e-06,
      "loss": 0.981,
      "step": 224
    },
    {
      "epoch": 0.01623318062118971,
      "grad_norm": 5.41529363200645,
      "learning_rate": 3.5923615101543413e-06,
      "loss": 1.0233,
      "step": 225
    },
    {
      "epoch": 0.016305328090617223,
      "grad_norm": 6.6971445859760825,
      "learning_rate": 3.5953028654905827e-06,
      "loss": 1.0226,
      "step": 226
    },
    {
      "epoch": 0.016377475560044732,
      "grad_norm": 3.3558154743010817,
      "learning_rate": 3.598231234668268e-06,
      "loss": 1.088,
      "step": 227
    },
    {
      "epoch": 0.01644962302947224,
      "grad_norm": 12.518624430887028,
      "learning_rate": 3.6011467318519812e-06,
      "loss": 1.1113,
      "step": 228
    },
    {
      "epoch": 0.01652177049889975,
      "grad_norm": 3.9159615605681903,
      "learning_rate": 3.6040494697074194e-06,
      "loss": 1.1687,
      "step": 229
    },
    {
      "epoch": 0.016593917968327262,
      "grad_norm": 5.166793862694744,
      "learning_rate": 3.6069395594275168e-06,
      "loss": 1.1345,
      "step": 230
    },
    {
      "epoch": 0.01666606543775477,
      "grad_norm": 3.4949627458717982,
      "learning_rate": 3.609817110758e-06,
      "loss": 1.0693,
      "step": 231
    },
    {
      "epoch": 0.01673821290718228,
      "grad_norm": 4.435771551787079,
      "learning_rate": 3.612682232022398e-06,
      "loss": 1.0389,
      "step": 232
    },
    {
      "epoch": 0.016810360376609792,
      "grad_norm": 5.568489022352961,
      "learning_rate": 3.6155350301465064e-06,
      "loss": 1.1106,
      "step": 233
    },
    {
      "epoch": 0.0168825078460373,
      "grad_norm": 3.3595359586258047,
      "learning_rate": 3.618375610682335e-06,
      "loss": 1.1569,
      "step": 234
    },
    {
      "epoch": 0.01695465531546481,
      "grad_norm": 3.2186360755684515,
      "learning_rate": 3.6212040778315378e-06,
      "loss": 1.0609,
      "step": 235
    },
    {
      "epoch": 0.01702680278489232,
      "grad_norm": 4.879650147826742,
      "learning_rate": 3.6240205344683526e-06,
      "loss": 1.0736,
      "step": 236
    },
    {
      "epoch": 0.01709895025431983,
      "grad_norm": 14.263725229471962,
      "learning_rate": 3.6268250821620504e-06,
      "loss": 1.0929,
      "step": 237
    },
    {
      "epoch": 0.01717109772374734,
      "grad_norm": 16.442836808903103,
      "learning_rate": 3.629617821198915e-06,
      "loss": 0.9739,
      "step": 238
    },
    {
      "epoch": 0.01724324519317485,
      "grad_norm": 5.401060579488142,
      "learning_rate": 3.6323988506037587e-06,
      "loss": 1.1041,
      "step": 239
    },
    {
      "epoch": 0.01731539266260236,
      "grad_norm": 7.7586351094897505,
      "learning_rate": 3.6351682681609926e-06,
      "loss": 1.1053,
      "step": 240
    },
    {
      "epoch": 0.01738754013202987,
      "grad_norm": 4.00829551715949,
      "learning_rate": 3.6379261704352583e-06,
      "loss": 1.0699,
      "step": 241
    },
    {
      "epoch": 0.01745968760145738,
      "grad_norm": 4.286061722432492,
      "learning_rate": 3.640672652791628e-06,
      "loss": 1.04,
      "step": 242
    },
    {
      "epoch": 0.017531835070884887,
      "grad_norm": 5.0550118155110235,
      "learning_rate": 3.6434078094153934e-06,
      "loss": 1.0486,
      "step": 243
    },
    {
      "epoch": 0.0176039825403124,
      "grad_norm": 3.984136117992137,
      "learning_rate": 3.6461317333314465e-06,
      "loss": 1.1559,
      "step": 244
    },
    {
      "epoch": 0.01767613000973991,
      "grad_norm": 2.441082693403469,
      "learning_rate": 3.6488445164232626e-06,
      "loss": 1.0325,
      "step": 245
    },
    {
      "epoch": 0.017748277479167417,
      "grad_norm": 6.058126858319669,
      "learning_rate": 3.6515462494514983e-06,
      "loss": 1.0651,
      "step": 246
    },
    {
      "epoch": 0.01782042494859493,
      "grad_norm": 3.529356997663394,
      "learning_rate": 3.654237022072213e-06,
      "loss": 1.0702,
      "step": 247
    },
    {
      "epoch": 0.01789257241802244,
      "grad_norm": 8.30551932790192,
      "learning_rate": 3.6569169228547184e-06,
      "loss": 1.0855,
      "step": 248
    },
    {
      "epoch": 0.017964719887449947,
      "grad_norm": 4.682911368985165,
      "learning_rate": 3.659586039299075e-06,
      "loss": 0.9635,
      "step": 249
    },
    {
      "epoch": 0.018036867356877456,
      "grad_norm": 6.535131441879841,
      "learning_rate": 3.662244457853231e-06,
      "loss": 1.2063,
      "step": 250
    },
    {
      "epoch": 0.01810901482630497,
      "grad_norm": 3.136583603243229,
      "learning_rate": 3.664892263929826e-06,
      "loss": 1.0376,
      "step": 251
    },
    {
      "epoch": 0.018181162295732477,
      "grad_norm": 6.91908445575067,
      "learning_rate": 3.667529541922654e-06,
      "loss": 1.0929,
      "step": 252
    },
    {
      "epoch": 0.018253309765159986,
      "grad_norm": 4.707494790453669,
      "learning_rate": 3.6701563752228055e-06,
      "loss": 1.0913,
      "step": 253
    },
    {
      "epoch": 0.018325457234587495,
      "grad_norm": 5.791817240049851,
      "learning_rate": 3.6727728462344894e-06,
      "loss": 0.9306,
      "step": 254
    },
    {
      "epoch": 0.018397604704015007,
      "grad_norm": 4.0913108955884,
      "learning_rate": 3.675379036390544e-06,
      "loss": 1.0395,
      "step": 255
    },
    {
      "epoch": 0.018469752173442516,
      "grad_norm": 1.026095251759301,
      "learning_rate": 3.6779750261676444e-06,
      "loss": 0.953,
      "step": 256
    },
    {
      "epoch": 0.018541899642870025,
      "grad_norm": 2.8903531333301755,
      "learning_rate": 3.6805608951012116e-06,
      "loss": 0.9763,
      "step": 257
    },
    {
      "epoch": 0.018614047112297537,
      "grad_norm": 5.6532053472018715,
      "learning_rate": 3.6831367218000362e-06,
      "loss": 1.0298,
      "step": 258
    },
    {
      "epoch": 0.018686194581725046,
      "grad_norm": 2.930874108652824,
      "learning_rate": 3.6857025839606174e-06,
      "loss": 1.0082,
      "step": 259
    },
    {
      "epoch": 0.018758342051152555,
      "grad_norm": 0.9363342666901984,
      "learning_rate": 3.6882585583812243e-06,
      "loss": 0.9393,
      "step": 260
    },
    {
      "epoch": 0.018830489520580064,
      "grad_norm": 10.800808092190877,
      "learning_rate": 3.6908047209756893e-06,
      "loss": 1.1176,
      "step": 261
    },
    {
      "epoch": 0.018902636990007576,
      "grad_norm": 4.098709014845883,
      "learning_rate": 3.6933411467869324e-06,
      "loss": 1.0657,
      "step": 262
    },
    {
      "epoch": 0.018974784459435085,
      "grad_norm": 3.8636520099833067,
      "learning_rate": 3.6958679100002413e-06,
      "loss": 1.0334,
      "step": 263
    },
    {
      "epoch": 0.019046931928862594,
      "grad_norm": 3.1776707507459743,
      "learning_rate": 3.6983850839562814e-06,
      "loss": 1.0188,
      "step": 264
    },
    {
      "epoch": 0.019119079398290106,
      "grad_norm": 3.924150270460321,
      "learning_rate": 3.7008927411638757e-06,
      "loss": 1.0496,
      "step": 265
    },
    {
      "epoch": 0.019191226867717615,
      "grad_norm": 3.648151133242564,
      "learning_rate": 3.7033909533125315e-06,
      "loss": 1.1293,
      "step": 266
    },
    {
      "epoch": 0.019263374337145124,
      "grad_norm": 3.079178528167082,
      "learning_rate": 3.7058797912847445e-06,
      "loss": 1.1243,
      "step": 267
    },
    {
      "epoch": 0.019335521806572633,
      "grad_norm": 3.6883291831656124,
      "learning_rate": 3.7083593251680595e-06,
      "loss": 1.0982,
      "step": 268
    },
    {
      "epoch": 0.019407669276000145,
      "grad_norm": 3.4284222121919683,
      "learning_rate": 3.7108296242669196e-06,
      "loss": 1.0004,
      "step": 269
    },
    {
      "epoch": 0.019479816745427654,
      "grad_norm": 3.247757504820769,
      "learning_rate": 3.7132907571142837e-06,
      "loss": 0.939,
      "step": 270
    },
    {
      "epoch": 0.019551964214855163,
      "grad_norm": 3.508911640288628,
      "learning_rate": 3.7157427914830385e-06,
      "loss": 1.0607,
      "step": 271
    },
    {
      "epoch": 0.019624111684282675,
      "grad_norm": 2.66746710645278,
      "learning_rate": 3.7181857943971964e-06,
      "loss": 1.1666,
      "step": 272
    },
    {
      "epoch": 0.019696259153710184,
      "grad_norm": 4.332424190714263,
      "learning_rate": 3.7206198321428855e-06,
      "loss": 1.0737,
      "step": 273
    },
    {
      "epoch": 0.019768406623137693,
      "grad_norm": 3.6789507280157316,
      "learning_rate": 3.723044970279146e-06,
      "loss": 1.0323,
      "step": 274
    },
    {
      "epoch": 0.019840554092565202,
      "grad_norm": 2.879399954752078,
      "learning_rate": 3.7254612736485204e-06,
      "loss": 0.8855,
      "step": 275
    },
    {
      "epoch": 0.019912701561992714,
      "grad_norm": 7.663358167973394,
      "learning_rate": 3.7278688063874583e-06,
      "loss": 1.1185,
      "step": 276
    },
    {
      "epoch": 0.019984849031420223,
      "grad_norm": 3.225333878141654,
      "learning_rate": 3.730267631936535e-06,
      "loss": 1.0552,
      "step": 277
    },
    {
      "epoch": 0.020056996500847732,
      "grad_norm": 2.763528320034873,
      "learning_rate": 3.732657813050475e-06,
      "loss": 1.0876,
      "step": 278
    },
    {
      "epoch": 0.020129143970275244,
      "grad_norm": 4.251230622684563,
      "learning_rate": 3.735039411808009e-06,
      "loss": 0.8315,
      "step": 279
    },
    {
      "epoch": 0.020201291439702753,
      "grad_norm": 4.039475912338419,
      "learning_rate": 3.7374124896215433e-06,
      "loss": 0.9203,
      "step": 280
    },
    {
      "epoch": 0.020273438909130262,
      "grad_norm": 7.607705687099968,
      "learning_rate": 3.739777107246667e-06,
      "loss": 1.0256,
      "step": 281
    },
    {
      "epoch": 0.02034558637855777,
      "grad_norm": 3.5977437260334275,
      "learning_rate": 3.7421333247914798e-06,
      "loss": 1.1482,
      "step": 282
    },
    {
      "epoch": 0.020417733847985283,
      "grad_norm": 4.231585607925491,
      "learning_rate": 3.7444812017257643e-06,
      "loss": 1.1564,
      "step": 283
    },
    {
      "epoch": 0.020489881317412792,
      "grad_norm": 4.58849153119632,
      "learning_rate": 3.7468207968899897e-06,
      "loss": 1.0439,
      "step": 284
    },
    {
      "epoch": 0.0205620287868403,
      "grad_norm": 4.367907415296097,
      "learning_rate": 3.749152168504162e-06,
      "loss": 1.0567,
      "step": 285
    },
    {
      "epoch": 0.020634176256267813,
      "grad_norm": 4.979792578180752,
      "learning_rate": 3.7514753741765134e-06,
      "loss": 1.0272,
      "step": 286
    },
    {
      "epoch": 0.020706323725695322,
      "grad_norm": 0.9540961274194286,
      "learning_rate": 3.7537904709120494e-06,
      "loss": 0.8618,
      "step": 287
    },
    {
      "epoch": 0.02077847119512283,
      "grad_norm": 7.569942374772284,
      "learning_rate": 3.756097515120935e-06,
      "loss": 1.074,
      "step": 288
    },
    {
      "epoch": 0.02085061866455034,
      "grad_norm": 4.335312366957172,
      "learning_rate": 3.758396562626748e-06,
      "loss": 1.1115,
      "step": 289
    },
    {
      "epoch": 0.020922766133977852,
      "grad_norm": 3.410500770111288,
      "learning_rate": 3.760687668674579e-06,
      "loss": 1.0044,
      "step": 290
    },
    {
      "epoch": 0.02099491360340536,
      "grad_norm": 5.720387375525736,
      "learning_rate": 3.7629708879390006e-06,
      "loss": 1.0303,
      "step": 291
    },
    {
      "epoch": 0.02106706107283287,
      "grad_norm": 0.9435651408654612,
      "learning_rate": 3.7652462745318934e-06,
      "loss": 0.9138,
      "step": 292
    },
    {
      "epoch": 0.02113920854226038,
      "grad_norm": 3.12009838976145,
      "learning_rate": 3.767513882010144e-06,
      "loss": 1.0997,
      "step": 293
    },
    {
      "epoch": 0.02121135601168789,
      "grad_norm": 5.797361944676054,
      "learning_rate": 3.7697737633832046e-06,
      "loss": 1.0963,
      "step": 294
    },
    {
      "epoch": 0.0212835034811154,
      "grad_norm": 4.621516178586896,
      "learning_rate": 3.772025971120533e-06,
      "loss": 1.0076,
      "step": 295
    },
    {
      "epoch": 0.02135565095054291,
      "grad_norm": 0.7744753261352078,
      "learning_rate": 3.774270557158899e-06,
      "loss": 0.8966,
      "step": 296
    },
    {
      "epoch": 0.02142779841997042,
      "grad_norm": 3.465435334528736,
      "learning_rate": 3.7765075729095725e-06,
      "loss": 1.043,
      "step": 297
    },
    {
      "epoch": 0.02149994588939793,
      "grad_norm": 3.8004663382034294,
      "learning_rate": 3.7787370692653873e-06,
      "loss": 1.082,
      "step": 298
    },
    {
      "epoch": 0.02157209335882544,
      "grad_norm": 3.8866955627056465,
      "learning_rate": 3.7809590966076912e-06,
      "loss": 0.931,
      "step": 299
    },
    {
      "epoch": 0.021644240828252947,
      "grad_norm": 4.118971378472398,
      "learning_rate": 3.7831737048131736e-06,
      "loss": 1.0143,
      "step": 300
    },
    {
      "epoch": 0.02171638829768046,
      "grad_norm": 5.344711306477176,
      "learning_rate": 3.785380943260587e-06,
      "loss": 1.0383,
      "step": 301
    },
    {
      "epoch": 0.02178853576710797,
      "grad_norm": 0.8440680434247272,
      "learning_rate": 3.7875808608373505e-06,
      "loss": 0.8414,
      "step": 302
    },
    {
      "epoch": 0.021860683236535478,
      "grad_norm": 3.827300313752553,
      "learning_rate": 3.7897735059460465e-06,
      "loss": 1.0206,
      "step": 303
    },
    {
      "epoch": 0.02193283070596299,
      "grad_norm": 3.5040328089079473,
      "learning_rate": 3.7919589265108135e-06,
      "loss": 1.0635,
      "step": 304
    },
    {
      "epoch": 0.0220049781753905,
      "grad_norm": 3.0619752395396196,
      "learning_rate": 3.7941371699836275e-06,
      "loss": 0.993,
      "step": 305
    },
    {
      "epoch": 0.022077125644818008,
      "grad_norm": 5.683150420317165,
      "learning_rate": 3.7963082833504866e-06,
      "loss": 1.0603,
      "step": 306
    },
    {
      "epoch": 0.022149273114245516,
      "grad_norm": 3.878376013096513,
      "learning_rate": 3.798472313137493e-06,
      "loss": 1.1454,
      "step": 307
    },
    {
      "epoch": 0.02222142058367303,
      "grad_norm": 3.6249604058569744,
      "learning_rate": 3.8006293054168325e-06,
      "loss": 1.0466,
      "step": 308
    },
    {
      "epoch": 0.022293568053100538,
      "grad_norm": 3.6758671263048974,
      "learning_rate": 3.802779305812662e-06,
      "loss": 1.0393,
      "step": 309
    },
    {
      "epoch": 0.022365715522528046,
      "grad_norm": 3.7376995530966304,
      "learning_rate": 3.8049223595068993e-06,
      "loss": 0.9058,
      "step": 310
    },
    {
      "epoch": 0.02243786299195556,
      "grad_norm": 3.25384543408991,
      "learning_rate": 3.8070585112449207e-06,
      "loss": 1.0442,
      "step": 311
    },
    {
      "epoch": 0.022510010461383068,
      "grad_norm": 4.327037247675762,
      "learning_rate": 3.809187805341167e-06,
      "loss": 1.1355,
      "step": 312
    },
    {
      "epoch": 0.022582157930810576,
      "grad_norm": 3.4481406854573877,
      "learning_rate": 3.8113102856846578e-06,
      "loss": 1.0259,
      "step": 313
    },
    {
      "epoch": 0.022654305400238085,
      "grad_norm": 4.058763399516157,
      "learning_rate": 3.813425995744421e-06,
      "loss": 1.0721,
      "step": 314
    },
    {
      "epoch": 0.022726452869665598,
      "grad_norm": 3.4543914543851852,
      "learning_rate": 3.8155349785748344e-06,
      "loss": 1.0529,
      "step": 315
    },
    {
      "epoch": 0.022798600339093106,
      "grad_norm": 2.6487627307969355,
      "learning_rate": 3.817637276820883e-06,
      "loss": 1.0054,
      "step": 316
    },
    {
      "epoch": 0.022870747808520615,
      "grad_norm": 3.9376456824100665,
      "learning_rate": 3.81973293272333e-06,
      "loss": 1.0649,
      "step": 317
    },
    {
      "epoch": 0.022942895277948128,
      "grad_norm": 2.918631972350549,
      "learning_rate": 3.821821988123818e-06,
      "loss": 1.1264,
      "step": 318
    },
    {
      "epoch": 0.023015042747375637,
      "grad_norm": 4.668922946080865,
      "learning_rate": 3.8239044844698675e-06,
      "loss": 1.0036,
      "step": 319
    },
    {
      "epoch": 0.023087190216803145,
      "grad_norm": 2.7053745148715644,
      "learning_rate": 3.8259804628198245e-06,
      "loss": 1.0566,
      "step": 320
    },
    {
      "epoch": 0.023159337686230654,
      "grad_norm": 3.2473112868759704,
      "learning_rate": 3.828049963847712e-06,
      "loss": 1.0699,
      "step": 321
    },
    {
      "epoch": 0.023231485155658167,
      "grad_norm": 3.1920247593352067,
      "learning_rate": 3.8301130278480095e-06,
      "loss": 1.0035,
      "step": 322
    },
    {
      "epoch": 0.023303632625085675,
      "grad_norm": 3.7856068377400005,
      "learning_rate": 3.832169694740365e-06,
      "loss": 1.1639,
      "step": 323
    },
    {
      "epoch": 0.023375780094513184,
      "grad_norm": 2.492504805213023,
      "learning_rate": 3.834220004074225e-06,
      "loss": 1.0912,
      "step": 324
    },
    {
      "epoch": 0.023447927563940693,
      "grad_norm": 3.640088086791589,
      "learning_rate": 3.836263995033405e-06,
      "loss": 0.8973,
      "step": 325
    },
    {
      "epoch": 0.023520075033368205,
      "grad_norm": 4.121155305375592,
      "learning_rate": 3.838301706440575e-06,
      "loss": 1.1729,
      "step": 326
    },
    {
      "epoch": 0.023592222502795714,
      "grad_norm": 3.3803208405394525,
      "learning_rate": 3.8403331767616854e-06,
      "loss": 1.0594,
      "step": 327
    },
    {
      "epoch": 0.023664369972223223,
      "grad_norm": 1.1344316158405834,
      "learning_rate": 3.8423584441103306e-06,
      "loss": 0.8383,
      "step": 328
    },
    {
      "epoch": 0.023736517441650735,
      "grad_norm": 3.241871572521745,
      "learning_rate": 3.844377546252031e-06,
      "loss": 0.9758,
      "step": 329
    },
    {
      "epoch": 0.023808664911078244,
      "grad_norm": 3.73310492077953,
      "learning_rate": 3.846390520608462e-06,
      "loss": 1.1328,
      "step": 330
    },
    {
      "epoch": 0.023880812380505753,
      "grad_norm": 4.667721742614859,
      "learning_rate": 3.848397404261619e-06,
      "loss": 1.0195,
      "step": 331
    },
    {
      "epoch": 0.023952959849933262,
      "grad_norm": 3.048770256397478,
      "learning_rate": 3.850398233957907e-06,
      "loss": 1.0622,
      "step": 332
    },
    {
      "epoch": 0.024025107319360774,
      "grad_norm": 3.5155734588235767,
      "learning_rate": 3.85239304611219e-06,
      "loss": 1.0353,
      "step": 333
    },
    {
      "epoch": 0.024097254788788283,
      "grad_norm": 3.1533183466721186,
      "learning_rate": 3.854381876811756e-06,
      "loss": 1.0974,
      "step": 334
    },
    {
      "epoch": 0.024169402258215792,
      "grad_norm": 3.571940712311797,
      "learning_rate": 3.856364761820241e-06,
      "loss": 1.0465,
      "step": 335
    },
    {
      "epoch": 0.024241549727643304,
      "grad_norm": 4.7478695157253945,
      "learning_rate": 3.858341736581486e-06,
      "loss": 0.9768,
      "step": 336
    },
    {
      "epoch": 0.024313697197070813,
      "grad_norm": 2.254306803899411,
      "learning_rate": 3.860312836223339e-06,
      "loss": 1.0622,
      "step": 337
    },
    {
      "epoch": 0.024385844666498322,
      "grad_norm": 3.649536098106755,
      "learning_rate": 3.862278095561399e-06,
      "loss": 1.0554,
      "step": 338
    },
    {
      "epoch": 0.02445799213592583,
      "grad_norm": 3.5674044662975444,
      "learning_rate": 3.864237549102705e-06,
      "loss": 1.132,
      "step": 339
    },
    {
      "epoch": 0.024530139605353343,
      "grad_norm": 3.1169471743687245,
      "learning_rate": 3.866191231049377e-06,
      "loss": 1.0338,
      "step": 340
    },
    {
      "epoch": 0.024602287074780852,
      "grad_norm": 5.820227623647073,
      "learning_rate": 3.868139175302188e-06,
      "loss": 1.1257,
      "step": 341
    },
    {
      "epoch": 0.02467443454420836,
      "grad_norm": 10.132715370342106,
      "learning_rate": 3.870081415464103e-06,
      "loss": 1.0521,
      "step": 342
    },
    {
      "epoch": 0.024746582013635873,
      "grad_norm": 3.193690702182742,
      "learning_rate": 3.872017984843755e-06,
      "loss": 0.9714,
      "step": 343
    },
    {
      "epoch": 0.024818729483063382,
      "grad_norm": 3.6113703773217813,
      "learning_rate": 3.8739489164588685e-06,
      "loss": 1.0034,
      "step": 344
    },
    {
      "epoch": 0.02489087695249089,
      "grad_norm": 5.513493544815808,
      "learning_rate": 3.87587424303964e-06,
      "loss": 1.0456,
      "step": 345
    },
    {
      "epoch": 0.0249630244219184,
      "grad_norm": 0.7221408965302605,
      "learning_rate": 3.877793997032067e-06,
      "loss": 0.8156,
      "step": 346
    },
    {
      "epoch": 0.025035171891345912,
      "grad_norm": 4.042155337860066,
      "learning_rate": 3.879708210601229e-06,
      "loss": 0.9501,
      "step": 347
    },
    {
      "epoch": 0.02510731936077342,
      "grad_norm": 9.952162571821113,
      "learning_rate": 3.881616915634521e-06,
      "loss": 1.0378,
      "step": 348
    },
    {
      "epoch": 0.02517946683020093,
      "grad_norm": 5.1727519141867475,
      "learning_rate": 3.883520143744839e-06,
      "loss": 0.9835,
      "step": 349
    },
    {
      "epoch": 0.025251614299628442,
      "grad_norm": 5.781015752222217,
      "learning_rate": 3.885417926273724e-06,
      "loss": 1.1563,
      "step": 350
    },
    {
      "epoch": 0.02532376176905595,
      "grad_norm": 3.813966812877564,
      "learning_rate": 3.887310294294458e-06,
      "loss": 1.1213,
      "step": 351
    },
    {
      "epoch": 0.02539590923848346,
      "grad_norm": 6.196115064169462,
      "learning_rate": 3.889197278615114e-06,
      "loss": 1.1189,
      "step": 352
    },
    {
      "epoch": 0.02546805670791097,
      "grad_norm": 3.6366167392211404,
      "learning_rate": 3.89107890978157e-06,
      "loss": 1.0289,
      "step": 353
    },
    {
      "epoch": 0.02554020417733848,
      "grad_norm": 4.620328038447197,
      "learning_rate": 3.8929552180804756e-06,
      "loss": 1.0654,
      "step": 354
    },
    {
      "epoch": 0.02561235164676599,
      "grad_norm": 0.8167823399801645,
      "learning_rate": 3.894826233542171e-06,
      "loss": 0.9048,
      "step": 355
    },
    {
      "epoch": 0.0256844991161935,
      "grad_norm": 4.266882854190145,
      "learning_rate": 3.896691985943577e-06,
      "loss": 1.0412,
      "step": 356
    },
    {
      "epoch": 0.025756646585621008,
      "grad_norm": 10.529952062658905,
      "learning_rate": 3.898552504811037e-06,
      "loss": 1.0566,
      "step": 357
    },
    {
      "epoch": 0.02582879405504852,
      "grad_norm": 7.332916190102967,
      "learning_rate": 3.900407819423122e-06,
      "loss": 0.863,
      "step": 358
    },
    {
      "epoch": 0.02590094152447603,
      "grad_norm": 13.577964860786103,
      "learning_rate": 3.90225795881339e-06,
      "loss": 1.0759,
      "step": 359
    },
    {
      "epoch": 0.025973088993903538,
      "grad_norm": 4.341671892530306,
      "learning_rate": 3.9041029517731164e-06,
      "loss": 1.0885,
      "step": 360
    },
    {
      "epoch": 0.02604523646333105,
      "grad_norm": 7.79812373556396,
      "learning_rate": 3.9059428268539815e-06,
      "loss": 1.0721,
      "step": 361
    },
    {
      "epoch": 0.02611738393275856,
      "grad_norm": 5.762615793516987,
      "learning_rate": 3.907777612370725e-06,
      "loss": 1.0833,
      "step": 362
    },
    {
      "epoch": 0.026189531402186068,
      "grad_norm": 9.381561424580857,
      "learning_rate": 3.909607336403751e-06,
      "loss": 1.0675,
      "step": 363
    },
    {
      "epoch": 0.026261678871613577,
      "grad_norm": 5.568072838504153,
      "learning_rate": 3.911432026801718e-06,
      "loss": 1.1295,
      "step": 364
    },
    {
      "epoch": 0.02633382634104109,
      "grad_norm": 6.128006800436126,
      "learning_rate": 3.913251711184074e-06,
      "loss": 1.171,
      "step": 365
    },
    {
      "epoch": 0.026405973810468598,
      "grad_norm": 1.1871349877986397,
      "learning_rate": 3.9150664169435694e-06,
      "loss": 0.8584,
      "step": 366
    },
    {
      "epoch": 0.026478121279896107,
      "grad_norm": 8.73983304769949,
      "learning_rate": 3.9168761712487285e-06,
      "loss": 1.1095,
      "step": 367
    },
    {
      "epoch": 0.02655026874932362,
      "grad_norm": 5.659722879587875,
      "learning_rate": 3.9186810010462915e-06,
      "loss": 1.053,
      "step": 368
    },
    {
      "epoch": 0.026622416218751128,
      "grad_norm": 4.060044705750826,
      "learning_rate": 3.920480933063622e-06,
      "loss": 1.1219,
      "step": 369
    },
    {
      "epoch": 0.026694563688178637,
      "grad_norm": 3.4758622396494654,
      "learning_rate": 3.92227599381108e-06,
      "loss": 1.049,
      "step": 370
    },
    {
      "epoch": 0.026766711157606145,
      "grad_norm": 4.908670399377176,
      "learning_rate": 3.924066209584369e-06,
      "loss": 1.0419,
      "step": 371
    },
    {
      "epoch": 0.026838858627033658,
      "grad_norm": 0.8299588465541962,
      "learning_rate": 3.925851606466841e-06,
      "loss": 0.8975,
      "step": 372
    },
    {
      "epoch": 0.026911006096461167,
      "grad_norm": 3.932820764107825,
      "learning_rate": 3.9276322103317845e-06,
      "loss": 0.9529,
      "step": 373
    },
    {
      "epoch": 0.026983153565888676,
      "grad_norm": 4.3255867797062555,
      "learning_rate": 3.929408046844666e-06,
      "loss": 1.0773,
      "step": 374
    },
    {
      "epoch": 0.027055301035316188,
      "grad_norm": 6.869020948541641,
      "learning_rate": 3.931179141465354e-06,
      "loss": 1.137,
      "step": 375
    },
    {
      "epoch": 0.027127448504743697,
      "grad_norm": 7.500989975760205,
      "learning_rate": 3.932945519450312e-06,
      "loss": 1.0264,
      "step": 376
    },
    {
      "epoch": 0.027199595974171206,
      "grad_norm": 6.138734327454349,
      "learning_rate": 3.934707205854753e-06,
      "loss": 1.0304,
      "step": 377
    },
    {
      "epoch": 0.027271743443598714,
      "grad_norm": 7.318280144520408,
      "learning_rate": 3.936464225534777e-06,
      "loss": 1.0404,
      "step": 378
    },
    {
      "epoch": 0.027343890913026227,
      "grad_norm": 9.835980549324974,
      "learning_rate": 3.9382166031494725e-06,
      "loss": 1.0685,
      "step": 379
    },
    {
      "epoch": 0.027416038382453736,
      "grad_norm": 8.31024368755119,
      "learning_rate": 3.939964363162994e-06,
      "loss": 1.055,
      "step": 380
    },
    {
      "epoch": 0.027488185851881244,
      "grad_norm": 16.553110361695907,
      "learning_rate": 3.941707529846612e-06,
      "loss": 1.0434,
      "step": 381
    },
    {
      "epoch": 0.027560333321308757,
      "grad_norm": 11.96285794000637,
      "learning_rate": 3.943446127280736e-06,
      "loss": 1.1239,
      "step": 382
    },
    {
      "epoch": 0.027632480790736266,
      "grad_norm": 7.659835287370002,
      "learning_rate": 3.945180179356906e-06,
      "loss": 1.0008,
      "step": 383
    },
    {
      "epoch": 0.027704628260163774,
      "grad_norm": 4.913837955360156,
      "learning_rate": 3.946909709779767e-06,
      "loss": 1.1052,
      "step": 384
    },
    {
      "epoch": 0.027776775729591283,
      "grad_norm": 0.7328689228126658,
      "learning_rate": 3.9486347420690135e-06,
      "loss": 0.8443,
      "step": 385
    },
    {
      "epoch": 0.027848923199018796,
      "grad_norm": 4.88376670363933,
      "learning_rate": 3.950355299561303e-06,
      "loss": 1.0363,
      "step": 386
    },
    {
      "epoch": 0.027921070668446304,
      "grad_norm": 10.607775560585264,
      "learning_rate": 3.95207140541216e-06,
      "loss": 1.0364,
      "step": 387
    },
    {
      "epoch": 0.027993218137873813,
      "grad_norm": 7.472927058687438,
      "learning_rate": 3.953783082597833e-06,
      "loss": 1.0118,
      "step": 388
    },
    {
      "epoch": 0.028065365607301322,
      "grad_norm": 5.598988046493347,
      "learning_rate": 3.955490353917153e-06,
      "loss": 1.0911,
      "step": 389
    },
    {
      "epoch": 0.028137513076728835,
      "grad_norm": 5.8881254281077355,
      "learning_rate": 3.957193241993348e-06,
      "loss": 1.0266,
      "step": 390
    },
    {
      "epoch": 0.028209660546156343,
      "grad_norm": 6.132199163199809,
      "learning_rate": 3.9588917692758435e-06,
      "loss": 1.0873,
      "step": 391
    },
    {
      "epoch": 0.028281808015583852,
      "grad_norm": 6.796655550516245,
      "learning_rate": 3.9605859580420365e-06,
      "loss": 1.0502,
      "step": 392
    },
    {
      "epoch": 0.028353955485011365,
      "grad_norm": 6.263413036435505,
      "learning_rate": 3.962275830399056e-06,
      "loss": 0.996,
      "step": 393
    },
    {
      "epoch": 0.028426102954438873,
      "grad_norm": 5.585992815504087,
      "learning_rate": 3.963961408285482e-06,
      "loss": 1.0228,
      "step": 394
    },
    {
      "epoch": 0.028498250423866382,
      "grad_norm": 5.437771319054798,
      "learning_rate": 3.965642713473064e-06,
      "loss": 1.0459,
      "step": 395
    },
    {
      "epoch": 0.02857039789329389,
      "grad_norm": 5.956601226794423,
      "learning_rate": 3.967319767568405e-06,
      "loss": 1.0477,
      "step": 396
    },
    {
      "epoch": 0.028642545362721403,
      "grad_norm": 5.448087185099045,
      "learning_rate": 3.968992592014627e-06,
      "loss": 1.0223,
      "step": 397
    },
    {
      "epoch": 0.028714692832148912,
      "grad_norm": 7.398310143604216,
      "learning_rate": 3.9706612080930185e-06,
      "loss": 1.1048,
      "step": 398
    },
    {
      "epoch": 0.02878684030157642,
      "grad_norm": 7.832170548948261,
      "learning_rate": 3.972325636924655e-06,
      "loss": 1.0044,
      "step": 399
    },
    {
      "epoch": 0.028858987771003933,
      "grad_norm": 8.687895621752869,
      "learning_rate": 3.973985899472006e-06,
      "loss": 0.9821,
      "step": 400
    },
    {
      "epoch": 0.028931135240431442,
      "grad_norm": 5.156464543358566,
      "learning_rate": 3.975642016540519e-06,
      "loss": 1.0869,
      "step": 401
    },
    {
      "epoch": 0.02900328270985895,
      "grad_norm": 8.188025537761302,
      "learning_rate": 3.977294008780183e-06,
      "loss": 1.0753,
      "step": 402
    },
    {
      "epoch": 0.02907543017928646,
      "grad_norm": 7.139738132151845,
      "learning_rate": 3.978941896687073e-06,
      "loss": 0.959,
      "step": 403
    },
    {
      "epoch": 0.029147577648713972,
      "grad_norm": 9.34820552476266,
      "learning_rate": 3.98058570060488e-06,
      "loss": 1.0045,
      "step": 404
    },
    {
      "epoch": 0.02921972511814148,
      "grad_norm": 9.45431980117195,
      "learning_rate": 3.982225440726407e-06,
      "loss": 1.0526,
      "step": 405
    },
    {
      "epoch": 0.02929187258756899,
      "grad_norm": 3.8840996740577007,
      "learning_rate": 3.983861137095072e-06,
      "loss": 1.0443,
      "step": 406
    },
    {
      "epoch": 0.029364020056996502,
      "grad_norm": 5.279331739213865,
      "learning_rate": 3.985492809606368e-06,
      "loss": 0.9679,
      "step": 407
    },
    {
      "epoch": 0.02943616752642401,
      "grad_norm": 4.346348705558782,
      "learning_rate": 3.987120478009319e-06,
      "loss": 1.1023,
      "step": 408
    },
    {
      "epoch": 0.02950831499585152,
      "grad_norm": 6.420778352608063,
      "learning_rate": 3.988744161907911e-06,
      "loss": 1.0446,
      "step": 409
    },
    {
      "epoch": 0.02958046246527903,
      "grad_norm": 5.524961010549379,
      "learning_rate": 3.990363880762511e-06,
      "loss": 1.0278,
      "step": 410
    },
    {
      "epoch": 0.02965260993470654,
      "grad_norm": 8.919944754801508,
      "learning_rate": 3.991979653891269e-06,
      "loss": 1.0193,
      "step": 411
    },
    {
      "epoch": 0.02972475740413405,
      "grad_norm": 1.2439335481551868,
      "learning_rate": 3.993591500471494e-06,
      "loss": 0.8622,
      "step": 412
    },
    {
      "epoch": 0.02979690487356156,
      "grad_norm": 3.699654542415458,
      "learning_rate": 3.995199439541027e-06,
      "loss": 1.0491,
      "step": 413
    },
    {
      "epoch": 0.02986905234298907,
      "grad_norm": 4.243367573694096,
      "learning_rate": 3.996803489999582e-06,
      "loss": 1.1008,
      "step": 414
    },
    {
      "epoch": 0.02994119981241658,
      "grad_norm": 7.808281331684734,
      "learning_rate": 3.998403670610088e-06,
      "loss": 1.2091,
      "step": 415
    },
    {
      "epoch": 0.03001334728184409,
      "grad_norm": 10.298480127287176,
      "learning_rate": 3.999999999999999e-06,
      "loss": 1.062,
      "step": 416
    },
    {
      "epoch": 0.030085494751271598,
      "grad_norm": 5.594634939766216,
      "learning_rate": 3.999999945393702e-06,
      "loss": 1.0604,
      "step": 417
    },
    {
      "epoch": 0.03015764222069911,
      "grad_norm": 5.312345435194058,
      "learning_rate": 3.999999781574811e-06,
      "loss": 0.9797,
      "step": 418
    },
    {
      "epoch": 0.03022978969012662,
      "grad_norm": 0.9015653811826897,
      "learning_rate": 3.9999995085433345e-06,
      "loss": 0.9574,
      "step": 419
    },
    {
      "epoch": 0.030301937159554128,
      "grad_norm": 8.194229272578479,
      "learning_rate": 3.999999126299289e-06,
      "loss": 1.0855,
      "step": 420
    },
    {
      "epoch": 0.03037408462898164,
      "grad_norm": 25.667316907432348,
      "learning_rate": 3.999998634842696e-06,
      "loss": 1.094,
      "step": 421
    },
    {
      "epoch": 0.03044623209840915,
      "grad_norm": 6.6672235574187555,
      "learning_rate": 3.99999803417358e-06,
      "loss": 1.0629,
      "step": 422
    },
    {
      "epoch": 0.030518379567836658,
      "grad_norm": 5.851230016004782,
      "learning_rate": 3.9999973242919755e-06,
      "loss": 1.0372,
      "step": 423
    },
    {
      "epoch": 0.030590527037264167,
      "grad_norm": 6.739280100333563,
      "learning_rate": 3.999996505197922e-06,
      "loss": 1.0197,
      "step": 424
    },
    {
      "epoch": 0.03066267450669168,
      "grad_norm": 3.3728513589198794,
      "learning_rate": 3.999995576891461e-06,
      "loss": 1.1059,
      "step": 425
    },
    {
      "epoch": 0.030734821976119188,
      "grad_norm": 15.579981043832237,
      "learning_rate": 3.999994539372647e-06,
      "loss": 1.1286,
      "step": 426
    },
    {
      "epoch": 0.030806969445546697,
      "grad_norm": 5.179342819414026,
      "learning_rate": 3.999993392641535e-06,
      "loss": 0.887,
      "step": 427
    },
    {
      "epoch": 0.030879116914974206,
      "grad_norm": 7.284776611254384,
      "learning_rate": 3.9999921366981865e-06,
      "loss": 1.0542,
      "step": 428
    },
    {
      "epoch": 0.030951264384401718,
      "grad_norm": 6.703462463278364,
      "learning_rate": 3.9999907715426715e-06,
      "loss": 1.1707,
      "step": 429
    },
    {
      "epoch": 0.031023411853829227,
      "grad_norm": 0.7359428308378907,
      "learning_rate": 3.999989297175064e-06,
      "loss": 0.8856,
      "step": 430
    },
    {
      "epoch": 0.031095559323256736,
      "grad_norm": 8.772342603374344,
      "learning_rate": 3.999987713595445e-06,
      "loss": 1.0974,
      "step": 431
    },
    {
      "epoch": 0.031167706792684248,
      "grad_norm": 7.373816443695388,
      "learning_rate": 3.999986020803901e-06,
      "loss": 1.0768,
      "step": 432
    },
    {
      "epoch": 0.031239854262111757,
      "grad_norm": 8.845360922600854,
      "learning_rate": 3.999984218800523e-06,
      "loss": 1.1194,
      "step": 433
    },
    {
      "epoch": 0.031312001731539266,
      "grad_norm": 4.738760047364608,
      "learning_rate": 3.999982307585412e-06,
      "loss": 1.055,
      "step": 434
    },
    {
      "epoch": 0.03138414920096678,
      "grad_norm": 8.197202568184323,
      "learning_rate": 3.99998028715867e-06,
      "loss": 1.0339,
      "step": 435
    },
    {
      "epoch": 0.03145629667039428,
      "grad_norm": 13.205926685916028,
      "learning_rate": 3.999978157520408e-06,
      "loss": 1.0382,
      "step": 436
    },
    {
      "epoch": 0.031528444139821796,
      "grad_norm": 13.193532133630207,
      "learning_rate": 3.9999759186707424e-06,
      "loss": 0.954,
      "step": 437
    },
    {
      "epoch": 0.03160059160924931,
      "grad_norm": 0.7800685341388225,
      "learning_rate": 3.999973570609796e-06,
      "loss": 0.8691,
      "step": 438
    },
    {
      "epoch": 0.03167273907867681,
      "grad_norm": 6.550558214934518,
      "learning_rate": 3.9999711133376955e-06,
      "loss": 1.0572,
      "step": 439
    },
    {
      "epoch": 0.031744886548104326,
      "grad_norm": 16.902330659949865,
      "learning_rate": 3.999968546854578e-06,
      "loss": 1.102,
      "step": 440
    },
    {
      "epoch": 0.03181703401753184,
      "grad_norm": 8.57741167059136,
      "learning_rate": 3.999965871160581e-06,
      "loss": 1.0078,
      "step": 441
    },
    {
      "epoch": 0.03188918148695934,
      "grad_norm": 7.416172511179616,
      "learning_rate": 3.9999630862558505e-06,
      "loss": 1.0366,
      "step": 442
    },
    {
      "epoch": 0.031961328956386856,
      "grad_norm": 9.413678350564705,
      "learning_rate": 3.99996019214054e-06,
      "loss": 1.1259,
      "step": 443
    },
    {
      "epoch": 0.03203347642581436,
      "grad_norm": 6.881961696315287,
      "learning_rate": 3.9999571888148075e-06,
      "loss": 1.0217,
      "step": 444
    },
    {
      "epoch": 0.032105623895241873,
      "grad_norm": 4.658723929219421,
      "learning_rate": 3.999954076278817e-06,
      "loss": 1.0636,
      "step": 445
    },
    {
      "epoch": 0.032177771364669386,
      "grad_norm": 6.242861218433976,
      "learning_rate": 3.999950854532737e-06,
      "loss": 1.0675,
      "step": 446
    },
    {
      "epoch": 0.03224991883409689,
      "grad_norm": 11.14244169641501,
      "learning_rate": 3.999947523576744e-06,
      "loss": 1.0438,
      "step": 447
    },
    {
      "epoch": 0.032322066303524404,
      "grad_norm": 11.492130406469327,
      "learning_rate": 3.999944083411021e-06,
      "loss": 0.9788,
      "step": 448
    },
    {
      "epoch": 0.032394213772951916,
      "grad_norm": 7.588157069807164,
      "learning_rate": 3.999940534035755e-06,
      "loss": 1.1453,
      "step": 449
    },
    {
      "epoch": 0.03246636124237942,
      "grad_norm": 14.477197541555705,
      "learning_rate": 3.9999368754511396e-06,
      "loss": 1.0593,
      "step": 450
    },
    {
      "epoch": 0.032538508711806934,
      "grad_norm": 4.522395038557073,
      "learning_rate": 3.999933107657376e-06,
      "loss": 0.9879,
      "step": 451
    },
    {
      "epoch": 0.032610656181234446,
      "grad_norm": 5.310979827595467,
      "learning_rate": 3.999929230654668e-06,
      "loss": 0.9912,
      "step": 452
    },
    {
      "epoch": 0.03268280365066195,
      "grad_norm": 0.7230389792960767,
      "learning_rate": 3.999925244443229e-06,
      "loss": 0.863,
      "step": 453
    },
    {
      "epoch": 0.032754951120089464,
      "grad_norm": 11.234072791267296,
      "learning_rate": 3.999921149023275e-06,
      "loss": 1.0394,
      "step": 454
    },
    {
      "epoch": 0.032827098589516976,
      "grad_norm": 9.017827074143757,
      "learning_rate": 3.99991694439503e-06,
      "loss": 1.056,
      "step": 455
    },
    {
      "epoch": 0.03289924605894448,
      "grad_norm": 11.671729233736707,
      "learning_rate": 3.999912630558725e-06,
      "loss": 0.9515,
      "step": 456
    },
    {
      "epoch": 0.032971393528371994,
      "grad_norm": 7.748823261139271,
      "learning_rate": 3.999908207514595e-06,
      "loss": 1.0663,
      "step": 457
    },
    {
      "epoch": 0.0330435409977995,
      "grad_norm": 9.18378814660267,
      "learning_rate": 3.99990367526288e-06,
      "loss": 1.0885,
      "step": 458
    },
    {
      "epoch": 0.03311568846722701,
      "grad_norm": 8.222908157513041,
      "learning_rate": 3.99989903380383e-06,
      "loss": 1.0463,
      "step": 459
    },
    {
      "epoch": 0.033187835936654524,
      "grad_norm": 12.499669795443285,
      "learning_rate": 3.999894283137696e-06,
      "loss": 1.0728,
      "step": 460
    },
    {
      "epoch": 0.03325998340608203,
      "grad_norm": 4.621122435918466,
      "learning_rate": 3.999889423264739e-06,
      "loss": 1.0844,
      "step": 461
    },
    {
      "epoch": 0.03333213087550954,
      "grad_norm": 19.596942647586957,
      "learning_rate": 3.999884454185224e-06,
      "loss": 1.1199,
      "step": 462
    },
    {
      "epoch": 0.033404278344937054,
      "grad_norm": 9.172151577628243,
      "learning_rate": 3.999879375899423e-06,
      "loss": 1.1394,
      "step": 463
    },
    {
      "epoch": 0.03347642581436456,
      "grad_norm": 5.147244999828242,
      "learning_rate": 3.999874188407612e-06,
      "loss": 1.1438,
      "step": 464
    },
    {
      "epoch": 0.03354857328379207,
      "grad_norm": 8.825742903176408,
      "learning_rate": 3.999868891710075e-06,
      "loss": 0.9955,
      "step": 465
    },
    {
      "epoch": 0.033620720753219584,
      "grad_norm": 6.5866276027638175,
      "learning_rate": 3.9998634858071004e-06,
      "loss": 1.0596,
      "step": 466
    },
    {
      "epoch": 0.03369286822264709,
      "grad_norm": 11.22226636603038,
      "learning_rate": 3.9998579706989845e-06,
      "loss": 1.0327,
      "step": 467
    },
    {
      "epoch": 0.0337650156920746,
      "grad_norm": 11.942157575726279,
      "learning_rate": 3.999852346386028e-06,
      "loss": 1.0284,
      "step": 468
    },
    {
      "epoch": 0.033837163161502114,
      "grad_norm": 12.531849682697546,
      "learning_rate": 3.999846612868539e-06,
      "loss": 0.9952,
      "step": 469
    },
    {
      "epoch": 0.03390931063092962,
      "grad_norm": 8.373781941215297,
      "learning_rate": 3.999840770146829e-06,
      "loss": 0.9451,
      "step": 470
    },
    {
      "epoch": 0.03398145810035713,
      "grad_norm": 9.915136352771098,
      "learning_rate": 3.999834818221216e-06,
      "loss": 1.0073,
      "step": 471
    },
    {
      "epoch": 0.03405360556978464,
      "grad_norm": 5.720821484884322,
      "learning_rate": 3.999828757092029e-06,
      "loss": 1.0417,
      "step": 472
    },
    {
      "epoch": 0.03412575303921215,
      "grad_norm": 6.829565110219432,
      "learning_rate": 3.999822586759595e-06,
      "loss": 0.9926,
      "step": 473
    },
    {
      "epoch": 0.03419790050863966,
      "grad_norm": 8.1676969332462,
      "learning_rate": 3.999816307224254e-06,
      "loss": 1.1015,
      "step": 474
    },
    {
      "epoch": 0.03427004797806717,
      "grad_norm": 25.965308887532995,
      "learning_rate": 3.999809918486347e-06,
      "loss": 1.0714,
      "step": 475
    },
    {
      "epoch": 0.03434219544749468,
      "grad_norm": 27.87126538358155,
      "learning_rate": 3.999803420546223e-06,
      "loss": 0.9804,
      "step": 476
    },
    {
      "epoch": 0.03441434291692219,
      "grad_norm": 24.08664069075688,
      "learning_rate": 3.999796813404238e-06,
      "loss": 1.0934,
      "step": 477
    },
    {
      "epoch": 0.0344864903863497,
      "grad_norm": 7.179394206468607,
      "learning_rate": 3.999790097060752e-06,
      "loss": 1.0326,
      "step": 478
    },
    {
      "epoch": 0.03455863785577721,
      "grad_norm": 6.331349246121808,
      "learning_rate": 3.999783271516131e-06,
      "loss": 1.0665,
      "step": 479
    },
    {
      "epoch": 0.03463078532520472,
      "grad_norm": 9.086555492740704,
      "learning_rate": 3.99977633677075e-06,
      "loss": 0.9915,
      "step": 480
    },
    {
      "epoch": 0.03470293279463223,
      "grad_norm": 16.768892024999555,
      "learning_rate": 3.999769292824984e-06,
      "loss": 1.1309,
      "step": 481
    },
    {
      "epoch": 0.03477508026405974,
      "grad_norm": 59.186283151430025,
      "learning_rate": 3.9997621396792215e-06,
      "loss": 1.1195,
      "step": 482
    },
    {
      "epoch": 0.034847227733487245,
      "grad_norm": 18.295732444578263,
      "learning_rate": 3.99975487733385e-06,
      "loss": 1.071,
      "step": 483
    },
    {
      "epoch": 0.03491937520291476,
      "grad_norm": 15.152386548502331,
      "learning_rate": 3.999747505789269e-06,
      "loss": 1.0027,
      "step": 484
    },
    {
      "epoch": 0.03499152267234227,
      "grad_norm": 3.34270853240925,
      "learning_rate": 3.999740025045879e-06,
      "loss": 1.021,
      "step": 485
    },
    {
      "epoch": 0.035063670141769775,
      "grad_norm": 5.755746706354014,
      "learning_rate": 3.9997324351040885e-06,
      "loss": 1.106,
      "step": 486
    },
    {
      "epoch": 0.03513581761119729,
      "grad_norm": 5.770451208286644,
      "learning_rate": 3.999724735964313e-06,
      "loss": 1.0147,
      "step": 487
    },
    {
      "epoch": 0.0352079650806248,
      "grad_norm": 0.8286190718368124,
      "learning_rate": 3.999716927626972e-06,
      "loss": 0.8646,
      "step": 488
    },
    {
      "epoch": 0.035280112550052305,
      "grad_norm": 8.964345880783076,
      "learning_rate": 3.999709010092493e-06,
      "loss": 0.9316,
      "step": 489
    },
    {
      "epoch": 0.03535226001947982,
      "grad_norm": 12.892378550237236,
      "learning_rate": 3.999700983361307e-06,
      "loss": 1.0963,
      "step": 490
    },
    {
      "epoch": 0.03542440748890733,
      "grad_norm": 6.962133987640947,
      "learning_rate": 3.999692847433853e-06,
      "loss": 1.081,
      "step": 491
    },
    {
      "epoch": 0.035496554958334835,
      "grad_norm": 6.903365002971943,
      "learning_rate": 3.999684602310575e-06,
      "loss": 1.044,
      "step": 492
    },
    {
      "epoch": 0.03556870242776235,
      "grad_norm": 9.671858829121174,
      "learning_rate": 3.999676247991924e-06,
      "loss": 0.9972,
      "step": 493
    },
    {
      "epoch": 0.03564084989718986,
      "grad_norm": 21.312202160675465,
      "learning_rate": 3.999667784478355e-06,
      "loss": 1.0906,
      "step": 494
    },
    {
      "epoch": 0.035712997366617365,
      "grad_norm": 7.268218990488741,
      "learning_rate": 3.999659211770331e-06,
      "loss": 1.0634,
      "step": 495
    },
    {
      "epoch": 0.03578514483604488,
      "grad_norm": 0.9477685596054551,
      "learning_rate": 3.99965052986832e-06,
      "loss": 0.8745,
      "step": 496
    },
    {
      "epoch": 0.03585729230547238,
      "grad_norm": 8.30591477290995,
      "learning_rate": 3.999641738772797e-06,
      "loss": 1.0786,
      "step": 497
    },
    {
      "epoch": 0.035929439774899895,
      "grad_norm": 8.59185303992497,
      "learning_rate": 3.99963283848424e-06,
      "loss": 1.0545,
      "step": 498
    },
    {
      "epoch": 0.03600158724432741,
      "grad_norm": 15.17279219678682,
      "learning_rate": 3.999623829003136e-06,
      "loss": 1.0606,
      "step": 499
    },
    {
      "epoch": 0.03607373471375491,
      "grad_norm": 13.849023094566206,
      "learning_rate": 3.999614710329977e-06,
      "loss": 1.1389,
      "step": 500
    },
    {
      "epoch": 0.036145882183182425,
      "grad_norm": 0.7648672908787838,
      "learning_rate": 3.999605482465262e-06,
      "loss": 0.8882,
      "step": 501
    },
    {
      "epoch": 0.03621802965260994,
      "grad_norm": 12.381377455292673,
      "learning_rate": 3.999596145409493e-06,
      "loss": 1.0489,
      "step": 502
    },
    {
      "epoch": 0.03629017712203744,
      "grad_norm": 5.918169538157305,
      "learning_rate": 3.999586699163181e-06,
      "loss": 1.0264,
      "step": 503
    },
    {
      "epoch": 0.036362324591464955,
      "grad_norm": 6.723353880039354,
      "learning_rate": 3.999577143726841e-06,
      "loss": 1.0854,
      "step": 504
    },
    {
      "epoch": 0.03643447206089247,
      "grad_norm": 11.338259879053794,
      "learning_rate": 3.999567479100996e-06,
      "loss": 1.1036,
      "step": 505
    },
    {
      "epoch": 0.03650661953031997,
      "grad_norm": 7.063441661071835,
      "learning_rate": 3.9995577052861735e-06,
      "loss": 0.935,
      "step": 506
    },
    {
      "epoch": 0.036578766999747485,
      "grad_norm": 10.168620020921365,
      "learning_rate": 3.999547822282906e-06,
      "loss": 0.9765,
      "step": 507
    },
    {
      "epoch": 0.03665091446917499,
      "grad_norm": 7.376161128305337,
      "learning_rate": 3.999537830091734e-06,
      "loss": 1.0269,
      "step": 508
    },
    {
      "epoch": 0.0367230619386025,
      "grad_norm": 16.861107200749192,
      "learning_rate": 3.999527728713204e-06,
      "loss": 1.0277,
      "step": 509
    },
    {
      "epoch": 0.036795209408030015,
      "grad_norm": 16.39747812373474,
      "learning_rate": 3.9995175181478655e-06,
      "loss": 0.957,
      "step": 510
    },
    {
      "epoch": 0.03686735687745752,
      "grad_norm": 15.504890408736015,
      "learning_rate": 3.999507198396278e-06,
      "loss": 1.0338,
      "step": 511
    },
    {
      "epoch": 0.03693950434688503,
      "grad_norm": 7.565406429973448,
      "learning_rate": 3.999496769459005e-06,
      "loss": 0.9954,
      "step": 512
    },
    {
      "epoch": 0.037011651816312545,
      "grad_norm": 8.832101652057972,
      "learning_rate": 3.999486231336614e-06,
      "loss": 0.9468,
      "step": 513
    },
    {
      "epoch": 0.03708379928574005,
      "grad_norm": 7.225375747149763,
      "learning_rate": 3.9994755840296825e-06,
      "loss": 1.0501,
      "step": 514
    },
    {
      "epoch": 0.03715594675516756,
      "grad_norm": 0.765115568432249,
      "learning_rate": 3.999464827538791e-06,
      "loss": 0.8609,
      "step": 515
    },
    {
      "epoch": 0.037228094224595075,
      "grad_norm": 9.72313961484286,
      "learning_rate": 3.999453961864527e-06,
      "loss": 1.069,
      "step": 516
    },
    {
      "epoch": 0.03730024169402258,
      "grad_norm": 7.298841975083862,
      "learning_rate": 3.999442987007484e-06,
      "loss": 0.979,
      "step": 517
    },
    {
      "epoch": 0.03737238916345009,
      "grad_norm": 8.78272036802882,
      "learning_rate": 3.999431902968261e-06,
      "loss": 1.0813,
      "step": 518
    },
    {
      "epoch": 0.037444536632877605,
      "grad_norm": 7.4466491331999976,
      "learning_rate": 3.999420709747464e-06,
      "loss": 1.0225,
      "step": 519
    },
    {
      "epoch": 0.03751668410230511,
      "grad_norm": 9.062851050582209,
      "learning_rate": 3.999409407345703e-06,
      "loss": 1.0735,
      "step": 520
    },
    {
      "epoch": 0.03758883157173262,
      "grad_norm": 8.636974481034152,
      "learning_rate": 3.999397995763596e-06,
      "loss": 0.9697,
      "step": 521
    },
    {
      "epoch": 0.03766097904116013,
      "grad_norm": 8.430940465992704,
      "learning_rate": 3.9993864750017664e-06,
      "loss": 1.0536,
      "step": 522
    },
    {
      "epoch": 0.03773312651058764,
      "grad_norm": 14.922704799407963,
      "learning_rate": 3.999374845060842e-06,
      "loss": 1.1522,
      "step": 523
    },
    {
      "epoch": 0.03780527398001515,
      "grad_norm": 7.477725457733804,
      "learning_rate": 3.999363105941459e-06,
      "loss": 1.075,
      "step": 524
    },
    {
      "epoch": 0.03787742144944266,
      "grad_norm": 10.364965527003772,
      "learning_rate": 3.999351257644258e-06,
      "loss": 1.0151,
      "step": 525
    },
    {
      "epoch": 0.03794956891887017,
      "grad_norm": 16.93772882398475,
      "learning_rate": 3.999339300169887e-06,
      "loss": 0.9722,
      "step": 526
    },
    {
      "epoch": 0.03802171638829768,
      "grad_norm": 9.724018006157847,
      "learning_rate": 3.999327233518997e-06,
      "loss": 1.0828,
      "step": 527
    },
    {
      "epoch": 0.03809386385772519,
      "grad_norm": 10.34181882227369,
      "learning_rate": 3.999315057692249e-06,
      "loss": 1.0299,
      "step": 528
    },
    {
      "epoch": 0.0381660113271527,
      "grad_norm": 25.33448360575586,
      "learning_rate": 3.999302772690307e-06,
      "loss": 1.0637,
      "step": 529
    },
    {
      "epoch": 0.03823815879658021,
      "grad_norm": 7.80324793952888,
      "learning_rate": 3.999290378513841e-06,
      "loss": 1.1176,
      "step": 530
    },
    {
      "epoch": 0.03831030626600772,
      "grad_norm": 8.527375064360735,
      "learning_rate": 3.9992778751635285e-06,
      "loss": 1.0521,
      "step": 531
    },
    {
      "epoch": 0.03838245373543523,
      "grad_norm": 7.732649300456954,
      "learning_rate": 3.999265262640053e-06,
      "loss": 0.955,
      "step": 532
    },
    {
      "epoch": 0.03845460120486274,
      "grad_norm": 10.446368258981018,
      "learning_rate": 3.999252540944102e-06,
      "loss": 1.0001,
      "step": 533
    },
    {
      "epoch": 0.03852674867429025,
      "grad_norm": 9.476153510341762,
      "learning_rate": 3.999239710076371e-06,
      "loss": 0.9849,
      "step": 534
    },
    {
      "epoch": 0.03859889614371776,
      "grad_norm": 15.617960816740519,
      "learning_rate": 3.99922677003756e-06,
      "loss": 0.8726,
      "step": 535
    },
    {
      "epoch": 0.038671043613145266,
      "grad_norm": 11.027881659192765,
      "learning_rate": 3.999213720828377e-06,
      "loss": 1.0038,
      "step": 536
    },
    {
      "epoch": 0.03874319108257278,
      "grad_norm": 26.198820084655054,
      "learning_rate": 3.999200562449532e-06,
      "loss": 0.9854,
      "step": 537
    },
    {
      "epoch": 0.03881533855200029,
      "grad_norm": 7.26017179217574,
      "learning_rate": 3.999187294901747e-06,
      "loss": 0.8732,
      "step": 538
    },
    {
      "epoch": 0.038887486021427796,
      "grad_norm": 10.82053284128089,
      "learning_rate": 3.9991739181857435e-06,
      "loss": 1.0268,
      "step": 539
    },
    {
      "epoch": 0.03895963349085531,
      "grad_norm": 11.921900597941065,
      "learning_rate": 3.999160432302252e-06,
      "loss": 1.0602,
      "step": 540
    },
    {
      "epoch": 0.03903178096028282,
      "grad_norm": 15.134556801251648,
      "learning_rate": 3.999146837252011e-06,
      "loss": 1.0645,
      "step": 541
    },
    {
      "epoch": 0.039103928429710326,
      "grad_norm": 6.931774079002014,
      "learning_rate": 3.999133133035762e-06,
      "loss": 1.0754,
      "step": 542
    },
    {
      "epoch": 0.03917607589913784,
      "grad_norm": 10.647520007589172,
      "learning_rate": 3.999119319654252e-06,
      "loss": 1.1322,
      "step": 543
    },
    {
      "epoch": 0.03924822336856535,
      "grad_norm": 11.104749787116893,
      "learning_rate": 3.999105397108238e-06,
      "loss": 1.1414,
      "step": 544
    },
    {
      "epoch": 0.039320370837992856,
      "grad_norm": 9.915525311949644,
      "learning_rate": 3.999091365398478e-06,
      "loss": 1.028,
      "step": 545
    },
    {
      "epoch": 0.03939251830742037,
      "grad_norm": 11.514173315257551,
      "learning_rate": 3.999077224525739e-06,
      "loss": 1.0706,
      "step": 546
    },
    {
      "epoch": 0.039464665776847874,
      "grad_norm": 5.579462484086449,
      "learning_rate": 3.999062974490793e-06,
      "loss": 1.0729,
      "step": 547
    },
    {
      "epoch": 0.039536813246275386,
      "grad_norm": 9.448066909307634,
      "learning_rate": 3.999048615294418e-06,
      "loss": 1.0993,
      "step": 548
    },
    {
      "epoch": 0.0396089607157029,
      "grad_norm": 6.286823893118522,
      "learning_rate": 3.999034146937399e-06,
      "loss": 1.0853,
      "step": 549
    },
    {
      "epoch": 0.039681108185130404,
      "grad_norm": 13.209752982268029,
      "learning_rate": 3.999019569420524e-06,
      "loss": 1.0948,
      "step": 550
    },
    {
      "epoch": 0.039753255654557916,
      "grad_norm": 8.327054010345492,
      "learning_rate": 3.999004882744592e-06,
      "loss": 1.1129,
      "step": 551
    },
    {
      "epoch": 0.03982540312398543,
      "grad_norm": 6.924738788842593,
      "learning_rate": 3.998990086910403e-06,
      "loss": 1.0029,
      "step": 552
    },
    {
      "epoch": 0.039897550593412934,
      "grad_norm": 7.6964803206512435,
      "learning_rate": 3.998975181918765e-06,
      "loss": 1.0368,
      "step": 553
    },
    {
      "epoch": 0.039969698062840446,
      "grad_norm": 8.534217625734561,
      "learning_rate": 3.998960167770493e-06,
      "loss": 1.1229,
      "step": 554
    },
    {
      "epoch": 0.04004184553226796,
      "grad_norm": 12.923150254485165,
      "learning_rate": 3.9989450444664054e-06,
      "loss": 1.1043,
      "step": 555
    },
    {
      "epoch": 0.040113993001695464,
      "grad_norm": 30.522596649859263,
      "learning_rate": 3.998929812007329e-06,
      "loss": 0.9699,
      "step": 556
    },
    {
      "epoch": 0.040186140471122976,
      "grad_norm": 0.7146238812285536,
      "learning_rate": 3.998914470394095e-06,
      "loss": 0.8623,
      "step": 557
    },
    {
      "epoch": 0.04025828794055049,
      "grad_norm": 5.608122364325432,
      "learning_rate": 3.998899019627542e-06,
      "loss": 1.0298,
      "step": 558
    },
    {
      "epoch": 0.040330435409977994,
      "grad_norm": 7.23846623060314,
      "learning_rate": 3.998883459708513e-06,
      "loss": 1.0059,
      "step": 559
    },
    {
      "epoch": 0.040402582879405506,
      "grad_norm": 31.09583653489141,
      "learning_rate": 3.9988677906378575e-06,
      "loss": 1.0546,
      "step": 560
    },
    {
      "epoch": 0.04047473034883301,
      "grad_norm": 9.490642556886131,
      "learning_rate": 3.998852012416432e-06,
      "loss": 1.0451,
      "step": 561
    },
    {
      "epoch": 0.040546877818260524,
      "grad_norm": 15.570796118775075,
      "learning_rate": 3.998836125045097e-06,
      "loss": 1.0166,
      "step": 562
    },
    {
      "epoch": 0.040619025287688036,
      "grad_norm": 6.61857898408372,
      "learning_rate": 3.9988201285247216e-06,
      "loss": 1.1312,
      "step": 563
    },
    {
      "epoch": 0.04069117275711554,
      "grad_norm": 24.60851064950333,
      "learning_rate": 3.998804022856178e-06,
      "loss": 1.1218,
      "step": 564
    },
    {
      "epoch": 0.040763320226543054,
      "grad_norm": 10.936376546654511,
      "learning_rate": 3.998787808040345e-06,
      "loss": 1.0679,
      "step": 565
    },
    {
      "epoch": 0.040835467695970566,
      "grad_norm": 6.7614451264322115,
      "learning_rate": 3.998771484078111e-06,
      "loss": 1.0705,
      "step": 566
    },
    {
      "epoch": 0.04090761516539807,
      "grad_norm": 9.333550541484694,
      "learning_rate": 3.998755050970364e-06,
      "loss": 1.0616,
      "step": 567
    },
    {
      "epoch": 0.040979762634825584,
      "grad_norm": 6.031308504917829,
      "learning_rate": 3.998738508718004e-06,
      "loss": 1.0522,
      "step": 568
    },
    {
      "epoch": 0.041051910104253096,
      "grad_norm": 8.644650395981898,
      "learning_rate": 3.998721857321932e-06,
      "loss": 1.2075,
      "step": 569
    },
    {
      "epoch": 0.0411240575736806,
      "grad_norm": 0.7974832494764361,
      "learning_rate": 3.998705096783059e-06,
      "loss": 0.7973,
      "step": 570
    },
    {
      "epoch": 0.041196205043108114,
      "grad_norm": 8.932020094647761,
      "learning_rate": 3.998688227102299e-06,
      "loss": 1.1358,
      "step": 571
    },
    {
      "epoch": 0.041268352512535626,
      "grad_norm": 12.054313451139402,
      "learning_rate": 3.998671248280575e-06,
      "loss": 1.0878,
      "step": 572
    },
    {
      "epoch": 0.04134049998196313,
      "grad_norm": 8.216277123967522,
      "learning_rate": 3.9986541603188114e-06,
      "loss": 1.0604,
      "step": 573
    },
    {
      "epoch": 0.041412647451390644,
      "grad_norm": 6.447772376233722,
      "learning_rate": 3.998636963217944e-06,
      "loss": 1.1277,
      "step": 574
    },
    {
      "epoch": 0.04148479492081815,
      "grad_norm": 6.2668224531789605,
      "learning_rate": 3.9986196569789095e-06,
      "loss": 1.0732,
      "step": 575
    },
    {
      "epoch": 0.04155694239024566,
      "grad_norm": 5.160679804356285,
      "learning_rate": 3.998602241602656e-06,
      "loss": 0.9647,
      "step": 576
    },
    {
      "epoch": 0.041629089859673174,
      "grad_norm": 7.493510681352431,
      "learning_rate": 3.998584717090132e-06,
      "loss": 1.088,
      "step": 577
    },
    {
      "epoch": 0.04170123732910068,
      "grad_norm": 38.51516702313924,
      "learning_rate": 3.998567083442295e-06,
      "loss": 1.0251,
      "step": 578
    },
    {
      "epoch": 0.04177338479852819,
      "grad_norm": 5.927919066315524,
      "learning_rate": 3.998549340660107e-06,
      "loss": 1.084,
      "step": 579
    },
    {
      "epoch": 0.041845532267955704,
      "grad_norm": 4.684762994884299,
      "learning_rate": 3.998531488744538e-06,
      "loss": 1.0406,
      "step": 580
    },
    {
      "epoch": 0.04191767973738321,
      "grad_norm": 10.450394574614718,
      "learning_rate": 3.998513527696564e-06,
      "loss": 1.1109,
      "step": 581
    },
    {
      "epoch": 0.04198982720681072,
      "grad_norm": 0.7930663100487875,
      "learning_rate": 3.9984954575171634e-06,
      "loss": 0.8783,
      "step": 582
    },
    {
      "epoch": 0.042061974676238234,
      "grad_norm": 11.857692594144384,
      "learning_rate": 3.998477278207325e-06,
      "loss": 1.0766,
      "step": 583
    },
    {
      "epoch": 0.04213412214566574,
      "grad_norm": 7.577098281365929,
      "learning_rate": 3.9984589897680405e-06,
      "loss": 1.0973,
      "step": 584
    },
    {
      "epoch": 0.04220626961509325,
      "grad_norm": 11.15271281952796,
      "learning_rate": 3.998440592200308e-06,
      "loss": 1.1284,
      "step": 585
    },
    {
      "epoch": 0.04227841708452076,
      "grad_norm": 13.043175818212552,
      "learning_rate": 3.998422085505132e-06,
      "loss": 0.8194,
      "step": 586
    },
    {
      "epoch": 0.04235056455394827,
      "grad_norm": 6.9551601217611045,
      "learning_rate": 3.998403469683525e-06,
      "loss": 1.0807,
      "step": 587
    },
    {
      "epoch": 0.04242271202337578,
      "grad_norm": 8.710608615568333,
      "learning_rate": 3.998384744736502e-06,
      "loss": 1.0907,
      "step": 588
    },
    {
      "epoch": 0.04249485949280329,
      "grad_norm": 9.053743328011443,
      "learning_rate": 3.998365910665087e-06,
      "loss": 1.0231,
      "step": 589
    },
    {
      "epoch": 0.0425670069622308,
      "grad_norm": 7.6790919029279285,
      "learning_rate": 3.998346967470306e-06,
      "loss": 1.0281,
      "step": 590
    },
    {
      "epoch": 0.04263915443165831,
      "grad_norm": 25.956625337778046,
      "learning_rate": 3.9983279151531944e-06,
      "loss": 1.0013,
      "step": 591
    },
    {
      "epoch": 0.04271130190108582,
      "grad_norm": 14.733921921884727,
      "learning_rate": 3.998308753714794e-06,
      "loss": 1.0324,
      "step": 592
    },
    {
      "epoch": 0.04278344937051333,
      "grad_norm": 6.998294895130455,
      "learning_rate": 3.998289483156149e-06,
      "loss": 0.9728,
      "step": 593
    },
    {
      "epoch": 0.04285559683994084,
      "grad_norm": 9.49614717758006,
      "learning_rate": 3.9982701034783136e-06,
      "loss": 0.9745,
      "step": 594
    },
    {
      "epoch": 0.04292774430936835,
      "grad_norm": 14.533056003362644,
      "learning_rate": 3.998250614682345e-06,
      "loss": 1.011,
      "step": 595
    },
    {
      "epoch": 0.04299989177879586,
      "grad_norm": 20.22883762310705,
      "learning_rate": 3.9982310167693075e-06,
      "loss": 1.0512,
      "step": 596
    },
    {
      "epoch": 0.04307203924822337,
      "grad_norm": 7.169116052595741,
      "learning_rate": 3.998211309740272e-06,
      "loss": 1.0678,
      "step": 597
    },
    {
      "epoch": 0.04314418671765088,
      "grad_norm": 9.739750096054506,
      "learning_rate": 3.9981914935963125e-06,
      "loss": 1.1215,
      "step": 598
    },
    {
      "epoch": 0.04321633418707839,
      "grad_norm": 5.298172963321184,
      "learning_rate": 3.998171568338514e-06,
      "loss": 1.0097,
      "step": 599
    },
    {
      "epoch": 0.043288481656505895,
      "grad_norm": 9.914891273721894,
      "learning_rate": 3.9981515339679625e-06,
      "loss": 0.9864,
      "step": 600
    },
    {
      "epoch": 0.04336062912593341,
      "grad_norm": 8.9764787821107,
      "learning_rate": 3.998131390485753e-06,
      "loss": 0.9469,
      "step": 601
    },
    {
      "epoch": 0.04343277659536092,
      "grad_norm": 5.307508030587516,
      "learning_rate": 3.9981111378929844e-06,
      "loss": 1.0804,
      "step": 602
    },
    {
      "epoch": 0.043504924064788425,
      "grad_norm": 11.669954399797772,
      "learning_rate": 3.998090776190763e-06,
      "loss": 1.1146,
      "step": 603
    },
    {
      "epoch": 0.04357707153421594,
      "grad_norm": 32.51776858023164,
      "learning_rate": 3.998070305380202e-06,
      "loss": 1.0702,
      "step": 604
    },
    {
      "epoch": 0.04364921900364345,
      "grad_norm": 10.808565162480265,
      "learning_rate": 3.9980497254624185e-06,
      "loss": 1.0114,
      "step": 605
    },
    {
      "epoch": 0.043721366473070955,
      "grad_norm": 12.547935431526176,
      "learning_rate": 3.998029036438536e-06,
      "loss": 1.0908,
      "step": 606
    },
    {
      "epoch": 0.04379351394249847,
      "grad_norm": 8.771780752688828,
      "learning_rate": 3.998008238309683e-06,
      "loss": 0.91,
      "step": 607
    },
    {
      "epoch": 0.04386566141192598,
      "grad_norm": 8.43506818977837,
      "learning_rate": 3.9979873310769965e-06,
      "loss": 1.0217,
      "step": 608
    },
    {
      "epoch": 0.043937808881353485,
      "grad_norm": 17.812902613322702,
      "learning_rate": 3.99796631474162e-06,
      "loss": 0.9989,
      "step": 609
    },
    {
      "epoch": 0.044009956350781,
      "grad_norm": 5.914638153642654,
      "learning_rate": 3.997945189304698e-06,
      "loss": 1.0619,
      "step": 610
    },
    {
      "epoch": 0.0440821038202085,
      "grad_norm": 6.344736708400577,
      "learning_rate": 3.997923954767385e-06,
      "loss": 1.0551,
      "step": 611
    },
    {
      "epoch": 0.044154251289636015,
      "grad_norm": 6.809671357188974,
      "learning_rate": 3.997902611130842e-06,
      "loss": 1.0574,
      "step": 612
    },
    {
      "epoch": 0.04422639875906353,
      "grad_norm": 0.787747335193985,
      "learning_rate": 3.997881158396233e-06,
      "loss": 0.8791,
      "step": 613
    },
    {
      "epoch": 0.04429854622849103,
      "grad_norm": 7.920090694486189,
      "learning_rate": 3.997859596564729e-06,
      "loss": 0.977,
      "step": 614
    },
    {
      "epoch": 0.044370693697918545,
      "grad_norm": 1.1612898300944927,
      "learning_rate": 3.997837925637509e-06,
      "loss": 0.8562,
      "step": 615
    },
    {
      "epoch": 0.04444284116734606,
      "grad_norm": 29.01753388813768,
      "learning_rate": 3.997816145615755e-06,
      "loss": 1.0295,
      "step": 616
    },
    {
      "epoch": 0.04451498863677356,
      "grad_norm": 21.399700326693026,
      "learning_rate": 3.997794256500658e-06,
      "loss": 1.014,
      "step": 617
    },
    {
      "epoch": 0.044587136106201075,
      "grad_norm": 8.27414706810254,
      "learning_rate": 3.997772258293411e-06,
      "loss": 1.0103,
      "step": 618
    },
    {
      "epoch": 0.04465928357562859,
      "grad_norm": 7.507152961080165,
      "learning_rate": 3.997750150995218e-06,
      "loss": 1.115,
      "step": 619
    },
    {
      "epoch": 0.04473143104505609,
      "grad_norm": 7.102671118087811,
      "learning_rate": 3.997727934607283e-06,
      "loss": 0.9663,
      "step": 620
    },
    {
      "epoch": 0.044803578514483605,
      "grad_norm": 8.649119390744445,
      "learning_rate": 3.997705609130822e-06,
      "loss": 1.1398,
      "step": 621
    },
    {
      "epoch": 0.04487572598391112,
      "grad_norm": 21.458242629763802,
      "learning_rate": 3.997683174567052e-06,
      "loss": 1.0421,
      "step": 622
    },
    {
      "epoch": 0.04494787345333862,
      "grad_norm": 11.196437909567175,
      "learning_rate": 3.9976606309172e-06,
      "loss": 1.0453,
      "step": 623
    },
    {
      "epoch": 0.045020020922766135,
      "grad_norm": 6.121508167408775,
      "learning_rate": 3.997637978182496e-06,
      "loss": 0.9752,
      "step": 624
    },
    {
      "epoch": 0.04509216839219364,
      "grad_norm": 10.855638478679417,
      "learning_rate": 3.9976152163641765e-06,
      "loss": 1.0122,
      "step": 625
    },
    {
      "epoch": 0.04516431586162115,
      "grad_norm": 62.80124277233692,
      "learning_rate": 3.9975923454634856e-06,
      "loss": 1.116,
      "step": 626
    },
    {
      "epoch": 0.045236463331048665,
      "grad_norm": 9.798046181988274,
      "learning_rate": 3.997569365481671e-06,
      "loss": 1.0566,
      "step": 627
    },
    {
      "epoch": 0.04530861080047617,
      "grad_norm": 7.791160553458989,
      "learning_rate": 3.997546276419988e-06,
      "loss": 0.9542,
      "step": 628
    },
    {
      "epoch": 0.04538075826990368,
      "grad_norm": 7.1940092383106125,
      "learning_rate": 3.997523078279697e-06,
      "loss": 1.1688,
      "step": 629
    },
    {
      "epoch": 0.045452905739331195,
      "grad_norm": 6.656828805893731,
      "learning_rate": 3.997499771062066e-06,
      "loss": 1.0687,
      "step": 630
    },
    {
      "epoch": 0.0455250532087587,
      "grad_norm": 14.68998638778274,
      "learning_rate": 3.997476354768367e-06,
      "loss": 1.155,
      "step": 631
    },
    {
      "epoch": 0.04559720067818621,
      "grad_norm": 7.584236119228293,
      "learning_rate": 3.9974528293998785e-06,
      "loss": 1.082,
      "step": 632
    },
    {
      "epoch": 0.045669348147613725,
      "grad_norm": 10.992869406820503,
      "learning_rate": 3.997429194957886e-06,
      "loss": 1.1292,
      "step": 633
    },
    {
      "epoch": 0.04574149561704123,
      "grad_norm": 7.8402385982874385,
      "learning_rate": 3.997405451443678e-06,
      "loss": 1.0397,
      "step": 634
    },
    {
      "epoch": 0.04581364308646874,
      "grad_norm": 6.382277579520557,
      "learning_rate": 3.997381598858554e-06,
      "loss": 1.0066,
      "step": 635
    },
    {
      "epoch": 0.045885790555896255,
      "grad_norm": 19.80228967239726,
      "learning_rate": 3.9973576372038135e-06,
      "loss": 0.927,
      "step": 636
    },
    {
      "epoch": 0.04595793802532376,
      "grad_norm": 7.996959585358881,
      "learning_rate": 3.997333566480767e-06,
      "loss": 1.0787,
      "step": 637
    },
    {
      "epoch": 0.04603008549475127,
      "grad_norm": 7.9313816309137035,
      "learning_rate": 3.997309386690729e-06,
      "loss": 0.9503,
      "step": 638
    },
    {
      "epoch": 0.04610223296417878,
      "grad_norm": 4.7479643224132575,
      "learning_rate": 3.997285097835018e-06,
      "loss": 1.0593,
      "step": 639
    },
    {
      "epoch": 0.04617438043360629,
      "grad_norm": 5.5979361613873175,
      "learning_rate": 3.997260699914962e-06,
      "loss": 1.0592,
      "step": 640
    },
    {
      "epoch": 0.0462465279030338,
      "grad_norm": 13.595182216218637,
      "learning_rate": 3.9972361929318925e-06,
      "loss": 1.0393,
      "step": 641
    },
    {
      "epoch": 0.04631867537246131,
      "grad_norm": 9.737348515675233,
      "learning_rate": 3.9972115768871485e-06,
      "loss": 1.0918,
      "step": 642
    },
    {
      "epoch": 0.04639082284188882,
      "grad_norm": 6.477618785904126,
      "learning_rate": 3.997186851782074e-06,
      "loss": 1.033,
      "step": 643
    },
    {
      "epoch": 0.04646297031131633,
      "grad_norm": 10.907622417678411,
      "learning_rate": 3.997162017618018e-06,
      "loss": 0.9645,
      "step": 644
    },
    {
      "epoch": 0.04653511778074384,
      "grad_norm": 9.422070553396615,
      "learning_rate": 3.997137074396338e-06,
      "loss": 1.0763,
      "step": 645
    },
    {
      "epoch": 0.04660726525017135,
      "grad_norm": 7.2458005120921705,
      "learning_rate": 3.997112022118396e-06,
      "loss": 1.118,
      "step": 646
    },
    {
      "epoch": 0.04667941271959886,
      "grad_norm": 11.429584703485101,
      "learning_rate": 3.997086860785558e-06,
      "loss": 0.9611,
      "step": 647
    },
    {
      "epoch": 0.04675156018902637,
      "grad_norm": 8.974023843453454,
      "learning_rate": 3.997061590399201e-06,
      "loss": 0.8698,
      "step": 648
    },
    {
      "epoch": 0.04682370765845388,
      "grad_norm": 8.52823661152459,
      "learning_rate": 3.997036210960703e-06,
      "loss": 1.0152,
      "step": 649
    },
    {
      "epoch": 0.046895855127881386,
      "grad_norm": 6.437152742075077,
      "learning_rate": 3.99701072247145e-06,
      "loss": 1.013,
      "step": 650
    },
    {
      "epoch": 0.0469680025973089,
      "grad_norm": 14.906568351880887,
      "learning_rate": 3.996985124932834e-06,
      "loss": 1.006,
      "step": 651
    },
    {
      "epoch": 0.04704015006673641,
      "grad_norm": 28.543670925484406,
      "learning_rate": 3.996959418346253e-06,
      "loss": 1.0244,
      "step": 652
    },
    {
      "epoch": 0.047112297536163916,
      "grad_norm": 4.854018617044968,
      "learning_rate": 3.9969336027131115e-06,
      "loss": 1.0466,
      "step": 653
    },
    {
      "epoch": 0.04718444500559143,
      "grad_norm": 8.435254963909054,
      "learning_rate": 3.996907678034817e-06,
      "loss": 1.1138,
      "step": 654
    },
    {
      "epoch": 0.04725659247501894,
      "grad_norm": 8.298606792397397,
      "learning_rate": 3.996881644312787e-06,
      "loss": 1.0425,
      "step": 655
    },
    {
      "epoch": 0.047328739944446446,
      "grad_norm": 1.0358572259931442,
      "learning_rate": 3.996855501548443e-06,
      "loss": 0.895,
      "step": 656
    },
    {
      "epoch": 0.04740088741387396,
      "grad_norm": 9.760253991746241,
      "learning_rate": 3.996829249743212e-06,
      "loss": 1.034,
      "step": 657
    },
    {
      "epoch": 0.04747303488330147,
      "grad_norm": 10.000085830320135,
      "learning_rate": 3.996802888898527e-06,
      "loss": 1.0254,
      "step": 658
    },
    {
      "epoch": 0.047545182352728976,
      "grad_norm": 8.682739270683866,
      "learning_rate": 3.99677641901583e-06,
      "loss": 1.0992,
      "step": 659
    },
    {
      "epoch": 0.04761732982215649,
      "grad_norm": 6.879260043783779,
      "learning_rate": 3.996749840096562e-06,
      "loss": 0.9655,
      "step": 660
    },
    {
      "epoch": 0.047689477291584,
      "grad_norm": 7.4264454471826875,
      "learning_rate": 3.996723152142179e-06,
      "loss": 0.9786,
      "step": 661
    },
    {
      "epoch": 0.047761624761011506,
      "grad_norm": 35.43249110324096,
      "learning_rate": 3.996696355154136e-06,
      "loss": 1.0499,
      "step": 662
    },
    {
      "epoch": 0.04783377223043902,
      "grad_norm": 8.13965212816728,
      "learning_rate": 3.996669449133896e-06,
      "loss": 1.0401,
      "step": 663
    },
    {
      "epoch": 0.047905919699866524,
      "grad_norm": 5.3354177375423975,
      "learning_rate": 3.996642434082929e-06,
      "loss": 1.019,
      "step": 664
    },
    {
      "epoch": 0.047978067169294036,
      "grad_norm": 12.256889430070606,
      "learning_rate": 3.9966153100027095e-06,
      "loss": 0.9837,
      "step": 665
    },
    {
      "epoch": 0.04805021463872155,
      "grad_norm": 6.319043082695119,
      "learning_rate": 3.9965880768947196e-06,
      "loss": 0.9999,
      "step": 666
    },
    {
      "epoch": 0.048122362108149054,
      "grad_norm": 7.0610447287305025,
      "learning_rate": 3.996560734760446e-06,
      "loss": 0.9952,
      "step": 667
    },
    {
      "epoch": 0.048194509577576566,
      "grad_norm": 7.3609951689924165,
      "learning_rate": 3.996533283601382e-06,
      "loss": 1.0478,
      "step": 668
    },
    {
      "epoch": 0.04826665704700408,
      "grad_norm": 7.322939992535604,
      "learning_rate": 3.996505723419025e-06,
      "loss": 0.9959,
      "step": 669
    },
    {
      "epoch": 0.048338804516431584,
      "grad_norm": 9.626345453724046,
      "learning_rate": 3.996478054214883e-06,
      "loss": 0.9834,
      "step": 670
    },
    {
      "epoch": 0.048410951985859096,
      "grad_norm": 47.922198503146745,
      "learning_rate": 3.9964502759904635e-06,
      "loss": 0.956,
      "step": 671
    },
    {
      "epoch": 0.04848309945528661,
      "grad_norm": 6.928755188922292,
      "learning_rate": 3.996422388747286e-06,
      "loss": 1.0044,
      "step": 672
    },
    {
      "epoch": 0.048555246924714114,
      "grad_norm": 51.06192386073379,
      "learning_rate": 3.996394392486872e-06,
      "loss": 1.0831,
      "step": 673
    },
    {
      "epoch": 0.048627394394141626,
      "grad_norm": 21.879001447637027,
      "learning_rate": 3.996366287210751e-06,
      "loss": 1.0279,
      "step": 674
    },
    {
      "epoch": 0.04869954186356914,
      "grad_norm": 6.866708367724163,
      "learning_rate": 3.996338072920457e-06,
      "loss": 0.9949,
      "step": 675
    },
    {
      "epoch": 0.048771689332996644,
      "grad_norm": 12.601458365192796,
      "learning_rate": 3.996309749617532e-06,
      "loss": 1.1647,
      "step": 676
    },
    {
      "epoch": 0.048843836802424156,
      "grad_norm": 36.487682197536714,
      "learning_rate": 3.996281317303521e-06,
      "loss": 0.9281,
      "step": 677
    },
    {
      "epoch": 0.04891598427185166,
      "grad_norm": 4.750073884088212,
      "learning_rate": 3.996252775979976e-06,
      "loss": 0.961,
      "step": 678
    },
    {
      "epoch": 0.048988131741279174,
      "grad_norm": 14.080446989593684,
      "learning_rate": 3.996224125648458e-06,
      "loss": 0.9972,
      "step": 679
    },
    {
      "epoch": 0.049060279210706687,
      "grad_norm": 7.416019997282751,
      "learning_rate": 3.99619536631053e-06,
      "loss": 0.9865,
      "step": 680
    },
    {
      "epoch": 0.04913242668013419,
      "grad_norm": 8.177769593681626,
      "learning_rate": 3.996166497967764e-06,
      "loss": 1.0482,
      "step": 681
    },
    {
      "epoch": 0.049204574149561704,
      "grad_norm": 7.749245145398765,
      "learning_rate": 3.9961375206217335e-06,
      "loss": 1.0881,
      "step": 682
    },
    {
      "epoch": 0.04927672161898922,
      "grad_norm": 6.815271496252145,
      "learning_rate": 3.996108434274022e-06,
      "loss": 1.1587,
      "step": 683
    },
    {
      "epoch": 0.04934886908841672,
      "grad_norm": 8.364962355853358,
      "learning_rate": 3.9960792389262195e-06,
      "loss": 0.9133,
      "step": 684
    },
    {
      "epoch": 0.049421016557844234,
      "grad_norm": 6.913880702500457,
      "learning_rate": 3.996049934579919e-06,
      "loss": 1.143,
      "step": 685
    },
    {
      "epoch": 0.04949316402727175,
      "grad_norm": 8.27858429236196,
      "learning_rate": 3.99602052123672e-06,
      "loss": 0.9737,
      "step": 686
    },
    {
      "epoch": 0.04956531149669925,
      "grad_norm": 8.54269323615514,
      "learning_rate": 3.99599099889823e-06,
      "loss": 1.0201,
      "step": 687
    },
    {
      "epoch": 0.049637458966126764,
      "grad_norm": 8.19109841979413,
      "learning_rate": 3.995961367566061e-06,
      "loss": 1.0882,
      "step": 688
    },
    {
      "epoch": 0.04970960643555427,
      "grad_norm": 10.087593964962762,
      "learning_rate": 3.9959316272418285e-06,
      "loss": 0.9969,
      "step": 689
    },
    {
      "epoch": 0.04978175390498178,
      "grad_norm": 4.8752278983814925,
      "learning_rate": 3.99590177792716e-06,
      "loss": 1.0986,
      "step": 690
    },
    {
      "epoch": 0.049853901374409294,
      "grad_norm": 8.006987381270694,
      "learning_rate": 3.995871819623684e-06,
      "loss": 1.1259,
      "step": 691
    },
    {
      "epoch": 0.0499260488438368,
      "grad_norm": 8.8665684152002,
      "learning_rate": 3.995841752333035e-06,
      "loss": 0.8906,
      "step": 692
    },
    {
      "epoch": 0.04999819631326431,
      "grad_norm": 8.011889682324531,
      "learning_rate": 3.9958115760568575e-06,
      "loss": 0.9672,
      "step": 693
    },
    {
      "epoch": 0.050070343782691824,
      "grad_norm": 6.61819511587539,
      "learning_rate": 3.995781290796798e-06,
      "loss": 0.9081,
      "step": 694
    },
    {
      "epoch": 0.05014249125211933,
      "grad_norm": 4.681356447687745,
      "learning_rate": 3.995750896554511e-06,
      "loss": 1.0323,
      "step": 695
    },
    {
      "epoch": 0.05021463872154684,
      "grad_norm": 8.014638858685789,
      "learning_rate": 3.995720393331654e-06,
      "loss": 1.0822,
      "step": 696
    },
    {
      "epoch": 0.050286786190974354,
      "grad_norm": 4.459097300639701,
      "learning_rate": 3.995689781129896e-06,
      "loss": 1.012,
      "step": 697
    },
    {
      "epoch": 0.05035893366040186,
      "grad_norm": 4.7879390425598105,
      "learning_rate": 3.995659059950905e-06,
      "loss": 1.0066,
      "step": 698
    },
    {
      "epoch": 0.05043108112982937,
      "grad_norm": 6.235365251184074,
      "learning_rate": 3.995628229796361e-06,
      "loss": 1.0364,
      "step": 699
    },
    {
      "epoch": 0.050503228599256884,
      "grad_norm": 5.180186804271487,
      "learning_rate": 3.995597290667948e-06,
      "loss": 1.0304,
      "step": 700
    },
    {
      "epoch": 0.05057537606868439,
      "grad_norm": 20.58078609051451,
      "learning_rate": 3.995566242567353e-06,
      "loss": 1.0231,
      "step": 701
    },
    {
      "epoch": 0.0506475235381119,
      "grad_norm": 5.321963330004547,
      "learning_rate": 3.9955350854962734e-06,
      "loss": 0.9064,
      "step": 702
    },
    {
      "epoch": 0.05071967100753941,
      "grad_norm": 3.7706416100645646,
      "learning_rate": 3.995503819456411e-06,
      "loss": 1.0034,
      "step": 703
    },
    {
      "epoch": 0.05079181847696692,
      "grad_norm": 7.06756356263699,
      "learning_rate": 3.99547244444947e-06,
      "loss": 1.0991,
      "step": 704
    },
    {
      "epoch": 0.05086396594639443,
      "grad_norm": 7.049376050244015,
      "learning_rate": 3.995440960477167e-06,
      "loss": 1.1492,
      "step": 705
    },
    {
      "epoch": 0.05093611341582194,
      "grad_norm": 0.9018040153394566,
      "learning_rate": 3.995409367541221e-06,
      "loss": 0.8247,
      "step": 706
    },
    {
      "epoch": 0.05100826088524945,
      "grad_norm": 10.689171732832454,
      "learning_rate": 3.995377665643355e-06,
      "loss": 0.8897,
      "step": 707
    },
    {
      "epoch": 0.05108040835467696,
      "grad_norm": 6.425303933639905,
      "learning_rate": 3.995345854785301e-06,
      "loss": 1.043,
      "step": 708
    },
    {
      "epoch": 0.05115255582410447,
      "grad_norm": 6.9303697940425755,
      "learning_rate": 3.995313934968797e-06,
      "loss": 0.942,
      "step": 709
    },
    {
      "epoch": 0.05122470329353198,
      "grad_norm": 6.200355507287307,
      "learning_rate": 3.995281906195586e-06,
      "loss": 1.077,
      "step": 710
    },
    {
      "epoch": 0.05129685076295949,
      "grad_norm": 9.105348511781198,
      "learning_rate": 3.995249768467415e-06,
      "loss": 1.1241,
      "step": 711
    },
    {
      "epoch": 0.051368998232387,
      "grad_norm": 11.296700419531906,
      "learning_rate": 3.995217521786041e-06,
      "loss": 0.9931,
      "step": 712
    },
    {
      "epoch": 0.05144114570181451,
      "grad_norm": 6.856085820696946,
      "learning_rate": 3.995185166153223e-06,
      "loss": 1.0587,
      "step": 713
    },
    {
      "epoch": 0.051513293171242015,
      "grad_norm": 7.666927692558469,
      "learning_rate": 3.995152701570731e-06,
      "loss": 1.0817,
      "step": 714
    },
    {
      "epoch": 0.05158544064066953,
      "grad_norm": 6.246778344479544,
      "learning_rate": 3.995120128040335e-06,
      "loss": 1.0005,
      "step": 715
    },
    {
      "epoch": 0.05165758811009704,
      "grad_norm": 0.8792652329488464,
      "learning_rate": 3.995087445563814e-06,
      "loss": 0.8896,
      "step": 716
    },
    {
      "epoch": 0.051729735579524545,
      "grad_norm": 6.429107528615613,
      "learning_rate": 3.995054654142954e-06,
      "loss": 0.954,
      "step": 717
    },
    {
      "epoch": 0.05180188304895206,
      "grad_norm": 6.446852621390056,
      "learning_rate": 3.995021753779544e-06,
      "loss": 1.1208,
      "step": 718
    },
    {
      "epoch": 0.05187403051837957,
      "grad_norm": 8.303688829005173,
      "learning_rate": 3.994988744475382e-06,
      "loss": 1.1227,
      "step": 719
    },
    {
      "epoch": 0.051946177987807075,
      "grad_norm": 7.3368243086692,
      "learning_rate": 3.99495562623227e-06,
      "loss": 1.1024,
      "step": 720
    },
    {
      "epoch": 0.05201832545723459,
      "grad_norm": 11.242523082313314,
      "learning_rate": 3.994922399052015e-06,
      "loss": 1.028,
      "step": 721
    },
    {
      "epoch": 0.0520904729266621,
      "grad_norm": 8.556903875986402,
      "learning_rate": 3.994889062936434e-06,
      "loss": 1.0044,
      "step": 722
    },
    {
      "epoch": 0.052162620396089605,
      "grad_norm": 0.9769166228054026,
      "learning_rate": 3.994855617887346e-06,
      "loss": 0.8787,
      "step": 723
    },
    {
      "epoch": 0.05223476786551712,
      "grad_norm": 7.634872798796886,
      "learning_rate": 3.994822063906577e-06,
      "loss": 0.9559,
      "step": 724
    },
    {
      "epoch": 0.05230691533494463,
      "grad_norm": 7.813211393392094,
      "learning_rate": 3.99478840099596e-06,
      "loss": 0.9541,
      "step": 725
    },
    {
      "epoch": 0.052379062804372135,
      "grad_norm": 6.131912379206795,
      "learning_rate": 3.994754629157332e-06,
      "loss": 0.9452,
      "step": 726
    },
    {
      "epoch": 0.05245121027379965,
      "grad_norm": 4.6293487723601405,
      "learning_rate": 3.994720748392539e-06,
      "loss": 1.0057,
      "step": 727
    },
    {
      "epoch": 0.05252335774322715,
      "grad_norm": 20.40658199863925,
      "learning_rate": 3.994686758703429e-06,
      "loss": 1.0968,
      "step": 728
    },
    {
      "epoch": 0.052595505212654665,
      "grad_norm": 6.088640310653893,
      "learning_rate": 3.994652660091861e-06,
      "loss": 1.0849,
      "step": 729
    },
    {
      "epoch": 0.05266765268208218,
      "grad_norm": 7.0885996670187135,
      "learning_rate": 3.994618452559693e-06,
      "loss": 0.9448,
      "step": 730
    },
    {
      "epoch": 0.05273980015150968,
      "grad_norm": 5.491881360655618,
      "learning_rate": 3.994584136108797e-06,
      "loss": 1.0671,
      "step": 731
    },
    {
      "epoch": 0.052811947620937195,
      "grad_norm": 6.8476094550597315,
      "learning_rate": 3.994549710741044e-06,
      "loss": 0.9359,
      "step": 732
    },
    {
      "epoch": 0.05288409509036471,
      "grad_norm": 6.61782304218139,
      "learning_rate": 3.994515176458316e-06,
      "loss": 1.0135,
      "step": 733
    },
    {
      "epoch": 0.05295624255979221,
      "grad_norm": 5.2795120178484725,
      "learning_rate": 3.994480533262496e-06,
      "loss": 1.0901,
      "step": 734
    },
    {
      "epoch": 0.053028390029219725,
      "grad_norm": 8.784902661413376,
      "learning_rate": 3.994445781155478e-06,
      "loss": 1.1068,
      "step": 735
    },
    {
      "epoch": 0.05310053749864724,
      "grad_norm": 23.901578638779934,
      "learning_rate": 3.99441092013916e-06,
      "loss": 1.0088,
      "step": 736
    },
    {
      "epoch": 0.05317268496807474,
      "grad_norm": 5.99904116280651,
      "learning_rate": 3.994375950215443e-06,
      "loss": 1.0945,
      "step": 737
    },
    {
      "epoch": 0.053244832437502256,
      "grad_norm": 5.370996980662359,
      "learning_rate": 3.994340871386241e-06,
      "loss": 1.0802,
      "step": 738
    },
    {
      "epoch": 0.05331697990692977,
      "grad_norm": 8.508776453368512,
      "learning_rate": 3.994305683653464e-06,
      "loss": 1.0079,
      "step": 739
    },
    {
      "epoch": 0.05338912737635727,
      "grad_norm": 6.725515250849135,
      "learning_rate": 3.994270387019038e-06,
      "loss": 1.0247,
      "step": 740
    },
    {
      "epoch": 0.053461274845784786,
      "grad_norm": 6.41408539743553,
      "learning_rate": 3.994234981484888e-06,
      "loss": 1.0589,
      "step": 741
    },
    {
      "epoch": 0.05353342231521229,
      "grad_norm": 4.580673602014595,
      "learning_rate": 3.9941994670529485e-06,
      "loss": 1.05,
      "step": 742
    },
    {
      "epoch": 0.0536055697846398,
      "grad_norm": 0.784044189908307,
      "learning_rate": 3.994163843725158e-06,
      "loss": 0.8209,
      "step": 743
    },
    {
      "epoch": 0.053677717254067316,
      "grad_norm": 6.596769825554599,
      "learning_rate": 3.9941281115034624e-06,
      "loss": 1.0643,
      "step": 744
    },
    {
      "epoch": 0.05374986472349482,
      "grad_norm": 4.453189889535449,
      "learning_rate": 3.994092270389812e-06,
      "loss": 1.1182,
      "step": 745
    },
    {
      "epoch": 0.05382201219292233,
      "grad_norm": 6.752487571956266,
      "learning_rate": 3.994056320386165e-06,
      "loss": 0.9931,
      "step": 746
    },
    {
      "epoch": 0.053894159662349846,
      "grad_norm": 4.571620030310709,
      "learning_rate": 3.994020261494484e-06,
      "loss": 0.9272,
      "step": 747
    },
    {
      "epoch": 0.05396630713177735,
      "grad_norm": 6.87506824806324,
      "learning_rate": 3.993984093716738e-06,
      "loss": 1.0392,
      "step": 748
    },
    {
      "epoch": 0.05403845460120486,
      "grad_norm": 12.401006940750063,
      "learning_rate": 3.993947817054903e-06,
      "loss": 1.0215,
      "step": 749
    },
    {
      "epoch": 0.054110602070632376,
      "grad_norm": 0.8189122556114266,
      "learning_rate": 3.993911431510958e-06,
      "loss": 0.928,
      "step": 750
    },
    {
      "epoch": 0.05418274954005988,
      "grad_norm": 7.646037113755195,
      "learning_rate": 3.993874937086891e-06,
      "loss": 1.087,
      "step": 751
    },
    {
      "epoch": 0.05425489700948739,
      "grad_norm": 6.376253191871046,
      "learning_rate": 3.993838333784695e-06,
      "loss": 1.1169,
      "step": 752
    },
    {
      "epoch": 0.0543270444789149,
      "grad_norm": 3.8568618561046746,
      "learning_rate": 3.993801621606369e-06,
      "loss": 1.0084,
      "step": 753
    },
    {
      "epoch": 0.05439919194834241,
      "grad_norm": 7.15605436499102,
      "learning_rate": 3.993764800553917e-06,
      "loss": 1.0679,
      "step": 754
    },
    {
      "epoch": 0.05447133941776992,
      "grad_norm": 10.51784416011034,
      "learning_rate": 3.99372787062935e-06,
      "loss": 1.1248,
      "step": 755
    },
    {
      "epoch": 0.05454348688719743,
      "grad_norm": 4.873890872865394,
      "learning_rate": 3.993690831834685e-06,
      "loss": 1.0423,
      "step": 756
    },
    {
      "epoch": 0.05461563435662494,
      "grad_norm": 5.355541225975216,
      "learning_rate": 3.993653684171942e-06,
      "loss": 1.0472,
      "step": 757
    },
    {
      "epoch": 0.05468778182605245,
      "grad_norm": 0.6341760802687726,
      "learning_rate": 3.9936164276431536e-06,
      "loss": 0.8127,
      "step": 758
    },
    {
      "epoch": 0.05475992929547996,
      "grad_norm": 5.995644896168157,
      "learning_rate": 3.9935790622503514e-06,
      "loss": 1.0446,
      "step": 759
    },
    {
      "epoch": 0.05483207676490747,
      "grad_norm": 5.583410670565664,
      "learning_rate": 3.993541587995577e-06,
      "loss": 0.9436,
      "step": 760
    },
    {
      "epoch": 0.05490422423433498,
      "grad_norm": 7.518769746343823,
      "learning_rate": 3.993504004880876e-06,
      "loss": 1.0674,
      "step": 761
    },
    {
      "epoch": 0.05497637170376249,
      "grad_norm": 6.297534993230827,
      "learning_rate": 3.9934663129083e-06,
      "loss": 1.0442,
      "step": 762
    },
    {
      "epoch": 0.05504851917319,
      "grad_norm": 6.919904785824228,
      "learning_rate": 3.99342851207991e-06,
      "loss": 1.0816,
      "step": 763
    },
    {
      "epoch": 0.055120666642617513,
      "grad_norm": 4.83022166531305,
      "learning_rate": 3.993390602397768e-06,
      "loss": 1.0415,
      "step": 764
    },
    {
      "epoch": 0.05519281411204502,
      "grad_norm": 6.080915972326866,
      "learning_rate": 3.993352583863943e-06,
      "loss": 0.9593,
      "step": 765
    },
    {
      "epoch": 0.05526496158147253,
      "grad_norm": 0.8667294435200744,
      "learning_rate": 3.9933144564805145e-06,
      "loss": 0.8124,
      "step": 766
    },
    {
      "epoch": 0.05533710905090004,
      "grad_norm": 6.451992572872383,
      "learning_rate": 3.993276220249561e-06,
      "loss": 1.0389,
      "step": 767
    },
    {
      "epoch": 0.05540925652032755,
      "grad_norm": 11.581994361239737,
      "learning_rate": 3.993237875173173e-06,
      "loss": 1.0379,
      "step": 768
    },
    {
      "epoch": 0.05548140398975506,
      "grad_norm": 4.122398943567538,
      "learning_rate": 3.993199421253443e-06,
      "loss": 1.0363,
      "step": 769
    },
    {
      "epoch": 0.05555355145918257,
      "grad_norm": 9.30524914997788,
      "learning_rate": 3.993160858492471e-06,
      "loss": 0.9575,
      "step": 770
    },
    {
      "epoch": 0.05562569892861008,
      "grad_norm": 10.145632596285388,
      "learning_rate": 3.993122186892363e-06,
      "loss": 1.0634,
      "step": 771
    },
    {
      "epoch": 0.05569784639803759,
      "grad_norm": 8.441405346193843,
      "learning_rate": 3.993083406455231e-06,
      "loss": 1.0788,
      "step": 772
    },
    {
      "epoch": 0.0557699938674651,
      "grad_norm": 24.34261545615207,
      "learning_rate": 3.993044517183192e-06,
      "loss": 1.078,
      "step": 773
    },
    {
      "epoch": 0.05584214133689261,
      "grad_norm": 0.7776665712131137,
      "learning_rate": 3.99300551907837e-06,
      "loss": 0.8644,
      "step": 774
    },
    {
      "epoch": 0.05591428880632012,
      "grad_norm": 4.288858822292778,
      "learning_rate": 3.992966412142895e-06,
      "loss": 1.0927,
      "step": 775
    },
    {
      "epoch": 0.05598643627574763,
      "grad_norm": 4.9909930165753496,
      "learning_rate": 3.992927196378901e-06,
      "loss": 0.9794,
      "step": 776
    },
    {
      "epoch": 0.05605858374517514,
      "grad_norm": 9.766488243096342,
      "learning_rate": 3.992887871788531e-06,
      "loss": 1.0358,
      "step": 777
    },
    {
      "epoch": 0.056130731214602644,
      "grad_norm": 7.002651802223468,
      "learning_rate": 3.992848438373931e-06,
      "loss": 1.0099,
      "step": 778
    },
    {
      "epoch": 0.05620287868403016,
      "grad_norm": 10.46647633017051,
      "learning_rate": 3.992808896137256e-06,
      "loss": 1.0223,
      "step": 779
    },
    {
      "epoch": 0.05627502615345767,
      "grad_norm": 4.198789040513739,
      "learning_rate": 3.992769245080663e-06,
      "loss": 1.0667,
      "step": 780
    },
    {
      "epoch": 0.056347173622885174,
      "grad_norm": 6.40076967618034,
      "learning_rate": 3.99272948520632e-06,
      "loss": 0.9934,
      "step": 781
    },
    {
      "epoch": 0.05641932109231269,
      "grad_norm": 4.206914505184371,
      "learning_rate": 3.992689616516396e-06,
      "loss": 1.123,
      "step": 782
    },
    {
      "epoch": 0.0564914685617402,
      "grad_norm": 13.437679981534492,
      "learning_rate": 3.992649639013069e-06,
      "loss": 1.0041,
      "step": 783
    },
    {
      "epoch": 0.056563616031167704,
      "grad_norm": 4.226999764378588,
      "learning_rate": 3.992609552698521e-06,
      "loss": 0.9237,
      "step": 784
    },
    {
      "epoch": 0.05663576350059522,
      "grad_norm": 3.7656919861213263,
      "learning_rate": 3.9925693575749425e-06,
      "loss": 1.0728,
      "step": 785
    },
    {
      "epoch": 0.05670791097002273,
      "grad_norm": 6.830220265446182,
      "learning_rate": 3.992529053644527e-06,
      "loss": 1.0012,
      "step": 786
    },
    {
      "epoch": 0.056780058439450234,
      "grad_norm": 7.9595065492049155,
      "learning_rate": 3.992488640909476e-06,
      "loss": 1.0934,
      "step": 787
    },
    {
      "epoch": 0.05685220590887775,
      "grad_norm": 6.279517091297892,
      "learning_rate": 3.992448119371997e-06,
      "loss": 1.0406,
      "step": 788
    },
    {
      "epoch": 0.05692435337830526,
      "grad_norm": 0.8780361361355598,
      "learning_rate": 3.992407489034302e-06,
      "loss": 0.8962,
      "step": 789
    },
    {
      "epoch": 0.056996500847732764,
      "grad_norm": 3.3994416956557862,
      "learning_rate": 3.992366749898609e-06,
      "loss": 0.9657,
      "step": 790
    },
    {
      "epoch": 0.05706864831716028,
      "grad_norm": 3.1203652153471455,
      "learning_rate": 3.992325901967144e-06,
      "loss": 0.9721,
      "step": 791
    },
    {
      "epoch": 0.05714079578658778,
      "grad_norm": 5.462343659771535,
      "learning_rate": 3.992284945242136e-06,
      "loss": 0.9395,
      "step": 792
    },
    {
      "epoch": 0.057212943256015295,
      "grad_norm": 5.004963890349377,
      "learning_rate": 3.992243879725822e-06,
      "loss": 1.0299,
      "step": 793
    },
    {
      "epoch": 0.05728509072544281,
      "grad_norm": 0.8547783421528425,
      "learning_rate": 3.992202705420446e-06,
      "loss": 0.8576,
      "step": 794
    },
    {
      "epoch": 0.05735723819487031,
      "grad_norm": 6.268441116770886,
      "learning_rate": 3.992161422328255e-06,
      "loss": 1.0532,
      "step": 795
    },
    {
      "epoch": 0.057429385664297825,
      "grad_norm": 4.518330011332571,
      "learning_rate": 3.992120030451503e-06,
      "loss": 1.0398,
      "step": 796
    },
    {
      "epoch": 0.05750153313372534,
      "grad_norm": 7.103226704840647,
      "learning_rate": 3.99207852979245e-06,
      "loss": 1.0337,
      "step": 797
    },
    {
      "epoch": 0.05757368060315284,
      "grad_norm": 4.490449519435296,
      "learning_rate": 3.992036920353364e-06,
      "loss": 1.026,
      "step": 798
    },
    {
      "epoch": 0.057645828072580355,
      "grad_norm": 3.8588263859184124,
      "learning_rate": 3.991995202136516e-06,
      "loss": 1.0342,
      "step": 799
    },
    {
      "epoch": 0.05771797554200787,
      "grad_norm": 6.919416348314239,
      "learning_rate": 3.991953375144183e-06,
      "loss": 0.9568,
      "step": 800
    },
    {
      "epoch": 0.05779012301143537,
      "grad_norm": 5.627708461159276,
      "learning_rate": 3.99191143937865e-06,
      "loss": 1.0398,
      "step": 801
    },
    {
      "epoch": 0.057862270480862885,
      "grad_norm": 7.461717175393772,
      "learning_rate": 3.991869394842209e-06,
      "loss": 1.0881,
      "step": 802
    },
    {
      "epoch": 0.0579344179502904,
      "grad_norm": 7.117307291373191,
      "learning_rate": 3.991827241537153e-06,
      "loss": 1.069,
      "step": 803
    },
    {
      "epoch": 0.0580065654197179,
      "grad_norm": 5.521865383586703,
      "learning_rate": 3.9917849794657846e-06,
      "loss": 0.9794,
      "step": 804
    },
    {
      "epoch": 0.058078712889145415,
      "grad_norm": 5.606381438382658,
      "learning_rate": 3.991742608630412e-06,
      "loss": 1.0262,
      "step": 805
    },
    {
      "epoch": 0.05815086035857292,
      "grad_norm": 4.351152438390806,
      "learning_rate": 3.991700129033349e-06,
      "loss": 1.0314,
      "step": 806
    },
    {
      "epoch": 0.05822300782800043,
      "grad_norm": 10.174785519900484,
      "learning_rate": 3.991657540676915e-06,
      "loss": 0.9562,
      "step": 807
    },
    {
      "epoch": 0.058295155297427945,
      "grad_norm": 38.8368718140981,
      "learning_rate": 3.991614843563435e-06,
      "loss": 1.0606,
      "step": 808
    },
    {
      "epoch": 0.05836730276685545,
      "grad_norm": 5.618607003468371,
      "learning_rate": 3.991572037695242e-06,
      "loss": 1.0416,
      "step": 809
    },
    {
      "epoch": 0.05843945023628296,
      "grad_norm": 6.321452524849473,
      "learning_rate": 3.9915291230746716e-06,
      "loss": 1.0409,
      "step": 810
    },
    {
      "epoch": 0.058511597705710475,
      "grad_norm": 5.582417522116682,
      "learning_rate": 3.9914860997040685e-06,
      "loss": 1.0478,
      "step": 811
    },
    {
      "epoch": 0.05858374517513798,
      "grad_norm": 6.783876534472081,
      "learning_rate": 3.991442967585782e-06,
      "loss": 1.1055,
      "step": 812
    },
    {
      "epoch": 0.05865589264456549,
      "grad_norm": 5.231405298259094,
      "learning_rate": 3.991399726722167e-06,
      "loss": 0.8972,
      "step": 813
    },
    {
      "epoch": 0.058728040113993005,
      "grad_norm": 17.7425366681846,
      "learning_rate": 3.9913563771155845e-06,
      "loss": 1.015,
      "step": 814
    },
    {
      "epoch": 0.05880018758342051,
      "grad_norm": 23.62402212551281,
      "learning_rate": 3.991312918768402e-06,
      "loss": 0.9548,
      "step": 815
    },
    {
      "epoch": 0.05887233505284802,
      "grad_norm": 7.8495904220348445,
      "learning_rate": 3.991269351682993e-06,
      "loss": 1.1388,
      "step": 816
    },
    {
      "epoch": 0.05894448252227553,
      "grad_norm": 18.31853669431851,
      "learning_rate": 3.991225675861736e-06,
      "loss": 1.057,
      "step": 817
    },
    {
      "epoch": 0.05901662999170304,
      "grad_norm": 8.14255352936139,
      "learning_rate": 3.991181891307017e-06,
      "loss": 1.0049,
      "step": 818
    },
    {
      "epoch": 0.05908877746113055,
      "grad_norm": 4.443653192552127,
      "learning_rate": 3.991137998021225e-06,
      "loss": 0.9965,
      "step": 819
    },
    {
      "epoch": 0.05916092493055806,
      "grad_norm": 7.529542341572637,
      "learning_rate": 3.991093996006757e-06,
      "loss": 0.9878,
      "step": 820
    },
    {
      "epoch": 0.05923307239998557,
      "grad_norm": 5.3459459221099515,
      "learning_rate": 3.991049885266018e-06,
      "loss": 1.0662,
      "step": 821
    },
    {
      "epoch": 0.05930521986941308,
      "grad_norm": 11.991707797816087,
      "learning_rate": 3.991005665801415e-06,
      "loss": 0.9139,
      "step": 822
    },
    {
      "epoch": 0.05937736733884059,
      "grad_norm": 5.001234093001739,
      "learning_rate": 3.990961337615363e-06,
      "loss": 1.0244,
      "step": 823
    },
    {
      "epoch": 0.0594495148082681,
      "grad_norm": 19.064193700372236,
      "learning_rate": 3.990916900710282e-06,
      "loss": 1.0496,
      "step": 824
    },
    {
      "epoch": 0.05952166227769561,
      "grad_norm": 10.548906416055981,
      "learning_rate": 3.9908723550886005e-06,
      "loss": 0.9693,
      "step": 825
    },
    {
      "epoch": 0.05959380974712312,
      "grad_norm": 4.629905161535123,
      "learning_rate": 3.990827700752749e-06,
      "loss": 1.0677,
      "step": 826
    },
    {
      "epoch": 0.05966595721655063,
      "grad_norm": 16.726078088793614,
      "learning_rate": 3.990782937705166e-06,
      "loss": 1.1145,
      "step": 827
    },
    {
      "epoch": 0.05973810468597814,
      "grad_norm": 4.767896567355667,
      "learning_rate": 3.9907380659482975e-06,
      "loss": 1.01,
      "step": 828
    },
    {
      "epoch": 0.05981025215540565,
      "grad_norm": 12.795368334114656,
      "learning_rate": 3.990693085484592e-06,
      "loss": 1.0161,
      "step": 829
    },
    {
      "epoch": 0.05988239962483316,
      "grad_norm": 0.8630671516164372,
      "learning_rate": 3.990647996316506e-06,
      "loss": 0.8792,
      "step": 830
    },
    {
      "epoch": 0.059954547094260666,
      "grad_norm": 14.971371606283691,
      "learning_rate": 3.990602798446503e-06,
      "loss": 1.1298,
      "step": 831
    },
    {
      "epoch": 0.06002669456368818,
      "grad_norm": 10.170322282123887,
      "learning_rate": 3.99055749187705e-06,
      "loss": 1.0845,
      "step": 832
    },
    {
      "epoch": 0.06009884203311569,
      "grad_norm": 7.334280068523096,
      "learning_rate": 3.9905120766106205e-06,
      "loss": 0.986,
      "step": 833
    },
    {
      "epoch": 0.060170989502543196,
      "grad_norm": 8.498348300000712,
      "learning_rate": 3.990466552649696e-06,
      "loss": 1.0377,
      "step": 834
    },
    {
      "epoch": 0.06024313697197071,
      "grad_norm": 13.495016237738877,
      "learning_rate": 3.9904209199967605e-06,
      "loss": 1.0577,
      "step": 835
    },
    {
      "epoch": 0.06031528444139822,
      "grad_norm": 0.9282939949695337,
      "learning_rate": 3.990375178654307e-06,
      "loss": 0.8627,
      "step": 836
    },
    {
      "epoch": 0.060387431910825726,
      "grad_norm": 6.097017165520398,
      "learning_rate": 3.9903293286248336e-06,
      "loss": 1.0395,
      "step": 837
    },
    {
      "epoch": 0.06045957938025324,
      "grad_norm": 5.703411941625723,
      "learning_rate": 3.990283369910844e-06,
      "loss": 1.057,
      "step": 838
    },
    {
      "epoch": 0.06053172684968075,
      "grad_norm": 5.571366861721255,
      "learning_rate": 3.9902373025148465e-06,
      "loss": 1.0861,
      "step": 839
    },
    {
      "epoch": 0.060603874319108256,
      "grad_norm": 9.67170540163298,
      "learning_rate": 3.990191126439357e-06,
      "loss": 1.0663,
      "step": 840
    },
    {
      "epoch": 0.06067602178853577,
      "grad_norm": 11.305515219459988,
      "learning_rate": 3.990144841686899e-06,
      "loss": 1.0075,
      "step": 841
    },
    {
      "epoch": 0.06074816925796328,
      "grad_norm": 13.103978352287772,
      "learning_rate": 3.990098448259997e-06,
      "loss": 1.0616,
      "step": 842
    },
    {
      "epoch": 0.060820316727390786,
      "grad_norm": 6.794248156633375,
      "learning_rate": 3.990051946161187e-06,
      "loss": 1.0362,
      "step": 843
    },
    {
      "epoch": 0.0608924641968183,
      "grad_norm": 7.7505945777465195,
      "learning_rate": 3.990005335393007e-06,
      "loss": 0.9909,
      "step": 844
    },
    {
      "epoch": 0.0609646116662458,
      "grad_norm": 4.960747758736143,
      "learning_rate": 3.989958615958002e-06,
      "loss": 0.9743,
      "step": 845
    },
    {
      "epoch": 0.061036759135673316,
      "grad_norm": 18.54174960114294,
      "learning_rate": 3.989911787858724e-06,
      "loss": 1.1379,
      "step": 846
    },
    {
      "epoch": 0.06110890660510083,
      "grad_norm": 27.234464644143916,
      "learning_rate": 3.989864851097729e-06,
      "loss": 1.049,
      "step": 847
    },
    {
      "epoch": 0.061181054074528333,
      "grad_norm": 6.197409968674351,
      "learning_rate": 3.989817805677581e-06,
      "loss": 0.9258,
      "step": 848
    },
    {
      "epoch": 0.061253201543955846,
      "grad_norm": 5.203198864844974,
      "learning_rate": 3.989770651600849e-06,
      "loss": 1.0423,
      "step": 849
    },
    {
      "epoch": 0.06132534901338336,
      "grad_norm": 4.291475211505572,
      "learning_rate": 3.989723388870107e-06,
      "loss": 0.9629,
      "step": 850
    },
    {
      "epoch": 0.061397496482810864,
      "grad_norm": 42.58951539157334,
      "learning_rate": 3.9896760174879365e-06,
      "loss": 1.1279,
      "step": 851
    },
    {
      "epoch": 0.061469643952238376,
      "grad_norm": 0.7838499675312973,
      "learning_rate": 3.989628537456925e-06,
      "loss": 0.8188,
      "step": 852
    },
    {
      "epoch": 0.06154179142166589,
      "grad_norm": 11.925097348757525,
      "learning_rate": 3.989580948779664e-06,
      "loss": 0.9662,
      "step": 853
    },
    {
      "epoch": 0.061613938891093394,
      "grad_norm": 6.849850239474047,
      "learning_rate": 3.989533251458752e-06,
      "loss": 1.0908,
      "step": 854
    },
    {
      "epoch": 0.061686086360520906,
      "grad_norm": 6.094723354755004,
      "learning_rate": 3.989485445496794e-06,
      "loss": 0.9037,
      "step": 855
    },
    {
      "epoch": 0.06175823382994841,
      "grad_norm": 9.219787746146023,
      "learning_rate": 3.989437530896402e-06,
      "loss": 0.9011,
      "step": 856
    },
    {
      "epoch": 0.061830381299375924,
      "grad_norm": 13.24592383682677,
      "learning_rate": 3.989389507660191e-06,
      "loss": 1.0303,
      "step": 857
    },
    {
      "epoch": 0.061902528768803436,
      "grad_norm": 7.32480336208508,
      "learning_rate": 3.989341375790782e-06,
      "loss": 1.1365,
      "step": 858
    },
    {
      "epoch": 0.06197467623823094,
      "grad_norm": 6.780026373797858,
      "learning_rate": 3.989293135290805e-06,
      "loss": 1.0436,
      "step": 859
    },
    {
      "epoch": 0.062046823707658454,
      "grad_norm": 7.291526255391023,
      "learning_rate": 3.989244786162895e-06,
      "loss": 1.0199,
      "step": 860
    },
    {
      "epoch": 0.062118971177085966,
      "grad_norm": 5.663479226433358,
      "learning_rate": 3.9891963284096904e-06,
      "loss": 0.9608,
      "step": 861
    },
    {
      "epoch": 0.06219111864651347,
      "grad_norm": 0.8659456313380249,
      "learning_rate": 3.989147762033838e-06,
      "loss": 0.8428,
      "step": 862
    },
    {
      "epoch": 0.062263266115940984,
      "grad_norm": 4.271343268988481,
      "learning_rate": 3.9890990870379905e-06,
      "loss": 1.0258,
      "step": 863
    },
    {
      "epoch": 0.062335413585368496,
      "grad_norm": 4.643809243975639,
      "learning_rate": 3.989050303424805e-06,
      "loss": 0.9935,
      "step": 864
    },
    {
      "epoch": 0.062407561054796,
      "grad_norm": 7.577721024590023,
      "learning_rate": 3.989001411196946e-06,
      "loss": 1.0844,
      "step": 865
    },
    {
      "epoch": 0.062479708524223514,
      "grad_norm": 5.849492642825256,
      "learning_rate": 3.988952410357081e-06,
      "loss": 0.9394,
      "step": 866
    },
    {
      "epoch": 0.06255185599365103,
      "grad_norm": 7.680518725281796,
      "learning_rate": 3.98890330090789e-06,
      "loss": 0.9695,
      "step": 867
    },
    {
      "epoch": 0.06262400346307853,
      "grad_norm": 18.01068137397049,
      "learning_rate": 3.988854082852052e-06,
      "loss": 1.0092,
      "step": 868
    },
    {
      "epoch": 0.06269615093250604,
      "grad_norm": 3.6372132912606387,
      "learning_rate": 3.988804756192254e-06,
      "loss": 1.0919,
      "step": 869
    },
    {
      "epoch": 0.06276829840193356,
      "grad_norm": 8.065544086037134,
      "learning_rate": 3.988755320931192e-06,
      "loss": 1.0302,
      "step": 870
    },
    {
      "epoch": 0.06284044587136106,
      "grad_norm": 9.054434719292944,
      "learning_rate": 3.988705777071563e-06,
      "loss": 1.0678,
      "step": 871
    },
    {
      "epoch": 0.06291259334078857,
      "grad_norm": 0.7288114954289935,
      "learning_rate": 3.988656124616074e-06,
      "loss": 0.8777,
      "step": 872
    },
    {
      "epoch": 0.06298474081021609,
      "grad_norm": 3.3231304103942874,
      "learning_rate": 3.9886063635674355e-06,
      "loss": 1.0388,
      "step": 873
    },
    {
      "epoch": 0.06305688827964359,
      "grad_norm": 0.752587860529159,
      "learning_rate": 3.988556493928365e-06,
      "loss": 0.8723,
      "step": 874
    },
    {
      "epoch": 0.0631290357490711,
      "grad_norm": 7.97213086981572,
      "learning_rate": 3.988506515701586e-06,
      "loss": 1.0155,
      "step": 875
    },
    {
      "epoch": 0.06320118321849862,
      "grad_norm": 10.232895885482515,
      "learning_rate": 3.9884564288898275e-06,
      "loss": 1.0518,
      "step": 876
    },
    {
      "epoch": 0.06327333068792612,
      "grad_norm": 9.208505712308229,
      "learning_rate": 3.988406233495824e-06,
      "loss": 0.9622,
      "step": 877
    },
    {
      "epoch": 0.06334547815735363,
      "grad_norm": 5.289026257434449,
      "learning_rate": 3.988355929522317e-06,
      "loss": 1.0104,
      "step": 878
    },
    {
      "epoch": 0.06341762562678115,
      "grad_norm": 0.8897599486426936,
      "learning_rate": 3.988305516972053e-06,
      "loss": 0.8874,
      "step": 879
    },
    {
      "epoch": 0.06348977309620865,
      "grad_norm": 7.990010700624484,
      "learning_rate": 3.988254995847786e-06,
      "loss": 1.1352,
      "step": 880
    },
    {
      "epoch": 0.06356192056563616,
      "grad_norm": 11.873970709915108,
      "learning_rate": 3.988204366152273e-06,
      "loss": 1.046,
      "step": 881
    },
    {
      "epoch": 0.06363406803506368,
      "grad_norm": 8.31290153738241,
      "learning_rate": 3.988153627888281e-06,
      "loss": 1.0495,
      "step": 882
    },
    {
      "epoch": 0.06370621550449118,
      "grad_norm": 11.35045549596228,
      "learning_rate": 3.988102781058578e-06,
      "loss": 1.0543,
      "step": 883
    },
    {
      "epoch": 0.06377836297391869,
      "grad_norm": 7.260096655588431,
      "learning_rate": 3.988051825665942e-06,
      "loss": 1.1212,
      "step": 884
    },
    {
      "epoch": 0.0638505104433462,
      "grad_norm": 4.85272841641649,
      "learning_rate": 3.988000761713156e-06,
      "loss": 1.1185,
      "step": 885
    },
    {
      "epoch": 0.06392265791277371,
      "grad_norm": 11.752070467722712,
      "learning_rate": 3.987949589203007e-06,
      "loss": 1.0537,
      "step": 886
    },
    {
      "epoch": 0.06399480538220122,
      "grad_norm": 14.252886563451804,
      "learning_rate": 3.987898308138291e-06,
      "loss": 1.0662,
      "step": 887
    },
    {
      "epoch": 0.06406695285162872,
      "grad_norm": 37.66805604605927,
      "learning_rate": 3.9878469185218065e-06,
      "loss": 1.0969,
      "step": 888
    },
    {
      "epoch": 0.06413910032105624,
      "grad_norm": 6.251489080423659,
      "learning_rate": 3.98779542035636e-06,
      "loss": 0.9752,
      "step": 889
    },
    {
      "epoch": 0.06421124779048375,
      "grad_norm": 6.752964357851165,
      "learning_rate": 3.987743813644765e-06,
      "loss": 1.2275,
      "step": 890
    },
    {
      "epoch": 0.06428339525991125,
      "grad_norm": 3.768820518732219,
      "learning_rate": 3.987692098389839e-06,
      "loss": 0.9676,
      "step": 891
    },
    {
      "epoch": 0.06435554272933877,
      "grad_norm": 7.377256920952013,
      "learning_rate": 3.987640274594405e-06,
      "loss": 1.0241,
      "step": 892
    },
    {
      "epoch": 0.06442769019876628,
      "grad_norm": 5.067942578298289,
      "learning_rate": 3.9875883422612944e-06,
      "loss": 1.129,
      "step": 893
    },
    {
      "epoch": 0.06449983766819378,
      "grad_norm": 11.122633092885215,
      "learning_rate": 3.9875363013933416e-06,
      "loss": 0.9709,
      "step": 894
    },
    {
      "epoch": 0.0645719851376213,
      "grad_norm": 4.945551428062629,
      "learning_rate": 3.987484151993389e-06,
      "loss": 0.9883,
      "step": 895
    },
    {
      "epoch": 0.06464413260704881,
      "grad_norm": 13.618122009671984,
      "learning_rate": 3.987431894064285e-06,
      "loss": 1.0526,
      "step": 896
    },
    {
      "epoch": 0.06471628007647631,
      "grad_norm": 25.024975317638926,
      "learning_rate": 3.987379527608881e-06,
      "loss": 1.0338,
      "step": 897
    },
    {
      "epoch": 0.06478842754590383,
      "grad_norm": 5.09308624623584,
      "learning_rate": 3.987327052630039e-06,
      "loss": 0.9747,
      "step": 898
    },
    {
      "epoch": 0.06486057501533134,
      "grad_norm": 14.926506050839299,
      "learning_rate": 3.987274469130624e-06,
      "loss": 0.9876,
      "step": 899
    },
    {
      "epoch": 0.06493272248475884,
      "grad_norm": 12.258477216680504,
      "learning_rate": 3.987221777113507e-06,
      "loss": 1.0159,
      "step": 900
    },
    {
      "epoch": 0.06500486995418636,
      "grad_norm": 61.252776164763546,
      "learning_rate": 3.987168976581564e-06,
      "loss": 0.9652,
      "step": 901
    },
    {
      "epoch": 0.06507701742361387,
      "grad_norm": 13.899870156978423,
      "learning_rate": 3.98711606753768e-06,
      "loss": 1.0093,
      "step": 902
    },
    {
      "epoch": 0.06514916489304137,
      "grad_norm": 7.208982522944086,
      "learning_rate": 3.987063049984743e-06,
      "loss": 1.0458,
      "step": 903
    },
    {
      "epoch": 0.06522131236246889,
      "grad_norm": 19.633735802005912,
      "learning_rate": 3.987009923925649e-06,
      "loss": 0.95,
      "step": 904
    },
    {
      "epoch": 0.0652934598318964,
      "grad_norm": 13.995006760748579,
      "learning_rate": 3.9869566893633e-06,
      "loss": 0.9509,
      "step": 905
    },
    {
      "epoch": 0.0653656073013239,
      "grad_norm": 10.77850926157488,
      "learning_rate": 3.9869033463006e-06,
      "loss": 1.0055,
      "step": 906
    },
    {
      "epoch": 0.06543775477075142,
      "grad_norm": 7.4642739390386605,
      "learning_rate": 3.986849894740464e-06,
      "loss": 0.8768,
      "step": 907
    },
    {
      "epoch": 0.06550990224017893,
      "grad_norm": 13.18471349411412,
      "learning_rate": 3.9867963346858105e-06,
      "loss": 1.0141,
      "step": 908
    },
    {
      "epoch": 0.06558204970960643,
      "grad_norm": 6.830499231906851,
      "learning_rate": 3.986742666139565e-06,
      "loss": 0.8709,
      "step": 909
    },
    {
      "epoch": 0.06565419717903395,
      "grad_norm": 9.213665093255972,
      "learning_rate": 3.986688889104655e-06,
      "loss": 0.9972,
      "step": 910
    },
    {
      "epoch": 0.06572634464846146,
      "grad_norm": 12.13196858985525,
      "learning_rate": 3.986635003584021e-06,
      "loss": 1.0533,
      "step": 911
    },
    {
      "epoch": 0.06579849211788896,
      "grad_norm": 5.507763844809473,
      "learning_rate": 3.986581009580603e-06,
      "loss": 1.0393,
      "step": 912
    },
    {
      "epoch": 0.06587063958731647,
      "grad_norm": 8.437616192052975,
      "learning_rate": 3.98652690709735e-06,
      "loss": 1.0343,
      "step": 913
    },
    {
      "epoch": 0.06594278705674399,
      "grad_norm": 10.232610698952081,
      "learning_rate": 3.986472696137217e-06,
      "loss": 1.1637,
      "step": 914
    },
    {
      "epoch": 0.06601493452617149,
      "grad_norm": 42.86941216092702,
      "learning_rate": 3.986418376703163e-06,
      "loss": 1.0788,
      "step": 915
    },
    {
      "epoch": 0.066087081995599,
      "grad_norm": 5.440338095649084,
      "learning_rate": 3.9863639487981555e-06,
      "loss": 1.0454,
      "step": 916
    },
    {
      "epoch": 0.06615922946502652,
      "grad_norm": 22.74469535940606,
      "learning_rate": 3.986309412425166e-06,
      "loss": 1.0495,
      "step": 917
    },
    {
      "epoch": 0.06623137693445402,
      "grad_norm": 8.620434865935247,
      "learning_rate": 3.986254767587173e-06,
      "loss": 1.0144,
      "step": 918
    },
    {
      "epoch": 0.06630352440388153,
      "grad_norm": 7.2529945766124175,
      "learning_rate": 3.986200014287159e-06,
      "loss": 1.0512,
      "step": 919
    },
    {
      "epoch": 0.06637567187330905,
      "grad_norm": 6.47824564415217,
      "learning_rate": 3.986145152528115e-06,
      "loss": 1.0136,
      "step": 920
    },
    {
      "epoch": 0.06644781934273655,
      "grad_norm": 5.637975688750649,
      "learning_rate": 3.986090182313037e-06,
      "loss": 0.9769,
      "step": 921
    },
    {
      "epoch": 0.06651996681216406,
      "grad_norm": 27.107679611971818,
      "learning_rate": 3.986035103644926e-06,
      "loss": 1.0794,
      "step": 922
    },
    {
      "epoch": 0.06659211428159158,
      "grad_norm": 5.421217469022576,
      "learning_rate": 3.985979916526791e-06,
      "loss": 1.1123,
      "step": 923
    },
    {
      "epoch": 0.06666426175101908,
      "grad_norm": 18.69909512375574,
      "learning_rate": 3.985924620961644e-06,
      "loss": 1.1418,
      "step": 924
    },
    {
      "epoch": 0.06673640922044659,
      "grad_norm": 0.7102398226236061,
      "learning_rate": 3.985869216952505e-06,
      "loss": 0.819,
      "step": 925
    },
    {
      "epoch": 0.06680855668987411,
      "grad_norm": 24.548195281966503,
      "learning_rate": 3.9858137045023996e-06,
      "loss": 1.0671,
      "step": 926
    },
    {
      "epoch": 0.06688070415930161,
      "grad_norm": 11.817002180220825,
      "learning_rate": 3.9857580836143595e-06,
      "loss": 1.0324,
      "step": 927
    },
    {
      "epoch": 0.06695285162872912,
      "grad_norm": 7.187572445711662,
      "learning_rate": 3.9857023542914205e-06,
      "loss": 1.0071,
      "step": 928
    },
    {
      "epoch": 0.06702499909815664,
      "grad_norm": 12.19143566261271,
      "learning_rate": 3.985646516536628e-06,
      "loss": 1.0611,
      "step": 929
    },
    {
      "epoch": 0.06709714656758414,
      "grad_norm": 23.440585734369193,
      "learning_rate": 3.985590570353029e-06,
      "loss": 1.0604,
      "step": 930
    },
    {
      "epoch": 0.06716929403701165,
      "grad_norm": 9.984997749309258,
      "learning_rate": 3.985534515743679e-06,
      "loss": 0.9454,
      "step": 931
    },
    {
      "epoch": 0.06724144150643917,
      "grad_norm": 7.690465339699106,
      "learning_rate": 3.98547835271164e-06,
      "loss": 1.1264,
      "step": 932
    },
    {
      "epoch": 0.06731358897586667,
      "grad_norm": 11.145095677194009,
      "learning_rate": 3.985422081259978e-06,
      "loss": 1.0524,
      "step": 933
    },
    {
      "epoch": 0.06738573644529418,
      "grad_norm": 10.354436879818454,
      "learning_rate": 3.985365701391766e-06,
      "loss": 1.0926,
      "step": 934
    },
    {
      "epoch": 0.0674578839147217,
      "grad_norm": 14.786794513760833,
      "learning_rate": 3.985309213110082e-06,
      "loss": 1.0464,
      "step": 935
    },
    {
      "epoch": 0.0675300313841492,
      "grad_norm": 11.07714185131879,
      "learning_rate": 3.985252616418012e-06,
      "loss": 1.015,
      "step": 936
    },
    {
      "epoch": 0.06760217885357671,
      "grad_norm": 7.638561758107469,
      "learning_rate": 3.985195911318646e-06,
      "loss": 0.9823,
      "step": 937
    },
    {
      "epoch": 0.06767432632300423,
      "grad_norm": 6.017070329110254,
      "learning_rate": 3.9851390978150795e-06,
      "loss": 0.969,
      "step": 938
    },
    {
      "epoch": 0.06774647379243173,
      "grad_norm": 13.468472435872577,
      "learning_rate": 3.985082175910416e-06,
      "loss": 1.0672,
      "step": 939
    },
    {
      "epoch": 0.06781862126185924,
      "grad_norm": 35.27677949976443,
      "learning_rate": 3.985025145607763e-06,
      "loss": 1.1138,
      "step": 940
    },
    {
      "epoch": 0.06789076873128674,
      "grad_norm": 6.77461952182758,
      "learning_rate": 3.984968006910235e-06,
      "loss": 0.9062,
      "step": 941
    },
    {
      "epoch": 0.06796291620071426,
      "grad_norm": 35.194130633596956,
      "learning_rate": 3.984910759820953e-06,
      "loss": 0.9311,
      "step": 942
    },
    {
      "epoch": 0.06803506367014177,
      "grad_norm": 8.192551779701239,
      "learning_rate": 3.984853404343042e-06,
      "loss": 0.9654,
      "step": 943
    },
    {
      "epoch": 0.06810721113956927,
      "grad_norm": 8.579864492346674,
      "learning_rate": 3.984795940479634e-06,
      "loss": 0.9126,
      "step": 944
    },
    {
      "epoch": 0.06817935860899679,
      "grad_norm": 15.884800086223839,
      "learning_rate": 3.984738368233867e-06,
      "loss": 1.045,
      "step": 945
    },
    {
      "epoch": 0.0682515060784243,
      "grad_norm": 10.12482066054521,
      "learning_rate": 3.9846806876088855e-06,
      "loss": 1.0755,
      "step": 946
    },
    {
      "epoch": 0.0683236535478518,
      "grad_norm": 32.63741232992821,
      "learning_rate": 3.984622898607838e-06,
      "loss": 1.1221,
      "step": 947
    },
    {
      "epoch": 0.06839580101727932,
      "grad_norm": 28.841682533569145,
      "learning_rate": 3.984565001233881e-06,
      "loss": 1.0654,
      "step": 948
    },
    {
      "epoch": 0.06846794848670683,
      "grad_norm": 18.69697580617986,
      "learning_rate": 3.984506995490177e-06,
      "loss": 1.0481,
      "step": 949
    },
    {
      "epoch": 0.06854009595613433,
      "grad_norm": 0.80583553454609,
      "learning_rate": 3.984448881379891e-06,
      "loss": 0.8422,
      "step": 950
    },
    {
      "epoch": 0.06861224342556185,
      "grad_norm": 7.236901651529887,
      "learning_rate": 3.984390658906198e-06,
      "loss": 1.0718,
      "step": 951
    },
    {
      "epoch": 0.06868439089498936,
      "grad_norm": 7.4578174718124535,
      "learning_rate": 3.984332328072278e-06,
      "loss": 1.1076,
      "step": 952
    },
    {
      "epoch": 0.06875653836441686,
      "grad_norm": 18.716169023337983,
      "learning_rate": 3.984273888881314e-06,
      "loss": 1.0057,
      "step": 953
    },
    {
      "epoch": 0.06882868583384438,
      "grad_norm": 10.423186395542894,
      "learning_rate": 3.9842153413364995e-06,
      "loss": 1.0895,
      "step": 954
    },
    {
      "epoch": 0.06890083330327189,
      "grad_norm": 20.937791509164715,
      "learning_rate": 3.984156685441031e-06,
      "loss": 0.9481,
      "step": 955
    },
    {
      "epoch": 0.0689729807726994,
      "grad_norm": 9.237053961818924,
      "learning_rate": 3.984097921198109e-06,
      "loss": 0.9504,
      "step": 956
    },
    {
      "epoch": 0.06904512824212691,
      "grad_norm": 12.753665172874694,
      "learning_rate": 3.984039048610946e-06,
      "loss": 0.9796,
      "step": 957
    },
    {
      "epoch": 0.06911727571155442,
      "grad_norm": 58.80055587402002,
      "learning_rate": 3.983980067682755e-06,
      "loss": 1.0091,
      "step": 958
    },
    {
      "epoch": 0.06918942318098192,
      "grad_norm": 6.296605712584825,
      "learning_rate": 3.983920978416757e-06,
      "loss": 0.9433,
      "step": 959
    },
    {
      "epoch": 0.06926157065040944,
      "grad_norm": 18.479196759887184,
      "learning_rate": 3.983861780816178e-06,
      "loss": 1.0224,
      "step": 960
    },
    {
      "epoch": 0.06933371811983695,
      "grad_norm": 7.373253308368378,
      "learning_rate": 3.983802474884252e-06,
      "loss": 1.0796,
      "step": 961
    },
    {
      "epoch": 0.06940586558926445,
      "grad_norm": 6.283858180028332,
      "learning_rate": 3.9837430606242154e-06,
      "loss": 1.0568,
      "step": 962
    },
    {
      "epoch": 0.06947801305869197,
      "grad_norm": 6.524143276753314,
      "learning_rate": 3.983683538039314e-06,
      "loss": 1.047,
      "step": 963
    },
    {
      "epoch": 0.06955016052811948,
      "grad_norm": 30.774497593761584,
      "learning_rate": 3.983623907132799e-06,
      "loss": 1.1346,
      "step": 964
    },
    {
      "epoch": 0.06962230799754698,
      "grad_norm": 21.518955125365636,
      "learning_rate": 3.983564167907925e-06,
      "loss": 0.9576,
      "step": 965
    },
    {
      "epoch": 0.06969445546697449,
      "grad_norm": 0.8191628892862562,
      "learning_rate": 3.983504320367955e-06,
      "loss": 0.8825,
      "step": 966
    },
    {
      "epoch": 0.06976660293640201,
      "grad_norm": 9.772884411183602,
      "learning_rate": 3.983444364516156e-06,
      "loss": 1.05,
      "step": 967
    },
    {
      "epoch": 0.06983875040582951,
      "grad_norm": 0.7651961935818086,
      "learning_rate": 3.983384300355803e-06,
      "loss": 0.895,
      "step": 968
    },
    {
      "epoch": 0.06991089787525702,
      "grad_norm": 0.8544808174466753,
      "learning_rate": 3.983324127890176e-06,
      "loss": 0.8997,
      "step": 969
    },
    {
      "epoch": 0.06998304534468454,
      "grad_norm": 4.861330086756333,
      "learning_rate": 3.98326384712256e-06,
      "loss": 1.0225,
      "step": 970
    },
    {
      "epoch": 0.07005519281411204,
      "grad_norm": 5.568102811631484,
      "learning_rate": 3.983203458056247e-06,
      "loss": 0.9816,
      "step": 971
    },
    {
      "epoch": 0.07012734028353955,
      "grad_norm": 9.277314144689605,
      "learning_rate": 3.983142960694535e-06,
      "loss": 0.9567,
      "step": 972
    },
    {
      "epoch": 0.07019948775296707,
      "grad_norm": 1.5154011354584886,
      "learning_rate": 3.9830823550407275e-06,
      "loss": 0.855,
      "step": 973
    },
    {
      "epoch": 0.07027163522239457,
      "grad_norm": 8.040031411788082,
      "learning_rate": 3.983021641098134e-06,
      "loss": 0.9976,
      "step": 974
    },
    {
      "epoch": 0.07034378269182208,
      "grad_norm": 18.21512228703504,
      "learning_rate": 3.982960818870069e-06,
      "loss": 1.035,
      "step": 975
    },
    {
      "epoch": 0.0704159301612496,
      "grad_norm": 6.727023260660163,
      "learning_rate": 3.982899888359854e-06,
      "loss": 1.0682,
      "step": 976
    },
    {
      "epoch": 0.0704880776306771,
      "grad_norm": 12.669492975099427,
      "learning_rate": 3.982838849570817e-06,
      "loss": 1.0388,
      "step": 977
    },
    {
      "epoch": 0.07056022510010461,
      "grad_norm": 12.431062959133676,
      "learning_rate": 3.982777702506291e-06,
      "loss": 0.8978,
      "step": 978
    },
    {
      "epoch": 0.07063237256953213,
      "grad_norm": 6.184855811274018,
      "learning_rate": 3.982716447169614e-06,
      "loss": 1.0045,
      "step": 979
    },
    {
      "epoch": 0.07070452003895963,
      "grad_norm": 11.448993515355331,
      "learning_rate": 3.982655083564131e-06,
      "loss": 0.9751,
      "step": 980
    },
    {
      "epoch": 0.07077666750838714,
      "grad_norm": 5.729902888302297,
      "learning_rate": 3.982593611693195e-06,
      "loss": 1.0801,
      "step": 981
    },
    {
      "epoch": 0.07084881497781466,
      "grad_norm": 38.85549211602042,
      "learning_rate": 3.98253203156016e-06,
      "loss": 0.986,
      "step": 982
    },
    {
      "epoch": 0.07092096244724216,
      "grad_norm": 33.98187311193002,
      "learning_rate": 3.98247034316839e-06,
      "loss": 1.0781,
      "step": 983
    },
    {
      "epoch": 0.07099310991666967,
      "grad_norm": 7.512166137385647,
      "learning_rate": 3.982408546521254e-06,
      "loss": 0.946,
      "step": 984
    },
    {
      "epoch": 0.07106525738609719,
      "grad_norm": 11.369906112833494,
      "learning_rate": 3.9823466416221254e-06,
      "loss": 1.0318,
      "step": 985
    },
    {
      "epoch": 0.0711374048555247,
      "grad_norm": 14.277629101115613,
      "learning_rate": 3.9822846284743856e-06,
      "loss": 1.0449,
      "step": 986
    },
    {
      "epoch": 0.0712095523249522,
      "grad_norm": 8.06811328033369,
      "learning_rate": 3.9822225070814195e-06,
      "loss": 1.134,
      "step": 987
    },
    {
      "epoch": 0.07128169979437972,
      "grad_norm": 10.214915451469192,
      "learning_rate": 3.982160277446621e-06,
      "loss": 1.0514,
      "step": 988
    },
    {
      "epoch": 0.07135384726380722,
      "grad_norm": 11.816432722077785,
      "learning_rate": 3.982097939573388e-06,
      "loss": 1.048,
      "step": 989
    },
    {
      "epoch": 0.07142599473323473,
      "grad_norm": 8.840639895240853,
      "learning_rate": 3.982035493465123e-06,
      "loss": 1.0139,
      "step": 990
    },
    {
      "epoch": 0.07149814220266223,
      "grad_norm": 17.99724727350903,
      "learning_rate": 3.981972939125237e-06,
      "loss": 1.0999,
      "step": 991
    },
    {
      "epoch": 0.07157028967208975,
      "grad_norm": 20.322828654051904,
      "learning_rate": 3.981910276557146e-06,
      "loss": 0.9558,
      "step": 992
    },
    {
      "epoch": 0.07164243714151726,
      "grad_norm": 61.63412117145928,
      "learning_rate": 3.9818475057642715e-06,
      "loss": 0.9317,
      "step": 993
    },
    {
      "epoch": 0.07171458461094476,
      "grad_norm": 11.009734961206632,
      "learning_rate": 3.981784626750041e-06,
      "loss": 1.0564,
      "step": 994
    },
    {
      "epoch": 0.07178673208037228,
      "grad_norm": 12.139174523231336,
      "learning_rate": 3.981721639517889e-06,
      "loss": 1.0286,
      "step": 995
    },
    {
      "epoch": 0.07185887954979979,
      "grad_norm": 15.797198102686052,
      "learning_rate": 3.981658544071254e-06,
      "loss": 1.0137,
      "step": 996
    },
    {
      "epoch": 0.0719310270192273,
      "grad_norm": 10.311724639412288,
      "learning_rate": 3.981595340413582e-06,
      "loss": 1.1434,
      "step": 997
    },
    {
      "epoch": 0.07200317448865481,
      "grad_norm": 14.892141515116286,
      "learning_rate": 3.981532028548324e-06,
      "loss": 1.024,
      "step": 998
    },
    {
      "epoch": 0.07207532195808232,
      "grad_norm": 7.8503336841587,
      "learning_rate": 3.9814686084789365e-06,
      "loss": 0.9779,
      "step": 999
    },
    {
      "epoch": 0.07214746942750982,
      "grad_norm": 17.718756028162556,
      "learning_rate": 3.981405080208885e-06,
      "loss": 0.9829,
      "step": 1000
    },
    {
      "epoch": 0.07221961689693734,
      "grad_norm": 6.051742920573294,
      "learning_rate": 3.981341443741636e-06,
      "loss": 1.0623,
      "step": 1001
    },
    {
      "epoch": 0.07229176436636485,
      "grad_norm": 12.344207986954665,
      "learning_rate": 3.981277699080665e-06,
      "loss": 0.9022,
      "step": 1002
    },
    {
      "epoch": 0.07236391183579235,
      "grad_norm": 7.956508194179339,
      "learning_rate": 3.981213846229454e-06,
      "loss": 1.0897,
      "step": 1003
    },
    {
      "epoch": 0.07243605930521987,
      "grad_norm": 13.430310122274495,
      "learning_rate": 3.981149885191489e-06,
      "loss": 0.9779,
      "step": 1004
    },
    {
      "epoch": 0.07250820677464738,
      "grad_norm": 15.34855846880457,
      "learning_rate": 3.9810858159702625e-06,
      "loss": 1.0687,
      "step": 1005
    },
    {
      "epoch": 0.07258035424407489,
      "grad_norm": 16.805978637638155,
      "learning_rate": 3.981021638569273e-06,
      "loss": 1.0801,
      "step": 1006
    },
    {
      "epoch": 0.0726525017135024,
      "grad_norm": 42.52401378100232,
      "learning_rate": 3.9809573529920255e-06,
      "loss": 1.0417,
      "step": 1007
    },
    {
      "epoch": 0.07272464918292991,
      "grad_norm": 9.465485768696837,
      "learning_rate": 3.980892959242031e-06,
      "loss": 1.1156,
      "step": 1008
    },
    {
      "epoch": 0.07279679665235742,
      "grad_norm": 12.331512179248461,
      "learning_rate": 3.9808284573228045e-06,
      "loss": 1.1305,
      "step": 1009
    },
    {
      "epoch": 0.07286894412178493,
      "grad_norm": 17.56737692239777,
      "learning_rate": 3.980763847237869e-06,
      "loss": 0.9189,
      "step": 1010
    },
    {
      "epoch": 0.07294109159121244,
      "grad_norm": 8.604682765257685,
      "learning_rate": 3.980699128990752e-06,
      "loss": 1.099,
      "step": 1011
    },
    {
      "epoch": 0.07301323906063995,
      "grad_norm": 13.223015712433,
      "learning_rate": 3.980634302584987e-06,
      "loss": 1.0849,
      "step": 1012
    },
    {
      "epoch": 0.07308538653006746,
      "grad_norm": 11.42429477510093,
      "learning_rate": 3.980569368024116e-06,
      "loss": 1.062,
      "step": 1013
    },
    {
      "epoch": 0.07315753399949497,
      "grad_norm": 11.350304929695515,
      "learning_rate": 3.980504325311683e-06,
      "loss": 1.0707,
      "step": 1014
    },
    {
      "epoch": 0.07322968146892248,
      "grad_norm": 11.725684797058163,
      "learning_rate": 3.980439174451241e-06,
      "loss": 0.9809,
      "step": 1015
    },
    {
      "epoch": 0.07330182893834998,
      "grad_norm": 11.796002528788287,
      "learning_rate": 3.980373915446346e-06,
      "loss": 1.1393,
      "step": 1016
    },
    {
      "epoch": 0.0733739764077775,
      "grad_norm": 65.12467010473067,
      "learning_rate": 3.980308548300564e-06,
      "loss": 0.9944,
      "step": 1017
    },
    {
      "epoch": 0.073446123877205,
      "grad_norm": 0.6793763555672886,
      "learning_rate": 3.980243073017462e-06,
      "loss": 0.7454,
      "step": 1018
    },
    {
      "epoch": 0.07351827134663251,
      "grad_norm": 9.114889317780586,
      "learning_rate": 3.980177489600616e-06,
      "loss": 0.9707,
      "step": 1019
    },
    {
      "epoch": 0.07359041881606003,
      "grad_norm": 0.876686616430057,
      "learning_rate": 3.980111798053609e-06,
      "loss": 0.8772,
      "step": 1020
    },
    {
      "epoch": 0.07366256628548754,
      "grad_norm": 9.728235062645332,
      "learning_rate": 3.980045998380026e-06,
      "loss": 1.0372,
      "step": 1021
    },
    {
      "epoch": 0.07373471375491504,
      "grad_norm": 32.86223165780513,
      "learning_rate": 3.9799800905834604e-06,
      "loss": 0.9711,
      "step": 1022
    },
    {
      "epoch": 0.07380686122434256,
      "grad_norm": 22.29893478791714,
      "learning_rate": 3.979914074667513e-06,
      "loss": 1.0418,
      "step": 1023
    },
    {
      "epoch": 0.07387900869377007,
      "grad_norm": 17.709865704225958,
      "learning_rate": 3.979847950635786e-06,
      "loss": 1.0213,
      "step": 1024
    },
    {
      "epoch": 0.07395115616319757,
      "grad_norm": 8.057235536747632,
      "learning_rate": 3.979781718491894e-06,
      "loss": 0.9797,
      "step": 1025
    },
    {
      "epoch": 0.07402330363262509,
      "grad_norm": 7.270640489995749,
      "learning_rate": 3.9797153782394485e-06,
      "loss": 1.0207,
      "step": 1026
    },
    {
      "epoch": 0.0740954511020526,
      "grad_norm": 6.6290268086230695,
      "learning_rate": 3.979648929882076e-06,
      "loss": 1.0763,
      "step": 1027
    },
    {
      "epoch": 0.0741675985714801,
      "grad_norm": 14.928728281423904,
      "learning_rate": 3.979582373423404e-06,
      "loss": 1.0537,
      "step": 1028
    },
    {
      "epoch": 0.07423974604090762,
      "grad_norm": 7.05500976907877,
      "learning_rate": 3.9795157088670665e-06,
      "loss": 1.0037,
      "step": 1029
    },
    {
      "epoch": 0.07431189351033513,
      "grad_norm": 11.341493324302373,
      "learning_rate": 3.979448936216703e-06,
      "loss": 1.0646,
      "step": 1030
    },
    {
      "epoch": 0.07438404097976263,
      "grad_norm": 13.501341576496552,
      "learning_rate": 3.979382055475962e-06,
      "loss": 0.9531,
      "step": 1031
    },
    {
      "epoch": 0.07445618844919015,
      "grad_norm": 10.646804874727787,
      "learning_rate": 3.979315066648495e-06,
      "loss": 1.0435,
      "step": 1032
    },
    {
      "epoch": 0.07452833591861766,
      "grad_norm": 15.125428248876593,
      "learning_rate": 3.979247969737958e-06,
      "loss": 0.9587,
      "step": 1033
    },
    {
      "epoch": 0.07460048338804516,
      "grad_norm": 8.769756139237808,
      "learning_rate": 3.979180764748017e-06,
      "loss": 1.0588,
      "step": 1034
    },
    {
      "epoch": 0.07467263085747268,
      "grad_norm": 14.781156575913798,
      "learning_rate": 3.979113451682341e-06,
      "loss": 0.9769,
      "step": 1035
    },
    {
      "epoch": 0.07474477832690019,
      "grad_norm": 16.187548073951298,
      "learning_rate": 3.9790460305446064e-06,
      "loss": 0.9927,
      "step": 1036
    },
    {
      "epoch": 0.07481692579632769,
      "grad_norm": 12.083481640837343,
      "learning_rate": 3.978978501338493e-06,
      "loss": 1.0567,
      "step": 1037
    },
    {
      "epoch": 0.07488907326575521,
      "grad_norm": 11.768553674552379,
      "learning_rate": 3.97891086406769e-06,
      "loss": 1.023,
      "step": 1038
    },
    {
      "epoch": 0.07496122073518272,
      "grad_norm": 22.39214328854403,
      "learning_rate": 3.978843118735891e-06,
      "loss": 0.9543,
      "step": 1039
    },
    {
      "epoch": 0.07503336820461022,
      "grad_norm": 13.543434066854765,
      "learning_rate": 3.978775265346794e-06,
      "loss": 0.9297,
      "step": 1040
    },
    {
      "epoch": 0.07510551567403774,
      "grad_norm": 13.121722293482119,
      "learning_rate": 3.978707303904106e-06,
      "loss": 1.0429,
      "step": 1041
    },
    {
      "epoch": 0.07517766314346525,
      "grad_norm": 9.706408352122041,
      "learning_rate": 3.978639234411536e-06,
      "loss": 1.0177,
      "step": 1042
    },
    {
      "epoch": 0.07524981061289275,
      "grad_norm": 9.001087123063117,
      "learning_rate": 3.978571056872803e-06,
      "loss": 1.0517,
      "step": 1043
    },
    {
      "epoch": 0.07532195808232026,
      "grad_norm": 19.671663242257804,
      "learning_rate": 3.978502771291628e-06,
      "loss": 1.0579,
      "step": 1044
    },
    {
      "epoch": 0.07539410555174778,
      "grad_norm": 23.020995798587727,
      "learning_rate": 3.978434377671741e-06,
      "loss": 1.0688,
      "step": 1045
    },
    {
      "epoch": 0.07546625302117528,
      "grad_norm": 14.74288561956585,
      "learning_rate": 3.9783658760168765e-06,
      "loss": 0.9675,
      "step": 1046
    },
    {
      "epoch": 0.07553840049060279,
      "grad_norm": 13.657332667680873,
      "learning_rate": 3.978297266330775e-06,
      "loss": 1.0399,
      "step": 1047
    },
    {
      "epoch": 0.0756105479600303,
      "grad_norm": 10.530565369762249,
      "learning_rate": 3.978228548617184e-06,
      "loss": 0.9857,
      "step": 1048
    },
    {
      "epoch": 0.07568269542945781,
      "grad_norm": 28.20065886897913,
      "learning_rate": 3.978159722879855e-06,
      "loss": 1.0425,
      "step": 1049
    },
    {
      "epoch": 0.07575484289888532,
      "grad_norm": 15.68239507405642,
      "learning_rate": 3.978090789122546e-06,
      "loss": 1.0243,
      "step": 1050
    },
    {
      "epoch": 0.07582699036831284,
      "grad_norm": 13.271898132117233,
      "learning_rate": 3.978021747349022e-06,
      "loss": 1.0091,
      "step": 1051
    },
    {
      "epoch": 0.07589913783774034,
      "grad_norm": 18.422664350402147,
      "learning_rate": 3.977952597563052e-06,
      "loss": 1.0438,
      "step": 1052
    },
    {
      "epoch": 0.07597128530716785,
      "grad_norm": 37.48512725902101,
      "learning_rate": 3.977883339768414e-06,
      "loss": 1.0701,
      "step": 1053
    },
    {
      "epoch": 0.07604343277659537,
      "grad_norm": 20.616400018117403,
      "learning_rate": 3.977813973968888e-06,
      "loss": 1.1487,
      "step": 1054
    },
    {
      "epoch": 0.07611558024602287,
      "grad_norm": 10.155322223248646,
      "learning_rate": 3.977744500168262e-06,
      "loss": 1.0569,
      "step": 1055
    },
    {
      "epoch": 0.07618772771545038,
      "grad_norm": 18.490018884247895,
      "learning_rate": 3.977674918370331e-06,
      "loss": 0.9868,
      "step": 1056
    },
    {
      "epoch": 0.0762598751848779,
      "grad_norm": 13.44497373465586,
      "learning_rate": 3.977605228578894e-06,
      "loss": 0.9799,
      "step": 1057
    },
    {
      "epoch": 0.0763320226543054,
      "grad_norm": 9.110182763955715,
      "learning_rate": 3.977535430797756e-06,
      "loss": 1.0125,
      "step": 1058
    },
    {
      "epoch": 0.0764041701237329,
      "grad_norm": 17.249112755998222,
      "learning_rate": 3.9774655250307285e-06,
      "loss": 1.1586,
      "step": 1059
    },
    {
      "epoch": 0.07647631759316043,
      "grad_norm": 18.940038731181936,
      "learning_rate": 3.977395511281629e-06,
      "loss": 1.029,
      "step": 1060
    },
    {
      "epoch": 0.07654846506258793,
      "grad_norm": 22.308877944805964,
      "learning_rate": 3.97732538955428e-06,
      "loss": 1.1109,
      "step": 1061
    },
    {
      "epoch": 0.07662061253201544,
      "grad_norm": 11.69435807833238,
      "learning_rate": 3.977255159852513e-06,
      "loss": 1.0662,
      "step": 1062
    },
    {
      "epoch": 0.07669276000144296,
      "grad_norm": 20.418626756065542,
      "learning_rate": 3.9771848221801605e-06,
      "loss": 0.9271,
      "step": 1063
    },
    {
      "epoch": 0.07676490747087046,
      "grad_norm": 21.570937830486617,
      "learning_rate": 3.9771143765410635e-06,
      "loss": 0.954,
      "step": 1064
    },
    {
      "epoch": 0.07683705494029797,
      "grad_norm": 17.37770290071357,
      "learning_rate": 3.97704382293907e-06,
      "loss": 1.0609,
      "step": 1065
    },
    {
      "epoch": 0.07690920240972549,
      "grad_norm": 97.0425114237956,
      "learning_rate": 3.976973161378032e-06,
      "loss": 1.0121,
      "step": 1066
    },
    {
      "epoch": 0.07698134987915299,
      "grad_norm": 9.195790397521915,
      "learning_rate": 3.9769023918618085e-06,
      "loss": 1.0542,
      "step": 1067
    },
    {
      "epoch": 0.0770534973485805,
      "grad_norm": 20.19116026584058,
      "learning_rate": 3.976831514394263e-06,
      "loss": 0.9923,
      "step": 1068
    },
    {
      "epoch": 0.077125644818008,
      "grad_norm": 10.708919365626588,
      "learning_rate": 3.976760528979267e-06,
      "loss": 0.977,
      "step": 1069
    },
    {
      "epoch": 0.07719779228743552,
      "grad_norm": 14.744948217094718,
      "learning_rate": 3.976689435620696e-06,
      "loss": 0.9898,
      "step": 1070
    },
    {
      "epoch": 0.07726993975686303,
      "grad_norm": 14.412879990450923,
      "learning_rate": 3.9766182343224325e-06,
      "loss": 1.0423,
      "step": 1071
    },
    {
      "epoch": 0.07734208722629053,
      "grad_norm": 10.695484056901108,
      "learning_rate": 3.976546925088364e-06,
      "loss": 1.0289,
      "step": 1072
    },
    {
      "epoch": 0.07741423469571805,
      "grad_norm": 0.8713393955533025,
      "learning_rate": 3.976475507922385e-06,
      "loss": 0.8708,
      "step": 1073
    },
    {
      "epoch": 0.07748638216514556,
      "grad_norm": 14.08927843161717,
      "learning_rate": 3.976403982828396e-06,
      "loss": 1.1175,
      "step": 1074
    },
    {
      "epoch": 0.07755852963457306,
      "grad_norm": 14.645756525495758,
      "learning_rate": 3.9763323498103e-06,
      "loss": 1.0102,
      "step": 1075
    },
    {
      "epoch": 0.07763067710400058,
      "grad_norm": 8.274557842876995,
      "learning_rate": 3.9762606088720126e-06,
      "loss": 1.0088,
      "step": 1076
    },
    {
      "epoch": 0.07770282457342809,
      "grad_norm": 12.905994625197431,
      "learning_rate": 3.976188760017448e-06,
      "loss": 1.0237,
      "step": 1077
    },
    {
      "epoch": 0.07777497204285559,
      "grad_norm": 10.064082998543013,
      "learning_rate": 3.976116803250531e-06,
      "loss": 1.0873,
      "step": 1078
    },
    {
      "epoch": 0.07784711951228311,
      "grad_norm": 16.144661205773836,
      "learning_rate": 3.976044738575191e-06,
      "loss": 1.1314,
      "step": 1079
    },
    {
      "epoch": 0.07791926698171062,
      "grad_norm": 13.4755592923703,
      "learning_rate": 3.975972565995363e-06,
      "loss": 1.0311,
      "step": 1080
    },
    {
      "epoch": 0.07799141445113812,
      "grad_norm": 14.43824984101881,
      "learning_rate": 3.975900285514988e-06,
      "loss": 1.0632,
      "step": 1081
    },
    {
      "epoch": 0.07806356192056564,
      "grad_norm": 10.13353830765999,
      "learning_rate": 3.975827897138012e-06,
      "loss": 0.9484,
      "step": 1082
    },
    {
      "epoch": 0.07813570938999315,
      "grad_norm": 11.59419813086923,
      "learning_rate": 3.97575540086839e-06,
      "loss": 1.0056,
      "step": 1083
    },
    {
      "epoch": 0.07820785685942065,
      "grad_norm": 8.815935047083464,
      "learning_rate": 3.975682796710079e-06,
      "loss": 1.0083,
      "step": 1084
    },
    {
      "epoch": 0.07828000432884817,
      "grad_norm": 33.310726383366365,
      "learning_rate": 3.975610084667044e-06,
      "loss": 0.9068,
      "step": 1085
    },
    {
      "epoch": 0.07835215179827568,
      "grad_norm": 66.89520357782574,
      "learning_rate": 3.975537264743255e-06,
      "loss": 1.1105,
      "step": 1086
    },
    {
      "epoch": 0.07842429926770318,
      "grad_norm": 14.66804625062748,
      "learning_rate": 3.975464336942691e-06,
      "loss": 1.0315,
      "step": 1087
    },
    {
      "epoch": 0.0784964467371307,
      "grad_norm": 16.007598979786327,
      "learning_rate": 3.975391301269332e-06,
      "loss": 0.9385,
      "step": 1088
    },
    {
      "epoch": 0.0785685942065582,
      "grad_norm": 20.888668131281804,
      "learning_rate": 3.975318157727166e-06,
      "loss": 1.0641,
      "step": 1089
    },
    {
      "epoch": 0.07864074167598571,
      "grad_norm": 8.008043060705242,
      "learning_rate": 3.975244906320188e-06,
      "loss": 0.919,
      "step": 1090
    },
    {
      "epoch": 0.07871288914541323,
      "grad_norm": 8.465299797359789,
      "learning_rate": 3.975171547052398e-06,
      "loss": 0.9303,
      "step": 1091
    },
    {
      "epoch": 0.07878503661484074,
      "grad_norm": 10.97413820259025,
      "learning_rate": 3.975098079927801e-06,
      "loss": 0.9313,
      "step": 1092
    },
    {
      "epoch": 0.07885718408426824,
      "grad_norm": 9.717279549883596,
      "learning_rate": 3.97502450495041e-06,
      "loss": 1.0061,
      "step": 1093
    },
    {
      "epoch": 0.07892933155369575,
      "grad_norm": 13.469992737403503,
      "learning_rate": 3.974950822124242e-06,
      "loss": 0.8956,
      "step": 1094
    },
    {
      "epoch": 0.07900147902312327,
      "grad_norm": 5.7342090218869295,
      "learning_rate": 3.974877031453321e-06,
      "loss": 1.0643,
      "step": 1095
    },
    {
      "epoch": 0.07907362649255077,
      "grad_norm": 0.7450021077452676,
      "learning_rate": 3.974803132941676e-06,
      "loss": 0.8679,
      "step": 1096
    },
    {
      "epoch": 0.07914577396197828,
      "grad_norm": 11.417120114716635,
      "learning_rate": 3.974729126593342e-06,
      "loss": 1.0507,
      "step": 1097
    },
    {
      "epoch": 0.0792179214314058,
      "grad_norm": 51.052487160886685,
      "learning_rate": 3.97465501241236e-06,
      "loss": 1.0047,
      "step": 1098
    },
    {
      "epoch": 0.0792900689008333,
      "grad_norm": 8.785743731051735,
      "learning_rate": 3.974580790402779e-06,
      "loss": 1.0146,
      "step": 1099
    },
    {
      "epoch": 0.07936221637026081,
      "grad_norm": 8.316120019146643,
      "learning_rate": 3.97450646056865e-06,
      "loss": 1.1407,
      "step": 1100
    },
    {
      "epoch": 0.07943436383968833,
      "grad_norm": 16.405946797793515,
      "learning_rate": 3.974432022914032e-06,
      "loss": 1.0194,
      "step": 1101
    },
    {
      "epoch": 0.07950651130911583,
      "grad_norm": 14.148268717070286,
      "learning_rate": 3.974357477442991e-06,
      "loss": 1.0639,
      "step": 1102
    },
    {
      "epoch": 0.07957865877854334,
      "grad_norm": 11.181398230178369,
      "learning_rate": 3.974282824159597e-06,
      "loss": 1.0587,
      "step": 1103
    },
    {
      "epoch": 0.07965080624797086,
      "grad_norm": 9.546212081244459,
      "learning_rate": 3.974208063067926e-06,
      "loss": 0.9868,
      "step": 1104
    },
    {
      "epoch": 0.07972295371739836,
      "grad_norm": 12.390350992990157,
      "learning_rate": 3.974133194172062e-06,
      "loss": 1.0992,
      "step": 1105
    },
    {
      "epoch": 0.07979510118682587,
      "grad_norm": 11.136176164001002,
      "learning_rate": 3.974058217476091e-06,
      "loss": 1.0737,
      "step": 1106
    },
    {
      "epoch": 0.07986724865625339,
      "grad_norm": 8.352137546707494,
      "learning_rate": 3.97398313298411e-06,
      "loss": 0.9606,
      "step": 1107
    },
    {
      "epoch": 0.07993939612568089,
      "grad_norm": 9.082686468300297,
      "learning_rate": 3.973907940700216e-06,
      "loss": 1.0033,
      "step": 1108
    },
    {
      "epoch": 0.0800115435951084,
      "grad_norm": 5.037012058603357,
      "learning_rate": 3.973832640628517e-06,
      "loss": 1.0223,
      "step": 1109
    },
    {
      "epoch": 0.08008369106453592,
      "grad_norm": 5.624795867606761,
      "learning_rate": 3.973757232773124e-06,
      "loss": 1.001,
      "step": 1110
    },
    {
      "epoch": 0.08015583853396342,
      "grad_norm": 6.770113330608089,
      "learning_rate": 3.9736817171381555e-06,
      "loss": 0.9683,
      "step": 1111
    },
    {
      "epoch": 0.08022798600339093,
      "grad_norm": 0.7374627556529137,
      "learning_rate": 3.973606093727735e-06,
      "loss": 0.7919,
      "step": 1112
    },
    {
      "epoch": 0.08030013347281845,
      "grad_norm": 7.687167524886138,
      "learning_rate": 3.973530362545992e-06,
      "loss": 1.127,
      "step": 1113
    },
    {
      "epoch": 0.08037228094224595,
      "grad_norm": 7.017805614441729,
      "learning_rate": 3.973454523597061e-06,
      "loss": 1.0301,
      "step": 1114
    },
    {
      "epoch": 0.08044442841167346,
      "grad_norm": 0.8224043921584354,
      "learning_rate": 3.973378576885084e-06,
      "loss": 0.8793,
      "step": 1115
    },
    {
      "epoch": 0.08051657588110098,
      "grad_norm": 38.55101378773975,
      "learning_rate": 3.973302522414209e-06,
      "loss": 0.9741,
      "step": 1116
    },
    {
      "epoch": 0.08058872335052848,
      "grad_norm": 19.024539559967018,
      "learning_rate": 3.973226360188587e-06,
      "loss": 0.9549,
      "step": 1117
    },
    {
      "epoch": 0.08066087081995599,
      "grad_norm": 12.19809360305241,
      "learning_rate": 3.973150090212379e-06,
      "loss": 1.1106,
      "step": 1118
    },
    {
      "epoch": 0.08073301828938349,
      "grad_norm": 9.512071669530727,
      "learning_rate": 3.973073712489748e-06,
      "loss": 1.0315,
      "step": 1119
    },
    {
      "epoch": 0.08080516575881101,
      "grad_norm": 22.19485257245226,
      "learning_rate": 3.9729972270248665e-06,
      "loss": 1.1609,
      "step": 1120
    },
    {
      "epoch": 0.08087731322823852,
      "grad_norm": 9.942370486759991,
      "learning_rate": 3.972920633821909e-06,
      "loss": 1.0521,
      "step": 1121
    },
    {
      "epoch": 0.08094946069766602,
      "grad_norm": 4.687762240385886,
      "learning_rate": 3.9728439328850605e-06,
      "loss": 0.9543,
      "step": 1122
    },
    {
      "epoch": 0.08102160816709354,
      "grad_norm": 12.060095038347932,
      "learning_rate": 3.972767124218508e-06,
      "loss": 0.9704,
      "step": 1123
    },
    {
      "epoch": 0.08109375563652105,
      "grad_norm": 5.709630480031598,
      "learning_rate": 3.972690207826446e-06,
      "loss": 1.0906,
      "step": 1124
    },
    {
      "epoch": 0.08116590310594855,
      "grad_norm": 8.375235483075844,
      "learning_rate": 3.972613183713073e-06,
      "loss": 1.0272,
      "step": 1125
    },
    {
      "epoch": 0.08123805057537607,
      "grad_norm": 5.749215445690551,
      "learning_rate": 3.972536051882597e-06,
      "loss": 1.0165,
      "step": 1126
    },
    {
      "epoch": 0.08131019804480358,
      "grad_norm": 8.08044329494202,
      "learning_rate": 3.97245881233923e-06,
      "loss": 0.9136,
      "step": 1127
    },
    {
      "epoch": 0.08138234551423108,
      "grad_norm": 28.077621769627875,
      "learning_rate": 3.972381465087189e-06,
      "loss": 1.0749,
      "step": 1128
    },
    {
      "epoch": 0.0814544929836586,
      "grad_norm": 6.741080713761935,
      "learning_rate": 3.972304010130696e-06,
      "loss": 0.9981,
      "step": 1129
    },
    {
      "epoch": 0.08152664045308611,
      "grad_norm": 11.70009713417879,
      "learning_rate": 3.972226447473984e-06,
      "loss": 1.0771,
      "step": 1130
    },
    {
      "epoch": 0.08159878792251361,
      "grad_norm": 6.057778948675465,
      "learning_rate": 3.972148777121286e-06,
      "loss": 1.0008,
      "step": 1131
    },
    {
      "epoch": 0.08167093539194113,
      "grad_norm": 7.3298993595807325,
      "learning_rate": 3.972070999076844e-06,
      "loss": 1.0394,
      "step": 1132
    },
    {
      "epoch": 0.08174308286136864,
      "grad_norm": 5.574800927409787,
      "learning_rate": 3.971993113344905e-06,
      "loss": 1.0403,
      "step": 1133
    },
    {
      "epoch": 0.08181523033079614,
      "grad_norm": 126.56075230274787,
      "learning_rate": 3.971915119929722e-06,
      "loss": 1.022,
      "step": 1134
    },
    {
      "epoch": 0.08188737780022366,
      "grad_norm": 7.939639546550611,
      "learning_rate": 3.971837018835555e-06,
      "loss": 1.0689,
      "step": 1135
    },
    {
      "epoch": 0.08195952526965117,
      "grad_norm": 17.851300941738113,
      "learning_rate": 3.9717588100666675e-06,
      "loss": 1.1279,
      "step": 1136
    },
    {
      "epoch": 0.08203167273907867,
      "grad_norm": 7.503485569041527,
      "learning_rate": 3.97168049362733e-06,
      "loss": 0.8949,
      "step": 1137
    },
    {
      "epoch": 0.08210382020850619,
      "grad_norm": 4.742643783714322,
      "learning_rate": 3.97160206952182e-06,
      "loss": 1.0283,
      "step": 1138
    },
    {
      "epoch": 0.0821759676779337,
      "grad_norm": 12.033512368459801,
      "learning_rate": 3.971523537754419e-06,
      "loss": 1.2145,
      "step": 1139
    },
    {
      "epoch": 0.0822481151473612,
      "grad_norm": 6.426863095195413,
      "learning_rate": 3.971444898329417e-06,
      "loss": 0.9808,
      "step": 1140
    },
    {
      "epoch": 0.08232026261678872,
      "grad_norm": 6.207910032583933,
      "learning_rate": 3.971366151251107e-06,
      "loss": 1.0368,
      "step": 1141
    },
    {
      "epoch": 0.08239241008621623,
      "grad_norm": 13.877533002706603,
      "learning_rate": 3.971287296523788e-06,
      "loss": 1.0661,
      "step": 1142
    },
    {
      "epoch": 0.08246455755564373,
      "grad_norm": 10.981342445119024,
      "learning_rate": 3.971208334151768e-06,
      "loss": 1.0784,
      "step": 1143
    },
    {
      "epoch": 0.08253670502507125,
      "grad_norm": 12.08668804735613,
      "learning_rate": 3.971129264139358e-06,
      "loss": 0.9842,
      "step": 1144
    },
    {
      "epoch": 0.08260885249449876,
      "grad_norm": 14.371386264520988,
      "learning_rate": 3.971050086490876e-06,
      "loss": 1.051,
      "step": 1145
    },
    {
      "epoch": 0.08268099996392626,
      "grad_norm": 11.136575570151154,
      "learning_rate": 3.970970801210645e-06,
      "loss": 1.0647,
      "step": 1146
    },
    {
      "epoch": 0.08275314743335377,
      "grad_norm": 16.796630630489837,
      "learning_rate": 3.970891408302995e-06,
      "loss": 1.0438,
      "step": 1147
    },
    {
      "epoch": 0.08282529490278129,
      "grad_norm": 9.149598293137315,
      "learning_rate": 3.970811907772262e-06,
      "loss": 0.9522,
      "step": 1148
    },
    {
      "epoch": 0.0828974423722088,
      "grad_norm": 10.369185323469774,
      "learning_rate": 3.9707322996227855e-06,
      "loss": 0.9715,
      "step": 1149
    },
    {
      "epoch": 0.0829695898416363,
      "grad_norm": 6.448357176228438,
      "learning_rate": 3.970652583858913e-06,
      "loss": 1.0769,
      "step": 1150
    },
    {
      "epoch": 0.08304173731106382,
      "grad_norm": 8.411373123036537,
      "learning_rate": 3.970572760484999e-06,
      "loss": 1.0593,
      "step": 1151
    },
    {
      "epoch": 0.08311388478049132,
      "grad_norm": 5.178375308076736,
      "learning_rate": 3.970492829505401e-06,
      "loss": 1.0476,
      "step": 1152
    },
    {
      "epoch": 0.08318603224991883,
      "grad_norm": 8.790301127998884,
      "learning_rate": 3.970412790924484e-06,
      "loss": 0.9295,
      "step": 1153
    },
    {
      "epoch": 0.08325817971934635,
      "grad_norm": 8.523321135204759,
      "learning_rate": 3.970332644746619e-06,
      "loss": 0.9578,
      "step": 1154
    },
    {
      "epoch": 0.08333032718877385,
      "grad_norm": 5.574304976996011,
      "learning_rate": 3.970252390976182e-06,
      "loss": 1.103,
      "step": 1155
    },
    {
      "epoch": 0.08340247465820136,
      "grad_norm": 5.896250176493432,
      "learning_rate": 3.970172029617556e-06,
      "loss": 1.0767,
      "step": 1156
    },
    {
      "epoch": 0.08347462212762888,
      "grad_norm": 11.459144665286365,
      "learning_rate": 3.970091560675128e-06,
      "loss": 0.9702,
      "step": 1157
    },
    {
      "epoch": 0.08354676959705638,
      "grad_norm": 5.605665760366324,
      "learning_rate": 3.970010984153293e-06,
      "loss": 1.0731,
      "step": 1158
    },
    {
      "epoch": 0.08361891706648389,
      "grad_norm": 0.6680974334439679,
      "learning_rate": 3.9699303000564515e-06,
      "loss": 0.8266,
      "step": 1159
    },
    {
      "epoch": 0.08369106453591141,
      "grad_norm": 14.74249749193822,
      "learning_rate": 3.969849508389008e-06,
      "loss": 0.9843,
      "step": 1160
    },
    {
      "epoch": 0.08376321200533891,
      "grad_norm": 12.445273265788753,
      "learning_rate": 3.969768609155376e-06,
      "loss": 0.9087,
      "step": 1161
    },
    {
      "epoch": 0.08383535947476642,
      "grad_norm": 14.76461585093178,
      "learning_rate": 3.969687602359971e-06,
      "loss": 1.0697,
      "step": 1162
    },
    {
      "epoch": 0.08390750694419394,
      "grad_norm": 8.86511496635544,
      "learning_rate": 3.969606488007218e-06,
      "loss": 0.9571,
      "step": 1163
    },
    {
      "epoch": 0.08397965441362144,
      "grad_norm": 7.113031339994925,
      "learning_rate": 3.9695252661015455e-06,
      "loss": 1.1327,
      "step": 1164
    },
    {
      "epoch": 0.08405180188304895,
      "grad_norm": 6.116413559557837,
      "learning_rate": 3.9694439366473904e-06,
      "loss": 1.0269,
      "step": 1165
    },
    {
      "epoch": 0.08412394935247647,
      "grad_norm": 27.059062642555634,
      "learning_rate": 3.9693624996491915e-06,
      "loss": 1.0534,
      "step": 1166
    },
    {
      "epoch": 0.08419609682190397,
      "grad_norm": 15.35331723034426,
      "learning_rate": 3.9692809551113974e-06,
      "loss": 0.9818,
      "step": 1167
    },
    {
      "epoch": 0.08426824429133148,
      "grad_norm": 13.408919346539161,
      "learning_rate": 3.969199303038459e-06,
      "loss": 0.9941,
      "step": 1168
    },
    {
      "epoch": 0.084340391760759,
      "grad_norm": 10.77384008550286,
      "learning_rate": 3.969117543434838e-06,
      "loss": 1.1079,
      "step": 1169
    },
    {
      "epoch": 0.0844125392301865,
      "grad_norm": 9.000818427278702,
      "learning_rate": 3.969035676304997e-06,
      "loss": 1.1343,
      "step": 1170
    },
    {
      "epoch": 0.08448468669961401,
      "grad_norm": 18.638494121971487,
      "learning_rate": 3.968953701653406e-06,
      "loss": 1.085,
      "step": 1171
    },
    {
      "epoch": 0.08455683416904151,
      "grad_norm": 12.227774874994823,
      "learning_rate": 3.968871619484543e-06,
      "loss": 1.0017,
      "step": 1172
    },
    {
      "epoch": 0.08462898163846903,
      "grad_norm": 21.796642583886342,
      "learning_rate": 3.9687894298028895e-06,
      "loss": 1.0881,
      "step": 1173
    },
    {
      "epoch": 0.08470112910789654,
      "grad_norm": 11.299996394604133,
      "learning_rate": 3.968707132612933e-06,
      "loss": 1.0886,
      "step": 1174
    },
    {
      "epoch": 0.08477327657732404,
      "grad_norm": 12.980829041606775,
      "learning_rate": 3.968624727919169e-06,
      "loss": 1.058,
      "step": 1175
    },
    {
      "epoch": 0.08484542404675156,
      "grad_norm": 7.224386285501934,
      "learning_rate": 3.968542215726095e-06,
      "loss": 0.9548,
      "step": 1176
    },
    {
      "epoch": 0.08491757151617907,
      "grad_norm": 7.806800411174357,
      "learning_rate": 3.968459596038218e-06,
      "loss": 1.0801,
      "step": 1177
    },
    {
      "epoch": 0.08498971898560657,
      "grad_norm": 10.86709588106149,
      "learning_rate": 3.968376868860049e-06,
      "loss": 0.9922,
      "step": 1178
    },
    {
      "epoch": 0.0850618664550341,
      "grad_norm": 7.135552873262793,
      "learning_rate": 3.968294034196107e-06,
      "loss": 1.0265,
      "step": 1179
    },
    {
      "epoch": 0.0851340139244616,
      "grad_norm": 9.986728635560898,
      "learning_rate": 3.968211092050914e-06,
      "loss": 0.9581,
      "step": 1180
    },
    {
      "epoch": 0.0852061613938891,
      "grad_norm": 5.922496913975128,
      "learning_rate": 3.968128042428998e-06,
      "loss": 0.9676,
      "step": 1181
    },
    {
      "epoch": 0.08527830886331662,
      "grad_norm": 6.885018279938543,
      "learning_rate": 3.9680448853348965e-06,
      "loss": 1.0152,
      "step": 1182
    },
    {
      "epoch": 0.08535045633274413,
      "grad_norm": 4.464964003124139,
      "learning_rate": 3.967961620773149e-06,
      "loss": 1.0989,
      "step": 1183
    },
    {
      "epoch": 0.08542260380217163,
      "grad_norm": 8.509506295179252,
      "learning_rate": 3.967878248748302e-06,
      "loss": 1.0406,
      "step": 1184
    },
    {
      "epoch": 0.08549475127159915,
      "grad_norm": 16.39975969091992,
      "learning_rate": 3.96779476926491e-06,
      "loss": 0.8912,
      "step": 1185
    },
    {
      "epoch": 0.08556689874102666,
      "grad_norm": 8.660531978345743,
      "learning_rate": 3.967711182327529e-06,
      "loss": 1.068,
      "step": 1186
    },
    {
      "epoch": 0.08563904621045416,
      "grad_norm": 4.812881850909643,
      "learning_rate": 3.967627487940725e-06,
      "loss": 0.9812,
      "step": 1187
    },
    {
      "epoch": 0.08571119367988168,
      "grad_norm": 8.19606420453282,
      "learning_rate": 3.967543686109068e-06,
      "loss": 1.0184,
      "step": 1188
    },
    {
      "epoch": 0.08578334114930919,
      "grad_norm": 5.122325850840845,
      "learning_rate": 3.967459776837134e-06,
      "loss": 1.0233,
      "step": 1189
    },
    {
      "epoch": 0.0858554886187367,
      "grad_norm": 8.205539544199747,
      "learning_rate": 3.967375760129504e-06,
      "loss": 0.9917,
      "step": 1190
    },
    {
      "epoch": 0.08592763608816421,
      "grad_norm": 6.344506129410156,
      "learning_rate": 3.967291635990768e-06,
      "loss": 0.969,
      "step": 1191
    },
    {
      "epoch": 0.08599978355759172,
      "grad_norm": 3.781934061517533,
      "learning_rate": 3.967207404425518e-06,
      "loss": 1.0433,
      "step": 1192
    },
    {
      "epoch": 0.08607193102701922,
      "grad_norm": 13.653349614233006,
      "learning_rate": 3.967123065438353e-06,
      "loss": 1.107,
      "step": 1193
    },
    {
      "epoch": 0.08614407849644674,
      "grad_norm": 7.852467590090262,
      "learning_rate": 3.96703861903388e-06,
      "loss": 0.9569,
      "step": 1194
    },
    {
      "epoch": 0.08621622596587425,
      "grad_norm": 1.0570668261758696,
      "learning_rate": 3.96695406521671e-06,
      "loss": 1.0145,
      "step": 1195
    },
    {
      "epoch": 0.08628837343530175,
      "grad_norm": 7.694839075697354,
      "learning_rate": 3.96686940399146e-06,
      "loss": 1.0102,
      "step": 1196
    },
    {
      "epoch": 0.08636052090472926,
      "grad_norm": 9.563288637439962,
      "learning_rate": 3.966784635362753e-06,
      "loss": 1.0479,
      "step": 1197
    },
    {
      "epoch": 0.08643266837415678,
      "grad_norm": 0.727883563308878,
      "learning_rate": 3.966699759335218e-06,
      "loss": 0.8228,
      "step": 1198
    },
    {
      "epoch": 0.08650481584358428,
      "grad_norm": 8.298319028203704,
      "learning_rate": 3.966614775913489e-06,
      "loss": 0.9789,
      "step": 1199
    },
    {
      "epoch": 0.08657696331301179,
      "grad_norm": 6.782456813608679,
      "learning_rate": 3.966529685102207e-06,
      "loss": 1.0524,
      "step": 1200
    },
    {
      "epoch": 0.08664911078243931,
      "grad_norm": 7.32222186319945,
      "learning_rate": 3.966444486906019e-06,
      "loss": 0.9953,
      "step": 1201
    },
    {
      "epoch": 0.08672125825186681,
      "grad_norm": 10.495546395525341,
      "learning_rate": 3.966359181329578e-06,
      "loss": 1.072,
      "step": 1202
    },
    {
      "epoch": 0.08679340572129432,
      "grad_norm": 10.953035493014275,
      "learning_rate": 3.9662737683775405e-06,
      "loss": 1.0236,
      "step": 1203
    },
    {
      "epoch": 0.08686555319072184,
      "grad_norm": 7.65652851454515,
      "learning_rate": 3.966188248054571e-06,
      "loss": 1.0677,
      "step": 1204
    },
    {
      "epoch": 0.08693770066014934,
      "grad_norm": 9.760433385638585,
      "learning_rate": 3.96610262036534e-06,
      "loss": 1.0363,
      "step": 1205
    },
    {
      "epoch": 0.08700984812957685,
      "grad_norm": 6.753415974107914,
      "learning_rate": 3.9660168853145245e-06,
      "loss": 1.0683,
      "step": 1206
    },
    {
      "epoch": 0.08708199559900437,
      "grad_norm": 7.524459880315565,
      "learning_rate": 3.965931042906804e-06,
      "loss": 1.0441,
      "step": 1207
    },
    {
      "epoch": 0.08715414306843187,
      "grad_norm": 0.9999183979118521,
      "learning_rate": 3.965845093146866e-06,
      "loss": 0.9384,
      "step": 1208
    },
    {
      "epoch": 0.08722629053785938,
      "grad_norm": 17.07890498133086,
      "learning_rate": 3.965759036039406e-06,
      "loss": 1.0068,
      "step": 1209
    },
    {
      "epoch": 0.0872984380072869,
      "grad_norm": 10.903772285820864,
      "learning_rate": 3.965672871589121e-06,
      "loss": 0.9957,
      "step": 1210
    },
    {
      "epoch": 0.0873705854767144,
      "grad_norm": 23.05795814016822,
      "learning_rate": 3.965586599800718e-06,
      "loss": 1.1314,
      "step": 1211
    },
    {
      "epoch": 0.08744273294614191,
      "grad_norm": 20.789024332922498,
      "learning_rate": 3.965500220678907e-06,
      "loss": 1.0265,
      "step": 1212
    },
    {
      "epoch": 0.08751488041556943,
      "grad_norm": 13.622280024339664,
      "learning_rate": 3.965413734228405e-06,
      "loss": 0.9967,
      "step": 1213
    },
    {
      "epoch": 0.08758702788499693,
      "grad_norm": 8.458828114537443,
      "learning_rate": 3.965327140453934e-06,
      "loss": 0.9673,
      "step": 1214
    },
    {
      "epoch": 0.08765917535442444,
      "grad_norm": 14.220479805493115,
      "learning_rate": 3.965240439360224e-06,
      "loss": 1.0937,
      "step": 1215
    },
    {
      "epoch": 0.08773132282385196,
      "grad_norm": 11.799195239539909,
      "learning_rate": 3.965153630952009e-06,
      "loss": 1.0377,
      "step": 1216
    },
    {
      "epoch": 0.08780347029327946,
      "grad_norm": 14.226145805779748,
      "learning_rate": 3.9650667152340284e-06,
      "loss": 1.0441,
      "step": 1217
    },
    {
      "epoch": 0.08787561776270697,
      "grad_norm": 13.999464297263362,
      "learning_rate": 3.964979692211029e-06,
      "loss": 1.0782,
      "step": 1218
    },
    {
      "epoch": 0.08794776523213449,
      "grad_norm": 11.030132291109432,
      "learning_rate": 3.964892561887762e-06,
      "loss": 1.0852,
      "step": 1219
    },
    {
      "epoch": 0.088019912701562,
      "grad_norm": 10.873165282777668,
      "learning_rate": 3.964805324268987e-06,
      "loss": 0.9761,
      "step": 1220
    },
    {
      "epoch": 0.0880920601709895,
      "grad_norm": 16.625472671558374,
      "learning_rate": 3.964717979359466e-06,
      "loss": 1.0788,
      "step": 1221
    },
    {
      "epoch": 0.088164207640417,
      "grad_norm": 16.042451807930284,
      "learning_rate": 3.96463052716397e-06,
      "loss": 0.9848,
      "step": 1222
    },
    {
      "epoch": 0.08823635510984452,
      "grad_norm": 18.190517529155557,
      "learning_rate": 3.964542967687273e-06,
      "loss": 1.0669,
      "step": 1223
    },
    {
      "epoch": 0.08830850257927203,
      "grad_norm": 11.796893755158932,
      "learning_rate": 3.964455300934158e-06,
      "loss": 0.9848,
      "step": 1224
    },
    {
      "epoch": 0.08838065004869954,
      "grad_norm": 16.866001470865417,
      "learning_rate": 3.96436752690941e-06,
      "loss": 0.9785,
      "step": 1225
    },
    {
      "epoch": 0.08845279751812705,
      "grad_norm": 9.604128712698415,
      "learning_rate": 3.964279645617824e-06,
      "loss": 0.917,
      "step": 1226
    },
    {
      "epoch": 0.08852494498755456,
      "grad_norm": 32.4810416459548,
      "learning_rate": 3.964191657064197e-06,
      "loss": 0.999,
      "step": 1227
    },
    {
      "epoch": 0.08859709245698207,
      "grad_norm": 13.498988926077656,
      "learning_rate": 3.964103561253336e-06,
      "loss": 1.1534,
      "step": 1228
    },
    {
      "epoch": 0.08866923992640958,
      "grad_norm": 15.373500843927586,
      "learning_rate": 3.96401535819005e-06,
      "loss": 1.0769,
      "step": 1229
    },
    {
      "epoch": 0.08874138739583709,
      "grad_norm": 10.212006656678925,
      "learning_rate": 3.963927047879156e-06,
      "loss": 1.0296,
      "step": 1230
    },
    {
      "epoch": 0.0888135348652646,
      "grad_norm": 52.29490724935603,
      "learning_rate": 3.9638386303254764e-06,
      "loss": 1.0111,
      "step": 1231
    },
    {
      "epoch": 0.08888568233469211,
      "grad_norm": 31.163474941189666,
      "learning_rate": 3.963750105533838e-06,
      "loss": 0.9564,
      "step": 1232
    },
    {
      "epoch": 0.08895782980411962,
      "grad_norm": 11.772961845484524,
      "learning_rate": 3.963661473509076e-06,
      "loss": 1.0278,
      "step": 1233
    },
    {
      "epoch": 0.08902997727354713,
      "grad_norm": 12.943985202776133,
      "learning_rate": 3.963572734256031e-06,
      "loss": 1.0382,
      "step": 1234
    },
    {
      "epoch": 0.08910212474297464,
      "grad_norm": 14.63999861086646,
      "learning_rate": 3.963483887779547e-06,
      "loss": 0.942,
      "step": 1235
    },
    {
      "epoch": 0.08917427221240215,
      "grad_norm": 0.7010561281689878,
      "learning_rate": 3.963394934084478e-06,
      "loss": 0.9073,
      "step": 1236
    },
    {
      "epoch": 0.08924641968182966,
      "grad_norm": 8.329411562941484,
      "learning_rate": 3.963305873175678e-06,
      "loss": 1.0303,
      "step": 1237
    },
    {
      "epoch": 0.08931856715125717,
      "grad_norm": 17.77508920896654,
      "learning_rate": 3.963216705058014e-06,
      "loss": 1.0951,
      "step": 1238
    },
    {
      "epoch": 0.08939071462068468,
      "grad_norm": 32.684921250454885,
      "learning_rate": 3.9631274297363525e-06,
      "loss": 1.0744,
      "step": 1239
    },
    {
      "epoch": 0.08946286209011219,
      "grad_norm": 40.855556704956,
      "learning_rate": 3.963038047215569e-06,
      "loss": 1.1466,
      "step": 1240
    },
    {
      "epoch": 0.0895350095595397,
      "grad_norm": 16.273579084537708,
      "learning_rate": 3.9629485575005444e-06,
      "loss": 0.9889,
      "step": 1241
    },
    {
      "epoch": 0.08960715702896721,
      "grad_norm": 16.67654756745326,
      "learning_rate": 3.962858960596166e-06,
      "loss": 1.058,
      "step": 1242
    },
    {
      "epoch": 0.08967930449839472,
      "grad_norm": 17.41034511971259,
      "learning_rate": 3.962769256507326e-06,
      "loss": 1.0342,
      "step": 1243
    },
    {
      "epoch": 0.08975145196782224,
      "grad_norm": 10.310121574092063,
      "learning_rate": 3.9626794452389235e-06,
      "loss": 1.1067,
      "step": 1244
    },
    {
      "epoch": 0.08982359943724974,
      "grad_norm": 18.265260769465147,
      "learning_rate": 3.9625895267958615e-06,
      "loss": 1.0048,
      "step": 1245
    },
    {
      "epoch": 0.08989574690667725,
      "grad_norm": 25.172910341060337,
      "learning_rate": 3.962499501183051e-06,
      "loss": 0.9563,
      "step": 1246
    },
    {
      "epoch": 0.08996789437610477,
      "grad_norm": 17.659051561473547,
      "learning_rate": 3.9624093684054065e-06,
      "loss": 0.9799,
      "step": 1247
    },
    {
      "epoch": 0.09004004184553227,
      "grad_norm": 22.496200579816666,
      "learning_rate": 3.962319128467852e-06,
      "loss": 1.0374,
      "step": 1248
    },
    {
      "epoch": 0.09011218931495978,
      "grad_norm": 6.19010361888232,
      "learning_rate": 3.9622287813753144e-06,
      "loss": 1.0806,
      "step": 1249
    },
    {
      "epoch": 0.09018433678438728,
      "grad_norm": 15.138890973167781,
      "learning_rate": 3.962138327132726e-06,
      "loss": 0.9885,
      "step": 1250
    },
    {
      "epoch": 0.0902564842538148,
      "grad_norm": 29.813465286718078,
      "learning_rate": 3.962047765745028e-06,
      "loss": 1.017,
      "step": 1251
    },
    {
      "epoch": 0.0903286317232423,
      "grad_norm": 15.967500581778651,
      "learning_rate": 3.961957097217165e-06,
      "loss": 0.9751,
      "step": 1252
    },
    {
      "epoch": 0.09040077919266981,
      "grad_norm": 131.8949925603698,
      "learning_rate": 3.961866321554087e-06,
      "loss": 1.016,
      "step": 1253
    },
    {
      "epoch": 0.09047292666209733,
      "grad_norm": 13.556733342546,
      "learning_rate": 3.9617754387607524e-06,
      "loss": 1.0501,
      "step": 1254
    },
    {
      "epoch": 0.09054507413152484,
      "grad_norm": 38.3983461023836,
      "learning_rate": 3.961684448842123e-06,
      "loss": 1.0557,
      "step": 1255
    },
    {
      "epoch": 0.09061722160095234,
      "grad_norm": 12.20663436854819,
      "learning_rate": 3.961593351803167e-06,
      "loss": 1.047,
      "step": 1256
    },
    {
      "epoch": 0.09068936907037986,
      "grad_norm": 14.573568566306466,
      "learning_rate": 3.9615021476488615e-06,
      "loss": 0.9273,
      "step": 1257
    },
    {
      "epoch": 0.09076151653980737,
      "grad_norm": 84.40325915656338,
      "learning_rate": 3.9614108363841835e-06,
      "loss": 1.0134,
      "step": 1258
    },
    {
      "epoch": 0.09083366400923487,
      "grad_norm": 7.942243465306703,
      "learning_rate": 3.961319418014121e-06,
      "loss": 1.0608,
      "step": 1259
    },
    {
      "epoch": 0.09090581147866239,
      "grad_norm": 18.45377575456062,
      "learning_rate": 3.961227892543666e-06,
      "loss": 0.9788,
      "step": 1260
    },
    {
      "epoch": 0.0909779589480899,
      "grad_norm": 19.312740188635843,
      "learning_rate": 3.961136259977815e-06,
      "loss": 0.9148,
      "step": 1261
    },
    {
      "epoch": 0.0910501064175174,
      "grad_norm": 10.367155119580179,
      "learning_rate": 3.961044520321573e-06,
      "loss": 1.0599,
      "step": 1262
    },
    {
      "epoch": 0.09112225388694492,
      "grad_norm": 7.969907669049487,
      "learning_rate": 3.96095267357995e-06,
      "loss": 1.0709,
      "step": 1263
    },
    {
      "epoch": 0.09119440135637243,
      "grad_norm": 10.819542856765924,
      "learning_rate": 3.960860719757961e-06,
      "loss": 1.0461,
      "step": 1264
    },
    {
      "epoch": 0.09126654882579993,
      "grad_norm": 12.205073593024043,
      "learning_rate": 3.960768658860626e-06,
      "loss": 1.1083,
      "step": 1265
    },
    {
      "epoch": 0.09133869629522745,
      "grad_norm": 9.751040280916627,
      "learning_rate": 3.960676490892973e-06,
      "loss": 1.0522,
      "step": 1266
    },
    {
      "epoch": 0.09141084376465496,
      "grad_norm": 17.808362835062475,
      "learning_rate": 3.960584215860036e-06,
      "loss": 1.0618,
      "step": 1267
    },
    {
      "epoch": 0.09148299123408246,
      "grad_norm": 13.444561189787397,
      "learning_rate": 3.960491833766852e-06,
      "loss": 1.0275,
      "step": 1268
    },
    {
      "epoch": 0.09155513870350998,
      "grad_norm": 43.10944720680796,
      "learning_rate": 3.960399344618467e-06,
      "loss": 1.0678,
      "step": 1269
    },
    {
      "epoch": 0.09162728617293749,
      "grad_norm": 9.315528735533942,
      "learning_rate": 3.96030674841993e-06,
      "loss": 1.1001,
      "step": 1270
    },
    {
      "epoch": 0.09169943364236499,
      "grad_norm": 15.981769666401837,
      "learning_rate": 3.960214045176299e-06,
      "loss": 1.0271,
      "step": 1271
    },
    {
      "epoch": 0.09177158111179251,
      "grad_norm": 0.8220128919142732,
      "learning_rate": 3.960121234892635e-06,
      "loss": 0.9084,
      "step": 1272
    },
    {
      "epoch": 0.09184372858122002,
      "grad_norm": 7.256697619396544,
      "learning_rate": 3.960028317574007e-06,
      "loss": 1.0123,
      "step": 1273
    },
    {
      "epoch": 0.09191587605064752,
      "grad_norm": 30.42129517616422,
      "learning_rate": 3.959935293225488e-06,
      "loss": 1.0706,
      "step": 1274
    },
    {
      "epoch": 0.09198802352007503,
      "grad_norm": 11.069436177934415,
      "learning_rate": 3.959842161852158e-06,
      "loss": 1.1279,
      "step": 1275
    },
    {
      "epoch": 0.09206017098950255,
      "grad_norm": 18.573852766504768,
      "learning_rate": 3.959748923459103e-06,
      "loss": 1.0236,
      "step": 1276
    },
    {
      "epoch": 0.09213231845893005,
      "grad_norm": 23.781566963640007,
      "learning_rate": 3.959655578051413e-06,
      "loss": 0.9438,
      "step": 1277
    },
    {
      "epoch": 0.09220446592835756,
      "grad_norm": 8.906166049494692,
      "learning_rate": 3.959562125634188e-06,
      "loss": 0.9284,
      "step": 1278
    },
    {
      "epoch": 0.09227661339778508,
      "grad_norm": 58.39522618678902,
      "learning_rate": 3.959468566212529e-06,
      "loss": 0.984,
      "step": 1279
    },
    {
      "epoch": 0.09234876086721258,
      "grad_norm": 62.139552417195404,
      "learning_rate": 3.959374899791544e-06,
      "loss": 0.9485,
      "step": 1280
    },
    {
      "epoch": 0.09242090833664009,
      "grad_norm": 10.045704727478292,
      "learning_rate": 3.95928112637635e-06,
      "loss": 1.0569,
      "step": 1281
    },
    {
      "epoch": 0.0924930558060676,
      "grad_norm": 10.261502394702463,
      "learning_rate": 3.959187245972066e-06,
      "loss": 1.1279,
      "step": 1282
    },
    {
      "epoch": 0.09256520327549511,
      "grad_norm": 33.614944259050354,
      "learning_rate": 3.95909325858382e-06,
      "loss": 1.0545,
      "step": 1283
    },
    {
      "epoch": 0.09263735074492262,
      "grad_norm": 0.841634854260013,
      "learning_rate": 3.9589991642167435e-06,
      "loss": 0.8303,
      "step": 1284
    },
    {
      "epoch": 0.09270949821435014,
      "grad_norm": 15.303573809339131,
      "learning_rate": 3.958904962875975e-06,
      "loss": 1.0649,
      "step": 1285
    },
    {
      "epoch": 0.09278164568377764,
      "grad_norm": 20.79747736378779,
      "learning_rate": 3.958810654566657e-06,
      "loss": 1.0418,
      "step": 1286
    },
    {
      "epoch": 0.09285379315320515,
      "grad_norm": 14.602450209606038,
      "learning_rate": 3.958716239293941e-06,
      "loss": 1.0418,
      "step": 1287
    },
    {
      "epoch": 0.09292594062263267,
      "grad_norm": 10.159336219791026,
      "learning_rate": 3.9586217170629816e-06,
      "loss": 0.9873,
      "step": 1288
    },
    {
      "epoch": 0.09299808809206017,
      "grad_norm": 24.278152955641005,
      "learning_rate": 3.958527087878941e-06,
      "loss": 1.0101,
      "step": 1289
    },
    {
      "epoch": 0.09307023556148768,
      "grad_norm": 16.42585974639389,
      "learning_rate": 3.958432351746986e-06,
      "loss": 1.0043,
      "step": 1290
    },
    {
      "epoch": 0.0931423830309152,
      "grad_norm": 7.181347411236219,
      "learning_rate": 3.958337508672291e-06,
      "loss": 0.9832,
      "step": 1291
    },
    {
      "epoch": 0.0932145305003427,
      "grad_norm": 11.23479739948264,
      "learning_rate": 3.9582425586600336e-06,
      "loss": 0.9582,
      "step": 1292
    },
    {
      "epoch": 0.09328667796977021,
      "grad_norm": 7.743415742792237,
      "learning_rate": 3.9581475017153995e-06,
      "loss": 1.038,
      "step": 1293
    },
    {
      "epoch": 0.09335882543919773,
      "grad_norm": 9.952099709872982,
      "learning_rate": 3.958052337843579e-06,
      "loss": 1.0595,
      "step": 1294
    },
    {
      "epoch": 0.09343097290862523,
      "grad_norm": 13.982981555925738,
      "learning_rate": 3.957957067049769e-06,
      "loss": 1.065,
      "step": 1295
    },
    {
      "epoch": 0.09350312037805274,
      "grad_norm": 17.513614263489774,
      "learning_rate": 3.9578616893391716e-06,
      "loss": 1.0912,
      "step": 1296
    },
    {
      "epoch": 0.09357526784748026,
      "grad_norm": 10.054912478665008,
      "learning_rate": 3.957766204716995e-06,
      "loss": 1.0059,
      "step": 1297
    },
    {
      "epoch": 0.09364741531690776,
      "grad_norm": 11.767839241138258,
      "learning_rate": 3.957670613188453e-06,
      "loss": 1.0009,
      "step": 1298
    },
    {
      "epoch": 0.09371956278633527,
      "grad_norm": 12.836153257278285,
      "learning_rate": 3.9575749147587665e-06,
      "loss": 0.928,
      "step": 1299
    },
    {
      "epoch": 0.09379171025576277,
      "grad_norm": 27.36902291875513,
      "learning_rate": 3.95747910943316e-06,
      "loss": 0.9693,
      "step": 1300
    },
    {
      "epoch": 0.09386385772519029,
      "grad_norm": 5.086299947782619,
      "learning_rate": 3.957383197216867e-06,
      "loss": 1.0329,
      "step": 1301
    },
    {
      "epoch": 0.0939360051946178,
      "grad_norm": 8.567288361588249,
      "learning_rate": 3.957287178115122e-06,
      "loss": 1.1666,
      "step": 1302
    },
    {
      "epoch": 0.0940081526640453,
      "grad_norm": 7.406019247460136,
      "learning_rate": 3.957191052133171e-06,
      "loss": 1.0407,
      "step": 1303
    },
    {
      "epoch": 0.09408030013347282,
      "grad_norm": 11.006055032366879,
      "learning_rate": 3.957094819276261e-06,
      "loss": 1.0135,
      "step": 1304
    },
    {
      "epoch": 0.09415244760290033,
      "grad_norm": 8.550802217739253,
      "learning_rate": 3.956998479549649e-06,
      "loss": 1.0727,
      "step": 1305
    },
    {
      "epoch": 0.09422459507232783,
      "grad_norm": 0.750222848527386,
      "learning_rate": 3.956902032958593e-06,
      "loss": 0.834,
      "step": 1306
    },
    {
      "epoch": 0.09429674254175535,
      "grad_norm": 8.331020288527183,
      "learning_rate": 3.956805479508362e-06,
      "loss": 1.0559,
      "step": 1307
    },
    {
      "epoch": 0.09436889001118286,
      "grad_norm": 23.887361052791206,
      "learning_rate": 3.956708819204228e-06,
      "loss": 1.0351,
      "step": 1308
    },
    {
      "epoch": 0.09444103748061036,
      "grad_norm": 10.289044333103742,
      "learning_rate": 3.956612052051468e-06,
      "loss": 1.0988,
      "step": 1309
    },
    {
      "epoch": 0.09451318495003788,
      "grad_norm": 46.46417365663302,
      "learning_rate": 3.956515178055367e-06,
      "loss": 1.062,
      "step": 1310
    },
    {
      "epoch": 0.09458533241946539,
      "grad_norm": 11.662198837006132,
      "learning_rate": 3.956418197221215e-06,
      "loss": 0.9853,
      "step": 1311
    },
    {
      "epoch": 0.09465747988889289,
      "grad_norm": 11.400224489377528,
      "learning_rate": 3.9563211095543085e-06,
      "loss": 0.9871,
      "step": 1312
    },
    {
      "epoch": 0.09472962735832041,
      "grad_norm": 32.17804017542476,
      "learning_rate": 3.956223915059948e-06,
      "loss": 1.0927,
      "step": 1313
    },
    {
      "epoch": 0.09480177482774792,
      "grad_norm": 10.67311100204219,
      "learning_rate": 3.956126613743441e-06,
      "loss": 0.9292,
      "step": 1314
    },
    {
      "epoch": 0.09487392229717542,
      "grad_norm": 7.486234621040776,
      "learning_rate": 3.956029205610101e-06,
      "loss": 1.0796,
      "step": 1315
    },
    {
      "epoch": 0.09494606976660294,
      "grad_norm": 22.44187370324724,
      "learning_rate": 3.955931690665247e-06,
      "loss": 0.9783,
      "step": 1316
    },
    {
      "epoch": 0.09501821723603045,
      "grad_norm": 10.3154514597452,
      "learning_rate": 3.955834068914203e-06,
      "loss": 1.0292,
      "step": 1317
    },
    {
      "epoch": 0.09509036470545795,
      "grad_norm": 9.436190684863956,
      "learning_rate": 3.9557363403623024e-06,
      "loss": 0.9792,
      "step": 1318
    },
    {
      "epoch": 0.09516251217488547,
      "grad_norm": 16.28222091034996,
      "learning_rate": 3.955638505014879e-06,
      "loss": 1.0742,
      "step": 1319
    },
    {
      "epoch": 0.09523465964431298,
      "grad_norm": 11.514252828059517,
      "learning_rate": 3.955540562877277e-06,
      "loss": 0.9268,
      "step": 1320
    },
    {
      "epoch": 0.09530680711374048,
      "grad_norm": 24.82197601247859,
      "learning_rate": 3.955442513954843e-06,
      "loss": 1.1241,
      "step": 1321
    },
    {
      "epoch": 0.095378954583168,
      "grad_norm": 10.095708981728327,
      "learning_rate": 3.955344358252934e-06,
      "loss": 1.1179,
      "step": 1322
    },
    {
      "epoch": 0.09545110205259551,
      "grad_norm": 11.250131563900847,
      "learning_rate": 3.955246095776907e-06,
      "loss": 0.8482,
      "step": 1323
    },
    {
      "epoch": 0.09552324952202301,
      "grad_norm": 42.956847783378784,
      "learning_rate": 3.955147726532128e-06,
      "loss": 0.9436,
      "step": 1324
    },
    {
      "epoch": 0.09559539699145052,
      "grad_norm": 13.601851370963198,
      "learning_rate": 3.95504925052397e-06,
      "loss": 1.0481,
      "step": 1325
    },
    {
      "epoch": 0.09566754446087804,
      "grad_norm": 13.414100606927526,
      "learning_rate": 3.95495066775781e-06,
      "loss": 1.1732,
      "step": 1326
    },
    {
      "epoch": 0.09573969193030554,
      "grad_norm": 27.002879801571627,
      "learning_rate": 3.954851978239031e-06,
      "loss": 0.9891,
      "step": 1327
    },
    {
      "epoch": 0.09581183939973305,
      "grad_norm": 27.72812319501572,
      "learning_rate": 3.9547531819730224e-06,
      "loss": 1.0029,
      "step": 1328
    },
    {
      "epoch": 0.09588398686916057,
      "grad_norm": 13.141957400308504,
      "learning_rate": 3.954654278965179e-06,
      "loss": 1.0353,
      "step": 1329
    },
    {
      "epoch": 0.09595613433858807,
      "grad_norm": 13.65902688711497,
      "learning_rate": 3.9545552692209e-06,
      "loss": 0.9999,
      "step": 1330
    },
    {
      "epoch": 0.09602828180801558,
      "grad_norm": 12.83735441814361,
      "learning_rate": 3.954456152745594e-06,
      "loss": 0.9358,
      "step": 1331
    },
    {
      "epoch": 0.0961004292774431,
      "grad_norm": 15.379697004689485,
      "learning_rate": 3.954356929544672e-06,
      "loss": 1.1875,
      "step": 1332
    },
    {
      "epoch": 0.0961725767468706,
      "grad_norm": 19.91132150001484,
      "learning_rate": 3.954257599623554e-06,
      "loss": 1.0874,
      "step": 1333
    },
    {
      "epoch": 0.09624472421629811,
      "grad_norm": 9.76540273184556,
      "learning_rate": 3.954158162987662e-06,
      "loss": 1.0425,
      "step": 1334
    },
    {
      "epoch": 0.09631687168572563,
      "grad_norm": 8.148748346180229,
      "learning_rate": 3.954058619642427e-06,
      "loss": 0.8794,
      "step": 1335
    },
    {
      "epoch": 0.09638901915515313,
      "grad_norm": 8.38138393946505,
      "learning_rate": 3.953958969593285e-06,
      "loss": 1.0955,
      "step": 1336
    },
    {
      "epoch": 0.09646116662458064,
      "grad_norm": 5.720482735704408,
      "learning_rate": 3.953859212845676e-06,
      "loss": 1.0008,
      "step": 1337
    },
    {
      "epoch": 0.09653331409400816,
      "grad_norm": 6.582898218125153,
      "learning_rate": 3.9537593494050494e-06,
      "loss": 1.0209,
      "step": 1338
    },
    {
      "epoch": 0.09660546156343566,
      "grad_norm": 13.63603785270166,
      "learning_rate": 3.953659379276856e-06,
      "loss": 1.0131,
      "step": 1339
    },
    {
      "epoch": 0.09667760903286317,
      "grad_norm": 8.878146191484081,
      "learning_rate": 3.953559302466557e-06,
      "loss": 0.9068,
      "step": 1340
    },
    {
      "epoch": 0.09674975650229069,
      "grad_norm": 7.906156154874736,
      "learning_rate": 3.953459118979617e-06,
      "loss": 1.0285,
      "step": 1341
    },
    {
      "epoch": 0.09682190397171819,
      "grad_norm": 5.129111338899854,
      "learning_rate": 3.9533588288215045e-06,
      "loss": 1.0366,
      "step": 1342
    },
    {
      "epoch": 0.0968940514411457,
      "grad_norm": 13.04288685770937,
      "learning_rate": 3.9532584319976985e-06,
      "loss": 0.9572,
      "step": 1343
    },
    {
      "epoch": 0.09696619891057322,
      "grad_norm": 8.410568999683523,
      "learning_rate": 3.953157928513679e-06,
      "loss": 1.0172,
      "step": 1344
    },
    {
      "epoch": 0.09703834638000072,
      "grad_norm": 7.857522856205447,
      "learning_rate": 3.9530573183749365e-06,
      "loss": 1.0599,
      "step": 1345
    },
    {
      "epoch": 0.09711049384942823,
      "grad_norm": 10.4694386725442,
      "learning_rate": 3.952956601586964e-06,
      "loss": 1.0442,
      "step": 1346
    },
    {
      "epoch": 0.09718264131885575,
      "grad_norm": 10.40221026482135,
      "learning_rate": 3.9528557781552605e-06,
      "loss": 0.9468,
      "step": 1347
    },
    {
      "epoch": 0.09725478878828325,
      "grad_norm": 9.349548377895344,
      "learning_rate": 3.952754848085332e-06,
      "loss": 1.1696,
      "step": 1348
    },
    {
      "epoch": 0.09732693625771076,
      "grad_norm": 5.569715470558045,
      "learning_rate": 3.95265381138269e-06,
      "loss": 1.0316,
      "step": 1349
    },
    {
      "epoch": 0.09739908372713828,
      "grad_norm": 13.413010819831976,
      "learning_rate": 3.952552668052853e-06,
      "loss": 0.9793,
      "step": 1350
    },
    {
      "epoch": 0.09747123119656578,
      "grad_norm": 4.4465360382277135,
      "learning_rate": 3.952451418101342e-06,
      "loss": 0.9589,
      "step": 1351
    },
    {
      "epoch": 0.09754337866599329,
      "grad_norm": 28.858553343406157,
      "learning_rate": 3.952350061533687e-06,
      "loss": 0.9785,
      "step": 1352
    },
    {
      "epoch": 0.0976155261354208,
      "grad_norm": 28.318365234778515,
      "learning_rate": 3.952248598355422e-06,
      "loss": 0.981,
      "step": 1353
    },
    {
      "epoch": 0.09768767360484831,
      "grad_norm": 11.546683386011624,
      "learning_rate": 3.952147028572088e-06,
      "loss": 1.0293,
      "step": 1354
    },
    {
      "epoch": 0.09775982107427582,
      "grad_norm": 7.025182657819825,
      "learning_rate": 3.952045352189232e-06,
      "loss": 1.0859,
      "step": 1355
    },
    {
      "epoch": 0.09783196854370332,
      "grad_norm": 6.869730369018684,
      "learning_rate": 3.9519435692124056e-06,
      "loss": 0.9958,
      "step": 1356
    },
    {
      "epoch": 0.09790411601313084,
      "grad_norm": 7.293174834728829,
      "learning_rate": 3.951841679647167e-06,
      "loss": 1.0613,
      "step": 1357
    },
    {
      "epoch": 0.09797626348255835,
      "grad_norm": 4.531541907017269,
      "learning_rate": 3.9517396834990784e-06,
      "loss": 1.0534,
      "step": 1358
    },
    {
      "epoch": 0.09804841095198585,
      "grad_norm": 6.507157419854508,
      "learning_rate": 3.951637580773712e-06,
      "loss": 1.0065,
      "step": 1359
    },
    {
      "epoch": 0.09812055842141337,
      "grad_norm": 6.349860795064662,
      "learning_rate": 3.951535371476641e-06,
      "loss": 1.068,
      "step": 1360
    },
    {
      "epoch": 0.09819270589084088,
      "grad_norm": 6.888663291655412,
      "learning_rate": 3.951433055613449e-06,
      "loss": 1.0645,
      "step": 1361
    },
    {
      "epoch": 0.09826485336026838,
      "grad_norm": 3.9596509781589906,
      "learning_rate": 3.951330633189721e-06,
      "loss": 0.9693,
      "step": 1362
    },
    {
      "epoch": 0.0983370008296959,
      "grad_norm": 5.592942163537697,
      "learning_rate": 3.951228104211051e-06,
      "loss": 0.9577,
      "step": 1363
    },
    {
      "epoch": 0.09840914829912341,
      "grad_norm": 12.657711142508644,
      "learning_rate": 3.951125468683037e-06,
      "loss": 0.971,
      "step": 1364
    },
    {
      "epoch": 0.09848129576855091,
      "grad_norm": 7.286480379832474,
      "learning_rate": 3.951022726611284e-06,
      "loss": 1.0514,
      "step": 1365
    },
    {
      "epoch": 0.09855344323797843,
      "grad_norm": 6.339990262767161,
      "learning_rate": 3.950919878001403e-06,
      "loss": 1.0183,
      "step": 1366
    },
    {
      "epoch": 0.09862559070740594,
      "grad_norm": 10.322432480025602,
      "learning_rate": 3.950816922859009e-06,
      "loss": 0.9935,
      "step": 1367
    },
    {
      "epoch": 0.09869773817683344,
      "grad_norm": 6.515705501507342,
      "learning_rate": 3.950713861189724e-06,
      "loss": 1.0003,
      "step": 1368
    },
    {
      "epoch": 0.09876988564626096,
      "grad_norm": 10.354114514429986,
      "learning_rate": 3.950610692999177e-06,
      "loss": 1.032,
      "step": 1369
    },
    {
      "epoch": 0.09884203311568847,
      "grad_norm": 8.976021932156945,
      "learning_rate": 3.950507418293001e-06,
      "loss": 1.103,
      "step": 1370
    },
    {
      "epoch": 0.09891418058511597,
      "grad_norm": 10.07742699859986,
      "learning_rate": 3.950404037076836e-06,
      "loss": 1.0101,
      "step": 1371
    },
    {
      "epoch": 0.0989863280545435,
      "grad_norm": 14.989097702222033,
      "learning_rate": 3.950300549356326e-06,
      "loss": 1.1343,
      "step": 1372
    },
    {
      "epoch": 0.099058475523971,
      "grad_norm": 9.146151340896669,
      "learning_rate": 3.950196955137122e-06,
      "loss": 1.0353,
      "step": 1373
    },
    {
      "epoch": 0.0991306229933985,
      "grad_norm": 12.429448114052445,
      "learning_rate": 3.950093254424883e-06,
      "loss": 0.8869,
      "step": 1374
    },
    {
      "epoch": 0.09920277046282602,
      "grad_norm": 8.758506781411947,
      "learning_rate": 3.94998944722527e-06,
      "loss": 1.022,
      "step": 1375
    },
    {
      "epoch": 0.09927491793225353,
      "grad_norm": 8.964915236643083,
      "learning_rate": 3.949885533543951e-06,
      "loss": 0.9957,
      "step": 1376
    },
    {
      "epoch": 0.09934706540168103,
      "grad_norm": 5.830058168412243,
      "learning_rate": 3.949781513386602e-06,
      "loss": 1.0547,
      "step": 1377
    },
    {
      "epoch": 0.09941921287110854,
      "grad_norm": 6.427698469684736,
      "learning_rate": 3.949677386758902e-06,
      "loss": 1.0118,
      "step": 1378
    },
    {
      "epoch": 0.09949136034053606,
      "grad_norm": 12.114208964595692,
      "learning_rate": 3.949573153666538e-06,
      "loss": 0.9218,
      "step": 1379
    },
    {
      "epoch": 0.09956350780996356,
      "grad_norm": 13.109431451941921,
      "learning_rate": 3.9494688141152005e-06,
      "loss": 1.0821,
      "step": 1380
    },
    {
      "epoch": 0.09963565527939107,
      "grad_norm": 3.851618087398375,
      "learning_rate": 3.949364368110589e-06,
      "loss": 0.9953,
      "step": 1381
    },
    {
      "epoch": 0.09970780274881859,
      "grad_norm": 5.541882484459375,
      "learning_rate": 3.949259815658404e-06,
      "loss": 0.979,
      "step": 1382
    },
    {
      "epoch": 0.0997799502182461,
      "grad_norm": 7.9500889707181255,
      "learning_rate": 3.949155156764357e-06,
      "loss": 0.9705,
      "step": 1383
    },
    {
      "epoch": 0.0998520976876736,
      "grad_norm": 7.487316087301385,
      "learning_rate": 3.949050391434163e-06,
      "loss": 0.9811,
      "step": 1384
    },
    {
      "epoch": 0.09992424515710112,
      "grad_norm": 5.792241715166423,
      "learning_rate": 3.948945519673541e-06,
      "loss": 0.902,
      "step": 1385
    },
    {
      "epoch": 0.09999639262652862,
      "grad_norm": 7.760810941029446,
      "learning_rate": 3.94884054148822e-06,
      "loss": 0.9456,
      "step": 1386
    },
    {
      "epoch": 0.10006854009595613,
      "grad_norm": 6.748197526809648,
      "learning_rate": 3.948735456883931e-06,
      "loss": 1.0144,
      "step": 1387
    },
    {
      "epoch": 0.10014068756538365,
      "grad_norm": 8.362462922656967,
      "learning_rate": 3.948630265866414e-06,
      "loss": 0.8829,
      "step": 1388
    },
    {
      "epoch": 0.10021283503481115,
      "grad_norm": 8.34645842409208,
      "learning_rate": 3.9485249684414095e-06,
      "loss": 1.0758,
      "step": 1389
    },
    {
      "epoch": 0.10028498250423866,
      "grad_norm": 6.413160216492517,
      "learning_rate": 3.948419564614671e-06,
      "loss": 1.0661,
      "step": 1390
    },
    {
      "epoch": 0.10035712997366618,
      "grad_norm": 6.267576603400407,
      "learning_rate": 3.948314054391953e-06,
      "loss": 0.9317,
      "step": 1391
    },
    {
      "epoch": 0.10042927744309368,
      "grad_norm": 8.17841703063332,
      "learning_rate": 3.948208437779017e-06,
      "loss": 1.002,
      "step": 1392
    },
    {
      "epoch": 0.10050142491252119,
      "grad_norm": 5.912479728034915,
      "learning_rate": 3.948102714781629e-06,
      "loss": 0.9564,
      "step": 1393
    },
    {
      "epoch": 0.10057357238194871,
      "grad_norm": 4.241418364826016,
      "learning_rate": 3.947996885405564e-06,
      "loss": 0.9094,
      "step": 1394
    },
    {
      "epoch": 0.10064571985137621,
      "grad_norm": 7.307349322837698,
      "learning_rate": 3.9478909496566005e-06,
      "loss": 1.0175,
      "step": 1395
    },
    {
      "epoch": 0.10071786732080372,
      "grad_norm": 4.19122126331939,
      "learning_rate": 3.947784907540523e-06,
      "loss": 1.0243,
      "step": 1396
    },
    {
      "epoch": 0.10079001479023124,
      "grad_norm": 11.765539404569124,
      "learning_rate": 3.947678759063122e-06,
      "loss": 1.0369,
      "step": 1397
    },
    {
      "epoch": 0.10086216225965874,
      "grad_norm": 4.390117140751655,
      "learning_rate": 3.947572504230194e-06,
      "loss": 1.0008,
      "step": 1398
    },
    {
      "epoch": 0.10093430972908625,
      "grad_norm": 8.982972886516068,
      "learning_rate": 3.947466143047542e-06,
      "loss": 1.054,
      "step": 1399
    },
    {
      "epoch": 0.10100645719851377,
      "grad_norm": 5.567951745427065,
      "learning_rate": 3.947359675520973e-06,
      "loss": 1.0567,
      "step": 1400
    },
    {
      "epoch": 0.10107860466794127,
      "grad_norm": 7.322881648783479,
      "learning_rate": 3.9472531016563e-06,
      "loss": 1.1177,
      "step": 1401
    },
    {
      "epoch": 0.10115075213736878,
      "grad_norm": 4.502337378530977,
      "learning_rate": 3.947146421459344e-06,
      "loss": 1.0328,
      "step": 1402
    },
    {
      "epoch": 0.10122289960679628,
      "grad_norm": 6.443328293063105,
      "learning_rate": 3.947039634935931e-06,
      "loss": 1.0804,
      "step": 1403
    },
    {
      "epoch": 0.1012950470762238,
      "grad_norm": 4.254735888522236,
      "learning_rate": 3.94693274209189e-06,
      "loss": 0.9665,
      "step": 1404
    },
    {
      "epoch": 0.10136719454565131,
      "grad_norm": 6.188935392712224,
      "learning_rate": 3.94682574293306e-06,
      "loss": 0.9351,
      "step": 1405
    },
    {
      "epoch": 0.10143934201507881,
      "grad_norm": 8.736790740951271,
      "learning_rate": 3.946718637465282e-06,
      "loss": 1.1084,
      "step": 1406
    },
    {
      "epoch": 0.10151148948450633,
      "grad_norm": 9.959292526549993,
      "learning_rate": 3.946611425694407e-06,
      "loss": 1.0248,
      "step": 1407
    },
    {
      "epoch": 0.10158363695393384,
      "grad_norm": 4.866300988938252,
      "learning_rate": 3.946504107626288e-06,
      "loss": 1.0616,
      "step": 1408
    },
    {
      "epoch": 0.10165578442336135,
      "grad_norm": 6.801970016992176,
      "learning_rate": 3.946396683266785e-06,
      "loss": 1.1191,
      "step": 1409
    },
    {
      "epoch": 0.10172793189278886,
      "grad_norm": 6.257090704315485,
      "learning_rate": 3.946289152621764e-06,
      "loss": 1.0061,
      "step": 1410
    },
    {
      "epoch": 0.10180007936221637,
      "grad_norm": 7.29801902543126,
      "learning_rate": 3.9461815156970975e-06,
      "loss": 1.0099,
      "step": 1411
    },
    {
      "epoch": 0.10187222683164388,
      "grad_norm": 8.738384275208485,
      "learning_rate": 3.946073772498664e-06,
      "loss": 1.0755,
      "step": 1412
    },
    {
      "epoch": 0.1019443743010714,
      "grad_norm": 7.361281486154989,
      "learning_rate": 3.945965923032345e-06,
      "loss": 1.1991,
      "step": 1413
    },
    {
      "epoch": 0.1020165217704989,
      "grad_norm": 4.838840088061099,
      "learning_rate": 3.945857967304031e-06,
      "loss": 0.9724,
      "step": 1414
    },
    {
      "epoch": 0.1020886692399264,
      "grad_norm": 7.13092223207369,
      "learning_rate": 3.945749905319616e-06,
      "loss": 1.0725,
      "step": 1415
    },
    {
      "epoch": 0.10216081670935392,
      "grad_norm": 0.8741682731776087,
      "learning_rate": 3.9456417370850015e-06,
      "loss": 0.9252,
      "step": 1416
    },
    {
      "epoch": 0.10223296417878143,
      "grad_norm": 5.5812092436768665,
      "learning_rate": 3.9455334626060955e-06,
      "loss": 0.965,
      "step": 1417
    },
    {
      "epoch": 0.10230511164820894,
      "grad_norm": 4.627618821492458,
      "learning_rate": 3.945425081888809e-06,
      "loss": 1.0396,
      "step": 1418
    },
    {
      "epoch": 0.10237725911763645,
      "grad_norm": 3.876759283567218,
      "learning_rate": 3.94531659493906e-06,
      "loss": 1.0012,
      "step": 1419
    },
    {
      "epoch": 0.10244940658706396,
      "grad_norm": 4.070453546170648,
      "learning_rate": 3.945208001762773e-06,
      "loss": 0.9915,
      "step": 1420
    },
    {
      "epoch": 0.10252155405649147,
      "grad_norm": 5.329955561239704,
      "learning_rate": 3.9450993023658775e-06,
      "loss": 0.9839,
      "step": 1421
    },
    {
      "epoch": 0.10259370152591898,
      "grad_norm": 4.140461929277091,
      "learning_rate": 3.9449904967543105e-06,
      "loss": 0.9915,
      "step": 1422
    },
    {
      "epoch": 0.10266584899534649,
      "grad_norm": 4.168621494766103,
      "learning_rate": 3.9448815849340125e-06,
      "loss": 0.9182,
      "step": 1423
    },
    {
      "epoch": 0.102737996464774,
      "grad_norm": 5.004359633478246,
      "learning_rate": 3.944772566910931e-06,
      "loss": 0.975,
      "step": 1424
    },
    {
      "epoch": 0.10281014393420151,
      "grad_norm": 4.918462823921003,
      "learning_rate": 3.944663442691018e-06,
      "loss": 0.9747,
      "step": 1425
    },
    {
      "epoch": 0.10288229140362902,
      "grad_norm": 0.7240541701152556,
      "learning_rate": 3.944554212280234e-06,
      "loss": 0.8372,
      "step": 1426
    },
    {
      "epoch": 0.10295443887305653,
      "grad_norm": 6.702393960992565,
      "learning_rate": 3.9444448756845425e-06,
      "loss": 1.1011,
      "step": 1427
    },
    {
      "epoch": 0.10302658634248403,
      "grad_norm": 5.690665569296972,
      "learning_rate": 3.944335432909915e-06,
      "loss": 1.0404,
      "step": 1428
    },
    {
      "epoch": 0.10309873381191155,
      "grad_norm": 4.5621504976366065,
      "learning_rate": 3.944225883962327e-06,
      "loss": 1.0485,
      "step": 1429
    },
    {
      "epoch": 0.10317088128133906,
      "grad_norm": 8.57732005416164,
      "learning_rate": 3.944116228847761e-06,
      "loss": 0.987,
      "step": 1430
    },
    {
      "epoch": 0.10324302875076656,
      "grad_norm": 4.281839371926791,
      "learning_rate": 3.9440064675722035e-06,
      "loss": 1.0195,
      "step": 1431
    },
    {
      "epoch": 0.10331517622019408,
      "grad_norm": 5.434561801571278,
      "learning_rate": 3.94389660014165e-06,
      "loss": 1.0383,
      "step": 1432
    },
    {
      "epoch": 0.10338732368962159,
      "grad_norm": 5.763735412433591,
      "learning_rate": 3.9437866265621e-06,
      "loss": 0.9306,
      "step": 1433
    },
    {
      "epoch": 0.10345947115904909,
      "grad_norm": 3.1634873609222534,
      "learning_rate": 3.943676546839557e-06,
      "loss": 1.03,
      "step": 1434
    },
    {
      "epoch": 0.10353161862847661,
      "grad_norm": 2.595780095207092,
      "learning_rate": 3.943566360980033e-06,
      "loss": 1.0778,
      "step": 1435
    },
    {
      "epoch": 0.10360376609790412,
      "grad_norm": 4.66889618794095,
      "learning_rate": 3.943456068989546e-06,
      "loss": 1.0855,
      "step": 1436
    },
    {
      "epoch": 0.10367591356733162,
      "grad_norm": 4.376859215371969,
      "learning_rate": 3.943345670874116e-06,
      "loss": 1.0831,
      "step": 1437
    },
    {
      "epoch": 0.10374806103675914,
      "grad_norm": 3.7279425091137504,
      "learning_rate": 3.943235166639774e-06,
      "loss": 0.9969,
      "step": 1438
    },
    {
      "epoch": 0.10382020850618665,
      "grad_norm": 3.7409778144951518,
      "learning_rate": 3.943124556292553e-06,
      "loss": 1.0421,
      "step": 1439
    },
    {
      "epoch": 0.10389235597561415,
      "grad_norm": 3.5722546793674574,
      "learning_rate": 3.943013839838492e-06,
      "loss": 1.0235,
      "step": 1440
    },
    {
      "epoch": 0.10396450344504167,
      "grad_norm": 3.792889799480847,
      "learning_rate": 3.942903017283639e-06,
      "loss": 0.8769,
      "step": 1441
    },
    {
      "epoch": 0.10403665091446918,
      "grad_norm": 7.932574809166778,
      "learning_rate": 3.942792088634045e-06,
      "loss": 0.9069,
      "step": 1442
    },
    {
      "epoch": 0.10410879838389668,
      "grad_norm": 3.273916277775636,
      "learning_rate": 3.942681053895766e-06,
      "loss": 1.0447,
      "step": 1443
    },
    {
      "epoch": 0.1041809458533242,
      "grad_norm": 4.005097479044722,
      "learning_rate": 3.9425699130748665e-06,
      "loss": 0.9987,
      "step": 1444
    },
    {
      "epoch": 0.1042530933227517,
      "grad_norm": 16.37563922595371,
      "learning_rate": 3.942458666177416e-06,
      "loss": 1.1316,
      "step": 1445
    },
    {
      "epoch": 0.10432524079217921,
      "grad_norm": 3.5848948307120008,
      "learning_rate": 3.9423473132094885e-06,
      "loss": 1.0014,
      "step": 1446
    },
    {
      "epoch": 0.10439738826160673,
      "grad_norm": 29.56351061135955,
      "learning_rate": 3.942235854177164e-06,
      "loss": 1.0704,
      "step": 1447
    },
    {
      "epoch": 0.10446953573103424,
      "grad_norm": 11.431600414513682,
      "learning_rate": 3.942124289086529e-06,
      "loss": 1.12,
      "step": 1448
    },
    {
      "epoch": 0.10454168320046174,
      "grad_norm": 6.199479038132017,
      "learning_rate": 3.942012617943676e-06,
      "loss": 0.9979,
      "step": 1449
    },
    {
      "epoch": 0.10461383066988926,
      "grad_norm": 4.003461293873461,
      "learning_rate": 3.941900840754704e-06,
      "loss": 1.031,
      "step": 1450
    },
    {
      "epoch": 0.10468597813931677,
      "grad_norm": 8.602409522044445,
      "learning_rate": 3.941788957525715e-06,
      "loss": 1.0032,
      "step": 1451
    },
    {
      "epoch": 0.10475812560874427,
      "grad_norm": 1.036998275869439,
      "learning_rate": 3.941676968262819e-06,
      "loss": 0.8736,
      "step": 1452
    },
    {
      "epoch": 0.10483027307817178,
      "grad_norm": 3.941591952195806,
      "learning_rate": 3.941564872972132e-06,
      "loss": 1.0431,
      "step": 1453
    },
    {
      "epoch": 0.1049024205475993,
      "grad_norm": 4.869957883993466,
      "learning_rate": 3.941452671659775e-06,
      "loss": 1.0005,
      "step": 1454
    },
    {
      "epoch": 0.1049745680170268,
      "grad_norm": 3.3550588870388025,
      "learning_rate": 3.941340364331875e-06,
      "loss": 1.081,
      "step": 1455
    },
    {
      "epoch": 0.1050467154864543,
      "grad_norm": 3.748514644183572,
      "learning_rate": 3.941227950994564e-06,
      "loss": 0.9879,
      "step": 1456
    },
    {
      "epoch": 0.10511886295588183,
      "grad_norm": 3.256823785350267,
      "learning_rate": 3.94111543165398e-06,
      "loss": 1.0567,
      "step": 1457
    },
    {
      "epoch": 0.10519101042530933,
      "grad_norm": 2.89708042506915,
      "learning_rate": 3.941002806316269e-06,
      "loss": 0.9964,
      "step": 1458
    },
    {
      "epoch": 0.10526315789473684,
      "grad_norm": 3.002118316453622,
      "learning_rate": 3.9408900749875795e-06,
      "loss": 1.0025,
      "step": 1459
    },
    {
      "epoch": 0.10533530536416436,
      "grad_norm": 2.8006227413850278,
      "learning_rate": 3.940777237674068e-06,
      "loss": 1.1007,
      "step": 1460
    },
    {
      "epoch": 0.10540745283359186,
      "grad_norm": 4.305568708721019,
      "learning_rate": 3.940664294381896e-06,
      "loss": 0.9935,
      "step": 1461
    },
    {
      "epoch": 0.10547960030301937,
      "grad_norm": 0.7540016706888323,
      "learning_rate": 3.940551245117231e-06,
      "loss": 0.808,
      "step": 1462
    },
    {
      "epoch": 0.10555174777244689,
      "grad_norm": 4.182210969760021,
      "learning_rate": 3.940438089886246e-06,
      "loss": 1.005,
      "step": 1463
    },
    {
      "epoch": 0.10562389524187439,
      "grad_norm": 5.05191572507758,
      "learning_rate": 3.940324828695121e-06,
      "loss": 1.0336,
      "step": 1464
    },
    {
      "epoch": 0.1056960427113019,
      "grad_norm": 38.53253622143565,
      "learning_rate": 3.9402114615500394e-06,
      "loss": 0.9784,
      "step": 1465
    },
    {
      "epoch": 0.10576819018072942,
      "grad_norm": 5.693181145366365,
      "learning_rate": 3.9400979884571925e-06,
      "loss": 1.0181,
      "step": 1466
    },
    {
      "epoch": 0.10584033765015692,
      "grad_norm": 6.946756871394985,
      "learning_rate": 3.939984409422776e-06,
      "loss": 1.081,
      "step": 1467
    },
    {
      "epoch": 0.10591248511958443,
      "grad_norm": 3.3961924887608084,
      "learning_rate": 3.939870724452993e-06,
      "loss": 1.1088,
      "step": 1468
    },
    {
      "epoch": 0.10598463258901195,
      "grad_norm": 4.499965243735136,
      "learning_rate": 3.9397569335540505e-06,
      "loss": 0.9266,
      "step": 1469
    },
    {
      "epoch": 0.10605678005843945,
      "grad_norm": 3.5587231877358363,
      "learning_rate": 3.939643036732163e-06,
      "loss": 1.0284,
      "step": 1470
    },
    {
      "epoch": 0.10612892752786696,
      "grad_norm": 7.981952098055606,
      "learning_rate": 3.939529033993549e-06,
      "loss": 1.0806,
      "step": 1471
    },
    {
      "epoch": 0.10620107499729448,
      "grad_norm": 5.66522987700642,
      "learning_rate": 3.939414925344435e-06,
      "loss": 1.0537,
      "step": 1472
    },
    {
      "epoch": 0.10627322246672198,
      "grad_norm": 2.6326470478718207,
      "learning_rate": 3.939300710791051e-06,
      "loss": 0.9845,
      "step": 1473
    },
    {
      "epoch": 0.10634536993614949,
      "grad_norm": 5.026046810905183,
      "learning_rate": 3.939186390339633e-06,
      "loss": 0.9934,
      "step": 1474
    },
    {
      "epoch": 0.106417517405577,
      "grad_norm": 3.6320746215935227,
      "learning_rate": 3.939071963996427e-06,
      "loss": 1.1895,
      "step": 1475
    },
    {
      "epoch": 0.10648966487500451,
      "grad_norm": 9.17267226957016,
      "learning_rate": 3.938957431767679e-06,
      "loss": 0.9928,
      "step": 1476
    },
    {
      "epoch": 0.10656181234443202,
      "grad_norm": 3.517958210135777,
      "learning_rate": 3.938842793659642e-06,
      "loss": 0.97,
      "step": 1477
    },
    {
      "epoch": 0.10663395981385954,
      "grad_norm": 4.03734866587937,
      "learning_rate": 3.938728049678579e-06,
      "loss": 1.1224,
      "step": 1478
    },
    {
      "epoch": 0.10670610728328704,
      "grad_norm": 12.583932083402528,
      "learning_rate": 3.938613199830753e-06,
      "loss": 1.0533,
      "step": 1479
    },
    {
      "epoch": 0.10677825475271455,
      "grad_norm": 5.072381066384621,
      "learning_rate": 3.9384982441224365e-06,
      "loss": 1.0215,
      "step": 1480
    },
    {
      "epoch": 0.10685040222214205,
      "grad_norm": 6.2321381726991,
      "learning_rate": 3.938383182559907e-06,
      "loss": 0.9693,
      "step": 1481
    },
    {
      "epoch": 0.10692254969156957,
      "grad_norm": 3.1196271866802467,
      "learning_rate": 3.9382680151494486e-06,
      "loss": 1.0921,
      "step": 1482
    },
    {
      "epoch": 0.10699469716099708,
      "grad_norm": 5.087829519186801,
      "learning_rate": 3.938152741897349e-06,
      "loss": 1.0374,
      "step": 1483
    },
    {
      "epoch": 0.10706684463042458,
      "grad_norm": 3.468604832266408,
      "learning_rate": 3.938037362809902e-06,
      "loss": 1.0097,
      "step": 1484
    },
    {
      "epoch": 0.1071389920998521,
      "grad_norm": 11.208517949924916,
      "learning_rate": 3.937921877893409e-06,
      "loss": 1.0466,
      "step": 1485
    },
    {
      "epoch": 0.1072111395692796,
      "grad_norm": 4.705221627637273,
      "learning_rate": 3.937806287154177e-06,
      "loss": 0.9872,
      "step": 1486
    },
    {
      "epoch": 0.10728328703870711,
      "grad_norm": 4.823840168913117,
      "learning_rate": 3.937690590598516e-06,
      "loss": 0.9974,
      "step": 1487
    },
    {
      "epoch": 0.10735543450813463,
      "grad_norm": 4.724681435800704,
      "learning_rate": 3.937574788232747e-06,
      "loss": 1.027,
      "step": 1488
    },
    {
      "epoch": 0.10742758197756214,
      "grad_norm": 6.899567145439893,
      "learning_rate": 3.937458880063189e-06,
      "loss": 1.0659,
      "step": 1489
    },
    {
      "epoch": 0.10749972944698964,
      "grad_norm": 5.681039011849012,
      "learning_rate": 3.937342866096175e-06,
      "loss": 0.9524,
      "step": 1490
    },
    {
      "epoch": 0.10757187691641716,
      "grad_norm": 8.961068189843628,
      "learning_rate": 3.937226746338039e-06,
      "loss": 1.0411,
      "step": 1491
    },
    {
      "epoch": 0.10764402438584467,
      "grad_norm": 6.519917124122048,
      "learning_rate": 3.937110520795122e-06,
      "loss": 0.9954,
      "step": 1492
    },
    {
      "epoch": 0.10771617185527217,
      "grad_norm": 16.799162271593897,
      "learning_rate": 3.93699418947377e-06,
      "loss": 1.082,
      "step": 1493
    },
    {
      "epoch": 0.10778831932469969,
      "grad_norm": 4.600896764501205,
      "learning_rate": 3.936877752380336e-06,
      "loss": 1.0239,
      "step": 1494
    },
    {
      "epoch": 0.1078604667941272,
      "grad_norm": 6.3219319497983255,
      "learning_rate": 3.936761209521178e-06,
      "loss": 1.0,
      "step": 1495
    },
    {
      "epoch": 0.1079326142635547,
      "grad_norm": 4.055176222202339,
      "learning_rate": 3.93664456090266e-06,
      "loss": 0.918,
      "step": 1496
    },
    {
      "epoch": 0.10800476173298222,
      "grad_norm": 11.688896218198051,
      "learning_rate": 3.936527806531151e-06,
      "loss": 1.0202,
      "step": 1497
    },
    {
      "epoch": 0.10807690920240973,
      "grad_norm": 4.740972473244854,
      "learning_rate": 3.936410946413028e-06,
      "loss": 0.9921,
      "step": 1498
    },
    {
      "epoch": 0.10814905667183723,
      "grad_norm": 4.190696748692507,
      "learning_rate": 3.9362939805546716e-06,
      "loss": 1.0277,
      "step": 1499
    },
    {
      "epoch": 0.10822120414126475,
      "grad_norm": 36.02706061670432,
      "learning_rate": 3.9361769089624685e-06,
      "loss": 1.0084,
      "step": 1500
    },
    {
      "epoch": 0.10829335161069226,
      "grad_norm": 4.138476004832953,
      "learning_rate": 3.9360597316428125e-06,
      "loss": 0.9988,
      "step": 1501
    },
    {
      "epoch": 0.10836549908011976,
      "grad_norm": 4.077319069221383,
      "learning_rate": 3.935942448602101e-06,
      "loss": 1.0137,
      "step": 1502
    },
    {
      "epoch": 0.10843764654954728,
      "grad_norm": 0.8144204112556588,
      "learning_rate": 3.935825059846739e-06,
      "loss": 0.8643,
      "step": 1503
    },
    {
      "epoch": 0.10850979401897479,
      "grad_norm": 5.103571871401207,
      "learning_rate": 3.935707565383137e-06,
      "loss": 1.083,
      "step": 1504
    },
    {
      "epoch": 0.10858194148840229,
      "grad_norm": 3.6315038805666053,
      "learning_rate": 3.93558996521771e-06,
      "loss": 0.9107,
      "step": 1505
    },
    {
      "epoch": 0.1086540889578298,
      "grad_norm": 7.816883047832818,
      "learning_rate": 3.935472259356881e-06,
      "loss": 1.0111,
      "step": 1506
    },
    {
      "epoch": 0.10872623642725732,
      "grad_norm": 4.658764461068078,
      "learning_rate": 3.935354447807077e-06,
      "loss": 0.9732,
      "step": 1507
    },
    {
      "epoch": 0.10879838389668482,
      "grad_norm": 5.4059920938527855,
      "learning_rate": 3.93523653057473e-06,
      "loss": 1.017,
      "step": 1508
    },
    {
      "epoch": 0.10887053136611233,
      "grad_norm": 8.699207348724006,
      "learning_rate": 3.9351185076662805e-06,
      "loss": 1.0691,
      "step": 1509
    },
    {
      "epoch": 0.10894267883553985,
      "grad_norm": 7.687038997070359,
      "learning_rate": 3.935000379088173e-06,
      "loss": 1.0406,
      "step": 1510
    },
    {
      "epoch": 0.10901482630496735,
      "grad_norm": 6.582980214881515,
      "learning_rate": 3.934882144846858e-06,
      "loss": 0.9973,
      "step": 1511
    },
    {
      "epoch": 0.10908697377439486,
      "grad_norm": 10.302509439191448,
      "learning_rate": 3.934763804948792e-06,
      "loss": 1.012,
      "step": 1512
    },
    {
      "epoch": 0.10915912124382238,
      "grad_norm": 4.577630026036774,
      "learning_rate": 3.934645359400436e-06,
      "loss": 1.0258,
      "step": 1513
    },
    {
      "epoch": 0.10923126871324988,
      "grad_norm": 3.6622688767348155,
      "learning_rate": 3.934526808208259e-06,
      "loss": 0.9959,
      "step": 1514
    },
    {
      "epoch": 0.10930341618267739,
      "grad_norm": 5.98510833310753,
      "learning_rate": 3.934408151378735e-06,
      "loss": 0.8713,
      "step": 1515
    },
    {
      "epoch": 0.1093755636521049,
      "grad_norm": 50.254761123947254,
      "learning_rate": 3.934289388918343e-06,
      "loss": 1.0991,
      "step": 1516
    },
    {
      "epoch": 0.10944771112153241,
      "grad_norm": 3.890329617850197,
      "learning_rate": 3.934170520833567e-06,
      "loss": 1.0125,
      "step": 1517
    },
    {
      "epoch": 0.10951985859095992,
      "grad_norm": 6.614094233014326,
      "learning_rate": 3.934051547130899e-06,
      "loss": 1.0507,
      "step": 1518
    },
    {
      "epoch": 0.10959200606038744,
      "grad_norm": 0.9175594959665876,
      "learning_rate": 3.933932467816836e-06,
      "loss": 0.8715,
      "step": 1519
    },
    {
      "epoch": 0.10966415352981494,
      "grad_norm": 5.985080607675313,
      "learning_rate": 3.9338132828978795e-06,
      "loss": 1.0483,
      "step": 1520
    },
    {
      "epoch": 0.10973630099924245,
      "grad_norm": 6.075057405428203,
      "learning_rate": 3.933693992380539e-06,
      "loss": 1.0066,
      "step": 1521
    },
    {
      "epoch": 0.10980844846866997,
      "grad_norm": 5.249865030642771,
      "learning_rate": 3.933574596271328e-06,
      "loss": 1.0191,
      "step": 1522
    },
    {
      "epoch": 0.10988059593809747,
      "grad_norm": 5.924034509984583,
      "learning_rate": 3.933455094576766e-06,
      "loss": 1.0311,
      "step": 1523
    },
    {
      "epoch": 0.10995274340752498,
      "grad_norm": 10.885739340250929,
      "learning_rate": 3.933335487303378e-06,
      "loss": 0.9679,
      "step": 1524
    },
    {
      "epoch": 0.1100248908769525,
      "grad_norm": 4.458638736330127,
      "learning_rate": 3.933215774457697e-06,
      "loss": 1.0547,
      "step": 1525
    },
    {
      "epoch": 0.11009703834638,
      "grad_norm": 17.50537462807182,
      "learning_rate": 3.933095956046259e-06,
      "loss": 0.9908,
      "step": 1526
    },
    {
      "epoch": 0.11016918581580751,
      "grad_norm": 5.203662924696632,
      "learning_rate": 3.932976032075607e-06,
      "loss": 1.1184,
      "step": 1527
    },
    {
      "epoch": 0.11024133328523503,
      "grad_norm": 6.779462868898353,
      "learning_rate": 3.932856002552289e-06,
      "loss": 0.9744,
      "step": 1528
    },
    {
      "epoch": 0.11031348075466253,
      "grad_norm": 3.5878911566072005,
      "learning_rate": 3.93273586748286e-06,
      "loss": 1.1107,
      "step": 1529
    },
    {
      "epoch": 0.11038562822409004,
      "grad_norm": 8.455127528225384,
      "learning_rate": 3.93261562687388e-06,
      "loss": 1.0213,
      "step": 1530
    },
    {
      "epoch": 0.11045777569351754,
      "grad_norm": 3.7831616574182587,
      "learning_rate": 3.932495280731914e-06,
      "loss": 0.9741,
      "step": 1531
    },
    {
      "epoch": 0.11052992316294506,
      "grad_norm": 0.8061239883362084,
      "learning_rate": 3.932374829063536e-06,
      "loss": 0.8534,
      "step": 1532
    },
    {
      "epoch": 0.11060207063237257,
      "grad_norm": 4.8719878060700355,
      "learning_rate": 3.9322542718753216e-06,
      "loss": 1.0931,
      "step": 1533
    },
    {
      "epoch": 0.11067421810180007,
      "grad_norm": 8.480721545921263,
      "learning_rate": 3.932133609173854e-06,
      "loss": 1.0651,
      "step": 1534
    },
    {
      "epoch": 0.11074636557122759,
      "grad_norm": 4.3642729995610114,
      "learning_rate": 3.932012840965723e-06,
      "loss": 1.1333,
      "step": 1535
    },
    {
      "epoch": 0.1108185130406551,
      "grad_norm": 4.4420971764966435,
      "learning_rate": 3.931891967257522e-06,
      "loss": 1.018,
      "step": 1536
    },
    {
      "epoch": 0.1108906605100826,
      "grad_norm": 6.582722631529627,
      "learning_rate": 3.931770988055853e-06,
      "loss": 1.0512,
      "step": 1537
    },
    {
      "epoch": 0.11096280797951012,
      "grad_norm": 47.34767929238982,
      "learning_rate": 3.931649903367321e-06,
      "loss": 0.9511,
      "step": 1538
    },
    {
      "epoch": 0.11103495544893763,
      "grad_norm": 16.04474486835565,
      "learning_rate": 3.931528713198539e-06,
      "loss": 1.1024,
      "step": 1539
    },
    {
      "epoch": 0.11110710291836513,
      "grad_norm": 0.8607949749899191,
      "learning_rate": 3.931407417556125e-06,
      "loss": 0.9474,
      "step": 1540
    },
    {
      "epoch": 0.11117925038779265,
      "grad_norm": 4.933635696389873,
      "learning_rate": 3.931286016446701e-06,
      "loss": 1.0694,
      "step": 1541
    },
    {
      "epoch": 0.11125139785722016,
      "grad_norm": 5.72638830151574,
      "learning_rate": 3.931164509876896e-06,
      "loss": 1.0437,
      "step": 1542
    },
    {
      "epoch": 0.11132354532664766,
      "grad_norm": 4.336028996782367,
      "learning_rate": 3.931042897853348e-06,
      "loss": 1.1115,
      "step": 1543
    },
    {
      "epoch": 0.11139569279607518,
      "grad_norm": 7.102686961907967,
      "learning_rate": 3.930921180382694e-06,
      "loss": 1.0154,
      "step": 1544
    },
    {
      "epoch": 0.11146784026550269,
      "grad_norm": 5.85921516709086,
      "learning_rate": 3.930799357471583e-06,
      "loss": 0.9636,
      "step": 1545
    },
    {
      "epoch": 0.1115399877349302,
      "grad_norm": 5.1228327937303915,
      "learning_rate": 3.9306774291266675e-06,
      "loss": 0.9877,
      "step": 1546
    },
    {
      "epoch": 0.11161213520435771,
      "grad_norm": 9.887506605066536,
      "learning_rate": 3.930555395354604e-06,
      "loss": 1.109,
      "step": 1547
    },
    {
      "epoch": 0.11168428267378522,
      "grad_norm": 4.247484696996648,
      "learning_rate": 3.930433256162057e-06,
      "loss": 0.9229,
      "step": 1548
    },
    {
      "epoch": 0.11175643014321272,
      "grad_norm": 9.26653482174103,
      "learning_rate": 3.930311011555696e-06,
      "loss": 0.913,
      "step": 1549
    },
    {
      "epoch": 0.11182857761264024,
      "grad_norm": 3.97820735105994,
      "learning_rate": 3.930188661542196e-06,
      "loss": 1.0957,
      "step": 1550
    },
    {
      "epoch": 0.11190072508206775,
      "grad_norm": 3.453383216006686,
      "learning_rate": 3.930066206128239e-06,
      "loss": 1.1141,
      "step": 1551
    },
    {
      "epoch": 0.11197287255149525,
      "grad_norm": 7.017061150245115,
      "learning_rate": 3.929943645320512e-06,
      "loss": 1.0476,
      "step": 1552
    },
    {
      "epoch": 0.11204502002092277,
      "grad_norm": 14.143904466919166,
      "learning_rate": 3.929820979125706e-06,
      "loss": 1.063,
      "step": 1553
    },
    {
      "epoch": 0.11211716749035028,
      "grad_norm": 6.557371624334624,
      "learning_rate": 3.92969820755052e-06,
      "loss": 1.059,
      "step": 1554
    },
    {
      "epoch": 0.11218931495977778,
      "grad_norm": 3.9815216019717603,
      "learning_rate": 3.92957533060166e-06,
      "loss": 1.086,
      "step": 1555
    },
    {
      "epoch": 0.11226146242920529,
      "grad_norm": 6.278606408869726,
      "learning_rate": 3.9294523482858336e-06,
      "loss": 1.0882,
      "step": 1556
    },
    {
      "epoch": 0.11233360989863281,
      "grad_norm": 0.7157281709620698,
      "learning_rate": 3.929329260609757e-06,
      "loss": 0.8262,
      "step": 1557
    },
    {
      "epoch": 0.11240575736806031,
      "grad_norm": 4.6270188745196545,
      "learning_rate": 3.929206067580151e-06,
      "loss": 0.9249,
      "step": 1558
    },
    {
      "epoch": 0.11247790483748782,
      "grad_norm": 0.7558996697461582,
      "learning_rate": 3.9290827692037435e-06,
      "loss": 0.8579,
      "step": 1559
    },
    {
      "epoch": 0.11255005230691534,
      "grad_norm": 11.085639847461973,
      "learning_rate": 3.9289593654872674e-06,
      "loss": 0.9915,
      "step": 1560
    },
    {
      "epoch": 0.11262219977634284,
      "grad_norm": 4.765983593066074,
      "learning_rate": 3.92883585643746e-06,
      "loss": 1.0625,
      "step": 1561
    },
    {
      "epoch": 0.11269434724577035,
      "grad_norm": 33.073463742194434,
      "learning_rate": 3.928712242061069e-06,
      "loss": 1.0194,
      "step": 1562
    },
    {
      "epoch": 0.11276649471519787,
      "grad_norm": 7.273718872528156,
      "learning_rate": 3.928588522364841e-06,
      "loss": 1.0146,
      "step": 1563
    },
    {
      "epoch": 0.11283864218462537,
      "grad_norm": 4.604034329867199,
      "learning_rate": 3.9284646973555335e-06,
      "loss": 0.9258,
      "step": 1564
    },
    {
      "epoch": 0.11291078965405288,
      "grad_norm": 10.059144023176321,
      "learning_rate": 3.9283407670399074e-06,
      "loss": 0.9324,
      "step": 1565
    },
    {
      "epoch": 0.1129829371234804,
      "grad_norm": 6.162291994124395,
      "learning_rate": 3.928216731424731e-06,
      "loss": 1.0321,
      "step": 1566
    },
    {
      "epoch": 0.1130550845929079,
      "grad_norm": 6.527936030785035,
      "learning_rate": 3.928092590516776e-06,
      "loss": 1.0525,
      "step": 1567
    },
    {
      "epoch": 0.11312723206233541,
      "grad_norm": 6.27059838075557,
      "learning_rate": 3.927968344322824e-06,
      "loss": 1.0384,
      "step": 1568
    },
    {
      "epoch": 0.11319937953176293,
      "grad_norm": 4.9889650645360435,
      "learning_rate": 3.927843992849656e-06,
      "loss": 1.0704,
      "step": 1569
    },
    {
      "epoch": 0.11327152700119043,
      "grad_norm": 6.346452109345875,
      "learning_rate": 3.927719536104065e-06,
      "loss": 0.9549,
      "step": 1570
    },
    {
      "epoch": 0.11334367447061794,
      "grad_norm": 23.24802447714308,
      "learning_rate": 3.927594974092846e-06,
      "loss": 0.9285,
      "step": 1571
    },
    {
      "epoch": 0.11341582194004546,
      "grad_norm": 19.65869901650372,
      "learning_rate": 3.927470306822802e-06,
      "loss": 1.0081,
      "step": 1572
    },
    {
      "epoch": 0.11348796940947296,
      "grad_norm": 4.17728989899774,
      "learning_rate": 3.927345534300738e-06,
      "loss": 0.9976,
      "step": 1573
    },
    {
      "epoch": 0.11356011687890047,
      "grad_norm": 3.6686649657082664,
      "learning_rate": 3.9272206565334705e-06,
      "loss": 1.0764,
      "step": 1574
    },
    {
      "epoch": 0.11363226434832799,
      "grad_norm": 7.413679685849329,
      "learning_rate": 3.927095673527817e-06,
      "loss": 1.0738,
      "step": 1575
    },
    {
      "epoch": 0.1137044118177555,
      "grad_norm": 5.388891676719617,
      "learning_rate": 3.926970585290603e-06,
      "loss": 0.9797,
      "step": 1576
    },
    {
      "epoch": 0.113776559287183,
      "grad_norm": 4.335076323896592,
      "learning_rate": 3.9268453918286585e-06,
      "loss": 1.0292,
      "step": 1577
    },
    {
      "epoch": 0.11384870675661052,
      "grad_norm": 14.261829485498609,
      "learning_rate": 3.926720093148819e-06,
      "loss": 1.0602,
      "step": 1578
    },
    {
      "epoch": 0.11392085422603802,
      "grad_norm": 4.519574085187566,
      "learning_rate": 3.9265946892579295e-06,
      "loss": 0.9578,
      "step": 1579
    },
    {
      "epoch": 0.11399300169546553,
      "grad_norm": 17.153368098113415,
      "learning_rate": 3.9264691801628345e-06,
      "loss": 0.9692,
      "step": 1580
    },
    {
      "epoch": 0.11406514916489305,
      "grad_norm": 5.310779068399898,
      "learning_rate": 3.92634356587039e-06,
      "loss": 1.0416,
      "step": 1581
    },
    {
      "epoch": 0.11413729663432055,
      "grad_norm": 0.7662326192206989,
      "learning_rate": 3.926217846387454e-06,
      "loss": 0.8158,
      "step": 1582
    },
    {
      "epoch": 0.11420944410374806,
      "grad_norm": 7.875431593920891,
      "learning_rate": 3.926092021720893e-06,
      "loss": 0.9897,
      "step": 1583
    },
    {
      "epoch": 0.11428159157317556,
      "grad_norm": 6.80509364145668,
      "learning_rate": 3.925966091877576e-06,
      "loss": 0.9558,
      "step": 1584
    },
    {
      "epoch": 0.11435373904260308,
      "grad_norm": 5.4954556851881,
      "learning_rate": 3.92584005686438e-06,
      "loss": 1.0867,
      "step": 1585
    },
    {
      "epoch": 0.11442588651203059,
      "grad_norm": 5.049721307901613,
      "learning_rate": 3.925713916688189e-06,
      "loss": 0.978,
      "step": 1586
    },
    {
      "epoch": 0.1144980339814581,
      "grad_norm": 82.39382964799306,
      "learning_rate": 3.925587671355889e-06,
      "loss": 0.9556,
      "step": 1587
    },
    {
      "epoch": 0.11457018145088561,
      "grad_norm": 5.243063385924611,
      "learning_rate": 3.9254613208743746e-06,
      "loss": 0.9983,
      "step": 1588
    },
    {
      "epoch": 0.11464232892031312,
      "grad_norm": 6.707177988712214,
      "learning_rate": 3.925334865250545e-06,
      "loss": 0.9993,
      "step": 1589
    },
    {
      "epoch": 0.11471447638974062,
      "grad_norm": 14.50254746782925,
      "learning_rate": 3.925208304491306e-06,
      "loss": 1.0774,
      "step": 1590
    },
    {
      "epoch": 0.11478662385916814,
      "grad_norm": 3.9988696170050546,
      "learning_rate": 3.925081638603568e-06,
      "loss": 1.026,
      "step": 1591
    },
    {
      "epoch": 0.11485877132859565,
      "grad_norm": 5.1748147313959425,
      "learning_rate": 3.924954867594249e-06,
      "loss": 0.9621,
      "step": 1592
    },
    {
      "epoch": 0.11493091879802315,
      "grad_norm": 3.895611734598709,
      "learning_rate": 3.92482799147027e-06,
      "loss": 0.847,
      "step": 1593
    },
    {
      "epoch": 0.11500306626745067,
      "grad_norm": 6.468831674548548,
      "learning_rate": 3.92470101023856e-06,
      "loss": 0.9433,
      "step": 1594
    },
    {
      "epoch": 0.11507521373687818,
      "grad_norm": 10.72056595347433,
      "learning_rate": 3.9245739239060525e-06,
      "loss": 0.9148,
      "step": 1595
    },
    {
      "epoch": 0.11514736120630568,
      "grad_norm": 5.493658571491504,
      "learning_rate": 3.924446732479688e-06,
      "loss": 1.1899,
      "step": 1596
    },
    {
      "epoch": 0.1152195086757332,
      "grad_norm": 13.062141377029864,
      "learning_rate": 3.924319435966411e-06,
      "loss": 1.092,
      "step": 1597
    },
    {
      "epoch": 0.11529165614516071,
      "grad_norm": 25.65173314952069,
      "learning_rate": 3.924192034373174e-06,
      "loss": 1.0144,
      "step": 1598
    },
    {
      "epoch": 0.11536380361458821,
      "grad_norm": 8.167812293007755,
      "learning_rate": 3.924064527706933e-06,
      "loss": 1.0198,
      "step": 1599
    },
    {
      "epoch": 0.11543595108401573,
      "grad_norm": 6.802999889395648,
      "learning_rate": 3.92393691597465e-06,
      "loss": 1.0003,
      "step": 1600
    },
    {
      "epoch": 0.11550809855344324,
      "grad_norm": 5.349165999390754,
      "learning_rate": 3.923809199183295e-06,
      "loss": 1.0355,
      "step": 1601
    },
    {
      "epoch": 0.11558024602287074,
      "grad_norm": 8.373899472934488,
      "learning_rate": 3.9236813773398405e-06,
      "loss": 1.0229,
      "step": 1602
    },
    {
      "epoch": 0.11565239349229826,
      "grad_norm": 0.8020631160087185,
      "learning_rate": 3.923553450451268e-06,
      "loss": 0.8812,
      "step": 1603
    },
    {
      "epoch": 0.11572454096172577,
      "grad_norm": 6.881762975547133,
      "learning_rate": 3.923425418524562e-06,
      "loss": 1.0185,
      "step": 1604
    },
    {
      "epoch": 0.11579668843115327,
      "grad_norm": 3.869199780277217,
      "learning_rate": 3.923297281566714e-06,
      "loss": 0.9122,
      "step": 1605
    },
    {
      "epoch": 0.1158688359005808,
      "grad_norm": 9.705991754578273,
      "learning_rate": 3.923169039584722e-06,
      "loss": 1.0404,
      "step": 1606
    },
    {
      "epoch": 0.1159409833700083,
      "grad_norm": 4.680892967390728,
      "learning_rate": 3.923040692585587e-06,
      "loss": 0.9213,
      "step": 1607
    },
    {
      "epoch": 0.1160131308394358,
      "grad_norm": 7.367629166407021,
      "learning_rate": 3.922912240576319e-06,
      "loss": 1.1023,
      "step": 1608
    },
    {
      "epoch": 0.11608527830886331,
      "grad_norm": 4.584276668687574,
      "learning_rate": 3.922783683563933e-06,
      "loss": 1.0724,
      "step": 1609
    },
    {
      "epoch": 0.11615742577829083,
      "grad_norm": 9.727347248086874,
      "learning_rate": 3.922655021555446e-06,
      "loss": 1.0132,
      "step": 1610
    },
    {
      "epoch": 0.11622957324771833,
      "grad_norm": 4.979334276429509,
      "learning_rate": 3.922526254557887e-06,
      "loss": 1.0778,
      "step": 1611
    },
    {
      "epoch": 0.11630172071714584,
      "grad_norm": 0.8680842462672624,
      "learning_rate": 3.9223973825782854e-06,
      "loss": 0.8088,
      "step": 1612
    },
    {
      "epoch": 0.11637386818657336,
      "grad_norm": 6.756954143012329,
      "learning_rate": 3.92226840562368e-06,
      "loss": 0.9095,
      "step": 1613
    },
    {
      "epoch": 0.11644601565600086,
      "grad_norm": 7.057361725369125,
      "learning_rate": 3.922139323701113e-06,
      "loss": 1.037,
      "step": 1614
    },
    {
      "epoch": 0.11651816312542837,
      "grad_norm": 5.39469141518659,
      "learning_rate": 3.922010136817633e-06,
      "loss": 0.9621,
      "step": 1615
    },
    {
      "epoch": 0.11659031059485589,
      "grad_norm": 6.181428338574479,
      "learning_rate": 3.921880844980293e-06,
      "loss": 0.9386,
      "step": 1616
    },
    {
      "epoch": 0.1166624580642834,
      "grad_norm": 4.555708337304951,
      "learning_rate": 3.921751448196156e-06,
      "loss": 1.0367,
      "step": 1617
    },
    {
      "epoch": 0.1167346055337109,
      "grad_norm": 4.4616384406485965,
      "learning_rate": 3.9216219464722866e-06,
      "loss": 1.0857,
      "step": 1618
    },
    {
      "epoch": 0.11680675300313842,
      "grad_norm": 5.605678009486891,
      "learning_rate": 3.921492339815756e-06,
      "loss": 1.0311,
      "step": 1619
    },
    {
      "epoch": 0.11687890047256592,
      "grad_norm": 10.489093747625882,
      "learning_rate": 3.921362628233642e-06,
      "loss": 1.1023,
      "step": 1620
    },
    {
      "epoch": 0.11695104794199343,
      "grad_norm": 2.851832379673274,
      "learning_rate": 3.921232811733027e-06,
      "loss": 1.006,
      "step": 1621
    },
    {
      "epoch": 0.11702319541142095,
      "grad_norm": 0.9657457590980248,
      "learning_rate": 3.921102890321001e-06,
      "loss": 0.9388,
      "step": 1622
    },
    {
      "epoch": 0.11709534288084845,
      "grad_norm": 8.810577399290397,
      "learning_rate": 3.920972864004657e-06,
      "loss": 1.0558,
      "step": 1623
    },
    {
      "epoch": 0.11716749035027596,
      "grad_norm": 6.130408876607234,
      "learning_rate": 3.920842732791097e-06,
      "loss": 0.947,
      "step": 1624
    },
    {
      "epoch": 0.11723963781970348,
      "grad_norm": 3.7672179085996635,
      "learning_rate": 3.920712496687425e-06,
      "loss": 0.9971,
      "step": 1625
    },
    {
      "epoch": 0.11731178528913098,
      "grad_norm": 6.4689112841268965,
      "learning_rate": 3.9205821557007545e-06,
      "loss": 0.925,
      "step": 1626
    },
    {
      "epoch": 0.11738393275855849,
      "grad_norm": 11.254124542311907,
      "learning_rate": 3.920451709838202e-06,
      "loss": 0.9485,
      "step": 1627
    },
    {
      "epoch": 0.11745608022798601,
      "grad_norm": 5.460427714771717,
      "learning_rate": 3.920321159106891e-06,
      "loss": 0.9485,
      "step": 1628
    },
    {
      "epoch": 0.11752822769741351,
      "grad_norm": 3.231202564592281,
      "learning_rate": 3.920190503513951e-06,
      "loss": 0.9937,
      "step": 1629
    },
    {
      "epoch": 0.11760037516684102,
      "grad_norm": 0.8329272751150881,
      "learning_rate": 3.920059743066515e-06,
      "loss": 0.8539,
      "step": 1630
    },
    {
      "epoch": 0.11767252263626854,
      "grad_norm": 19.546754783303385,
      "learning_rate": 3.919928877771725e-06,
      "loss": 1.1127,
      "step": 1631
    },
    {
      "epoch": 0.11774467010569604,
      "grad_norm": 6.565620770241974,
      "learning_rate": 3.919797907636726e-06,
      "loss": 1.0729,
      "step": 1632
    },
    {
      "epoch": 0.11781681757512355,
      "grad_norm": 13.90319927913401,
      "learning_rate": 3.919666832668669e-06,
      "loss": 1.1257,
      "step": 1633
    },
    {
      "epoch": 0.11788896504455106,
      "grad_norm": 13.541501150831422,
      "learning_rate": 3.919535652874714e-06,
      "loss": 1.0491,
      "step": 1634
    },
    {
      "epoch": 0.11796111251397857,
      "grad_norm": 9.35468298174582,
      "learning_rate": 3.919404368262023e-06,
      "loss": 1.0175,
      "step": 1635
    },
    {
      "epoch": 0.11803325998340608,
      "grad_norm": 5.070699197984959,
      "learning_rate": 3.919272978837764e-06,
      "loss": 1.0814,
      "step": 1636
    },
    {
      "epoch": 0.11810540745283359,
      "grad_norm": 5.938342225422047,
      "learning_rate": 3.919141484609113e-06,
      "loss": 1.0861,
      "step": 1637
    },
    {
      "epoch": 0.1181775549222611,
      "grad_norm": 4.931652614465624,
      "learning_rate": 3.919009885583249e-06,
      "loss": 0.9942,
      "step": 1638
    },
    {
      "epoch": 0.11824970239168861,
      "grad_norm": 9.048297987586835,
      "learning_rate": 3.91887818176736e-06,
      "loss": 1.043,
      "step": 1639
    },
    {
      "epoch": 0.11832184986111612,
      "grad_norm": 7.263245945767028,
      "learning_rate": 3.9187463731686365e-06,
      "loss": 1.0259,
      "step": 1640
    },
    {
      "epoch": 0.11839399733054363,
      "grad_norm": 10.025408698904904,
      "learning_rate": 3.918614459794277e-06,
      "loss": 0.922,
      "step": 1641
    },
    {
      "epoch": 0.11846614479997114,
      "grad_norm": 0.9398822716280595,
      "learning_rate": 3.918482441651484e-06,
      "loss": 0.8903,
      "step": 1642
    },
    {
      "epoch": 0.11853829226939865,
      "grad_norm": 7.995574442330069,
      "learning_rate": 3.918350318747468e-06,
      "loss": 1.0365,
      "step": 1643
    },
    {
      "epoch": 0.11861043973882616,
      "grad_norm": 11.599279683867024,
      "learning_rate": 3.9182180910894404e-06,
      "loss": 0.9449,
      "step": 1644
    },
    {
      "epoch": 0.11868258720825367,
      "grad_norm": 6.986387641275706,
      "learning_rate": 3.918085758684625e-06,
      "loss": 1.107,
      "step": 1645
    },
    {
      "epoch": 0.11875473467768118,
      "grad_norm": 13.125698107364197,
      "learning_rate": 3.917953321540248e-06,
      "loss": 1.0,
      "step": 1646
    },
    {
      "epoch": 0.1188268821471087,
      "grad_norm": 13.73325896258417,
      "learning_rate": 3.917820779663538e-06,
      "loss": 1.1222,
      "step": 1647
    },
    {
      "epoch": 0.1188990296165362,
      "grad_norm": 5.778042950210102,
      "learning_rate": 3.917688133061736e-06,
      "loss": 1.0095,
      "step": 1648
    },
    {
      "epoch": 0.1189711770859637,
      "grad_norm": 8.824758674412637,
      "learning_rate": 3.917555381742083e-06,
      "loss": 1.0432,
      "step": 1649
    },
    {
      "epoch": 0.11904332455539123,
      "grad_norm": 6.972495949662962,
      "learning_rate": 3.917422525711831e-06,
      "loss": 0.8869,
      "step": 1650
    },
    {
      "epoch": 0.11911547202481873,
      "grad_norm": 5.303224611892248,
      "learning_rate": 3.917289564978231e-06,
      "loss": 1.0634,
      "step": 1651
    },
    {
      "epoch": 0.11918761949424624,
      "grad_norm": 3.7956583719028627,
      "learning_rate": 3.917156499548546e-06,
      "loss": 1.0323,
      "step": 1652
    },
    {
      "epoch": 0.11925976696367376,
      "grad_norm": 11.365462444276767,
      "learning_rate": 3.917023329430042e-06,
      "loss": 0.9182,
      "step": 1653
    },
    {
      "epoch": 0.11933191443310126,
      "grad_norm": 10.979614094273973,
      "learning_rate": 3.916890054629991e-06,
      "loss": 0.9385,
      "step": 1654
    },
    {
      "epoch": 0.11940406190252877,
      "grad_norm": 9.835270017438157,
      "learning_rate": 3.916756675155669e-06,
      "loss": 1.0668,
      "step": 1655
    },
    {
      "epoch": 0.11947620937195629,
      "grad_norm": 10.320136924249304,
      "learning_rate": 3.916623191014362e-06,
      "loss": 0.8995,
      "step": 1656
    },
    {
      "epoch": 0.11954835684138379,
      "grad_norm": 7.266453193305386,
      "learning_rate": 3.9164896022133565e-06,
      "loss": 1.0356,
      "step": 1657
    },
    {
      "epoch": 0.1196205043108113,
      "grad_norm": 0.7983368199842392,
      "learning_rate": 3.9163559087599495e-06,
      "loss": 0.829,
      "step": 1658
    },
    {
      "epoch": 0.1196926517802388,
      "grad_norm": 8.30034373846998,
      "learning_rate": 3.916222110661439e-06,
      "loss": 1.1462,
      "step": 1659
    },
    {
      "epoch": 0.11976479924966632,
      "grad_norm": 5.10077804913554,
      "learning_rate": 3.916088207925133e-06,
      "loss": 1.0022,
      "step": 1660
    },
    {
      "epoch": 0.11983694671909383,
      "grad_norm": 4.3803649704996355,
      "learning_rate": 3.915954200558344e-06,
      "loss": 1.0319,
      "step": 1661
    },
    {
      "epoch": 0.11990909418852133,
      "grad_norm": 5.8373580490595165,
      "learning_rate": 3.915820088568388e-06,
      "loss": 1.0185,
      "step": 1662
    },
    {
      "epoch": 0.11998124165794885,
      "grad_norm": 13.55018218083392,
      "learning_rate": 3.91568587196259e-06,
      "loss": 0.9998,
      "step": 1663
    },
    {
      "epoch": 0.12005338912737636,
      "grad_norm": 5.631594204729525,
      "learning_rate": 3.915551550748277e-06,
      "loss": 0.9717,
      "step": 1664
    },
    {
      "epoch": 0.12012553659680386,
      "grad_norm": 4.893529449988553,
      "learning_rate": 3.915417124932785e-06,
      "loss": 1.0777,
      "step": 1665
    },
    {
      "epoch": 0.12019768406623138,
      "grad_norm": 6.8218613446632395,
      "learning_rate": 3.915282594523455e-06,
      "loss": 1.0612,
      "step": 1666
    },
    {
      "epoch": 0.12026983153565889,
      "grad_norm": 5.446115968117424,
      "learning_rate": 3.915147959527632e-06,
      "loss": 0.9519,
      "step": 1667
    },
    {
      "epoch": 0.12034197900508639,
      "grad_norm": 17.349212611028094,
      "learning_rate": 3.915013219952669e-06,
      "loss": 0.9802,
      "step": 1668
    },
    {
      "epoch": 0.12041412647451391,
      "grad_norm": 8.2600840581353,
      "learning_rate": 3.9148783758059235e-06,
      "loss": 0.9643,
      "step": 1669
    },
    {
      "epoch": 0.12048627394394142,
      "grad_norm": 0.8439915276029842,
      "learning_rate": 3.9147434270947585e-06,
      "loss": 0.8966,
      "step": 1670
    },
    {
      "epoch": 0.12055842141336892,
      "grad_norm": 5.808261761801332,
      "learning_rate": 3.914608373826544e-06,
      "loss": 1.0196,
      "step": 1671
    },
    {
      "epoch": 0.12063056888279644,
      "grad_norm": 13.103722755340488,
      "learning_rate": 3.914473216008651e-06,
      "loss": 0.9817,
      "step": 1672
    },
    {
      "epoch": 0.12070271635222395,
      "grad_norm": 6.493881280065862,
      "learning_rate": 3.9143379536484654e-06,
      "loss": 0.9935,
      "step": 1673
    },
    {
      "epoch": 0.12077486382165145,
      "grad_norm": 0.8001380384086378,
      "learning_rate": 3.914202586753369e-06,
      "loss": 0.8413,
      "step": 1674
    },
    {
      "epoch": 0.12084701129107897,
      "grad_norm": 8.224675932184388,
      "learning_rate": 3.914067115330757e-06,
      "loss": 0.9528,
      "step": 1675
    },
    {
      "epoch": 0.12091915876050648,
      "grad_norm": 15.832637945598524,
      "learning_rate": 3.913931539388024e-06,
      "loss": 0.9182,
      "step": 1676
    },
    {
      "epoch": 0.12099130622993398,
      "grad_norm": 8.498364459490316,
      "learning_rate": 3.913795858932576e-06,
      "loss": 1.0757,
      "step": 1677
    },
    {
      "epoch": 0.1210634536993615,
      "grad_norm": 0.8784234988773777,
      "learning_rate": 3.913660073971821e-06,
      "loss": 0.8659,
      "step": 1678
    },
    {
      "epoch": 0.121135601168789,
      "grad_norm": 9.828635392028417,
      "learning_rate": 3.913524184513173e-06,
      "loss": 1.1206,
      "step": 1679
    },
    {
      "epoch": 0.12120774863821651,
      "grad_norm": 0.7685345231431958,
      "learning_rate": 3.913388190564052e-06,
      "loss": 0.8691,
      "step": 1680
    },
    {
      "epoch": 0.12127989610764403,
      "grad_norm": 6.115021654716072,
      "learning_rate": 3.913252092131886e-06,
      "loss": 1.0622,
      "step": 1681
    },
    {
      "epoch": 0.12135204357707154,
      "grad_norm": 5.44619038974516,
      "learning_rate": 3.913115889224105e-06,
      "loss": 1.0975,
      "step": 1682
    },
    {
      "epoch": 0.12142419104649904,
      "grad_norm": 13.487819686933339,
      "learning_rate": 3.9129795818481485e-06,
      "loss": 1.1612,
      "step": 1683
    },
    {
      "epoch": 0.12149633851592656,
      "grad_norm": 8.868032379903426,
      "learning_rate": 3.912843170011458e-06,
      "loss": 1.0332,
      "step": 1684
    },
    {
      "epoch": 0.12156848598535407,
      "grad_norm": 6.417230193258731,
      "learning_rate": 3.912706653721483e-06,
      "loss": 0.8535,
      "step": 1685
    },
    {
      "epoch": 0.12164063345478157,
      "grad_norm": 5.520230799010203,
      "learning_rate": 3.912570032985679e-06,
      "loss": 1.0684,
      "step": 1686
    },
    {
      "epoch": 0.12171278092420908,
      "grad_norm": 11.962249504504067,
      "learning_rate": 3.912433307811505e-06,
      "loss": 0.9101,
      "step": 1687
    },
    {
      "epoch": 0.1217849283936366,
      "grad_norm": 4.5526357122111785,
      "learning_rate": 3.912296478206428e-06,
      "loss": 0.9816,
      "step": 1688
    },
    {
      "epoch": 0.1218570758630641,
      "grad_norm": 6.174014722543951,
      "learning_rate": 3.912159544177919e-06,
      "loss": 0.9451,
      "step": 1689
    },
    {
      "epoch": 0.1219292233324916,
      "grad_norm": 16.97913561235018,
      "learning_rate": 3.912022505733456e-06,
      "loss": 0.9124,
      "step": 1690
    },
    {
      "epoch": 0.12200137080191913,
      "grad_norm": 4.826102712854289,
      "learning_rate": 3.911885362880522e-06,
      "loss": 0.9808,
      "step": 1691
    },
    {
      "epoch": 0.12207351827134663,
      "grad_norm": 5.0302672759195906,
      "learning_rate": 3.9117481156266065e-06,
      "loss": 0.8773,
      "step": 1692
    },
    {
      "epoch": 0.12214566574077414,
      "grad_norm": 18.289619812098167,
      "learning_rate": 3.911610763979202e-06,
      "loss": 1.0009,
      "step": 1693
    },
    {
      "epoch": 0.12221781321020166,
      "grad_norm": 6.241914325413381,
      "learning_rate": 3.911473307945812e-06,
      "loss": 1.0748,
      "step": 1694
    },
    {
      "epoch": 0.12228996067962916,
      "grad_norm": 6.713960581736776,
      "learning_rate": 3.911335747533939e-06,
      "loss": 1.0549,
      "step": 1695
    },
    {
      "epoch": 0.12236210814905667,
      "grad_norm": 6.4057563405450875,
      "learning_rate": 3.9111980827510975e-06,
      "loss": 1.0017,
      "step": 1696
    },
    {
      "epoch": 0.12243425561848419,
      "grad_norm": 19.83517438961418,
      "learning_rate": 3.911060313604804e-06,
      "loss": 1.0043,
      "step": 1697
    },
    {
      "epoch": 0.12250640308791169,
      "grad_norm": 8.100753052167867,
      "learning_rate": 3.91092244010258e-06,
      "loss": 1.0231,
      "step": 1698
    },
    {
      "epoch": 0.1225785505573392,
      "grad_norm": 13.838373453335633,
      "learning_rate": 3.910784462251956e-06,
      "loss": 1.1348,
      "step": 1699
    },
    {
      "epoch": 0.12265069802676672,
      "grad_norm": 5.479180378286323,
      "learning_rate": 3.910646380060467e-06,
      "loss": 1.0352,
      "step": 1700
    },
    {
      "epoch": 0.12272284549619422,
      "grad_norm": 19.79123526119738,
      "learning_rate": 3.9105081935356506e-06,
      "loss": 1.1602,
      "step": 1701
    },
    {
      "epoch": 0.12279499296562173,
      "grad_norm": 10.605505078782162,
      "learning_rate": 3.910369902685055e-06,
      "loss": 1.0992,
      "step": 1702
    },
    {
      "epoch": 0.12286714043504925,
      "grad_norm": 19.911611939746678,
      "learning_rate": 3.910231507516231e-06,
      "loss": 0.9896,
      "step": 1703
    },
    {
      "epoch": 0.12293928790447675,
      "grad_norm": 0.8185671449267182,
      "learning_rate": 3.910093008036735e-06,
      "loss": 0.896,
      "step": 1704
    },
    {
      "epoch": 0.12301143537390426,
      "grad_norm": 6.073990462450484,
      "learning_rate": 3.909954404254132e-06,
      "loss": 1.0086,
      "step": 1705
    },
    {
      "epoch": 0.12308358284333178,
      "grad_norm": 7.834123639219647,
      "learning_rate": 3.9098156961759885e-06,
      "loss": 1.0203,
      "step": 1706
    },
    {
      "epoch": 0.12315573031275928,
      "grad_norm": 7.620649253439323,
      "learning_rate": 3.9096768838098805e-06,
      "loss": 1.0225,
      "step": 1707
    },
    {
      "epoch": 0.12322787778218679,
      "grad_norm": 3.8968868497103943,
      "learning_rate": 3.9095379671633865e-06,
      "loss": 1.0735,
      "step": 1708
    },
    {
      "epoch": 0.1233000252516143,
      "grad_norm": 7.442224108189277,
      "learning_rate": 3.909398946244094e-06,
      "loss": 1.0264,
      "step": 1709
    },
    {
      "epoch": 0.12337217272104181,
      "grad_norm": 5.304179238871108,
      "learning_rate": 3.909259821059592e-06,
      "loss": 0.997,
      "step": 1710
    },
    {
      "epoch": 0.12344432019046932,
      "grad_norm": 5.50135994483871,
      "learning_rate": 3.90912059161748e-06,
      "loss": 0.9625,
      "step": 1711
    },
    {
      "epoch": 0.12351646765989682,
      "grad_norm": 8.751659562904653,
      "learning_rate": 3.90898125792536e-06,
      "loss": 1.0152,
      "step": 1712
    },
    {
      "epoch": 0.12358861512932434,
      "grad_norm": 13.22449327157903,
      "learning_rate": 3.90884181999084e-06,
      "loss": 1.1198,
      "step": 1713
    },
    {
      "epoch": 0.12366076259875185,
      "grad_norm": 11.024094635688733,
      "learning_rate": 3.908702277821534e-06,
      "loss": 1.0482,
      "step": 1714
    },
    {
      "epoch": 0.12373291006817935,
      "grad_norm": 6.559760693627933,
      "learning_rate": 3.908562631425063e-06,
      "loss": 0.9326,
      "step": 1715
    },
    {
      "epoch": 0.12380505753760687,
      "grad_norm": 7.944958644252764,
      "learning_rate": 3.908422880809051e-06,
      "loss": 0.9991,
      "step": 1716
    },
    {
      "epoch": 0.12387720500703438,
      "grad_norm": 23.672697489836526,
      "learning_rate": 3.908283025981131e-06,
      "loss": 1.0254,
      "step": 1717
    },
    {
      "epoch": 0.12394935247646188,
      "grad_norm": 0.6903749738324056,
      "learning_rate": 3.9081430669489395e-06,
      "loss": 0.8179,
      "step": 1718
    },
    {
      "epoch": 0.1240214999458894,
      "grad_norm": 5.610942313463892,
      "learning_rate": 3.908003003720118e-06,
      "loss": 1.1258,
      "step": 1719
    },
    {
      "epoch": 0.12409364741531691,
      "grad_norm": 9.693126594965896,
      "learning_rate": 3.907862836302316e-06,
      "loss": 0.9849,
      "step": 1720
    },
    {
      "epoch": 0.12416579488474441,
      "grad_norm": 4.237849650403354,
      "learning_rate": 3.907722564703188e-06,
      "loss": 0.9583,
      "step": 1721
    },
    {
      "epoch": 0.12423794235417193,
      "grad_norm": 5.0351590910460375,
      "learning_rate": 3.907582188930392e-06,
      "loss": 1.1517,
      "step": 1722
    },
    {
      "epoch": 0.12431008982359944,
      "grad_norm": 6.600825703151001,
      "learning_rate": 3.907441708991594e-06,
      "loss": 1.089,
      "step": 1723
    },
    {
      "epoch": 0.12438223729302694,
      "grad_norm": 5.5730061271355895,
      "learning_rate": 3.907301124894465e-06,
      "loss": 1.0891,
      "step": 1724
    },
    {
      "epoch": 0.12445438476245446,
      "grad_norm": 6.08631670533153,
      "learning_rate": 3.907160436646684e-06,
      "loss": 1.0344,
      "step": 1725
    },
    {
      "epoch": 0.12452653223188197,
      "grad_norm": 7.758975584310078,
      "learning_rate": 3.90701964425593e-06,
      "loss": 0.9687,
      "step": 1726
    },
    {
      "epoch": 0.12459867970130947,
      "grad_norm": 4.786374007994902,
      "learning_rate": 3.906878747729893e-06,
      "loss": 0.9903,
      "step": 1727
    },
    {
      "epoch": 0.12467082717073699,
      "grad_norm": 3.3779401335123245,
      "learning_rate": 3.906737747076268e-06,
      "loss": 1.0896,
      "step": 1728
    },
    {
      "epoch": 0.1247429746401645,
      "grad_norm": 3.5539254849446498,
      "learning_rate": 3.906596642302752e-06,
      "loss": 0.944,
      "step": 1729
    },
    {
      "epoch": 0.124815122109592,
      "grad_norm": 7.0139207023008225,
      "learning_rate": 3.9064554334170504e-06,
      "loss": 1.02,
      "step": 1730
    },
    {
      "epoch": 0.12488726957901952,
      "grad_norm": 3.521283152962136,
      "learning_rate": 3.906314120426876e-06,
      "loss": 0.9664,
      "step": 1731
    },
    {
      "epoch": 0.12495941704844703,
      "grad_norm": 5.354866466976594,
      "learning_rate": 3.9061727033399446e-06,
      "loss": 0.9142,
      "step": 1732
    },
    {
      "epoch": 0.12503156451787453,
      "grad_norm": 5.056815641283716,
      "learning_rate": 3.906031182163978e-06,
      "loss": 0.9921,
      "step": 1733
    },
    {
      "epoch": 0.12510371198730205,
      "grad_norm": 4.49370091694009,
      "learning_rate": 3.9058895569067055e-06,
      "loss": 1.0554,
      "step": 1734
    },
    {
      "epoch": 0.12517585945672954,
      "grad_norm": 4.247448548021603,
      "learning_rate": 3.905747827575858e-06,
      "loss": 1.1214,
      "step": 1735
    },
    {
      "epoch": 0.12524800692615706,
      "grad_norm": 5.496367902499696,
      "learning_rate": 3.905605994179178e-06,
      "loss": 0.9717,
      "step": 1736
    },
    {
      "epoch": 0.12532015439558458,
      "grad_norm": 4.502972151143035,
      "learning_rate": 3.905464056724408e-06,
      "loss": 0.9362,
      "step": 1737
    },
    {
      "epoch": 0.12539230186501207,
      "grad_norm": 5.077722527620516,
      "learning_rate": 3.9053220152193e-06,
      "loss": 1.0941,
      "step": 1738
    },
    {
      "epoch": 0.1254644493344396,
      "grad_norm": 5.985925697627689,
      "learning_rate": 3.90517986967161e-06,
      "loss": 1.0408,
      "step": 1739
    },
    {
      "epoch": 0.1255365968038671,
      "grad_norm": 7.853823572518799,
      "learning_rate": 3.9050376200891e-06,
      "loss": 1.0356,
      "step": 1740
    },
    {
      "epoch": 0.1256087442732946,
      "grad_norm": 5.43018383803832,
      "learning_rate": 3.904895266479537e-06,
      "loss": 0.9086,
      "step": 1741
    },
    {
      "epoch": 0.12568089174272212,
      "grad_norm": 4.731903942071541,
      "learning_rate": 3.904752808850696e-06,
      "loss": 1.0966,
      "step": 1742
    },
    {
      "epoch": 0.12575303921214964,
      "grad_norm": 8.82271465504112,
      "learning_rate": 3.904610247210355e-06,
      "loss": 1.0111,
      "step": 1743
    },
    {
      "epoch": 0.12582518668157713,
      "grad_norm": 5.049855772168365,
      "learning_rate": 3.9044675815663e-06,
      "loss": 0.9482,
      "step": 1744
    },
    {
      "epoch": 0.12589733415100465,
      "grad_norm": 6.249629505620276,
      "learning_rate": 3.904324811926319e-06,
      "loss": 1.0548,
      "step": 1745
    },
    {
      "epoch": 0.12596948162043217,
      "grad_norm": 0.7456337674209828,
      "learning_rate": 3.90418193829821e-06,
      "loss": 0.89,
      "step": 1746
    },
    {
      "epoch": 0.12604162908985966,
      "grad_norm": 7.232710908517058,
      "learning_rate": 3.904038960689775e-06,
      "loss": 1.0438,
      "step": 1747
    },
    {
      "epoch": 0.12611377655928718,
      "grad_norm": 5.190180258441156,
      "learning_rate": 3.90389587910882e-06,
      "loss": 1.0612,
      "step": 1748
    },
    {
      "epoch": 0.1261859240287147,
      "grad_norm": 4.238388130492788,
      "learning_rate": 3.903752693563159e-06,
      "loss": 0.9636,
      "step": 1749
    },
    {
      "epoch": 0.1262580714981422,
      "grad_norm": 0.7726612241332244,
      "learning_rate": 3.903609404060612e-06,
      "loss": 0.8559,
      "step": 1750
    },
    {
      "epoch": 0.1263302189675697,
      "grad_norm": 0.7544828196592552,
      "learning_rate": 3.903466010609002e-06,
      "loss": 0.7939,
      "step": 1751
    },
    {
      "epoch": 0.12640236643699723,
      "grad_norm": 7.134867744957654,
      "learning_rate": 3.903322513216159e-06,
      "loss": 1.0094,
      "step": 1752
    },
    {
      "epoch": 0.12647451390642472,
      "grad_norm": 5.2694206706916,
      "learning_rate": 3.9031789118899195e-06,
      "loss": 0.9901,
      "step": 1753
    },
    {
      "epoch": 0.12654666137585224,
      "grad_norm": 4.463161361796217,
      "learning_rate": 3.903035206638125e-06,
      "loss": 0.9109,
      "step": 1754
    },
    {
      "epoch": 0.12661880884527976,
      "grad_norm": 3.5624451214762978,
      "learning_rate": 3.902891397468623e-06,
      "loss": 0.8974,
      "step": 1755
    },
    {
      "epoch": 0.12669095631470725,
      "grad_norm": 5.146148218978853,
      "learning_rate": 3.902747484389266e-06,
      "loss": 1.0619,
      "step": 1756
    },
    {
      "epoch": 0.12676310378413477,
      "grad_norm": 9.498411346916619,
      "learning_rate": 3.902603467407912e-06,
      "loss": 0.8408,
      "step": 1757
    },
    {
      "epoch": 0.1268352512535623,
      "grad_norm": 8.372310832982853,
      "learning_rate": 3.902459346532426e-06,
      "loss": 1.0866,
      "step": 1758
    },
    {
      "epoch": 0.12690739872298978,
      "grad_norm": 4.857346089703332,
      "learning_rate": 3.902315121770679e-06,
      "loss": 1.1299,
      "step": 1759
    },
    {
      "epoch": 0.1269795461924173,
      "grad_norm": 3.109499214796699,
      "learning_rate": 3.902170793130544e-06,
      "loss": 1.0114,
      "step": 1760
    },
    {
      "epoch": 0.12705169366184482,
      "grad_norm": 17.083234882846398,
      "learning_rate": 3.902026360619904e-06,
      "loss": 1.0602,
      "step": 1761
    },
    {
      "epoch": 0.1271238411312723,
      "grad_norm": 5.160701795064129,
      "learning_rate": 3.901881824246645e-06,
      "loss": 1.0302,
      "step": 1762
    },
    {
      "epoch": 0.12719598860069983,
      "grad_norm": 7.9044824365882445,
      "learning_rate": 3.901737184018661e-06,
      "loss": 1.0739,
      "step": 1763
    },
    {
      "epoch": 0.12726813607012735,
      "grad_norm": 5.651936609332512,
      "learning_rate": 3.901592439943849e-06,
      "loss": 1.1285,
      "step": 1764
    },
    {
      "epoch": 0.12734028353955484,
      "grad_norm": 4.222658056775747,
      "learning_rate": 3.901447592030113e-06,
      "loss": 0.9623,
      "step": 1765
    },
    {
      "epoch": 0.12741243100898236,
      "grad_norm": 9.785121943552635,
      "learning_rate": 3.901302640285363e-06,
      "loss": 1.152,
      "step": 1766
    },
    {
      "epoch": 0.12748457847840988,
      "grad_norm": 19.546678281384217,
      "learning_rate": 3.901157584717514e-06,
      "loss": 0.9654,
      "step": 1767
    },
    {
      "epoch": 0.12755672594783737,
      "grad_norm": 4.286433291294497,
      "learning_rate": 3.901012425334488e-06,
      "loss": 0.889,
      "step": 1768
    },
    {
      "epoch": 0.1276288734172649,
      "grad_norm": 4.144678053920991,
      "learning_rate": 3.90086716214421e-06,
      "loss": 1.0768,
      "step": 1769
    },
    {
      "epoch": 0.1277010208866924,
      "grad_norm": 0.7432030405657835,
      "learning_rate": 3.900721795154613e-06,
      "loss": 0.8327,
      "step": 1770
    },
    {
      "epoch": 0.1277731683561199,
      "grad_norm": 9.49611945954973,
      "learning_rate": 3.900576324373635e-06,
      "loss": 1.0123,
      "step": 1771
    },
    {
      "epoch": 0.12784531582554742,
      "grad_norm": 3.947703386784435,
      "learning_rate": 3.900430749809221e-06,
      "loss": 0.8838,
      "step": 1772
    },
    {
      "epoch": 0.12791746329497494,
      "grad_norm": 6.5319718687499435,
      "learning_rate": 3.9002850714693176e-06,
      "loss": 0.9206,
      "step": 1773
    },
    {
      "epoch": 0.12798961076440243,
      "grad_norm": 4.969796124580712,
      "learning_rate": 3.900139289361881e-06,
      "loss": 0.9457,
      "step": 1774
    },
    {
      "epoch": 0.12806175823382995,
      "grad_norm": 10.421448568744946,
      "learning_rate": 3.899993403494872e-06,
      "loss": 1.0452,
      "step": 1775
    },
    {
      "epoch": 0.12813390570325744,
      "grad_norm": 5.433988116698176,
      "learning_rate": 3.899847413876257e-06,
      "loss": 0.989,
      "step": 1776
    },
    {
      "epoch": 0.12820605317268496,
      "grad_norm": 6.230668651279814,
      "learning_rate": 3.899701320514007e-06,
      "loss": 1.0738,
      "step": 1777
    },
    {
      "epoch": 0.12827820064211248,
      "grad_norm": 5.951520568427536,
      "learning_rate": 3.899555123416101e-06,
      "loss": 1.04,
      "step": 1778
    },
    {
      "epoch": 0.12835034811153997,
      "grad_norm": 5.180147590743882,
      "learning_rate": 3.899408822590522e-06,
      "loss": 1.2282,
      "step": 1779
    },
    {
      "epoch": 0.1284224955809675,
      "grad_norm": 6.488897672121789,
      "learning_rate": 3.899262418045259e-06,
      "loss": 1.0586,
      "step": 1780
    },
    {
      "epoch": 0.128494643050395,
      "grad_norm": 6.0766869734053195,
      "learning_rate": 3.899115909788305e-06,
      "loss": 1.0261,
      "step": 1781
    },
    {
      "epoch": 0.1285667905198225,
      "grad_norm": 8.379049204782836,
      "learning_rate": 3.8989692978276615e-06,
      "loss": 0.9923,
      "step": 1782
    },
    {
      "epoch": 0.12863893798925002,
      "grad_norm": 6.599312844773564,
      "learning_rate": 3.898822582171335e-06,
      "loss": 1.0374,
      "step": 1783
    },
    {
      "epoch": 0.12871108545867754,
      "grad_norm": 5.701866031845819,
      "learning_rate": 3.8986757628273365e-06,
      "loss": 0.9825,
      "step": 1784
    },
    {
      "epoch": 0.12878323292810503,
      "grad_norm": 6.0697525532042,
      "learning_rate": 3.898528839803682e-06,
      "loss": 1.097,
      "step": 1785
    },
    {
      "epoch": 0.12885538039753255,
      "grad_norm": 12.137479985983756,
      "learning_rate": 3.898381813108397e-06,
      "loss": 0.9469,
      "step": 1786
    },
    {
      "epoch": 0.12892752786696007,
      "grad_norm": 3.8462505115685977,
      "learning_rate": 3.8982346827495086e-06,
      "loss": 1.0288,
      "step": 1787
    },
    {
      "epoch": 0.12899967533638756,
      "grad_norm": 5.3939999837398,
      "learning_rate": 3.898087448735051e-06,
      "loss": 1.0401,
      "step": 1788
    },
    {
      "epoch": 0.12907182280581508,
      "grad_norm": 4.870466227058853,
      "learning_rate": 3.897940111073064e-06,
      "loss": 1.0862,
      "step": 1789
    },
    {
      "epoch": 0.1291439702752426,
      "grad_norm": 6.395323725806327,
      "learning_rate": 3.897792669771593e-06,
      "loss": 0.9995,
      "step": 1790
    },
    {
      "epoch": 0.1292161177446701,
      "grad_norm": 3.3743559611377427,
      "learning_rate": 3.897645124838691e-06,
      "loss": 0.9127,
      "step": 1791
    },
    {
      "epoch": 0.12928826521409761,
      "grad_norm": 0.7634329653842212,
      "learning_rate": 3.897497476282413e-06,
      "loss": 0.7923,
      "step": 1792
    },
    {
      "epoch": 0.12936041268352513,
      "grad_norm": 6.013689953590071,
      "learning_rate": 3.897349724110822e-06,
      "loss": 0.9974,
      "step": 1793
    },
    {
      "epoch": 0.12943256015295262,
      "grad_norm": 6.1441441515264215,
      "learning_rate": 3.897201868331986e-06,
      "loss": 0.9616,
      "step": 1794
    },
    {
      "epoch": 0.12950470762238014,
      "grad_norm": 5.112319517797257,
      "learning_rate": 3.89705390895398e-06,
      "loss": 1.1398,
      "step": 1795
    },
    {
      "epoch": 0.12957685509180766,
      "grad_norm": 5.150820746343433,
      "learning_rate": 3.896905845984882e-06,
      "loss": 0.9433,
      "step": 1796
    },
    {
      "epoch": 0.12964900256123515,
      "grad_norm": 11.502575793183443,
      "learning_rate": 3.896757679432779e-06,
      "loss": 1.0237,
      "step": 1797
    },
    {
      "epoch": 0.12972115003066267,
      "grad_norm": 7.1047554853486545,
      "learning_rate": 3.896609409305759e-06,
      "loss": 1.0464,
      "step": 1798
    },
    {
      "epoch": 0.1297932975000902,
      "grad_norm": 4.650227998455782,
      "learning_rate": 3.896461035611921e-06,
      "loss": 0.9778,
      "step": 1799
    },
    {
      "epoch": 0.12986544496951768,
      "grad_norm": 5.846508986385025,
      "learning_rate": 3.896312558359367e-06,
      "loss": 0.9136,
      "step": 1800
    },
    {
      "epoch": 0.1299375924389452,
      "grad_norm": 5.168388920525764,
      "learning_rate": 3.8961639775562035e-06,
      "loss": 1.0131,
      "step": 1801
    },
    {
      "epoch": 0.13000973990837272,
      "grad_norm": 11.163200371824557,
      "learning_rate": 3.8960152932105435e-06,
      "loss": 0.9662,
      "step": 1802
    },
    {
      "epoch": 0.13008188737780021,
      "grad_norm": 0.9752411715243099,
      "learning_rate": 3.895866505330509e-06,
      "loss": 0.9796,
      "step": 1803
    },
    {
      "epoch": 0.13015403484722773,
      "grad_norm": 17.928531757412888,
      "learning_rate": 3.8957176139242214e-06,
      "loss": 1.1004,
      "step": 1804
    },
    {
      "epoch": 0.13022618231665525,
      "grad_norm": 7.950668825050192,
      "learning_rate": 3.895568618999814e-06,
      "loss": 0.9653,
      "step": 1805
    },
    {
      "epoch": 0.13029832978608274,
      "grad_norm": 7.0066981285446275,
      "learning_rate": 3.89541952056542e-06,
      "loss": 1.0146,
      "step": 1806
    },
    {
      "epoch": 0.13037047725551026,
      "grad_norm": 6.114359741628622,
      "learning_rate": 3.895270318629184e-06,
      "loss": 1.0666,
      "step": 1807
    },
    {
      "epoch": 0.13044262472493778,
      "grad_norm": 8.96436971105087,
      "learning_rate": 3.895121013199251e-06,
      "loss": 0.9457,
      "step": 1808
    },
    {
      "epoch": 0.13051477219436527,
      "grad_norm": 10.846921957052277,
      "learning_rate": 3.8949716042837746e-06,
      "loss": 0.9686,
      "step": 1809
    },
    {
      "epoch": 0.1305869196637928,
      "grad_norm": 12.061935174397481,
      "learning_rate": 3.8948220918909145e-06,
      "loss": 0.999,
      "step": 1810
    },
    {
      "epoch": 0.1306590671332203,
      "grad_norm": 7.69127084933511,
      "learning_rate": 3.8946724760288345e-06,
      "loss": 1.1,
      "step": 1811
    },
    {
      "epoch": 0.1307312146026478,
      "grad_norm": 5.7498626692584605,
      "learning_rate": 3.894522756705704e-06,
      "loss": 1.0543,
      "step": 1812
    },
    {
      "epoch": 0.13080336207207532,
      "grad_norm": 7.784797074472953,
      "learning_rate": 3.8943729339296994e-06,
      "loss": 1.0385,
      "step": 1813
    },
    {
      "epoch": 0.13087550954150284,
      "grad_norm": 10.432209067256522,
      "learning_rate": 3.894223007709001e-06,
      "loss": 0.9383,
      "step": 1814
    },
    {
      "epoch": 0.13094765701093034,
      "grad_norm": 4.986402906252306,
      "learning_rate": 3.8940729780517975e-06,
      "loss": 1.0512,
      "step": 1815
    },
    {
      "epoch": 0.13101980448035785,
      "grad_norm": 5.763118209280512,
      "learning_rate": 3.893922844966279e-06,
      "loss": 1.0044,
      "step": 1816
    },
    {
      "epoch": 0.13109195194978537,
      "grad_norm": 4.849988894105761,
      "learning_rate": 3.893772608460646e-06,
      "loss": 1.0922,
      "step": 1817
    },
    {
      "epoch": 0.13116409941921287,
      "grad_norm": 10.442676471208147,
      "learning_rate": 3.8936222685431e-06,
      "loss": 1.0629,
      "step": 1818
    },
    {
      "epoch": 0.13123624688864038,
      "grad_norm": 7.248394492388308,
      "learning_rate": 3.893471825221853e-06,
      "loss": 0.9604,
      "step": 1819
    },
    {
      "epoch": 0.1313083943580679,
      "grad_norm": 5.7091911766555485,
      "learning_rate": 3.893321278505119e-06,
      "loss": 0.9371,
      "step": 1820
    },
    {
      "epoch": 0.1313805418274954,
      "grad_norm": 5.966154962402535,
      "learning_rate": 3.893170628401118e-06,
      "loss": 1.0426,
      "step": 1821
    },
    {
      "epoch": 0.13145268929692291,
      "grad_norm": 13.08215547025974,
      "learning_rate": 3.893019874918078e-06,
      "loss": 1.0356,
      "step": 1822
    },
    {
      "epoch": 0.13152483676635043,
      "grad_norm": 4.130783650310656,
      "learning_rate": 3.89286901806423e-06,
      "loss": 1.0835,
      "step": 1823
    },
    {
      "epoch": 0.13159698423577793,
      "grad_norm": 5.567384782411728,
      "learning_rate": 3.892718057847813e-06,
      "loss": 1.0536,
      "step": 1824
    },
    {
      "epoch": 0.13166913170520544,
      "grad_norm": 7.383836786171161,
      "learning_rate": 3.892566994277069e-06,
      "loss": 1.0887,
      "step": 1825
    },
    {
      "epoch": 0.13174127917463294,
      "grad_norm": 4.141651184749002,
      "learning_rate": 3.892415827360247e-06,
      "loss": 1.0109,
      "step": 1826
    },
    {
      "epoch": 0.13181342664406046,
      "grad_norm": 5.468258906521344,
      "learning_rate": 3.892264557105602e-06,
      "loss": 1.0008,
      "step": 1827
    },
    {
      "epoch": 0.13188557411348797,
      "grad_norm": 6.047092325410702,
      "learning_rate": 3.892113183521395e-06,
      "loss": 1.0378,
      "step": 1828
    },
    {
      "epoch": 0.13195772158291547,
      "grad_norm": 4.497276329598352,
      "learning_rate": 3.891961706615891e-06,
      "loss": 0.9903,
      "step": 1829
    },
    {
      "epoch": 0.13202986905234299,
      "grad_norm": 4.365972523930025,
      "learning_rate": 3.891810126397362e-06,
      "loss": 0.9859,
      "step": 1830
    },
    {
      "epoch": 0.1321020165217705,
      "grad_norm": 6.648397048845067,
      "learning_rate": 3.8916584428740865e-06,
      "loss": 1.0855,
      "step": 1831
    },
    {
      "epoch": 0.132174163991198,
      "grad_norm": 4.778787215494802,
      "learning_rate": 3.891506656054344e-06,
      "loss": 0.9742,
      "step": 1832
    },
    {
      "epoch": 0.13224631146062552,
      "grad_norm": 4.23696516144479,
      "learning_rate": 3.891354765946427e-06,
      "loss": 1.0825,
      "step": 1833
    },
    {
      "epoch": 0.13231845893005303,
      "grad_norm": 5.235422149222201,
      "learning_rate": 3.891202772558626e-06,
      "loss": 1.1404,
      "step": 1834
    },
    {
      "epoch": 0.13239060639948053,
      "grad_norm": 7.3844843510470906,
      "learning_rate": 3.891050675899245e-06,
      "loss": 0.9485,
      "step": 1835
    },
    {
      "epoch": 0.13246275386890805,
      "grad_norm": 10.197306587796005,
      "learning_rate": 3.890898475976585e-06,
      "loss": 1.0218,
      "step": 1836
    },
    {
      "epoch": 0.13253490133833556,
      "grad_norm": 7.549338750113472,
      "learning_rate": 3.89074617279896e-06,
      "loss": 0.9886,
      "step": 1837
    },
    {
      "epoch": 0.13260704880776306,
      "grad_norm": 16.093656135952006,
      "learning_rate": 3.890593766374685e-06,
      "loss": 1.2165,
      "step": 1838
    },
    {
      "epoch": 0.13267919627719058,
      "grad_norm": 4.406546075491517,
      "learning_rate": 3.890441256712084e-06,
      "loss": 1.08,
      "step": 1839
    },
    {
      "epoch": 0.1327513437466181,
      "grad_norm": 5.983062361139583,
      "learning_rate": 3.8902886438194834e-06,
      "loss": 0.9472,
      "step": 1840
    },
    {
      "epoch": 0.13282349121604559,
      "grad_norm": 5.974808259704606,
      "learning_rate": 3.890135927705218e-06,
      "loss": 0.964,
      "step": 1841
    },
    {
      "epoch": 0.1328956386854731,
      "grad_norm": 4.6438718797751735,
      "learning_rate": 3.889983108377627e-06,
      "loss": 1.0538,
      "step": 1842
    },
    {
      "epoch": 0.13296778615490062,
      "grad_norm": 5.617814115290747,
      "learning_rate": 3.889830185845054e-06,
      "loss": 1.0576,
      "step": 1843
    },
    {
      "epoch": 0.13303993362432812,
      "grad_norm": 4.768299990478193,
      "learning_rate": 3.889677160115852e-06,
      "loss": 1.0555,
      "step": 1844
    },
    {
      "epoch": 0.13311208109375564,
      "grad_norm": 5.147048971343028,
      "learning_rate": 3.8895240311983745e-06,
      "loss": 1.0342,
      "step": 1845
    },
    {
      "epoch": 0.13318422856318315,
      "grad_norm": 5.768888547479678,
      "learning_rate": 3.889370799100984e-06,
      "loss": 1.0466,
      "step": 1846
    },
    {
      "epoch": 0.13325637603261065,
      "grad_norm": 7.754353438773472,
      "learning_rate": 3.889217463832049e-06,
      "loss": 0.9826,
      "step": 1847
    },
    {
      "epoch": 0.13332852350203817,
      "grad_norm": 0.8117127272352168,
      "learning_rate": 3.889064025399942e-06,
      "loss": 0.8516,
      "step": 1848
    },
    {
      "epoch": 0.13340067097146568,
      "grad_norm": 4.500754399112755,
      "learning_rate": 3.888910483813042e-06,
      "loss": 1.0053,
      "step": 1849
    },
    {
      "epoch": 0.13347281844089318,
      "grad_norm": 6.188180905262851,
      "learning_rate": 3.888756839079733e-06,
      "loss": 1.0688,
      "step": 1850
    },
    {
      "epoch": 0.1335449659103207,
      "grad_norm": 5.361578396651716,
      "learning_rate": 3.888603091208404e-06,
      "loss": 0.9832,
      "step": 1851
    },
    {
      "epoch": 0.13361711337974821,
      "grad_norm": 4.13299317857358,
      "learning_rate": 3.888449240207452e-06,
      "loss": 1.1025,
      "step": 1852
    },
    {
      "epoch": 0.1336892608491757,
      "grad_norm": 5.20315890830965,
      "learning_rate": 3.888295286085278e-06,
      "loss": 0.9297,
      "step": 1853
    },
    {
      "epoch": 0.13376140831860323,
      "grad_norm": 5.131304700887447,
      "learning_rate": 3.888141228850288e-06,
      "loss": 0.9811,
      "step": 1854
    },
    {
      "epoch": 0.13383355578803074,
      "grad_norm": 9.186908728848213,
      "learning_rate": 3.887987068510896e-06,
      "loss": 0.9835,
      "step": 1855
    },
    {
      "epoch": 0.13390570325745824,
      "grad_norm": 0.7627509142039673,
      "learning_rate": 3.887832805075519e-06,
      "loss": 0.8147,
      "step": 1856
    },
    {
      "epoch": 0.13397785072688576,
      "grad_norm": 4.683011563651947,
      "learning_rate": 3.8876784385525806e-06,
      "loss": 0.9929,
      "step": 1857
    },
    {
      "epoch": 0.13404999819631327,
      "grad_norm": 3.2219496443618145,
      "learning_rate": 3.887523968950512e-06,
      "loss": 1.1004,
      "step": 1858
    },
    {
      "epoch": 0.13412214566574077,
      "grad_norm": 5.819396818700035,
      "learning_rate": 3.887369396277745e-06,
      "loss": 0.9677,
      "step": 1859
    },
    {
      "epoch": 0.13419429313516829,
      "grad_norm": 5.328412448143377,
      "learning_rate": 3.8872147205427235e-06,
      "loss": 1.0735,
      "step": 1860
    },
    {
      "epoch": 0.1342664406045958,
      "grad_norm": 5.667661504745413,
      "learning_rate": 3.887059941753891e-06,
      "loss": 0.9957,
      "step": 1861
    },
    {
      "epoch": 0.1343385880740233,
      "grad_norm": 11.78262604385193,
      "learning_rate": 3.8869050599197014e-06,
      "loss": 1.0238,
      "step": 1862
    },
    {
      "epoch": 0.13441073554345082,
      "grad_norm": 4.614978758989347,
      "learning_rate": 3.886750075048612e-06,
      "loss": 0.9697,
      "step": 1863
    },
    {
      "epoch": 0.13448288301287833,
      "grad_norm": 3.7830681331185265,
      "learning_rate": 3.8865949871490845e-06,
      "loss": 0.9898,
      "step": 1864
    },
    {
      "epoch": 0.13455503048230583,
      "grad_norm": 10.920726884956,
      "learning_rate": 3.88643979622959e-06,
      "loss": 1.0403,
      "step": 1865
    },
    {
      "epoch": 0.13462717795173335,
      "grad_norm": 5.252943031477056,
      "learning_rate": 3.8862845022986006e-06,
      "loss": 1.0137,
      "step": 1866
    },
    {
      "epoch": 0.13469932542116086,
      "grad_norm": 4.578754472910003,
      "learning_rate": 3.886129105364598e-06,
      "loss": 1.0787,
      "step": 1867
    },
    {
      "epoch": 0.13477147289058836,
      "grad_norm": 4.8364577005903415,
      "learning_rate": 3.8859736054360665e-06,
      "loss": 0.9848,
      "step": 1868
    },
    {
      "epoch": 0.13484362036001588,
      "grad_norm": 5.905925116105925,
      "learning_rate": 3.885818002521499e-06,
      "loss": 1.0239,
      "step": 1869
    },
    {
      "epoch": 0.1349157678294434,
      "grad_norm": 3.1424391332299004,
      "learning_rate": 3.88566229662939e-06,
      "loss": 0.9833,
      "step": 1870
    },
    {
      "epoch": 0.1349879152988709,
      "grad_norm": 9.162635841767786,
      "learning_rate": 3.885506487768244e-06,
      "loss": 0.9794,
      "step": 1871
    },
    {
      "epoch": 0.1350600627682984,
      "grad_norm": 3.9796795872286497,
      "learning_rate": 3.8853505759465696e-06,
      "loss": 0.9054,
      "step": 1872
    },
    {
      "epoch": 0.13513221023772592,
      "grad_norm": 1.3100053148671433,
      "learning_rate": 3.885194561172878e-06,
      "loss": 0.8536,
      "step": 1873
    },
    {
      "epoch": 0.13520435770715342,
      "grad_norm": 4.79446698612372,
      "learning_rate": 3.885038443455692e-06,
      "loss": 0.9129,
      "step": 1874
    },
    {
      "epoch": 0.13527650517658094,
      "grad_norm": 5.052497306721086,
      "learning_rate": 3.884882222803533e-06,
      "loss": 0.9985,
      "step": 1875
    },
    {
      "epoch": 0.13534865264600845,
      "grad_norm": 7.5579939099035185,
      "learning_rate": 3.884725899224934e-06,
      "loss": 0.9719,
      "step": 1876
    },
    {
      "epoch": 0.13542080011543595,
      "grad_norm": 5.20993759879379,
      "learning_rate": 3.884569472728432e-06,
      "loss": 1.0313,
      "step": 1877
    },
    {
      "epoch": 0.13549294758486347,
      "grad_norm": 5.285286348465741,
      "learning_rate": 3.884412943322566e-06,
      "loss": 1.0731,
      "step": 1878
    },
    {
      "epoch": 0.13556509505429096,
      "grad_norm": 3.9521541081701965,
      "learning_rate": 3.884256311015885e-06,
      "loss": 0.9931,
      "step": 1879
    },
    {
      "epoch": 0.13563724252371848,
      "grad_norm": 10.82981694201516,
      "learning_rate": 3.884099575816943e-06,
      "loss": 0.9869,
      "step": 1880
    },
    {
      "epoch": 0.135709389993146,
      "grad_norm": 6.182069186862304,
      "learning_rate": 3.883942737734297e-06,
      "loss": 1.1057,
      "step": 1881
    },
    {
      "epoch": 0.1357815374625735,
      "grad_norm": 4.767710145140555,
      "learning_rate": 3.883785796776513e-06,
      "loss": 0.8877,
      "step": 1882
    },
    {
      "epoch": 0.135853684932001,
      "grad_norm": 4.7874689484627595,
      "learning_rate": 3.883628752952159e-06,
      "loss": 0.9817,
      "step": 1883
    },
    {
      "epoch": 0.13592583240142853,
      "grad_norm": 35.10343950161793,
      "learning_rate": 3.883471606269813e-06,
      "loss": 1.0141,
      "step": 1884
    },
    {
      "epoch": 0.13599797987085602,
      "grad_norm": 8.30508895578347,
      "learning_rate": 3.883314356738054e-06,
      "loss": 1.0141,
      "step": 1885
    },
    {
      "epoch": 0.13607012734028354,
      "grad_norm": 2.7260856320707476,
      "learning_rate": 3.88315700436547e-06,
      "loss": 1.0411,
      "step": 1886
    },
    {
      "epoch": 0.13614227480971106,
      "grad_norm": 3.1123993700298187,
      "learning_rate": 3.882999549160654e-06,
      "loss": 1.0505,
      "step": 1887
    },
    {
      "epoch": 0.13621442227913855,
      "grad_norm": 17.10642632981371,
      "learning_rate": 3.882841991132203e-06,
      "loss": 1.0283,
      "step": 1888
    },
    {
      "epoch": 0.13628656974856607,
      "grad_norm": 5.399693918383842,
      "learning_rate": 3.88268433028872e-06,
      "loss": 1.064,
      "step": 1889
    },
    {
      "epoch": 0.13635871721799359,
      "grad_norm": 5.609754969231663,
      "learning_rate": 3.882526566638816e-06,
      "loss": 1.057,
      "step": 1890
    },
    {
      "epoch": 0.13643086468742108,
      "grad_norm": 5.841675415612969,
      "learning_rate": 3.882368700191105e-06,
      "loss": 1.0243,
      "step": 1891
    },
    {
      "epoch": 0.1365030121568486,
      "grad_norm": 5.639232178225731,
      "learning_rate": 3.882210730954208e-06,
      "loss": 1.0085,
      "step": 1892
    },
    {
      "epoch": 0.13657515962627612,
      "grad_norm": 5.75783502582093,
      "learning_rate": 3.8820526589367495e-06,
      "loss": 1.0263,
      "step": 1893
    },
    {
      "epoch": 0.1366473070957036,
      "grad_norm": 6.056983561618989,
      "learning_rate": 3.8818944841473625e-06,
      "loss": 0.9822,
      "step": 1894
    },
    {
      "epoch": 0.13671945456513113,
      "grad_norm": 8.565252376830971,
      "learning_rate": 3.881736206594686e-06,
      "loss": 1.0927,
      "step": 1895
    },
    {
      "epoch": 0.13679160203455865,
      "grad_norm": 8.897535484352534,
      "learning_rate": 3.881577826287359e-06,
      "loss": 0.9993,
      "step": 1896
    },
    {
      "epoch": 0.13686374950398614,
      "grad_norm": 21.86496561148465,
      "learning_rate": 3.881419343234034e-06,
      "loss": 0.9836,
      "step": 1897
    },
    {
      "epoch": 0.13693589697341366,
      "grad_norm": 6.713393236112107,
      "learning_rate": 3.881260757443362e-06,
      "loss": 0.9583,
      "step": 1898
    },
    {
      "epoch": 0.13700804444284118,
      "grad_norm": 4.528446777367942,
      "learning_rate": 3.881102068924006e-06,
      "loss": 0.9754,
      "step": 1899
    },
    {
      "epoch": 0.13708019191226867,
      "grad_norm": 6.836580326918294,
      "learning_rate": 3.880943277684628e-06,
      "loss": 1.0156,
      "step": 1900
    },
    {
      "epoch": 0.1371523393816962,
      "grad_norm": 6.646252498342846,
      "learning_rate": 3.880784383733901e-06,
      "loss": 0.9634,
      "step": 1901
    },
    {
      "epoch": 0.1372244868511237,
      "grad_norm": 5.20791341487556,
      "learning_rate": 3.880625387080502e-06,
      "loss": 1.0018,
      "step": 1902
    },
    {
      "epoch": 0.1372966343205512,
      "grad_norm": 0.7917628397163976,
      "learning_rate": 3.880466287733111e-06,
      "loss": 0.7945,
      "step": 1903
    },
    {
      "epoch": 0.13736878178997872,
      "grad_norm": 3.8672263403589295,
      "learning_rate": 3.8803070857004184e-06,
      "loss": 0.9266,
      "step": 1904
    },
    {
      "epoch": 0.13744092925940624,
      "grad_norm": 2.987539321449107,
      "learning_rate": 3.880147780991116e-06,
      "loss": 1.0832,
      "step": 1905
    },
    {
      "epoch": 0.13751307672883373,
      "grad_norm": 5.302050557313151,
      "learning_rate": 3.879988373613903e-06,
      "loss": 1.0786,
      "step": 1906
    },
    {
      "epoch": 0.13758522419826125,
      "grad_norm": 5.674988787803833,
      "learning_rate": 3.879828863577486e-06,
      "loss": 0.9605,
      "step": 1907
    },
    {
      "epoch": 0.13765737166768877,
      "grad_norm": 7.399416317989131,
      "learning_rate": 3.8796692508905725e-06,
      "loss": 1.1316,
      "step": 1908
    },
    {
      "epoch": 0.13772951913711626,
      "grad_norm": 16.36522306976128,
      "learning_rate": 3.8795095355618796e-06,
      "loss": 0.9628,
      "step": 1909
    },
    {
      "epoch": 0.13780166660654378,
      "grad_norm": 3.6742546000304235,
      "learning_rate": 3.879349717600129e-06,
      "loss": 1.043,
      "step": 1910
    },
    {
      "epoch": 0.1378738140759713,
      "grad_norm": 5.59485534317333,
      "learning_rate": 3.8791897970140466e-06,
      "loss": 0.9965,
      "step": 1911
    },
    {
      "epoch": 0.1379459615453988,
      "grad_norm": 4.661016085937608,
      "learning_rate": 3.879029773812366e-06,
      "loss": 1.0084,
      "step": 1912
    },
    {
      "epoch": 0.1380181090148263,
      "grad_norm": 5.436096185581304,
      "learning_rate": 3.878869648003826e-06,
      "loss": 0.9712,
      "step": 1913
    },
    {
      "epoch": 0.13809025648425383,
      "grad_norm": 6.763015631515937,
      "learning_rate": 3.87870941959717e-06,
      "loss": 1.1043,
      "step": 1914
    },
    {
      "epoch": 0.13816240395368132,
      "grad_norm": 6.56830158826424,
      "learning_rate": 3.878549088601147e-06,
      "loss": 0.9892,
      "step": 1915
    },
    {
      "epoch": 0.13823455142310884,
      "grad_norm": 14.28086642161298,
      "learning_rate": 3.878388655024513e-06,
      "loss": 0.9666,
      "step": 1916
    },
    {
      "epoch": 0.13830669889253636,
      "grad_norm": 4.345112319390056,
      "learning_rate": 3.878228118876027e-06,
      "loss": 0.8696,
      "step": 1917
    },
    {
      "epoch": 0.13837884636196385,
      "grad_norm": 1.0657579736162996,
      "learning_rate": 3.878067480164457e-06,
      "loss": 1.0299,
      "step": 1918
    },
    {
      "epoch": 0.13845099383139137,
      "grad_norm": 8.468960805148447,
      "learning_rate": 3.877906738898574e-06,
      "loss": 1.0542,
      "step": 1919
    },
    {
      "epoch": 0.13852314130081889,
      "grad_norm": 5.23509880868959,
      "learning_rate": 3.877745895087157e-06,
      "loss": 1.0118,
      "step": 1920
    },
    {
      "epoch": 0.13859528877024638,
      "grad_norm": 4.272826658454359,
      "learning_rate": 3.877584948738986e-06,
      "loss": 1.014,
      "step": 1921
    },
    {
      "epoch": 0.1386674362396739,
      "grad_norm": 5.290858810314597,
      "learning_rate": 3.877423899862853e-06,
      "loss": 1.1112,
      "step": 1922
    },
    {
      "epoch": 0.13873958370910142,
      "grad_norm": 5.6543344868675645,
      "learning_rate": 3.87726274846755e-06,
      "loss": 1.0537,
      "step": 1923
    },
    {
      "epoch": 0.1388117311785289,
      "grad_norm": 6.825282153211489,
      "learning_rate": 3.877101494561878e-06,
      "loss": 0.9396,
      "step": 1924
    },
    {
      "epoch": 0.13888387864795643,
      "grad_norm": 4.240536194492021,
      "learning_rate": 3.876940138154642e-06,
      "loss": 1.0251,
      "step": 1925
    },
    {
      "epoch": 0.13895602611738395,
      "grad_norm": 9.358332759033983,
      "learning_rate": 3.876778679254653e-06,
      "loss": 1.0662,
      "step": 1926
    },
    {
      "epoch": 0.13902817358681144,
      "grad_norm": 5.541875773144048,
      "learning_rate": 3.876617117870728e-06,
      "loss": 1.0337,
      "step": 1927
    },
    {
      "epoch": 0.13910032105623896,
      "grad_norm": 6.249666128777713,
      "learning_rate": 3.8764554540116895e-06,
      "loss": 0.9428,
      "step": 1928
    },
    {
      "epoch": 0.13917246852566645,
      "grad_norm": 3.938586251726942,
      "learning_rate": 3.876293687686365e-06,
      "loss": 0.8972,
      "step": 1929
    },
    {
      "epoch": 0.13924461599509397,
      "grad_norm": 6.539044415484252,
      "learning_rate": 3.876131818903588e-06,
      "loss": 0.9896,
      "step": 1930
    },
    {
      "epoch": 0.1393167634645215,
      "grad_norm": 4.179148302326314,
      "learning_rate": 3.875969847672198e-06,
      "loss": 1.0164,
      "step": 1931
    },
    {
      "epoch": 0.13938891093394898,
      "grad_norm": 8.85326679518794,
      "learning_rate": 3.875807774001038e-06,
      "loss": 0.8427,
      "step": 1932
    },
    {
      "epoch": 0.1394610584033765,
      "grad_norm": 5.825694213888202,
      "learning_rate": 3.875645597898961e-06,
      "loss": 0.923,
      "step": 1933
    },
    {
      "epoch": 0.13953320587280402,
      "grad_norm": 5.731800634458913,
      "learning_rate": 3.8754833193748194e-06,
      "loss": 1.0155,
      "step": 1934
    },
    {
      "epoch": 0.1396053533422315,
      "grad_norm": 5.680184826597339,
      "learning_rate": 3.8753209384374785e-06,
      "loss": 0.9908,
      "step": 1935
    },
    {
      "epoch": 0.13967750081165903,
      "grad_norm": 8.442724348300537,
      "learning_rate": 3.875158455095802e-06,
      "loss": 1.0403,
      "step": 1936
    },
    {
      "epoch": 0.13974964828108655,
      "grad_norm": 4.823486272517731,
      "learning_rate": 3.8749958693586646e-06,
      "loss": 1.0838,
      "step": 1937
    },
    {
      "epoch": 0.13982179575051404,
      "grad_norm": 5.373092135339634,
      "learning_rate": 3.874833181234944e-06,
      "loss": 1.0823,
      "step": 1938
    },
    {
      "epoch": 0.13989394321994156,
      "grad_norm": 9.285270873161874,
      "learning_rate": 3.874670390733523e-06,
      "loss": 1.1757,
      "step": 1939
    },
    {
      "epoch": 0.13996609068936908,
      "grad_norm": 6.678472968440132,
      "learning_rate": 3.874507497863292e-06,
      "loss": 0.9995,
      "step": 1940
    },
    {
      "epoch": 0.14003823815879657,
      "grad_norm": 19.844591210024173,
      "learning_rate": 3.8743445026331455e-06,
      "loss": 0.8971,
      "step": 1941
    },
    {
      "epoch": 0.1401103856282241,
      "grad_norm": 4.827404573387353,
      "learning_rate": 3.874181405051984e-06,
      "loss": 0.9822,
      "step": 1942
    },
    {
      "epoch": 0.1401825330976516,
      "grad_norm": 0.7863019867829362,
      "learning_rate": 3.8740182051287155e-06,
      "loss": 0.8607,
      "step": 1943
    },
    {
      "epoch": 0.1402546805670791,
      "grad_norm": 7.846240617091032,
      "learning_rate": 3.8738549028722495e-06,
      "loss": 1.0374,
      "step": 1944
    },
    {
      "epoch": 0.14032682803650662,
      "grad_norm": 5.211203961515575,
      "learning_rate": 3.873691498291504e-06,
      "loss": 1.0302,
      "step": 1945
    },
    {
      "epoch": 0.14039897550593414,
      "grad_norm": 0.741853417947918,
      "learning_rate": 3.873527991395402e-06,
      "loss": 0.8729,
      "step": 1946
    },
    {
      "epoch": 0.14047112297536163,
      "grad_norm": 6.8708697310162785,
      "learning_rate": 3.873364382192871e-06,
      "loss": 0.9276,
      "step": 1947
    },
    {
      "epoch": 0.14054327044478915,
      "grad_norm": 5.715433461862925,
      "learning_rate": 3.873200670692846e-06,
      "loss": 1.0105,
      "step": 1948
    },
    {
      "epoch": 0.14061541791421667,
      "grad_norm": 5.6907801967785465,
      "learning_rate": 3.8730368569042685e-06,
      "loss": 1.0757,
      "step": 1949
    },
    {
      "epoch": 0.14068756538364416,
      "grad_norm": 5.79356038627914,
      "learning_rate": 3.872872940836081e-06,
      "loss": 1.0375,
      "step": 1950
    },
    {
      "epoch": 0.14075971285307168,
      "grad_norm": 0.8364548552849757,
      "learning_rate": 3.872708922497235e-06,
      "loss": 0.8785,
      "step": 1951
    },
    {
      "epoch": 0.1408318603224992,
      "grad_norm": 13.46733492965189,
      "learning_rate": 3.872544801896688e-06,
      "loss": 1.0106,
      "step": 1952
    },
    {
      "epoch": 0.1409040077919267,
      "grad_norm": 4.835165969856122,
      "learning_rate": 3.872380579043401e-06,
      "loss": 1.0594,
      "step": 1953
    },
    {
      "epoch": 0.1409761552613542,
      "grad_norm": 6.53261511741024,
      "learning_rate": 3.872216253946341e-06,
      "loss": 1.0238,
      "step": 1954
    },
    {
      "epoch": 0.14104830273078173,
      "grad_norm": 6.190400031602928,
      "learning_rate": 3.8720518266144836e-06,
      "loss": 0.8923,
      "step": 1955
    },
    {
      "epoch": 0.14112045020020922,
      "grad_norm": 0.7657339446738184,
      "learning_rate": 3.871887297056805e-06,
      "loss": 0.8068,
      "step": 1956
    },
    {
      "epoch": 0.14119259766963674,
      "grad_norm": 9.100143230275194,
      "learning_rate": 3.871722665282291e-06,
      "loss": 1.0769,
      "step": 1957
    },
    {
      "epoch": 0.14126474513906426,
      "grad_norm": 7.984767478657874,
      "learning_rate": 3.8715579312999304e-06,
      "loss": 1.0319,
      "step": 1958
    },
    {
      "epoch": 0.14133689260849175,
      "grad_norm": 5.693301082463158,
      "learning_rate": 3.87139309511872e-06,
      "loss": 0.9452,
      "step": 1959
    },
    {
      "epoch": 0.14140904007791927,
      "grad_norm": 5.077963301211633,
      "learning_rate": 3.87122815674766e-06,
      "loss": 1.0445,
      "step": 1960
    },
    {
      "epoch": 0.1414811875473468,
      "grad_norm": 5.2822403740979995,
      "learning_rate": 3.871063116195758e-06,
      "loss": 1.0177,
      "step": 1961
    },
    {
      "epoch": 0.14155333501677428,
      "grad_norm": 5.248086489778843,
      "learning_rate": 3.870897973472026e-06,
      "loss": 1.0234,
      "step": 1962
    },
    {
      "epoch": 0.1416254824862018,
      "grad_norm": 5.726868583544881,
      "learning_rate": 3.870732728585481e-06,
      "loss": 0.9722,
      "step": 1963
    },
    {
      "epoch": 0.14169762995562932,
      "grad_norm": 7.733329810218447,
      "learning_rate": 3.870567381545147e-06,
      "loss": 0.9739,
      "step": 1964
    },
    {
      "epoch": 0.1417697774250568,
      "grad_norm": 16.361673071636805,
      "learning_rate": 3.870401932360053e-06,
      "loss": 1.0032,
      "step": 1965
    },
    {
      "epoch": 0.14184192489448433,
      "grad_norm": 4.984000164993308,
      "learning_rate": 3.870236381039232e-06,
      "loss": 1.0714,
      "step": 1966
    },
    {
      "epoch": 0.14191407236391185,
      "grad_norm": 7.6751843557675885,
      "learning_rate": 3.870070727591727e-06,
      "loss": 1.0156,
      "step": 1967
    },
    {
      "epoch": 0.14198621983333934,
      "grad_norm": 7.407604238579429,
      "learning_rate": 3.869904972026582e-06,
      "loss": 1.0308,
      "step": 1968
    },
    {
      "epoch": 0.14205836730276686,
      "grad_norm": 7.884916812939288,
      "learning_rate": 3.869739114352849e-06,
      "loss": 0.9332,
      "step": 1969
    },
    {
      "epoch": 0.14213051477219438,
      "grad_norm": 12.984528358187305,
      "learning_rate": 3.869573154579585e-06,
      "loss": 0.9619,
      "step": 1970
    },
    {
      "epoch": 0.14220266224162187,
      "grad_norm": 5.939419727175092,
      "learning_rate": 3.869407092715851e-06,
      "loss": 1.0417,
      "step": 1971
    },
    {
      "epoch": 0.1422748097110494,
      "grad_norm": 7.023037467316687,
      "learning_rate": 3.869240928770716e-06,
      "loss": 0.9767,
      "step": 1972
    },
    {
      "epoch": 0.1423469571804769,
      "grad_norm": 8.438755874417023,
      "learning_rate": 3.869074662753253e-06,
      "loss": 1.0492,
      "step": 1973
    },
    {
      "epoch": 0.1424191046499044,
      "grad_norm": 4.529145692684924,
      "learning_rate": 3.8689082946725435e-06,
      "loss": 1.0339,
      "step": 1974
    },
    {
      "epoch": 0.14249125211933192,
      "grad_norm": 9.978001621438127,
      "learning_rate": 3.868741824537669e-06,
      "loss": 1.016,
      "step": 1975
    },
    {
      "epoch": 0.14256339958875944,
      "grad_norm": 15.415136532842382,
      "learning_rate": 3.868575252357722e-06,
      "loss": 1.0533,
      "step": 1976
    },
    {
      "epoch": 0.14263554705818693,
      "grad_norm": 6.193940173019381,
      "learning_rate": 3.868408578141798e-06,
      "loss": 1.0642,
      "step": 1977
    },
    {
      "epoch": 0.14270769452761445,
      "grad_norm": 7.385907141177427,
      "learning_rate": 3.868241801898997e-06,
      "loss": 1.0572,
      "step": 1978
    },
    {
      "epoch": 0.14277984199704197,
      "grad_norm": 4.997046933726091,
      "learning_rate": 3.868074923638428e-06,
      "loss": 1.0012,
      "step": 1979
    },
    {
      "epoch": 0.14285198946646946,
      "grad_norm": 6.603253932259192,
      "learning_rate": 3.8679079433692026e-06,
      "loss": 1.0007,
      "step": 1980
    },
    {
      "epoch": 0.14292413693589698,
      "grad_norm": 4.550699729352992,
      "learning_rate": 3.86774086110044e-06,
      "loss": 1.1899,
      "step": 1981
    },
    {
      "epoch": 0.14299628440532447,
      "grad_norm": 4.231279318410705,
      "learning_rate": 3.867573676841262e-06,
      "loss": 1.0976,
      "step": 1982
    },
    {
      "epoch": 0.143068431874752,
      "grad_norm": 5.5180755445031515,
      "learning_rate": 3.8674063906008e-06,
      "loss": 1.0396,
      "step": 1983
    },
    {
      "epoch": 0.1431405793441795,
      "grad_norm": 6.042040723424133,
      "learning_rate": 3.8672390023881865e-06,
      "loss": 0.9328,
      "step": 1984
    },
    {
      "epoch": 0.143212726813607,
      "grad_norm": 4.108411784838373,
      "learning_rate": 3.867071512212564e-06,
      "loss": 1.0039,
      "step": 1985
    },
    {
      "epoch": 0.14328487428303452,
      "grad_norm": 6.700522035217375,
      "learning_rate": 3.866903920083079e-06,
      "loss": 0.9634,
      "step": 1986
    },
    {
      "epoch": 0.14335702175246204,
      "grad_norm": 7.796978967007856,
      "learning_rate": 3.86673622600888e-06,
      "loss": 0.9427,
      "step": 1987
    },
    {
      "epoch": 0.14342916922188953,
      "grad_norm": 56.559513042679214,
      "learning_rate": 3.866568429999128e-06,
      "loss": 0.9552,
      "step": 1988
    },
    {
      "epoch": 0.14350131669131705,
      "grad_norm": 9.401620733262398,
      "learning_rate": 3.866400532062983e-06,
      "loss": 1.0762,
      "step": 1989
    },
    {
      "epoch": 0.14357346416074457,
      "grad_norm": 6.189447269936486,
      "learning_rate": 3.866232532209614e-06,
      "loss": 0.9724,
      "step": 1990
    },
    {
      "epoch": 0.14364561163017206,
      "grad_norm": 5.929817771234529,
      "learning_rate": 3.866064430448195e-06,
      "loss": 1.0621,
      "step": 1991
    },
    {
      "epoch": 0.14371775909959958,
      "grad_norm": 9.41446284159129,
      "learning_rate": 3.865896226787905e-06,
      "loss": 1.0706,
      "step": 1992
    },
    {
      "epoch": 0.1437899065690271,
      "grad_norm": 5.352682578381807,
      "learning_rate": 3.86572792123793e-06,
      "loss": 0.8978,
      "step": 1993
    },
    {
      "epoch": 0.1438620540384546,
      "grad_norm": 8.15716664509076,
      "learning_rate": 3.8655595138074594e-06,
      "loss": 1.0353,
      "step": 1994
    },
    {
      "epoch": 0.1439342015078821,
      "grad_norm": 6.433576554887464,
      "learning_rate": 3.865391004505691e-06,
      "loss": 1.089,
      "step": 1995
    },
    {
      "epoch": 0.14400634897730963,
      "grad_norm": 8.066658780853368,
      "learning_rate": 3.865222393341824e-06,
      "loss": 1.0636,
      "step": 1996
    },
    {
      "epoch": 0.14407849644673712,
      "grad_norm": 4.977288161808468,
      "learning_rate": 3.865053680325067e-06,
      "loss": 1.1127,
      "step": 1997
    },
    {
      "epoch": 0.14415064391616464,
      "grad_norm": 4.777008968173164,
      "learning_rate": 3.864884865464633e-06,
      "loss": 1.0314,
      "step": 1998
    },
    {
      "epoch": 0.14422279138559216,
      "grad_norm": 5.389134474638965,
      "learning_rate": 3.864715948769741e-06,
      "loss": 0.9593,
      "step": 1999
    },
    {
      "epoch": 0.14429493885501965,
      "grad_norm": 5.079731942260793,
      "learning_rate": 3.864546930249612e-06,
      "loss": 0.9306,
      "step": 2000
    },
    {
      "epoch": 0.14436708632444717,
      "grad_norm": 0.6523242650576979,
      "learning_rate": 3.8643778099134785e-06,
      "loss": 0.7539,
      "step": 2001
    },
    {
      "epoch": 0.1444392337938747,
      "grad_norm": 5.503135654281016,
      "learning_rate": 3.864208587770575e-06,
      "loss": 0.9947,
      "step": 2002
    },
    {
      "epoch": 0.14451138126330218,
      "grad_norm": 6.761753163726695,
      "learning_rate": 3.864039263830141e-06,
      "loss": 1.0686,
      "step": 2003
    },
    {
      "epoch": 0.1445835287327297,
      "grad_norm": 6.860626176109606,
      "learning_rate": 3.863869838101423e-06,
      "loss": 1.0745,
      "step": 2004
    },
    {
      "epoch": 0.14465567620215722,
      "grad_norm": 8.099367055770566,
      "learning_rate": 3.863700310593674e-06,
      "loss": 1.0018,
      "step": 2005
    },
    {
      "epoch": 0.1447278236715847,
      "grad_norm": 7.385002207100344,
      "learning_rate": 3.863530681316149e-06,
      "loss": 0.9789,
      "step": 2006
    },
    {
      "epoch": 0.14479997114101223,
      "grad_norm": 37.19855567015086,
      "learning_rate": 3.863360950278113e-06,
      "loss": 0.904,
      "step": 2007
    },
    {
      "epoch": 0.14487211861043975,
      "grad_norm": 5.824364151857051,
      "learning_rate": 3.863191117488833e-06,
      "loss": 0.9174,
      "step": 2008
    },
    {
      "epoch": 0.14494426607986724,
      "grad_norm": 5.089807025721313,
      "learning_rate": 3.8630211829575835e-06,
      "loss": 1.0409,
      "step": 2009
    },
    {
      "epoch": 0.14501641354929476,
      "grad_norm": 6.593588298243113,
      "learning_rate": 3.862851146693644e-06,
      "loss": 0.8408,
      "step": 2010
    },
    {
      "epoch": 0.14508856101872228,
      "grad_norm": 9.412661575612486,
      "learning_rate": 3.862681008706299e-06,
      "loss": 1.0526,
      "step": 2011
    },
    {
      "epoch": 0.14516070848814977,
      "grad_norm": 5.385559001446056,
      "learning_rate": 3.862510769004839e-06,
      "loss": 0.9957,
      "step": 2012
    },
    {
      "epoch": 0.1452328559575773,
      "grad_norm": 7.897362821763618,
      "learning_rate": 3.862340427598561e-06,
      "loss": 1.0565,
      "step": 2013
    },
    {
      "epoch": 0.1453050034270048,
      "grad_norm": 6.07526242038667,
      "learning_rate": 3.8621699844967675e-06,
      "loss": 0.9109,
      "step": 2014
    },
    {
      "epoch": 0.1453771508964323,
      "grad_norm": 4.50941901446544,
      "learning_rate": 3.861999439708764e-06,
      "loss": 0.9279,
      "step": 2015
    },
    {
      "epoch": 0.14544929836585982,
      "grad_norm": 6.017439927616286,
      "learning_rate": 3.861828793243864e-06,
      "loss": 1.0122,
      "step": 2016
    },
    {
      "epoch": 0.14552144583528734,
      "grad_norm": 4.974289978581428,
      "learning_rate": 3.861658045111386e-06,
      "loss": 0.9395,
      "step": 2017
    },
    {
      "epoch": 0.14559359330471483,
      "grad_norm": 6.4506837231107115,
      "learning_rate": 3.861487195320653e-06,
      "loss": 0.896,
      "step": 2018
    },
    {
      "epoch": 0.14566574077414235,
      "grad_norm": 5.802832983637073,
      "learning_rate": 3.861316243880996e-06,
      "loss": 0.9852,
      "step": 2019
    },
    {
      "epoch": 0.14573788824356987,
      "grad_norm": 3.256049175256641,
      "learning_rate": 3.861145190801749e-06,
      "loss": 1.084,
      "step": 2020
    },
    {
      "epoch": 0.14581003571299736,
      "grad_norm": 4.829559804313679,
      "learning_rate": 3.860974036092253e-06,
      "loss": 0.9457,
      "step": 2021
    },
    {
      "epoch": 0.14588218318242488,
      "grad_norm": 5.014105355664178,
      "learning_rate": 3.860802779761854e-06,
      "loss": 0.961,
      "step": 2022
    },
    {
      "epoch": 0.1459543306518524,
      "grad_norm": 0.7677920288536656,
      "learning_rate": 3.860631421819904e-06,
      "loss": 0.7562,
      "step": 2023
    },
    {
      "epoch": 0.1460264781212799,
      "grad_norm": 5.457393691547368,
      "learning_rate": 3.860459962275759e-06,
      "loss": 0.9728,
      "step": 2024
    },
    {
      "epoch": 0.1460986255907074,
      "grad_norm": 5.528410964219885,
      "learning_rate": 3.860288401138783e-06,
      "loss": 1.0028,
      "step": 2025
    },
    {
      "epoch": 0.14617077306013493,
      "grad_norm": 8.476110659040318,
      "learning_rate": 3.8601167384183445e-06,
      "loss": 0.9966,
      "step": 2026
    },
    {
      "epoch": 0.14624292052956242,
      "grad_norm": 6.745195727721579,
      "learning_rate": 3.859944974123816e-06,
      "loss": 0.8018,
      "step": 2027
    },
    {
      "epoch": 0.14631506799898994,
      "grad_norm": 5.251876858873548,
      "learning_rate": 3.8597731082645785e-06,
      "loss": 1.0271,
      "step": 2028
    },
    {
      "epoch": 0.14638721546841746,
      "grad_norm": 12.64524924591093,
      "learning_rate": 3.859601140850016e-06,
      "loss": 0.9449,
      "step": 2029
    },
    {
      "epoch": 0.14645936293784495,
      "grad_norm": 9.935245666077538,
      "learning_rate": 3.859429071889518e-06,
      "loss": 1.0031,
      "step": 2030
    },
    {
      "epoch": 0.14653151040727247,
      "grad_norm": 5.964736307971444,
      "learning_rate": 3.859256901392484e-06,
      "loss": 1.0568,
      "step": 2031
    },
    {
      "epoch": 0.14660365787669996,
      "grad_norm": 6.007442944176754,
      "learning_rate": 3.859084629368312e-06,
      "loss": 0.9654,
      "step": 2032
    },
    {
      "epoch": 0.14667580534612748,
      "grad_norm": 5.846398717176676,
      "learning_rate": 3.85891225582641e-06,
      "loss": 1.0467,
      "step": 2033
    },
    {
      "epoch": 0.146747952815555,
      "grad_norm": 4.947260420262264,
      "learning_rate": 3.858739780776192e-06,
      "loss": 0.9385,
      "step": 2034
    },
    {
      "epoch": 0.1468201002849825,
      "grad_norm": 16.12550719890308,
      "learning_rate": 3.858567204227075e-06,
      "loss": 0.9528,
      "step": 2035
    },
    {
      "epoch": 0.14689224775441,
      "grad_norm": 3.7461059697402366,
      "learning_rate": 3.8583945261884835e-06,
      "loss": 0.9947,
      "step": 2036
    },
    {
      "epoch": 0.14696439522383753,
      "grad_norm": 5.156955185265518,
      "learning_rate": 3.858221746669847e-06,
      "loss": 1.0298,
      "step": 2037
    },
    {
      "epoch": 0.14703654269326502,
      "grad_norm": 4.604224686523156,
      "learning_rate": 3.8580488656805985e-06,
      "loss": 0.9628,
      "step": 2038
    },
    {
      "epoch": 0.14710869016269254,
      "grad_norm": 4.712760509302588,
      "learning_rate": 3.85787588323018e-06,
      "loss": 0.9583,
      "step": 2039
    },
    {
      "epoch": 0.14718083763212006,
      "grad_norm": 6.156838326008522,
      "learning_rate": 3.857702799328038e-06,
      "loss": 1.1359,
      "step": 2040
    },
    {
      "epoch": 0.14725298510154755,
      "grad_norm": 7.932786878897959,
      "learning_rate": 3.857529613983623e-06,
      "loss": 0.9903,
      "step": 2041
    },
    {
      "epoch": 0.14732513257097507,
      "grad_norm": 6.22114003322448,
      "learning_rate": 3.857356327206391e-06,
      "loss": 0.9571,
      "step": 2042
    },
    {
      "epoch": 0.1473972800404026,
      "grad_norm": 4.325704069656428,
      "learning_rate": 3.8571829390058066e-06,
      "loss": 1.07,
      "step": 2043
    },
    {
      "epoch": 0.14746942750983008,
      "grad_norm": 4.873486993213634,
      "learning_rate": 3.857009449391337e-06,
      "loss": 1.0392,
      "step": 2044
    },
    {
      "epoch": 0.1475415749792576,
      "grad_norm": 10.972889871578456,
      "learning_rate": 3.856835858372455e-06,
      "loss": 0.8962,
      "step": 2045
    },
    {
      "epoch": 0.14761372244868512,
      "grad_norm": 4.825638312776086,
      "learning_rate": 3.856662165958641e-06,
      "loss": 0.9589,
      "step": 2046
    },
    {
      "epoch": 0.1476858699181126,
      "grad_norm": 4.321074875518126,
      "learning_rate": 3.856488372159379e-06,
      "loss": 1.0489,
      "step": 2047
    },
    {
      "epoch": 0.14775801738754013,
      "grad_norm": 5.5631775175520755,
      "learning_rate": 3.856314476984161e-06,
      "loss": 0.9917,
      "step": 2048
    },
    {
      "epoch": 0.14783016485696765,
      "grad_norm": 5.655906708979878,
      "learning_rate": 3.8561404804424795e-06,
      "loss": 1.0012,
      "step": 2049
    },
    {
      "epoch": 0.14790231232639514,
      "grad_norm": 8.857187367692198,
      "learning_rate": 3.855966382543838e-06,
      "loss": 0.9517,
      "step": 2050
    },
    {
      "epoch": 0.14797445979582266,
      "grad_norm": 6.930240991759047,
      "learning_rate": 3.855792183297743e-06,
      "loss": 0.9292,
      "step": 2051
    },
    {
      "epoch": 0.14804660726525018,
      "grad_norm": 4.467582710239078,
      "learning_rate": 3.855617882713707e-06,
      "loss": 1.0284,
      "step": 2052
    },
    {
      "epoch": 0.14811875473467767,
      "grad_norm": 0.9937798345633747,
      "learning_rate": 3.855443480801247e-06,
      "loss": 0.8732,
      "step": 2053
    },
    {
      "epoch": 0.1481909022041052,
      "grad_norm": 8.450765396756417,
      "learning_rate": 3.855268977569888e-06,
      "loss": 0.9666,
      "step": 2054
    },
    {
      "epoch": 0.1482630496735327,
      "grad_norm": 0.8718560821777844,
      "learning_rate": 3.855094373029158e-06,
      "loss": 0.919,
      "step": 2055
    },
    {
      "epoch": 0.1483351971429602,
      "grad_norm": 11.134477671520031,
      "learning_rate": 3.8549196671885915e-06,
      "loss": 0.9458,
      "step": 2056
    },
    {
      "epoch": 0.14840734461238772,
      "grad_norm": 7.089016986488098,
      "learning_rate": 3.854744860057728e-06,
      "loss": 1.0265,
      "step": 2057
    },
    {
      "epoch": 0.14847949208181524,
      "grad_norm": 8.992228649585472,
      "learning_rate": 3.8545699516461145e-06,
      "loss": 1.0957,
      "step": 2058
    },
    {
      "epoch": 0.14855163955124273,
      "grad_norm": 4.603376409912391,
      "learning_rate": 3.854394941963301e-06,
      "loss": 0.9616,
      "step": 2059
    },
    {
      "epoch": 0.14862378702067025,
      "grad_norm": 13.414231989683453,
      "learning_rate": 3.854219831018845e-06,
      "loss": 0.9963,
      "step": 2060
    },
    {
      "epoch": 0.14869593449009777,
      "grad_norm": 5.440804540661484,
      "learning_rate": 3.854044618822307e-06,
      "loss": 0.9591,
      "step": 2061
    },
    {
      "epoch": 0.14876808195952526,
      "grad_norm": 4.402795831032907,
      "learning_rate": 3.853869305383256e-06,
      "loss": 1.1063,
      "step": 2062
    },
    {
      "epoch": 0.14884022942895278,
      "grad_norm": 5.259188649077159,
      "learning_rate": 3.853693890711265e-06,
      "loss": 0.9894,
      "step": 2063
    },
    {
      "epoch": 0.1489123768983803,
      "grad_norm": 7.322784494986164,
      "learning_rate": 3.853518374815913e-06,
      "loss": 1.0572,
      "step": 2064
    },
    {
      "epoch": 0.1489845243678078,
      "grad_norm": 5.25523478788353,
      "learning_rate": 3.853342757706785e-06,
      "loss": 0.9057,
      "step": 2065
    },
    {
      "epoch": 0.1490566718372353,
      "grad_norm": 4.903285594909208,
      "learning_rate": 3.853167039393469e-06,
      "loss": 0.9172,
      "step": 2066
    },
    {
      "epoch": 0.14912881930666283,
      "grad_norm": 8.014465129136962,
      "learning_rate": 3.852991219885562e-06,
      "loss": 1.0394,
      "step": 2067
    },
    {
      "epoch": 0.14920096677609032,
      "grad_norm": 7.414920415573394,
      "learning_rate": 3.852815299192662e-06,
      "loss": 1.056,
      "step": 2068
    },
    {
      "epoch": 0.14927311424551784,
      "grad_norm": 5.305439464548371,
      "learning_rate": 3.852639277324379e-06,
      "loss": 1.1408,
      "step": 2069
    },
    {
      "epoch": 0.14934526171494536,
      "grad_norm": 6.915738453302885,
      "learning_rate": 3.852463154290323e-06,
      "loss": 1.0365,
      "step": 2070
    },
    {
      "epoch": 0.14941740918437285,
      "grad_norm": 11.98593205131215,
      "learning_rate": 3.852286930100112e-06,
      "loss": 1.0685,
      "step": 2071
    },
    {
      "epoch": 0.14948955665380037,
      "grad_norm": 7.469340919517107,
      "learning_rate": 3.852110604763368e-06,
      "loss": 0.9653,
      "step": 2072
    },
    {
      "epoch": 0.1495617041232279,
      "grad_norm": 5.745190474869331,
      "learning_rate": 3.8519341782897204e-06,
      "loss": 0.8999,
      "step": 2073
    },
    {
      "epoch": 0.14963385159265538,
      "grad_norm": 7.74108497013775,
      "learning_rate": 3.851757650688803e-06,
      "loss": 1.0148,
      "step": 2074
    },
    {
      "epoch": 0.1497059990620829,
      "grad_norm": 4.161840364810737,
      "learning_rate": 3.851581021970255e-06,
      "loss": 1.0444,
      "step": 2075
    },
    {
      "epoch": 0.14977814653151042,
      "grad_norm": 6.543836229807334,
      "learning_rate": 3.851404292143723e-06,
      "loss": 1.0836,
      "step": 2076
    },
    {
      "epoch": 0.1498502940009379,
      "grad_norm": 4.933576159472302,
      "learning_rate": 3.851227461218855e-06,
      "loss": 1.0834,
      "step": 2077
    },
    {
      "epoch": 0.14992244147036543,
      "grad_norm": 44.0613779777355,
      "learning_rate": 3.851050529205309e-06,
      "loss": 0.9874,
      "step": 2078
    },
    {
      "epoch": 0.14999458893979295,
      "grad_norm": 6.033638120648873,
      "learning_rate": 3.8508734961127454e-06,
      "loss": 1.037,
      "step": 2079
    },
    {
      "epoch": 0.15006673640922044,
      "grad_norm": 6.290236388405565,
      "learning_rate": 3.850696361950832e-06,
      "loss": 0.931,
      "step": 2080
    },
    {
      "epoch": 0.15013888387864796,
      "grad_norm": 8.038533392603647,
      "learning_rate": 3.850519126729241e-06,
      "loss": 1.0702,
      "step": 2081
    },
    {
      "epoch": 0.15021103134807548,
      "grad_norm": 7.676295356908445,
      "learning_rate": 3.8503417904576515e-06,
      "loss": 1.0548,
      "step": 2082
    },
    {
      "epoch": 0.15028317881750297,
      "grad_norm": 5.555667924804204,
      "learning_rate": 3.850164353145747e-06,
      "loss": 1.098,
      "step": 2083
    },
    {
      "epoch": 0.1503553262869305,
      "grad_norm": 1.2126768848347997,
      "learning_rate": 3.8499868148032145e-06,
      "loss": 0.8973,
      "step": 2084
    },
    {
      "epoch": 0.15042747375635798,
      "grad_norm": 5.775046819852145,
      "learning_rate": 3.849809175439752e-06,
      "loss": 0.9415,
      "step": 2085
    },
    {
      "epoch": 0.1504996212257855,
      "grad_norm": 5.8180741684881685,
      "learning_rate": 3.849631435065058e-06,
      "loss": 0.9171,
      "step": 2086
    },
    {
      "epoch": 0.15057176869521302,
      "grad_norm": 6.437560442298168,
      "learning_rate": 3.8494535936888375e-06,
      "loss": 0.9358,
      "step": 2087
    },
    {
      "epoch": 0.1506439161646405,
      "grad_norm": 5.939173091269241,
      "learning_rate": 3.849275651320804e-06,
      "loss": 0.8949,
      "step": 2088
    },
    {
      "epoch": 0.15071606363406803,
      "grad_norm": 5.705052047612432,
      "learning_rate": 3.849097607970672e-06,
      "loss": 0.9511,
      "step": 2089
    },
    {
      "epoch": 0.15078821110349555,
      "grad_norm": 6.700897771554511,
      "learning_rate": 3.848919463648165e-06,
      "loss": 0.9558,
      "step": 2090
    },
    {
      "epoch": 0.15086035857292304,
      "grad_norm": 4.386212830145501,
      "learning_rate": 3.848741218363011e-06,
      "loss": 1.018,
      "step": 2091
    },
    {
      "epoch": 0.15093250604235056,
      "grad_norm": 4.018970565319834,
      "learning_rate": 3.848562872124942e-06,
      "loss": 0.9597,
      "step": 2092
    },
    {
      "epoch": 0.15100465351177808,
      "grad_norm": 9.995346894114851,
      "learning_rate": 3.848384424943699e-06,
      "loss": 1.0547,
      "step": 2093
    },
    {
      "epoch": 0.15107680098120557,
      "grad_norm": 9.036256874030144,
      "learning_rate": 3.848205876829024e-06,
      "loss": 0.9781,
      "step": 2094
    },
    {
      "epoch": 0.1511489484506331,
      "grad_norm": 6.8584183914391845,
      "learning_rate": 3.848027227790668e-06,
      "loss": 1.0361,
      "step": 2095
    },
    {
      "epoch": 0.1512210959200606,
      "grad_norm": 4.069081770207711,
      "learning_rate": 3.847848477838387e-06,
      "loss": 0.9874,
      "step": 2096
    },
    {
      "epoch": 0.1512932433894881,
      "grad_norm": 7.679008446103705,
      "learning_rate": 3.84766962698194e-06,
      "loss": 1.125,
      "step": 2097
    },
    {
      "epoch": 0.15136539085891562,
      "grad_norm": 4.32176782746841,
      "learning_rate": 3.8474906752310945e-06,
      "loss": 0.9928,
      "step": 2098
    },
    {
      "epoch": 0.15143753832834314,
      "grad_norm": 9.679895578877323,
      "learning_rate": 3.847311622595623e-06,
      "loss": 0.9237,
      "step": 2099
    },
    {
      "epoch": 0.15150968579777063,
      "grad_norm": 9.541873002354748,
      "learning_rate": 3.847132469085302e-06,
      "loss": 0.9582,
      "step": 2100
    },
    {
      "epoch": 0.15158183326719815,
      "grad_norm": 7.237847238489154,
      "learning_rate": 3.846953214709916e-06,
      "loss": 0.9744,
      "step": 2101
    },
    {
      "epoch": 0.15165398073662567,
      "grad_norm": 5.8617068927517355,
      "learning_rate": 3.846773859479251e-06,
      "loss": 0.9429,
      "step": 2102
    },
    {
      "epoch": 0.15172612820605316,
      "grad_norm": 6.851993699462274,
      "learning_rate": 3.846594403403101e-06,
      "loss": 1.0757,
      "step": 2103
    },
    {
      "epoch": 0.15179827567548068,
      "grad_norm": 5.24729195642634,
      "learning_rate": 3.846414846491268e-06,
      "loss": 0.9567,
      "step": 2104
    },
    {
      "epoch": 0.1518704231449082,
      "grad_norm": 8.424546081811785,
      "learning_rate": 3.846235188753555e-06,
      "loss": 1.0741,
      "step": 2105
    },
    {
      "epoch": 0.1519425706143357,
      "grad_norm": 7.089653546432806,
      "learning_rate": 3.846055430199773e-06,
      "loss": 0.9951,
      "step": 2106
    },
    {
      "epoch": 0.1520147180837632,
      "grad_norm": 23.97088000843745,
      "learning_rate": 3.845875570839738e-06,
      "loss": 0.9292,
      "step": 2107
    },
    {
      "epoch": 0.15208686555319073,
      "grad_norm": 0.838561352011037,
      "learning_rate": 3.845695610683271e-06,
      "loss": 0.8998,
      "step": 2108
    },
    {
      "epoch": 0.15215901302261822,
      "grad_norm": 4.439848802071294,
      "learning_rate": 3.8455155497401995e-06,
      "loss": 1.0134,
      "step": 2109
    },
    {
      "epoch": 0.15223116049204574,
      "grad_norm": 5.470659195200273,
      "learning_rate": 3.8453353880203545e-06,
      "loss": 1.0272,
      "step": 2110
    },
    {
      "epoch": 0.15230330796147326,
      "grad_norm": 6.090200789325166,
      "learning_rate": 3.845155125533576e-06,
      "loss": 1.0076,
      "step": 2111
    },
    {
      "epoch": 0.15237545543090075,
      "grad_norm": 58.29496319901918,
      "learning_rate": 3.8449747622897075e-06,
      "loss": 0.9742,
      "step": 2112
    },
    {
      "epoch": 0.15244760290032827,
      "grad_norm": 175.59609971287801,
      "learning_rate": 3.844794298298596e-06,
      "loss": 0.999,
      "step": 2113
    },
    {
      "epoch": 0.1525197503697558,
      "grad_norm": 9.445217185726694,
      "learning_rate": 3.844613733570097e-06,
      "loss": 0.9961,
      "step": 2114
    },
    {
      "epoch": 0.15259189783918328,
      "grad_norm": 8.00145422116007,
      "learning_rate": 3.844433068114072e-06,
      "loss": 1.0692,
      "step": 2115
    },
    {
      "epoch": 0.1526640453086108,
      "grad_norm": 11.146474963582412,
      "learning_rate": 3.844252301940384e-06,
      "loss": 0.9966,
      "step": 2116
    },
    {
      "epoch": 0.15273619277803832,
      "grad_norm": 8.513624155707628,
      "learning_rate": 3.844071435058904e-06,
      "loss": 1.0841,
      "step": 2117
    },
    {
      "epoch": 0.1528083402474658,
      "grad_norm": 11.125176589078647,
      "learning_rate": 3.843890467479511e-06,
      "loss": 0.8854,
      "step": 2118
    },
    {
      "epoch": 0.15288048771689333,
      "grad_norm": 5.421270419225924,
      "learning_rate": 3.843709399212084e-06,
      "loss": 1.0892,
      "step": 2119
    },
    {
      "epoch": 0.15295263518632085,
      "grad_norm": 7.078078381382764,
      "learning_rate": 3.843528230266514e-06,
      "loss": 1.0047,
      "step": 2120
    },
    {
      "epoch": 0.15302478265574834,
      "grad_norm": 9.587481410917443,
      "learning_rate": 3.84334696065269e-06,
      "loss": 0.9826,
      "step": 2121
    },
    {
      "epoch": 0.15309693012517586,
      "grad_norm": 25.83892223649721,
      "learning_rate": 3.843165590380512e-06,
      "loss": 0.9738,
      "step": 2122
    },
    {
      "epoch": 0.15316907759460338,
      "grad_norm": 7.122658595806211,
      "learning_rate": 3.8429841194598855e-06,
      "loss": 1.0554,
      "step": 2123
    },
    {
      "epoch": 0.15324122506403087,
      "grad_norm": 9.166866462148517,
      "learning_rate": 3.842802547900719e-06,
      "loss": 0.998,
      "step": 2124
    },
    {
      "epoch": 0.1533133725334584,
      "grad_norm": 5.601493888232605,
      "learning_rate": 3.842620875712926e-06,
      "loss": 0.9561,
      "step": 2125
    },
    {
      "epoch": 0.1533855200028859,
      "grad_norm": 8.924794375831798,
      "learning_rate": 3.842439102906429e-06,
      "loss": 1.1519,
      "step": 2126
    },
    {
      "epoch": 0.1534576674723134,
      "grad_norm": 6.401822676991802,
      "learning_rate": 3.842257229491152e-06,
      "loss": 1.0554,
      "step": 2127
    },
    {
      "epoch": 0.15352981494174092,
      "grad_norm": 8.649535291239324,
      "learning_rate": 3.842075255477028e-06,
      "loss": 1.0117,
      "step": 2128
    },
    {
      "epoch": 0.15360196241116844,
      "grad_norm": 6.698056169622349,
      "learning_rate": 3.841893180873994e-06,
      "loss": 0.9724,
      "step": 2129
    },
    {
      "epoch": 0.15367410988059593,
      "grad_norm": 4.35509912949572,
      "learning_rate": 3.841711005691992e-06,
      "loss": 1.0129,
      "step": 2130
    },
    {
      "epoch": 0.15374625735002345,
      "grad_norm": 19.83204625331814,
      "learning_rate": 3.841528729940969e-06,
      "loss": 1.0229,
      "step": 2131
    },
    {
      "epoch": 0.15381840481945097,
      "grad_norm": 11.54138631285085,
      "learning_rate": 3.841346353630879e-06,
      "loss": 0.9374,
      "step": 2132
    },
    {
      "epoch": 0.15389055228887846,
      "grad_norm": 10.104397758646549,
      "learning_rate": 3.841163876771681e-06,
      "loss": 1.1158,
      "step": 2133
    },
    {
      "epoch": 0.15396269975830598,
      "grad_norm": 19.89960688677158,
      "learning_rate": 3.84098129937334e-06,
      "loss": 0.9974,
      "step": 2134
    },
    {
      "epoch": 0.15403484722773347,
      "grad_norm": 9.00409901840827,
      "learning_rate": 3.840798621445825e-06,
      "loss": 1.0372,
      "step": 2135
    },
    {
      "epoch": 0.154106994697161,
      "grad_norm": 7.519785912895476,
      "learning_rate": 3.840615842999112e-06,
      "loss": 0.9468,
      "step": 2136
    },
    {
      "epoch": 0.1541791421665885,
      "grad_norm": 8.379483972653043,
      "learning_rate": 3.840432964043182e-06,
      "loss": 1.1251,
      "step": 2137
    },
    {
      "epoch": 0.154251289636016,
      "grad_norm": 7.761402970347258,
      "learning_rate": 3.840249984588021e-06,
      "loss": 0.9947,
      "step": 2138
    },
    {
      "epoch": 0.15432343710544352,
      "grad_norm": 6.516633099845948,
      "learning_rate": 3.84006690464362e-06,
      "loss": 0.9282,
      "step": 2139
    },
    {
      "epoch": 0.15439558457487104,
      "grad_norm": 7.9758018257981,
      "learning_rate": 3.839883724219976e-06,
      "loss": 1.0412,
      "step": 2140
    },
    {
      "epoch": 0.15446773204429853,
      "grad_norm": 11.409373903339786,
      "learning_rate": 3.839700443327094e-06,
      "loss": 0.9387,
      "step": 2141
    },
    {
      "epoch": 0.15453987951372605,
      "grad_norm": 7.589083772424891,
      "learning_rate": 3.839517061974981e-06,
      "loss": 0.9342,
      "step": 2142
    },
    {
      "epoch": 0.15461202698315357,
      "grad_norm": 5.643353507645597,
      "learning_rate": 3.839333580173651e-06,
      "loss": 0.9296,
      "step": 2143
    },
    {
      "epoch": 0.15468417445258106,
      "grad_norm": 7.53175725851033,
      "learning_rate": 3.839149997933123e-06,
      "loss": 0.9597,
      "step": 2144
    },
    {
      "epoch": 0.15475632192200858,
      "grad_norm": 5.8975730818375816,
      "learning_rate": 3.838966315263422e-06,
      "loss": 0.9627,
      "step": 2145
    },
    {
      "epoch": 0.1548284693914361,
      "grad_norm": 6.836563587400494,
      "learning_rate": 3.838782532174579e-06,
      "loss": 1.0209,
      "step": 2146
    },
    {
      "epoch": 0.1549006168608636,
      "grad_norm": 0.7061481967910295,
      "learning_rate": 3.838598648676628e-06,
      "loss": 0.8861,
      "step": 2147
    },
    {
      "epoch": 0.1549727643302911,
      "grad_norm": 7.665605084548503,
      "learning_rate": 3.838414664779611e-06,
      "loss": 1.0468,
      "step": 2148
    },
    {
      "epoch": 0.15504491179971863,
      "grad_norm": 31.438907538694192,
      "learning_rate": 3.838230580493575e-06,
      "loss": 0.9663,
      "step": 2149
    },
    {
      "epoch": 0.15511705926914612,
      "grad_norm": 10.7073372898391,
      "learning_rate": 3.838046395828572e-06,
      "loss": 1.1029,
      "step": 2150
    },
    {
      "epoch": 0.15518920673857364,
      "grad_norm": 14.077972174184369,
      "learning_rate": 3.837862110794659e-06,
      "loss": 0.9594,
      "step": 2151
    },
    {
      "epoch": 0.15526135420800116,
      "grad_norm": 44.99030250327216,
      "learning_rate": 3.8376777254018994e-06,
      "loss": 0.9347,
      "step": 2152
    },
    {
      "epoch": 0.15533350167742865,
      "grad_norm": 4.44066474784012,
      "learning_rate": 3.837493239660363e-06,
      "loss": 1.0108,
      "step": 2153
    },
    {
      "epoch": 0.15540564914685617,
      "grad_norm": 7.136114185702659,
      "learning_rate": 3.837308653580122e-06,
      "loss": 0.8729,
      "step": 2154
    },
    {
      "epoch": 0.1554777966162837,
      "grad_norm": 10.18761832513723,
      "learning_rate": 3.837123967171258e-06,
      "loss": 1.0329,
      "step": 2155
    },
    {
      "epoch": 0.15554994408571118,
      "grad_norm": 12.500347895549373,
      "learning_rate": 3.8369391804438545e-06,
      "loss": 0.9194,
      "step": 2156
    },
    {
      "epoch": 0.1556220915551387,
      "grad_norm": 8.4927606608557,
      "learning_rate": 3.836754293408002e-06,
      "loss": 0.9198,
      "step": 2157
    },
    {
      "epoch": 0.15569423902456622,
      "grad_norm": 6.603497428158158,
      "learning_rate": 3.836569306073798e-06,
      "loss": 1.0112,
      "step": 2158
    },
    {
      "epoch": 0.1557663864939937,
      "grad_norm": 6.7636636970243655,
      "learning_rate": 3.836384218451342e-06,
      "loss": 1.0139,
      "step": 2159
    },
    {
      "epoch": 0.15583853396342123,
      "grad_norm": 11.178702363675514,
      "learning_rate": 3.8361990305507415e-06,
      "loss": 0.9256,
      "step": 2160
    },
    {
      "epoch": 0.15591068143284875,
      "grad_norm": 5.66494268484321,
      "learning_rate": 3.83601374238211e-06,
      "loss": 1.043,
      "step": 2161
    },
    {
      "epoch": 0.15598282890227624,
      "grad_norm": 5.27018400070103,
      "learning_rate": 3.835828353955565e-06,
      "loss": 0.9549,
      "step": 2162
    },
    {
      "epoch": 0.15605497637170376,
      "grad_norm": 4.385748819682973,
      "learning_rate": 3.8356428652812295e-06,
      "loss": 1.0594,
      "step": 2163
    },
    {
      "epoch": 0.15612712384113128,
      "grad_norm": 7.703994299564649,
      "learning_rate": 3.835457276369232e-06,
      "loss": 0.9475,
      "step": 2164
    },
    {
      "epoch": 0.15619927131055877,
      "grad_norm": 6.616141676874563,
      "learning_rate": 3.835271587229708e-06,
      "loss": 0.9772,
      "step": 2165
    },
    {
      "epoch": 0.1562714187799863,
      "grad_norm": 0.8267641582286036,
      "learning_rate": 3.835085797872795e-06,
      "loss": 0.8927,
      "step": 2166
    },
    {
      "epoch": 0.1563435662494138,
      "grad_norm": 4.54187344888199,
      "learning_rate": 3.8348999083086414e-06,
      "loss": 0.9927,
      "step": 2167
    },
    {
      "epoch": 0.1564157137188413,
      "grad_norm": 5.116113168474357,
      "learning_rate": 3.834713918547396e-06,
      "loss": 1.0061,
      "step": 2168
    },
    {
      "epoch": 0.15648786118826882,
      "grad_norm": 5.979407258017015,
      "learning_rate": 3.834527828599216e-06,
      "loss": 0.9673,
      "step": 2169
    },
    {
      "epoch": 0.15656000865769634,
      "grad_norm": 6.354335428300238,
      "learning_rate": 3.834341638474261e-06,
      "loss": 1.024,
      "step": 2170
    },
    {
      "epoch": 0.15663215612712383,
      "grad_norm": 8.078790810629648,
      "learning_rate": 3.834155348182701e-06,
      "loss": 1.0364,
      "step": 2171
    },
    {
      "epoch": 0.15670430359655135,
      "grad_norm": 6.53928271871937,
      "learning_rate": 3.833968957734706e-06,
      "loss": 1.0292,
      "step": 2172
    },
    {
      "epoch": 0.15677645106597887,
      "grad_norm": 0.7831347428527892,
      "learning_rate": 3.833782467140456e-06,
      "loss": 0.9023,
      "step": 2173
    },
    {
      "epoch": 0.15684859853540636,
      "grad_norm": 4.309856323184017,
      "learning_rate": 3.833595876410134e-06,
      "loss": 1.0347,
      "step": 2174
    },
    {
      "epoch": 0.15692074600483388,
      "grad_norm": 6.567886902703123,
      "learning_rate": 3.833409185553929e-06,
      "loss": 0.9615,
      "step": 2175
    },
    {
      "epoch": 0.1569928934742614,
      "grad_norm": 5.700755005742489,
      "learning_rate": 3.833222394582035e-06,
      "loss": 0.9566,
      "step": 2176
    },
    {
      "epoch": 0.1570650409436889,
      "grad_norm": 10.196897326534705,
      "learning_rate": 3.833035503504653e-06,
      "loss": 1.0546,
      "step": 2177
    },
    {
      "epoch": 0.1571371884131164,
      "grad_norm": 0.8797014496207124,
      "learning_rate": 3.832848512331987e-06,
      "loss": 0.8632,
      "step": 2178
    },
    {
      "epoch": 0.15720933588254393,
      "grad_norm": 6.732420458460107,
      "learning_rate": 3.832661421074249e-06,
      "loss": 0.9719,
      "step": 2179
    },
    {
      "epoch": 0.15728148335197142,
      "grad_norm": 4.447463120408922,
      "learning_rate": 3.832474229741655e-06,
      "loss": 0.9712,
      "step": 2180
    },
    {
      "epoch": 0.15735363082139894,
      "grad_norm": 4.0306821911161945,
      "learning_rate": 3.832286938344428e-06,
      "loss": 0.9682,
      "step": 2181
    },
    {
      "epoch": 0.15742577829082646,
      "grad_norm": 8.057853838578032,
      "learning_rate": 3.832099546892792e-06,
      "loss": 0.9722,
      "step": 2182
    },
    {
      "epoch": 0.15749792576025395,
      "grad_norm": 8.186646395149772,
      "learning_rate": 3.831912055396984e-06,
      "loss": 1.1039,
      "step": 2183
    },
    {
      "epoch": 0.15757007322968147,
      "grad_norm": 12.688201133259469,
      "learning_rate": 3.831724463867239e-06,
      "loss": 1.0508,
      "step": 2184
    },
    {
      "epoch": 0.157642220699109,
      "grad_norm": 5.46621279034096,
      "learning_rate": 3.8315367723138025e-06,
      "loss": 1.0527,
      "step": 2185
    },
    {
      "epoch": 0.15771436816853648,
      "grad_norm": 6.909130479617171,
      "learning_rate": 3.831348980746922e-06,
      "loss": 1.0608,
      "step": 2186
    },
    {
      "epoch": 0.157786515637964,
      "grad_norm": 7.423870517094829,
      "learning_rate": 3.831161089176855e-06,
      "loss": 1.0154,
      "step": 2187
    },
    {
      "epoch": 0.1578586631073915,
      "grad_norm": 6.623062714243077,
      "learning_rate": 3.830973097613859e-06,
      "loss": 1.05,
      "step": 2188
    },
    {
      "epoch": 0.15793081057681901,
      "grad_norm": 6.203196642867924,
      "learning_rate": 3.830785006068199e-06,
      "loss": 1.0382,
      "step": 2189
    },
    {
      "epoch": 0.15800295804624653,
      "grad_norm": 11.243477859299523,
      "learning_rate": 3.830596814550148e-06,
      "loss": 1.041,
      "step": 2190
    },
    {
      "epoch": 0.15807510551567402,
      "grad_norm": 6.73818500837067,
      "learning_rate": 3.830408523069981e-06,
      "loss": 0.9889,
      "step": 2191
    },
    {
      "epoch": 0.15814725298510154,
      "grad_norm": 34.33196347167334,
      "learning_rate": 3.830220131637981e-06,
      "loss": 0.9632,
      "step": 2192
    },
    {
      "epoch": 0.15821940045452906,
      "grad_norm": 4.224856330189488,
      "learning_rate": 3.8300316402644345e-06,
      "loss": 0.9469,
      "step": 2193
    },
    {
      "epoch": 0.15829154792395655,
      "grad_norm": 25.711795649990588,
      "learning_rate": 3.829843048959635e-06,
      "loss": 0.9885,
      "step": 2194
    },
    {
      "epoch": 0.15836369539338407,
      "grad_norm": 5.6228075629107535,
      "learning_rate": 3.829654357733881e-06,
      "loss": 1.0576,
      "step": 2195
    },
    {
      "epoch": 0.1584358428628116,
      "grad_norm": 6.558831633751726,
      "learning_rate": 3.829465566597474e-06,
      "loss": 1.0214,
      "step": 2196
    },
    {
      "epoch": 0.15850799033223908,
      "grad_norm": 4.252433304664533,
      "learning_rate": 3.829276675560727e-06,
      "loss": 1.0723,
      "step": 2197
    },
    {
      "epoch": 0.1585801378016666,
      "grad_norm": 7.083777320261195,
      "learning_rate": 3.829087684633951e-06,
      "loss": 1.1051,
      "step": 2198
    },
    {
      "epoch": 0.15865228527109412,
      "grad_norm": 5.848785026430736,
      "learning_rate": 3.8288985938274675e-06,
      "loss": 1.0178,
      "step": 2199
    },
    {
      "epoch": 0.15872443274052161,
      "grad_norm": 5.350988471437773,
      "learning_rate": 3.828709403151603e-06,
      "loss": 1.0715,
      "step": 2200
    },
    {
      "epoch": 0.15879658020994913,
      "grad_norm": 5.766019316631329,
      "learning_rate": 3.8285201126166865e-06,
      "loss": 1.0085,
      "step": 2201
    },
    {
      "epoch": 0.15886872767937665,
      "grad_norm": 6.836219860239946,
      "learning_rate": 3.8283307222330566e-06,
      "loss": 1.0188,
      "step": 2202
    },
    {
      "epoch": 0.15894087514880414,
      "grad_norm": 6.623399649167915,
      "learning_rate": 3.828141232011054e-06,
      "loss": 0.983,
      "step": 2203
    },
    {
      "epoch": 0.15901302261823166,
      "grad_norm": 8.904547706011396,
      "learning_rate": 3.827951641961026e-06,
      "loss": 1.0667,
      "step": 2204
    },
    {
      "epoch": 0.15908517008765918,
      "grad_norm": 25.59526666751589,
      "learning_rate": 3.827761952093326e-06,
      "loss": 1.0733,
      "step": 2205
    },
    {
      "epoch": 0.15915731755708667,
      "grad_norm": 6.65193258326369,
      "learning_rate": 3.827572162418312e-06,
      "loss": 1.0068,
      "step": 2206
    },
    {
      "epoch": 0.1592294650265142,
      "grad_norm": 17.524949952841787,
      "learning_rate": 3.827382272946348e-06,
      "loss": 0.963,
      "step": 2207
    },
    {
      "epoch": 0.1593016124959417,
      "grad_norm": 6.944613547067629,
      "learning_rate": 3.827192283687801e-06,
      "loss": 0.9314,
      "step": 2208
    },
    {
      "epoch": 0.1593737599653692,
      "grad_norm": 8.36275212887153,
      "learning_rate": 3.827002194653049e-06,
      "loss": 1.0522,
      "step": 2209
    },
    {
      "epoch": 0.15944590743479672,
      "grad_norm": 9.950922891573574,
      "learning_rate": 3.826812005852471e-06,
      "loss": 0.9612,
      "step": 2210
    },
    {
      "epoch": 0.15951805490422424,
      "grad_norm": 4.936526238717059,
      "learning_rate": 3.826621717296451e-06,
      "loss": 0.8983,
      "step": 2211
    },
    {
      "epoch": 0.15959020237365174,
      "grad_norm": 6.4771332161464015,
      "learning_rate": 3.826431328995381e-06,
      "loss": 1.0104,
      "step": 2212
    },
    {
      "epoch": 0.15966234984307925,
      "grad_norm": 10.208982319865395,
      "learning_rate": 3.826240840959658e-06,
      "loss": 1.1178,
      "step": 2213
    },
    {
      "epoch": 0.15973449731250677,
      "grad_norm": 6.03492206008,
      "learning_rate": 3.826050253199682e-06,
      "loss": 1.0205,
      "step": 2214
    },
    {
      "epoch": 0.15980664478193427,
      "grad_norm": 5.164022055971069,
      "learning_rate": 3.825859565725862e-06,
      "loss": 1.0453,
      "step": 2215
    },
    {
      "epoch": 0.15987879225136178,
      "grad_norm": 5.944373719615636,
      "learning_rate": 3.82566877854861e-06,
      "loss": 1.0211,
      "step": 2216
    },
    {
      "epoch": 0.1599509397207893,
      "grad_norm": 8.325510956717894,
      "learning_rate": 3.825477891678345e-06,
      "loss": 0.9869,
      "step": 2217
    },
    {
      "epoch": 0.1600230871902168,
      "grad_norm": 126.1716844038808,
      "learning_rate": 3.82528690512549e-06,
      "loss": 1.0174,
      "step": 2218
    },
    {
      "epoch": 0.16009523465964431,
      "grad_norm": 14.10893321480972,
      "learning_rate": 3.825095818900474e-06,
      "loss": 1.0115,
      "step": 2219
    },
    {
      "epoch": 0.16016738212907183,
      "grad_norm": 10.137975596455929,
      "learning_rate": 3.824904633013731e-06,
      "loss": 1.0059,
      "step": 2220
    },
    {
      "epoch": 0.16023952959849933,
      "grad_norm": 4.998660480358752,
      "learning_rate": 3.824713347475702e-06,
      "loss": 0.9667,
      "step": 2221
    },
    {
      "epoch": 0.16031167706792684,
      "grad_norm": 4.616501707212433,
      "learning_rate": 3.824521962296832e-06,
      "loss": 1.0288,
      "step": 2222
    },
    {
      "epoch": 0.16038382453735436,
      "grad_norm": 0.8626018629875205,
      "learning_rate": 3.824330477487572e-06,
      "loss": 0.9138,
      "step": 2223
    },
    {
      "epoch": 0.16045597200678186,
      "grad_norm": 7.212479944052488,
      "learning_rate": 3.824138893058377e-06,
      "loss": 0.9684,
      "step": 2224
    },
    {
      "epoch": 0.16052811947620937,
      "grad_norm": 6.822972638626173,
      "learning_rate": 3.823947209019712e-06,
      "loss": 1.0006,
      "step": 2225
    },
    {
      "epoch": 0.1606002669456369,
      "grad_norm": 8.358428958587954,
      "learning_rate": 3.82375542538204e-06,
      "loss": 0.9959,
      "step": 2226
    },
    {
      "epoch": 0.16067241441506439,
      "grad_norm": 6.850295189950074,
      "learning_rate": 3.8235635421558365e-06,
      "loss": 1.0587,
      "step": 2227
    },
    {
      "epoch": 0.1607445618844919,
      "grad_norm": 6.715255609278713,
      "learning_rate": 3.823371559351578e-06,
      "loss": 1.0269,
      "step": 2228
    },
    {
      "epoch": 0.16081670935391942,
      "grad_norm": 5.2679753251015615,
      "learning_rate": 3.823179476979748e-06,
      "loss": 1.0183,
      "step": 2229
    },
    {
      "epoch": 0.16088885682334692,
      "grad_norm": 5.930011403682352,
      "learning_rate": 3.822987295050837e-06,
      "loss": 1.0935,
      "step": 2230
    },
    {
      "epoch": 0.16096100429277443,
      "grad_norm": 7.5716295858263525,
      "learning_rate": 3.822795013575338e-06,
      "loss": 1.0051,
      "step": 2231
    },
    {
      "epoch": 0.16103315176220195,
      "grad_norm": 3.826824854949464,
      "learning_rate": 3.822602632563751e-06,
      "loss": 1.0848,
      "step": 2232
    },
    {
      "epoch": 0.16110529923162945,
      "grad_norm": 0.6933613521923834,
      "learning_rate": 3.822410152026581e-06,
      "loss": 0.8075,
      "step": 2233
    },
    {
      "epoch": 0.16117744670105696,
      "grad_norm": 7.685095434247781,
      "learning_rate": 3.822217571974339e-06,
      "loss": 1.0387,
      "step": 2234
    },
    {
      "epoch": 0.16124959417048448,
      "grad_norm": 4.423164745721987,
      "learning_rate": 3.822024892417542e-06,
      "loss": 1.148,
      "step": 2235
    },
    {
      "epoch": 0.16132174163991198,
      "grad_norm": 6.523343936454773,
      "learning_rate": 3.82183211336671e-06,
      "loss": 0.9395,
      "step": 2236
    },
    {
      "epoch": 0.1613938891093395,
      "grad_norm": 5.15136524363115,
      "learning_rate": 3.82163923483237e-06,
      "loss": 0.9613,
      "step": 2237
    },
    {
      "epoch": 0.16146603657876699,
      "grad_norm": 0.66109925132776,
      "learning_rate": 3.821446256825055e-06,
      "loss": 0.7991,
      "step": 2238
    },
    {
      "epoch": 0.1615381840481945,
      "grad_norm": 6.715992064316261,
      "learning_rate": 3.821253179355302e-06,
      "loss": 1.0108,
      "step": 2239
    },
    {
      "epoch": 0.16161033151762202,
      "grad_norm": 6.918458116778843,
      "learning_rate": 3.8210600024336556e-06,
      "loss": 1.008,
      "step": 2240
    },
    {
      "epoch": 0.16168247898704952,
      "grad_norm": 5.959831565166607,
      "learning_rate": 3.8208667260706645e-06,
      "loss": 1.0006,
      "step": 2241
    },
    {
      "epoch": 0.16175462645647704,
      "grad_norm": 4.5380807888460675,
      "learning_rate": 3.820673350276881e-06,
      "loss": 1.0602,
      "step": 2242
    },
    {
      "epoch": 0.16182677392590455,
      "grad_norm": 6.5072855100627525,
      "learning_rate": 3.820479875062867e-06,
      "loss": 1.0583,
      "step": 2243
    },
    {
      "epoch": 0.16189892139533205,
      "grad_norm": 9.43583290584107,
      "learning_rate": 3.820286300439184e-06,
      "loss": 0.8966,
      "step": 2244
    },
    {
      "epoch": 0.16197106886475957,
      "grad_norm": 5.657132074695347,
      "learning_rate": 3.820092626416407e-06,
      "loss": 1.0345,
      "step": 2245
    },
    {
      "epoch": 0.16204321633418708,
      "grad_norm": 11.820136291441067,
      "learning_rate": 3.819898853005108e-06,
      "loss": 0.9923,
      "step": 2246
    },
    {
      "epoch": 0.16211536380361458,
      "grad_norm": 13.98421460278126,
      "learning_rate": 3.8197049802158705e-06,
      "loss": 0.9699,
      "step": 2247
    },
    {
      "epoch": 0.1621875112730421,
      "grad_norm": 7.573697462238092,
      "learning_rate": 3.81951100805928e-06,
      "loss": 1.0498,
      "step": 2248
    },
    {
      "epoch": 0.16225965874246961,
      "grad_norm": 9.324495309053177,
      "learning_rate": 3.819316936545929e-06,
      "loss": 1.0133,
      "step": 2249
    },
    {
      "epoch": 0.1623318062118971,
      "grad_norm": 9.239058338088865,
      "learning_rate": 3.819122765686415e-06,
      "loss": 0.9482,
      "step": 2250
    },
    {
      "epoch": 0.16240395368132463,
      "grad_norm": 4.544505127675482,
      "learning_rate": 3.818928495491341e-06,
      "loss": 0.9361,
      "step": 2251
    },
    {
      "epoch": 0.16247610115075214,
      "grad_norm": 4.251679032707898,
      "learning_rate": 3.818734125971315e-06,
      "loss": 1.024,
      "step": 2252
    },
    {
      "epoch": 0.16254824862017964,
      "grad_norm": 5.647138644103105,
      "learning_rate": 3.818539657136951e-06,
      "loss": 0.9735,
      "step": 2253
    },
    {
      "epoch": 0.16262039608960716,
      "grad_norm": 5.500739654743494,
      "learning_rate": 3.818345088998868e-06,
      "loss": 0.9565,
      "step": 2254
    },
    {
      "epoch": 0.16269254355903467,
      "grad_norm": 0.9315940797322847,
      "learning_rate": 3.818150421567692e-06,
      "loss": 0.9035,
      "step": 2255
    },
    {
      "epoch": 0.16276469102846217,
      "grad_norm": 9.34982092374781,
      "learning_rate": 3.8179556548540514e-06,
      "loss": 0.9407,
      "step": 2256
    },
    {
      "epoch": 0.16283683849788969,
      "grad_norm": 7.548957742661452,
      "learning_rate": 3.817760788868583e-06,
      "loss": 1.0744,
      "step": 2257
    },
    {
      "epoch": 0.1629089859673172,
      "grad_norm": 5.796021432061649,
      "learning_rate": 3.817565823621926e-06,
      "loss": 1.0568,
      "step": 2258
    },
    {
      "epoch": 0.1629811334367447,
      "grad_norm": 5.683061146114699,
      "learning_rate": 3.8173707591247286e-06,
      "loss": 0.8835,
      "step": 2259
    },
    {
      "epoch": 0.16305328090617222,
      "grad_norm": 9.28120807676769,
      "learning_rate": 3.817175595387641e-06,
      "loss": 1.011,
      "step": 2260
    },
    {
      "epoch": 0.16312542837559973,
      "grad_norm": 8.842771658009898,
      "learning_rate": 3.816980332421322e-06,
      "loss": 1.1101,
      "step": 2261
    },
    {
      "epoch": 0.16319757584502723,
      "grad_norm": 9.035547204306445,
      "learning_rate": 3.816784970236432e-06,
      "loss": 0.9952,
      "step": 2262
    },
    {
      "epoch": 0.16326972331445475,
      "grad_norm": 6.628600671543955,
      "learning_rate": 3.816589508843642e-06,
      "loss": 1.0648,
      "step": 2263
    },
    {
      "epoch": 0.16334187078388226,
      "grad_norm": 7.7582267658305035,
      "learning_rate": 3.816393948253623e-06,
      "loss": 1.0906,
      "step": 2264
    },
    {
      "epoch": 0.16341401825330976,
      "grad_norm": 5.638490732769743,
      "learning_rate": 3.816198288477054e-06,
      "loss": 1.0397,
      "step": 2265
    },
    {
      "epoch": 0.16348616572273728,
      "grad_norm": 7.210432008030562,
      "learning_rate": 3.81600252952462e-06,
      "loss": 1.0945,
      "step": 2266
    },
    {
      "epoch": 0.1635583131921648,
      "grad_norm": 6.511846457580935,
      "learning_rate": 3.815806671407011e-06,
      "loss": 1.0353,
      "step": 2267
    },
    {
      "epoch": 0.1636304606615923,
      "grad_norm": 4.415286058293676,
      "learning_rate": 3.815610714134921e-06,
      "loss": 0.9689,
      "step": 2268
    },
    {
      "epoch": 0.1637026081310198,
      "grad_norm": 6.555343067726944,
      "learning_rate": 3.815414657719051e-06,
      "loss": 0.9993,
      "step": 2269
    },
    {
      "epoch": 0.16377475560044732,
      "grad_norm": 8.928327642109876,
      "learning_rate": 3.815218502170108e-06,
      "loss": 0.9496,
      "step": 2270
    },
    {
      "epoch": 0.16384690306987482,
      "grad_norm": 7.791732896601527,
      "learning_rate": 3.8150222474988006e-06,
      "loss": 0.9836,
      "step": 2271
    },
    {
      "epoch": 0.16391905053930234,
      "grad_norm": 6.952683886990827,
      "learning_rate": 3.814825893715849e-06,
      "loss": 1.034,
      "step": 2272
    },
    {
      "epoch": 0.16399119800872985,
      "grad_norm": 13.779148348109862,
      "learning_rate": 3.8146294408319726e-06,
      "loss": 1.0122,
      "step": 2273
    },
    {
      "epoch": 0.16406334547815735,
      "grad_norm": 9.345320859802314,
      "learning_rate": 3.8144328888579e-06,
      "loss": 1.0764,
      "step": 2274
    },
    {
      "epoch": 0.16413549294758487,
      "grad_norm": 4.273573404274651,
      "learning_rate": 3.814236237804364e-06,
      "loss": 0.9386,
      "step": 2275
    },
    {
      "epoch": 0.16420764041701238,
      "grad_norm": 5.941251834264754,
      "learning_rate": 3.814039487682104e-06,
      "loss": 1.0511,
      "step": 2276
    },
    {
      "epoch": 0.16427978788643988,
      "grad_norm": 6.458003899160503,
      "learning_rate": 3.813842638501862e-06,
      "loss": 1.0419,
      "step": 2277
    },
    {
      "epoch": 0.1643519353558674,
      "grad_norm": 5.668655869145081,
      "learning_rate": 3.813645690274388e-06,
      "loss": 0.9452,
      "step": 2278
    },
    {
      "epoch": 0.16442408282529491,
      "grad_norm": 7.105491566844207,
      "learning_rate": 3.8134486430104373e-06,
      "loss": 0.9127,
      "step": 2279
    },
    {
      "epoch": 0.1644962302947224,
      "grad_norm": 6.893404099613369,
      "learning_rate": 3.8132514967207693e-06,
      "loss": 1.0254,
      "step": 2280
    },
    {
      "epoch": 0.16456837776414993,
      "grad_norm": 4.572603717543259,
      "learning_rate": 3.813054251416149e-06,
      "loss": 0.9377,
      "step": 2281
    },
    {
      "epoch": 0.16464052523357744,
      "grad_norm": 9.87164017705043,
      "learning_rate": 3.8128569071073487e-06,
      "loss": 0.917,
      "step": 2282
    },
    {
      "epoch": 0.16471267270300494,
      "grad_norm": 5.005415463271849,
      "learning_rate": 3.812659463805143e-06,
      "loss": 0.9446,
      "step": 2283
    },
    {
      "epoch": 0.16478482017243246,
      "grad_norm": 5.717840070197782,
      "learning_rate": 3.8124619215203143e-06,
      "loss": 0.977,
      "step": 2284
    },
    {
      "epoch": 0.16485696764185997,
      "grad_norm": 0.7249622154255408,
      "learning_rate": 3.81226428026365e-06,
      "loss": 0.8725,
      "step": 2285
    },
    {
      "epoch": 0.16492911511128747,
      "grad_norm": 6.794240576921739,
      "learning_rate": 3.812066540045942e-06,
      "loss": 1.0224,
      "step": 2286
    },
    {
      "epoch": 0.16500126258071499,
      "grad_norm": 5.212486300004478,
      "learning_rate": 3.8118687008779876e-06,
      "loss": 1.06,
      "step": 2287
    },
    {
      "epoch": 0.1650734100501425,
      "grad_norm": 11.585416611057102,
      "learning_rate": 3.8116707627705907e-06,
      "loss": 1.0242,
      "step": 2288
    },
    {
      "epoch": 0.16514555751957,
      "grad_norm": 5.9414271166834425,
      "learning_rate": 3.811472725734561e-06,
      "loss": 0.9764,
      "step": 2289
    },
    {
      "epoch": 0.16521770498899752,
      "grad_norm": 0.7360908267587243,
      "learning_rate": 3.8112745897807114e-06,
      "loss": 0.8065,
      "step": 2290
    },
    {
      "epoch": 0.165289852458425,
      "grad_norm": 5.72412923662439,
      "learning_rate": 3.811076354919861e-06,
      "loss": 1.0167,
      "step": 2291
    },
    {
      "epoch": 0.16536199992785253,
      "grad_norm": 7.589083269769036,
      "learning_rate": 3.8108780211628353e-06,
      "loss": 0.9388,
      "step": 2292
    },
    {
      "epoch": 0.16543414739728005,
      "grad_norm": 6.724654756781227,
      "learning_rate": 3.8106795885204648e-06,
      "loss": 1.0579,
      "step": 2293
    },
    {
      "epoch": 0.16550629486670754,
      "grad_norm": 5.131179805373065,
      "learning_rate": 3.810481057003585e-06,
      "loss": 1.0251,
      "step": 2294
    },
    {
      "epoch": 0.16557844233613506,
      "grad_norm": 4.480190331639365,
      "learning_rate": 3.810282426623036e-06,
      "loss": 0.994,
      "step": 2295
    },
    {
      "epoch": 0.16565058980556258,
      "grad_norm": 6.614111823950859,
      "learning_rate": 3.8100836973896663e-06,
      "loss": 1.0399,
      "step": 2296
    },
    {
      "epoch": 0.16572273727499007,
      "grad_norm": 5.019041994942568,
      "learning_rate": 3.809884869314326e-06,
      "loss": 1.0144,
      "step": 2297
    },
    {
      "epoch": 0.1657948847444176,
      "grad_norm": 8.63560210160395,
      "learning_rate": 3.809685942407873e-06,
      "loss": 1.0304,
      "step": 2298
    },
    {
      "epoch": 0.1658670322138451,
      "grad_norm": 4.548322204784973,
      "learning_rate": 3.8094869166811696e-06,
      "loss": 1.0819,
      "step": 2299
    },
    {
      "epoch": 0.1659391796832726,
      "grad_norm": 9.567565772929703,
      "learning_rate": 3.8092877921450847e-06,
      "loss": 1.0345,
      "step": 2300
    },
    {
      "epoch": 0.16601132715270012,
      "grad_norm": 7.451942334020074,
      "learning_rate": 3.809088568810491e-06,
      "loss": 0.9099,
      "step": 2301
    },
    {
      "epoch": 0.16608347462212764,
      "grad_norm": 5.37087481798122,
      "learning_rate": 3.8088892466882674e-06,
      "loss": 1.1175,
      "step": 2302
    },
    {
      "epoch": 0.16615562209155513,
      "grad_norm": 5.357108568354272,
      "learning_rate": 3.8086898257892982e-06,
      "loss": 0.8889,
      "step": 2303
    },
    {
      "epoch": 0.16622776956098265,
      "grad_norm": 5.064655739776512,
      "learning_rate": 3.808490306124474e-06,
      "loss": 1.0281,
      "step": 2304
    },
    {
      "epoch": 0.16629991703041017,
      "grad_norm": 7.503978691168424,
      "learning_rate": 3.808290687704688e-06,
      "loss": 1.0814,
      "step": 2305
    },
    {
      "epoch": 0.16637206449983766,
      "grad_norm": 4.537941037436794,
      "learning_rate": 3.808090970540842e-06,
      "loss": 1.0146,
      "step": 2306
    },
    {
      "epoch": 0.16644421196926518,
      "grad_norm": 3.806754363520148,
      "learning_rate": 3.8078911546438415e-06,
      "loss": 0.9638,
      "step": 2307
    },
    {
      "epoch": 0.1665163594386927,
      "grad_norm": 16.7203197277108,
      "learning_rate": 3.807691240024598e-06,
      "loss": 1.1732,
      "step": 2308
    },
    {
      "epoch": 0.1665885069081202,
      "grad_norm": 3.7706330107419332,
      "learning_rate": 3.807491226694027e-06,
      "loss": 1.0809,
      "step": 2309
    },
    {
      "epoch": 0.1666606543775477,
      "grad_norm": 6.067225551363101,
      "learning_rate": 3.8072911146630516e-06,
      "loss": 1.0012,
      "step": 2310
    },
    {
      "epoch": 0.16673280184697523,
      "grad_norm": 4.449431376613793,
      "learning_rate": 3.807090903942599e-06,
      "loss": 0.9273,
      "step": 2311
    },
    {
      "epoch": 0.16680494931640272,
      "grad_norm": 5.513151224642876,
      "learning_rate": 3.806890594543601e-06,
      "loss": 1.0077,
      "step": 2312
    },
    {
      "epoch": 0.16687709678583024,
      "grad_norm": 6.432710513996784,
      "learning_rate": 3.8066901864769974e-06,
      "loss": 0.9667,
      "step": 2313
    },
    {
      "epoch": 0.16694924425525776,
      "grad_norm": 7.828584760310464,
      "learning_rate": 3.8064896797537304e-06,
      "loss": 0.9594,
      "step": 2314
    },
    {
      "epoch": 0.16702139172468525,
      "grad_norm": 6.534843952677052,
      "learning_rate": 3.8062890743847493e-06,
      "loss": 0.9367,
      "step": 2315
    },
    {
      "epoch": 0.16709353919411277,
      "grad_norm": 4.694083714717534,
      "learning_rate": 3.8060883703810088e-06,
      "loss": 0.8947,
      "step": 2316
    },
    {
      "epoch": 0.16716568666354029,
      "grad_norm": 6.7471570279789015,
      "learning_rate": 3.805887567753468e-06,
      "loss": 0.8763,
      "step": 2317
    },
    {
      "epoch": 0.16723783413296778,
      "grad_norm": 7.150676305837048,
      "learning_rate": 3.8056866665130923e-06,
      "loss": 1.0136,
      "step": 2318
    },
    {
      "epoch": 0.1673099816023953,
      "grad_norm": 4.698237011370823,
      "learning_rate": 3.8054856666708528e-06,
      "loss": 1.0291,
      "step": 2319
    },
    {
      "epoch": 0.16738212907182282,
      "grad_norm": 12.379627325953402,
      "learning_rate": 3.8052845682377238e-06,
      "loss": 1.0269,
      "step": 2320
    },
    {
      "epoch": 0.1674542765412503,
      "grad_norm": 9.901186060901878,
      "learning_rate": 3.805083371224688e-06,
      "loss": 0.9561,
      "step": 2321
    },
    {
      "epoch": 0.16752642401067783,
      "grad_norm": 21.172902707990495,
      "learning_rate": 3.8048820756427312e-06,
      "loss": 0.9698,
      "step": 2322
    },
    {
      "epoch": 0.16759857148010535,
      "grad_norm": 8.403466426769363,
      "learning_rate": 3.8046806815028456e-06,
      "loss": 0.9443,
      "step": 2323
    },
    {
      "epoch": 0.16767071894953284,
      "grad_norm": 9.514680862567273,
      "learning_rate": 3.804479188816029e-06,
      "loss": 1.0883,
      "step": 2324
    },
    {
      "epoch": 0.16774286641896036,
      "grad_norm": 5.566059645184812,
      "learning_rate": 3.804277597593284e-06,
      "loss": 0.9772,
      "step": 2325
    },
    {
      "epoch": 0.16781501388838788,
      "grad_norm": 15.230462237912175,
      "learning_rate": 3.804075907845618e-06,
      "loss": 0.9828,
      "step": 2326
    },
    {
      "epoch": 0.16788716135781537,
      "grad_norm": 5.932723844848863,
      "learning_rate": 3.803874119584046e-06,
      "loss": 1.0657,
      "step": 2327
    },
    {
      "epoch": 0.1679593088272429,
      "grad_norm": 7.8634279998259835,
      "learning_rate": 3.8036722328195854e-06,
      "loss": 1.045,
      "step": 2328
    },
    {
      "epoch": 0.1680314562966704,
      "grad_norm": 4.818737595043716,
      "learning_rate": 3.803470247563261e-06,
      "loss": 1.0763,
      "step": 2329
    },
    {
      "epoch": 0.1681036037660979,
      "grad_norm": 3.7146189849277995,
      "learning_rate": 3.803268163826103e-06,
      "loss": 0.9809,
      "step": 2330
    },
    {
      "epoch": 0.16817575123552542,
      "grad_norm": 8.053130153762417,
      "learning_rate": 3.8030659816191457e-06,
      "loss": 1.0712,
      "step": 2331
    },
    {
      "epoch": 0.16824789870495294,
      "grad_norm": 6.30262815364086,
      "learning_rate": 3.8028637009534305e-06,
      "loss": 0.9465,
      "step": 2332
    },
    {
      "epoch": 0.16832004617438043,
      "grad_norm": 5.28546281533048,
      "learning_rate": 3.802661321840002e-06,
      "loss": 0.9756,
      "step": 2333
    },
    {
      "epoch": 0.16839219364380795,
      "grad_norm": 11.850411426467964,
      "learning_rate": 3.802458844289912e-06,
      "loss": 0.9368,
      "step": 2334
    },
    {
      "epoch": 0.16846434111323547,
      "grad_norm": 10.516934498934697,
      "learning_rate": 3.8022562683142176e-06,
      "loss": 0.967,
      "step": 2335
    },
    {
      "epoch": 0.16853648858266296,
      "grad_norm": 6.238535322295554,
      "learning_rate": 3.8020535939239796e-06,
      "loss": 1.04,
      "step": 2336
    },
    {
      "epoch": 0.16860863605209048,
      "grad_norm": 6.360359584311884,
      "learning_rate": 3.801850821130266e-06,
      "loss": 1.0134,
      "step": 2337
    },
    {
      "epoch": 0.168680783521518,
      "grad_norm": 6.294498572584907,
      "learning_rate": 3.801647949944149e-06,
      "loss": 1.0853,
      "step": 2338
    },
    {
      "epoch": 0.1687529309909455,
      "grad_norm": 4.652160692176804,
      "learning_rate": 3.8014449803767067e-06,
      "loss": 1.0489,
      "step": 2339
    },
    {
      "epoch": 0.168825078460373,
      "grad_norm": 5.7687126513891185,
      "learning_rate": 3.801241912439023e-06,
      "loss": 0.9745,
      "step": 2340
    },
    {
      "epoch": 0.1688972259298005,
      "grad_norm": 5.507612681991534,
      "learning_rate": 3.8010387461421866e-06,
      "loss": 0.9475,
      "step": 2341
    },
    {
      "epoch": 0.16896937339922802,
      "grad_norm": 5.72197328209131,
      "learning_rate": 3.8008354814972913e-06,
      "loss": 1.0519,
      "step": 2342
    },
    {
      "epoch": 0.16904152086865554,
      "grad_norm": 6.598121098189028,
      "learning_rate": 3.8006321185154373e-06,
      "loss": 1.0095,
      "step": 2343
    },
    {
      "epoch": 0.16911366833808303,
      "grad_norm": 6.896334749952212,
      "learning_rate": 3.800428657207728e-06,
      "loss": 0.991,
      "step": 2344
    },
    {
      "epoch": 0.16918581580751055,
      "grad_norm": 4.8269223193613175,
      "learning_rate": 3.800225097585276e-06,
      "loss": 1.0897,
      "step": 2345
    },
    {
      "epoch": 0.16925796327693807,
      "grad_norm": 7.736185621005127,
      "learning_rate": 3.8000214396591945e-06,
      "loss": 1.1353,
      "step": 2346
    },
    {
      "epoch": 0.16933011074636556,
      "grad_norm": 6.220662958563417,
      "learning_rate": 3.799817683440606e-06,
      "loss": 0.974,
      "step": 2347
    },
    {
      "epoch": 0.16940225821579308,
      "grad_norm": 0.6691762261125527,
      "learning_rate": 3.7996138289406366e-06,
      "loss": 0.8125,
      "step": 2348
    },
    {
      "epoch": 0.1694744056852206,
      "grad_norm": 7.0705865796324225,
      "learning_rate": 3.799409876170418e-06,
      "loss": 1.0548,
      "step": 2349
    },
    {
      "epoch": 0.1695465531546481,
      "grad_norm": 13.641089222842393,
      "learning_rate": 3.7992058251410876e-06,
      "loss": 1.0245,
      "step": 2350
    },
    {
      "epoch": 0.1696187006240756,
      "grad_norm": 6.11727511561217,
      "learning_rate": 3.7990016758637866e-06,
      "loss": 1.0534,
      "step": 2351
    },
    {
      "epoch": 0.16969084809350313,
      "grad_norm": 4.48520517141496,
      "learning_rate": 3.7987974283496645e-06,
      "loss": 1.0694,
      "step": 2352
    },
    {
      "epoch": 0.16976299556293062,
      "grad_norm": 0.6359189633004001,
      "learning_rate": 3.7985930826098735e-06,
      "loss": 0.8195,
      "step": 2353
    },
    {
      "epoch": 0.16983514303235814,
      "grad_norm": 4.629082192726922,
      "learning_rate": 3.798388638655572e-06,
      "loss": 0.9842,
      "step": 2354
    },
    {
      "epoch": 0.16990729050178566,
      "grad_norm": 6.137257980267525,
      "learning_rate": 3.798184096497925e-06,
      "loss": 0.7868,
      "step": 2355
    },
    {
      "epoch": 0.16997943797121315,
      "grad_norm": 6.176494945617006,
      "learning_rate": 3.797979456148101e-06,
      "loss": 1.0196,
      "step": 2356
    },
    {
      "epoch": 0.17005158544064067,
      "grad_norm": 5.621154318265153,
      "learning_rate": 3.797774717617275e-06,
      "loss": 1.0716,
      "step": 2357
    },
    {
      "epoch": 0.1701237329100682,
      "grad_norm": 9.861046699229982,
      "learning_rate": 3.797569880916626e-06,
      "loss": 1.0168,
      "step": 2358
    },
    {
      "epoch": 0.17019588037949568,
      "grad_norm": 8.232857667513459,
      "learning_rate": 3.7973649460573407e-06,
      "loss": 1.0854,
      "step": 2359
    },
    {
      "epoch": 0.1702680278489232,
      "grad_norm": 7.800384404418121,
      "learning_rate": 3.79715991305061e-06,
      "loss": 1.0337,
      "step": 2360
    },
    {
      "epoch": 0.17034017531835072,
      "grad_norm": 6.061142287663179,
      "learning_rate": 3.796954781907628e-06,
      "loss": 0.8968,
      "step": 2361
    },
    {
      "epoch": 0.1704123227877782,
      "grad_norm": 8.860226652462554,
      "learning_rate": 3.796749552639598e-06,
      "loss": 1.0692,
      "step": 2362
    },
    {
      "epoch": 0.17048447025720573,
      "grad_norm": 11.432259781893654,
      "learning_rate": 3.7965442252577265e-06,
      "loss": 1.0225,
      "step": 2363
    },
    {
      "epoch": 0.17055661772663325,
      "grad_norm": 5.2390928045136445,
      "learning_rate": 3.7963387997732253e-06,
      "loss": 0.984,
      "step": 2364
    },
    {
      "epoch": 0.17062876519606074,
      "grad_norm": 5.382420585106711,
      "learning_rate": 3.796133276197312e-06,
      "loss": 1.0343,
      "step": 2365
    },
    {
      "epoch": 0.17070091266548826,
      "grad_norm": 6.262121119811518,
      "learning_rate": 3.7959276545412093e-06,
      "loss": 1.0739,
      "step": 2366
    },
    {
      "epoch": 0.17077306013491578,
      "grad_norm": 4.888343351925922,
      "learning_rate": 3.7957219348161464e-06,
      "loss": 1.0473,
      "step": 2367
    },
    {
      "epoch": 0.17084520760434327,
      "grad_norm": 6.2870034964096515,
      "learning_rate": 3.7955161170333556e-06,
      "loss": 0.9987,
      "step": 2368
    },
    {
      "epoch": 0.1709173550737708,
      "grad_norm": 9.11396979891888,
      "learning_rate": 3.7953102012040764e-06,
      "loss": 1.0522,
      "step": 2369
    },
    {
      "epoch": 0.1709895025431983,
      "grad_norm": 5.704599017562223,
      "learning_rate": 3.795104187339554e-06,
      "loss": 0.9311,
      "step": 2370
    },
    {
      "epoch": 0.1710616500126258,
      "grad_norm": 6.86391151239583,
      "learning_rate": 3.7948980754510363e-06,
      "loss": 1.0175,
      "step": 2371
    },
    {
      "epoch": 0.17113379748205332,
      "grad_norm": 8.669742576359003,
      "learning_rate": 3.794691865549779e-06,
      "loss": 1.0491,
      "step": 2372
    },
    {
      "epoch": 0.17120594495148084,
      "grad_norm": 6.899838885209256,
      "learning_rate": 3.794485557647043e-06,
      "loss": 1.1311,
      "step": 2373
    },
    {
      "epoch": 0.17127809242090833,
      "grad_norm": 5.916699082549937,
      "learning_rate": 3.794279151754094e-06,
      "loss": 0.9895,
      "step": 2374
    },
    {
      "epoch": 0.17135023989033585,
      "grad_norm": 0.9464740710104975,
      "learning_rate": 3.794072647882202e-06,
      "loss": 0.8725,
      "step": 2375
    },
    {
      "epoch": 0.17142238735976337,
      "grad_norm": 11.364100337873554,
      "learning_rate": 3.793866046042645e-06,
      "loss": 1.1243,
      "step": 2376
    },
    {
      "epoch": 0.17149453482919086,
      "grad_norm": 4.0478481462968094,
      "learning_rate": 3.793659346246703e-06,
      "loss": 0.9958,
      "step": 2377
    },
    {
      "epoch": 0.17156668229861838,
      "grad_norm": 0.8024355664040155,
      "learning_rate": 3.7934525485056643e-06,
      "loss": 0.8486,
      "step": 2378
    },
    {
      "epoch": 0.1716388297680459,
      "grad_norm": 5.95626820164208,
      "learning_rate": 3.793245652830821e-06,
      "loss": 1.0182,
      "step": 2379
    },
    {
      "epoch": 0.1717109772374734,
      "grad_norm": 8.71379095991328,
      "learning_rate": 3.793038659233471e-06,
      "loss": 1.0161,
      "step": 2380
    },
    {
      "epoch": 0.1717831247069009,
      "grad_norm": 6.5489603513343715,
      "learning_rate": 3.792831567724918e-06,
      "loss": 1.0568,
      "step": 2381
    },
    {
      "epoch": 0.17185527217632843,
      "grad_norm": 0.8035743198645514,
      "learning_rate": 3.7926243783164693e-06,
      "loss": 0.808,
      "step": 2382
    },
    {
      "epoch": 0.17192741964575592,
      "grad_norm": 6.82331423778064,
      "learning_rate": 3.792417091019439e-06,
      "loss": 0.9487,
      "step": 2383
    },
    {
      "epoch": 0.17199956711518344,
      "grad_norm": 7.25992352321898,
      "learning_rate": 3.7922097058451477e-06,
      "loss": 0.981,
      "step": 2384
    },
    {
      "epoch": 0.17207171458461096,
      "grad_norm": 10.125064802551112,
      "learning_rate": 3.7920022228049186e-06,
      "loss": 1.0554,
      "step": 2385
    },
    {
      "epoch": 0.17214386205403845,
      "grad_norm": 4.985760630713478,
      "learning_rate": 3.7917946419100816e-06,
      "loss": 0.9373,
      "step": 2386
    },
    {
      "epoch": 0.17221600952346597,
      "grad_norm": 6.405012206543153,
      "learning_rate": 3.791586963171972e-06,
      "loss": 1.0092,
      "step": 2387
    },
    {
      "epoch": 0.1722881569928935,
      "grad_norm": 5.591253309515407,
      "learning_rate": 3.791379186601931e-06,
      "loss": 1.0296,
      "step": 2388
    },
    {
      "epoch": 0.17236030446232098,
      "grad_norm": 5.23158595238083,
      "learning_rate": 3.791171312211304e-06,
      "loss": 1.0023,
      "step": 2389
    },
    {
      "epoch": 0.1724324519317485,
      "grad_norm": 7.690265685255254,
      "learning_rate": 3.790963340011442e-06,
      "loss": 1.0174,
      "step": 2390
    },
    {
      "epoch": 0.17250459940117602,
      "grad_norm": 5.5143929204959825,
      "learning_rate": 3.7907552700137024e-06,
      "loss": 0.8766,
      "step": 2391
    },
    {
      "epoch": 0.1725767468706035,
      "grad_norm": 0.8924876361279042,
      "learning_rate": 3.7905471022294463e-06,
      "loss": 0.8825,
      "step": 2392
    },
    {
      "epoch": 0.17264889434003103,
      "grad_norm": 8.464441308355658,
      "learning_rate": 3.790338836670042e-06,
      "loss": 1.0669,
      "step": 2393
    },
    {
      "epoch": 0.17272104180945852,
      "grad_norm": 6.45950113218575,
      "learning_rate": 3.7901304733468607e-06,
      "loss": 0.9132,
      "step": 2394
    },
    {
      "epoch": 0.17279318927888604,
      "grad_norm": 4.84641591203054,
      "learning_rate": 3.789922012271281e-06,
      "loss": 0.9473,
      "step": 2395
    },
    {
      "epoch": 0.17286533674831356,
      "grad_norm": 8.409851890829207,
      "learning_rate": 3.7897134534546863e-06,
      "loss": 1.0299,
      "step": 2396
    },
    {
      "epoch": 0.17293748421774105,
      "grad_norm": 4.612742496721928,
      "learning_rate": 3.789504796908465e-06,
      "loss": 0.9956,
      "step": 2397
    },
    {
      "epoch": 0.17300963168716857,
      "grad_norm": 6.790748237653732,
      "learning_rate": 3.789296042644012e-06,
      "loss": 0.9855,
      "step": 2398
    },
    {
      "epoch": 0.1730817791565961,
      "grad_norm": 4.530390796099257,
      "learning_rate": 3.7890871906727256e-06,
      "loss": 0.9244,
      "step": 2399
    },
    {
      "epoch": 0.17315392662602358,
      "grad_norm": 6.442292734822933,
      "learning_rate": 3.788878241006011e-06,
      "loss": 0.9563,
      "step": 2400
    },
    {
      "epoch": 0.1732260740954511,
      "grad_norm": 4.837743763856396,
      "learning_rate": 3.788669193655277e-06,
      "loss": 1.093,
      "step": 2401
    },
    {
      "epoch": 0.17329822156487862,
      "grad_norm": 7.159424723099102,
      "learning_rate": 3.7884600486319405e-06,
      "loss": 0.956,
      "step": 2402
    },
    {
      "epoch": 0.1733703690343061,
      "grad_norm": 8.47817524980501,
      "learning_rate": 3.788250805947421e-06,
      "loss": 0.9152,
      "step": 2403
    },
    {
      "epoch": 0.17344251650373363,
      "grad_norm": 0.8222600071368548,
      "learning_rate": 3.788041465613145e-06,
      "loss": 0.8813,
      "step": 2404
    },
    {
      "epoch": 0.17351466397316115,
      "grad_norm": 5.368192753499143,
      "learning_rate": 3.787832027640544e-06,
      "loss": 1.0996,
      "step": 2405
    },
    {
      "epoch": 0.17358681144258864,
      "grad_norm": 8.175779383659606,
      "learning_rate": 3.787622492041054e-06,
      "loss": 1.0354,
      "step": 2406
    },
    {
      "epoch": 0.17365895891201616,
      "grad_norm": 5.873673958977225,
      "learning_rate": 3.7874128588261174e-06,
      "loss": 0.9169,
      "step": 2407
    },
    {
      "epoch": 0.17373110638144368,
      "grad_norm": 9.306480561648208,
      "learning_rate": 3.787203128007181e-06,
      "loss": 1.0466,
      "step": 2408
    },
    {
      "epoch": 0.17380325385087117,
      "grad_norm": 8.153030169203436,
      "learning_rate": 3.7869932995956982e-06,
      "loss": 0.85,
      "step": 2409
    },
    {
      "epoch": 0.1738754013202987,
      "grad_norm": 5.540831118050772,
      "learning_rate": 3.786783373603126e-06,
      "loss": 1.0696,
      "step": 2410
    },
    {
      "epoch": 0.1739475487897262,
      "grad_norm": 9.414851010805432,
      "learning_rate": 3.7865733500409286e-06,
      "loss": 0.9934,
      "step": 2411
    },
    {
      "epoch": 0.1740196962591537,
      "grad_norm": 8.935132973678915,
      "learning_rate": 3.7863632289205743e-06,
      "loss": 0.9017,
      "step": 2412
    },
    {
      "epoch": 0.17409184372858122,
      "grad_norm": 5.552809772120457,
      "learning_rate": 3.786153010253537e-06,
      "loss": 1.0232,
      "step": 2413
    },
    {
      "epoch": 0.17416399119800874,
      "grad_norm": 4.844401850984623,
      "learning_rate": 3.7859426940512956e-06,
      "loss": 0.9278,
      "step": 2414
    },
    {
      "epoch": 0.17423613866743623,
      "grad_norm": 4.6249352527931995,
      "learning_rate": 3.785732280325335e-06,
      "loss": 0.9517,
      "step": 2415
    },
    {
      "epoch": 0.17430828613686375,
      "grad_norm": 4.790885043318303,
      "learning_rate": 3.785521769087145e-06,
      "loss": 1.0048,
      "step": 2416
    },
    {
      "epoch": 0.17438043360629127,
      "grad_norm": 4.612627130012499,
      "learning_rate": 3.785311160348222e-06,
      "loss": 1.0289,
      "step": 2417
    },
    {
      "epoch": 0.17445258107571876,
      "grad_norm": 6.260229056070765,
      "learning_rate": 3.7851004541200645e-06,
      "loss": 1.0562,
      "step": 2418
    },
    {
      "epoch": 0.17452472854514628,
      "grad_norm": 16.695253092723036,
      "learning_rate": 3.78488965041418e-06,
      "loss": 1.0199,
      "step": 2419
    },
    {
      "epoch": 0.1745968760145738,
      "grad_norm": 6.794480316297538,
      "learning_rate": 3.7846787492420785e-06,
      "loss": 0.9756,
      "step": 2420
    },
    {
      "epoch": 0.1746690234840013,
      "grad_norm": 5.485644766815514,
      "learning_rate": 3.7844677506152776e-06,
      "loss": 1.0565,
      "step": 2421
    },
    {
      "epoch": 0.1747411709534288,
      "grad_norm": 10.31775656157225,
      "learning_rate": 3.7842566545452982e-06,
      "loss": 0.9991,
      "step": 2422
    },
    {
      "epoch": 0.17481331842285633,
      "grad_norm": 4.778289277151154,
      "learning_rate": 3.7840454610436685e-06,
      "loss": 1.0289,
      "step": 2423
    },
    {
      "epoch": 0.17488546589228382,
      "grad_norm": 4.312540026492802,
      "learning_rate": 3.7838341701219204e-06,
      "loss": 0.9588,
      "step": 2424
    },
    {
      "epoch": 0.17495761336171134,
      "grad_norm": 9.354145098167876,
      "learning_rate": 3.7836227817915917e-06,
      "loss": 0.8841,
      "step": 2425
    },
    {
      "epoch": 0.17502976083113886,
      "grad_norm": 7.917307483101968,
      "learning_rate": 3.7834112960642254e-06,
      "loss": 1.0247,
      "step": 2426
    },
    {
      "epoch": 0.17510190830056635,
      "grad_norm": 6.430171016714813,
      "learning_rate": 3.7831997129513707e-06,
      "loss": 0.9497,
      "step": 2427
    },
    {
      "epoch": 0.17517405576999387,
      "grad_norm": 17.364618287244863,
      "learning_rate": 3.7829880324645804e-06,
      "loss": 0.993,
      "step": 2428
    },
    {
      "epoch": 0.1752462032394214,
      "grad_norm": 11.232360913969073,
      "learning_rate": 3.7827762546154144e-06,
      "loss": 1.0427,
      "step": 2429
    },
    {
      "epoch": 0.17531835070884888,
      "grad_norm": 8.802201117154501,
      "learning_rate": 3.7825643794154367e-06,
      "loss": 1.0386,
      "step": 2430
    },
    {
      "epoch": 0.1753904981782764,
      "grad_norm": 0.8372003873407714,
      "learning_rate": 3.7823524068762174e-06,
      "loss": 0.9102,
      "step": 2431
    },
    {
      "epoch": 0.17546264564770392,
      "grad_norm": 11.615538325485407,
      "learning_rate": 3.7821403370093303e-06,
      "loss": 0.9831,
      "step": 2432
    },
    {
      "epoch": 0.1755347931171314,
      "grad_norm": 13.217796489694065,
      "learning_rate": 3.781928169826357e-06,
      "loss": 1.0022,
      "step": 2433
    },
    {
      "epoch": 0.17560694058655893,
      "grad_norm": 27.007299496140273,
      "learning_rate": 3.7817159053388828e-06,
      "loss": 0.8993,
      "step": 2434
    },
    {
      "epoch": 0.17567908805598645,
      "grad_norm": 4.267986384576533,
      "learning_rate": 3.781503543558499e-06,
      "loss": 0.9845,
      "step": 2435
    },
    {
      "epoch": 0.17575123552541394,
      "grad_norm": 6.721486088964301,
      "learning_rate": 3.7812910844968017e-06,
      "loss": 1.0164,
      "step": 2436
    },
    {
      "epoch": 0.17582338299484146,
      "grad_norm": 5.632151656200531,
      "learning_rate": 3.7810785281653922e-06,
      "loss": 0.9496,
      "step": 2437
    },
    {
      "epoch": 0.17589553046426898,
      "grad_norm": 9.4842080819589,
      "learning_rate": 3.7808658745758775e-06,
      "loss": 1.0085,
      "step": 2438
    },
    {
      "epoch": 0.17596767793369647,
      "grad_norm": 12.79977037700683,
      "learning_rate": 3.7806531237398697e-06,
      "loss": 1.046,
      "step": 2439
    },
    {
      "epoch": 0.176039825403124,
      "grad_norm": 15.198972787781118,
      "learning_rate": 3.780440275668987e-06,
      "loss": 0.9333,
      "step": 2440
    },
    {
      "epoch": 0.1761119728725515,
      "grad_norm": 6.9236818456868,
      "learning_rate": 3.780227330374852e-06,
      "loss": 1.0307,
      "step": 2441
    },
    {
      "epoch": 0.176184120341979,
      "grad_norm": 10.633985601846254,
      "learning_rate": 3.780014287869092e-06,
      "loss": 0.917,
      "step": 2442
    },
    {
      "epoch": 0.17625626781140652,
      "grad_norm": 26.954646696175132,
      "learning_rate": 3.7798011481633416e-06,
      "loss": 1.0045,
      "step": 2443
    },
    {
      "epoch": 0.176328415280834,
      "grad_norm": 6.16343772825904,
      "learning_rate": 3.779587911269239e-06,
      "loss": 1.0301,
      "step": 2444
    },
    {
      "epoch": 0.17640056275026153,
      "grad_norm": 5.480719497393244,
      "learning_rate": 3.7793745771984282e-06,
      "loss": 0.8586,
      "step": 2445
    },
    {
      "epoch": 0.17647271021968905,
      "grad_norm": 6.340898145241735,
      "learning_rate": 3.7791611459625587e-06,
      "loss": 0.9876,
      "step": 2446
    },
    {
      "epoch": 0.17654485768911654,
      "grad_norm": 13.346554749508977,
      "learning_rate": 3.7789476175732852e-06,
      "loss": 0.9404,
      "step": 2447
    },
    {
      "epoch": 0.17661700515854406,
      "grad_norm": 27.382901658200073,
      "learning_rate": 3.7787339920422686e-06,
      "loss": 0.9604,
      "step": 2448
    },
    {
      "epoch": 0.17668915262797158,
      "grad_norm": 4.7132453389746445,
      "learning_rate": 3.7785202693811725e-06,
      "loss": 0.9441,
      "step": 2449
    },
    {
      "epoch": 0.17676130009739907,
      "grad_norm": 8.418005002822168,
      "learning_rate": 3.778306449601668e-06,
      "loss": 1.0157,
      "step": 2450
    },
    {
      "epoch": 0.1768334475668266,
      "grad_norm": 9.223334152208777,
      "learning_rate": 3.7780925327154323e-06,
      "loss": 1.0094,
      "step": 2451
    },
    {
      "epoch": 0.1769055950362541,
      "grad_norm": 11.130885935274945,
      "learning_rate": 3.7778785187341456e-06,
      "loss": 0.8663,
      "step": 2452
    },
    {
      "epoch": 0.1769777425056816,
      "grad_norm": 6.674427125043,
      "learning_rate": 3.7776644076694936e-06,
      "loss": 0.8977,
      "step": 2453
    },
    {
      "epoch": 0.17704988997510912,
      "grad_norm": 0.7044493600609986,
      "learning_rate": 3.7774501995331693e-06,
      "loss": 0.8485,
      "step": 2454
    },
    {
      "epoch": 0.17712203744453664,
      "grad_norm": 5.802208762693644,
      "learning_rate": 3.77723589433687e-06,
      "loss": 0.9901,
      "step": 2455
    },
    {
      "epoch": 0.17719418491396413,
      "grad_norm": 6.529070768968673,
      "learning_rate": 3.7770214920922974e-06,
      "loss": 0.9126,
      "step": 2456
    },
    {
      "epoch": 0.17726633238339165,
      "grad_norm": 7.352217577166902,
      "learning_rate": 3.776806992811159e-06,
      "loss": 1.1235,
      "step": 2457
    },
    {
      "epoch": 0.17733847985281917,
      "grad_norm": 10.584794031300685,
      "learning_rate": 3.776592396505169e-06,
      "loss": 0.9454,
      "step": 2458
    },
    {
      "epoch": 0.17741062732224666,
      "grad_norm": 20.62086103119772,
      "learning_rate": 3.7763777031860445e-06,
      "loss": 1.0056,
      "step": 2459
    },
    {
      "epoch": 0.17748277479167418,
      "grad_norm": 5.207963955814544,
      "learning_rate": 3.7761629128655094e-06,
      "loss": 0.9641,
      "step": 2460
    },
    {
      "epoch": 0.1775549222611017,
      "grad_norm": 7.904952474278733,
      "learning_rate": 3.7759480255552934e-06,
      "loss": 1.0404,
      "step": 2461
    },
    {
      "epoch": 0.1776270697305292,
      "grad_norm": 0.7244538502734283,
      "learning_rate": 3.775733041267129e-06,
      "loss": 0.8106,
      "step": 2462
    },
    {
      "epoch": 0.1776992171999567,
      "grad_norm": 6.9103411816336555,
      "learning_rate": 3.7755179600127576e-06,
      "loss": 0.9656,
      "step": 2463
    },
    {
      "epoch": 0.17777136466938423,
      "grad_norm": 7.149648761791882,
      "learning_rate": 3.7753027818039227e-06,
      "loss": 1.048,
      "step": 2464
    },
    {
      "epoch": 0.17784351213881172,
      "grad_norm": 5.748960981617814,
      "learning_rate": 3.775087506652375e-06,
      "loss": 0.9959,
      "step": 2465
    },
    {
      "epoch": 0.17791565960823924,
      "grad_norm": 8.026350018327086,
      "learning_rate": 3.7748721345698698e-06,
      "loss": 1.0312,
      "step": 2466
    },
    {
      "epoch": 0.17798780707766676,
      "grad_norm": 14.635635551725443,
      "learning_rate": 3.7746566655681675e-06,
      "loss": 1.0058,
      "step": 2467
    },
    {
      "epoch": 0.17805995454709425,
      "grad_norm": 12.055043183751328,
      "learning_rate": 3.7744410996590348e-06,
      "loss": 1.0688,
      "step": 2468
    },
    {
      "epoch": 0.17813210201652177,
      "grad_norm": 12.012407246528287,
      "learning_rate": 3.7742254368542418e-06,
      "loss": 1.1161,
      "step": 2469
    },
    {
      "epoch": 0.1782042494859493,
      "grad_norm": 8.121508508378241,
      "learning_rate": 3.7740096771655658e-06,
      "loss": 0.9671,
      "step": 2470
    },
    {
      "epoch": 0.17827639695537678,
      "grad_norm": 9.169247720492976,
      "learning_rate": 3.773793820604789e-06,
      "loss": 1.0945,
      "step": 2471
    },
    {
      "epoch": 0.1783485444248043,
      "grad_norm": 6.410528178004999,
      "learning_rate": 3.773577867183697e-06,
      "loss": 0.9066,
      "step": 2472
    },
    {
      "epoch": 0.17842069189423182,
      "grad_norm": 9.8031198167772,
      "learning_rate": 3.7733618169140842e-06,
      "loss": 0.9919,
      "step": 2473
    },
    {
      "epoch": 0.1784928393636593,
      "grad_norm": 6.637919370689685,
      "learning_rate": 3.7731456698077466e-06,
      "loss": 1.0458,
      "step": 2474
    },
    {
      "epoch": 0.17856498683308683,
      "grad_norm": 5.315291804344521,
      "learning_rate": 3.772929425876488e-06,
      "loss": 1.0266,
      "step": 2475
    },
    {
      "epoch": 0.17863713430251435,
      "grad_norm": 5.99625406951798,
      "learning_rate": 3.7727130851321166e-06,
      "loss": 1.0241,
      "step": 2476
    },
    {
      "epoch": 0.17870928177194184,
      "grad_norm": 6.201133119957638,
      "learning_rate": 3.7724966475864468e-06,
      "loss": 1.0037,
      "step": 2477
    },
    {
      "epoch": 0.17878142924136936,
      "grad_norm": 11.404708131018126,
      "learning_rate": 3.772280113251296e-06,
      "loss": 1.0019,
      "step": 2478
    },
    {
      "epoch": 0.17885357671079688,
      "grad_norm": 7.217894928727381,
      "learning_rate": 3.772063482138489e-06,
      "loss": 0.9104,
      "step": 2479
    },
    {
      "epoch": 0.17892572418022437,
      "grad_norm": 7.349262133040737,
      "learning_rate": 3.7718467542598553e-06,
      "loss": 1.0052,
      "step": 2480
    },
    {
      "epoch": 0.1789978716496519,
      "grad_norm": 15.91519147551007,
      "learning_rate": 3.7716299296272292e-06,
      "loss": 1.0032,
      "step": 2481
    },
    {
      "epoch": 0.1790700191190794,
      "grad_norm": 11.261031104352883,
      "learning_rate": 3.7714130082524516e-06,
      "loss": 1.0077,
      "step": 2482
    },
    {
      "epoch": 0.1791421665885069,
      "grad_norm": 8.986493890763542,
      "learning_rate": 3.771195990147367e-06,
      "loss": 1.0028,
      "step": 2483
    },
    {
      "epoch": 0.17921431405793442,
      "grad_norm": 14.807678210454704,
      "learning_rate": 3.7709788753238255e-06,
      "loss": 0.9771,
      "step": 2484
    },
    {
      "epoch": 0.17928646152736194,
      "grad_norm": 10.53479560005608,
      "learning_rate": 3.770761663793684e-06,
      "loss": 0.9085,
      "step": 2485
    },
    {
      "epoch": 0.17935860899678943,
      "grad_norm": 11.485036651858419,
      "learning_rate": 3.770544355568803e-06,
      "loss": 1.1245,
      "step": 2486
    },
    {
      "epoch": 0.17943075646621695,
      "grad_norm": 8.373610651724839,
      "learning_rate": 3.770326950661049e-06,
      "loss": 1.0048,
      "step": 2487
    },
    {
      "epoch": 0.17950290393564447,
      "grad_norm": 4.894828379411438,
      "learning_rate": 3.7701094490822943e-06,
      "loss": 0.9229,
      "step": 2488
    },
    {
      "epoch": 0.17957505140507196,
      "grad_norm": 5.43499996546769,
      "learning_rate": 3.769891850844415e-06,
      "loss": 1.0137,
      "step": 2489
    },
    {
      "epoch": 0.17964719887449948,
      "grad_norm": 8.629752356378592,
      "learning_rate": 3.7696741559592937e-06,
      "loss": 1.0245,
      "step": 2490
    },
    {
      "epoch": 0.179719346343927,
      "grad_norm": 10.16732118928165,
      "learning_rate": 3.7694563644388175e-06,
      "loss": 0.9561,
      "step": 2491
    },
    {
      "epoch": 0.1797914938133545,
      "grad_norm": 9.152350418798253,
      "learning_rate": 3.76923847629488e-06,
      "loss": 0.8775,
      "step": 2492
    },
    {
      "epoch": 0.179863641282782,
      "grad_norm": 7.4682831558463585,
      "learning_rate": 3.7690204915393784e-06,
      "loss": 0.9349,
      "step": 2493
    },
    {
      "epoch": 0.17993578875220953,
      "grad_norm": 4.482518679169831,
      "learning_rate": 3.7688024101842173e-06,
      "loss": 1.0011,
      "step": 2494
    },
    {
      "epoch": 0.18000793622163702,
      "grad_norm": 7.8623480510735915,
      "learning_rate": 3.7685842322413034e-06,
      "loss": 1.1506,
      "step": 2495
    },
    {
      "epoch": 0.18008008369106454,
      "grad_norm": 8.453611462146204,
      "learning_rate": 3.768365957722552e-06,
      "loss": 1.0221,
      "step": 2496
    },
    {
      "epoch": 0.18015223116049203,
      "grad_norm": 13.91611581652529,
      "learning_rate": 3.768147586639882e-06,
      "loss": 1.0007,
      "step": 2497
    },
    {
      "epoch": 0.18022437862991955,
      "grad_norm": 14.568521588996685,
      "learning_rate": 3.7679291190052175e-06,
      "loss": 0.9433,
      "step": 2498
    },
    {
      "epoch": 0.18029652609934707,
      "grad_norm": 7.914444487711836,
      "learning_rate": 3.767710554830489e-06,
      "loss": 1.0486,
      "step": 2499
    },
    {
      "epoch": 0.18036867356877456,
      "grad_norm": 9.00729074811587,
      "learning_rate": 3.767491894127631e-06,
      "loss": 1.0493,
      "step": 2500
    },
    {
      "epoch": 0.18044082103820208,
      "grad_norm": 13.406468529687452,
      "learning_rate": 3.7672731369085833e-06,
      "loss": 0.8883,
      "step": 2501
    },
    {
      "epoch": 0.1805129685076296,
      "grad_norm": 7.5587519666096075,
      "learning_rate": 3.767054283185292e-06,
      "loss": 1.0067,
      "step": 2502
    },
    {
      "epoch": 0.1805851159770571,
      "grad_norm": 27.481953195958322,
      "learning_rate": 3.766835332969707e-06,
      "loss": 1.0586,
      "step": 2503
    },
    {
      "epoch": 0.1806572634464846,
      "grad_norm": 7.473561289450162,
      "learning_rate": 3.7666162862737857e-06,
      "loss": 1.0229,
      "step": 2504
    },
    {
      "epoch": 0.18072941091591213,
      "grad_norm": 8.704415299325108,
      "learning_rate": 3.766397143109489e-06,
      "loss": 1.0075,
      "step": 2505
    },
    {
      "epoch": 0.18080155838533962,
      "grad_norm": 7.074163477748596,
      "learning_rate": 3.766177903488783e-06,
      "loss": 0.9621,
      "step": 2506
    },
    {
      "epoch": 0.18087370585476714,
      "grad_norm": 10.475528720085304,
      "learning_rate": 3.76595856742364e-06,
      "loss": 0.9974,
      "step": 2507
    },
    {
      "epoch": 0.18094585332419466,
      "grad_norm": 6.2827605144073795,
      "learning_rate": 3.7657391349260362e-06,
      "loss": 1.1288,
      "step": 2508
    },
    {
      "epoch": 0.18101800079362215,
      "grad_norm": 7.155425578109063,
      "learning_rate": 3.765519606007956e-06,
      "loss": 1.0605,
      "step": 2509
    },
    {
      "epoch": 0.18109014826304967,
      "grad_norm": 6.472661554199697,
      "learning_rate": 3.7652999806813856e-06,
      "loss": 0.9893,
      "step": 2510
    },
    {
      "epoch": 0.1811622957324772,
      "grad_norm": 9.316299380283269,
      "learning_rate": 3.7650802589583173e-06,
      "loss": 0.9556,
      "step": 2511
    },
    {
      "epoch": 0.18123444320190468,
      "grad_norm": 6.649513811967524,
      "learning_rate": 3.764860440850751e-06,
      "loss": 0.9643,
      "step": 2512
    },
    {
      "epoch": 0.1813065906713322,
      "grad_norm": 5.644234218503877,
      "learning_rate": 3.764640526370689e-06,
      "loss": 1.025,
      "step": 2513
    },
    {
      "epoch": 0.18137873814075972,
      "grad_norm": 9.682104958794108,
      "learning_rate": 3.7644205155301404e-06,
      "loss": 0.9823,
      "step": 2514
    },
    {
      "epoch": 0.1814508856101872,
      "grad_norm": 4.901178350647272,
      "learning_rate": 3.764200408341119e-06,
      "loss": 0.9937,
      "step": 2515
    },
    {
      "epoch": 0.18152303307961473,
      "grad_norm": 7.828337706457092,
      "learning_rate": 3.7639802048156445e-06,
      "loss": 0.9716,
      "step": 2516
    },
    {
      "epoch": 0.18159518054904225,
      "grad_norm": 5.547740428196202,
      "learning_rate": 3.763759904965741e-06,
      "loss": 1.0337,
      "step": 2517
    },
    {
      "epoch": 0.18166732801846974,
      "grad_norm": 6.9838577522744485,
      "learning_rate": 3.7635395088034376e-06,
      "loss": 1.039,
      "step": 2518
    },
    {
      "epoch": 0.18173947548789726,
      "grad_norm": 5.592537860494941,
      "learning_rate": 3.7633190163407702e-06,
      "loss": 1.0216,
      "step": 2519
    },
    {
      "epoch": 0.18181162295732478,
      "grad_norm": 6.705235431358202,
      "learning_rate": 3.7630984275897796e-06,
      "loss": 0.9775,
      "step": 2520
    },
    {
      "epoch": 0.18188377042675227,
      "grad_norm": 7.481182142059889,
      "learning_rate": 3.7628777425625103e-06,
      "loss": 0.9459,
      "step": 2521
    },
    {
      "epoch": 0.1819559178961798,
      "grad_norm": 6.997374450931075,
      "learning_rate": 3.762656961271013e-06,
      "loss": 1.0012,
      "step": 2522
    },
    {
      "epoch": 0.1820280653656073,
      "grad_norm": 6.6495536826517005,
      "learning_rate": 3.762436083727345e-06,
      "loss": 0.8173,
      "step": 2523
    },
    {
      "epoch": 0.1821002128350348,
      "grad_norm": 12.399743821974143,
      "learning_rate": 3.7622151099435666e-06,
      "loss": 0.9887,
      "step": 2524
    },
    {
      "epoch": 0.18217236030446232,
      "grad_norm": 9.946879249257849,
      "learning_rate": 3.7619940399317437e-06,
      "loss": 1.0678,
      "step": 2525
    },
    {
      "epoch": 0.18224450777388984,
      "grad_norm": 4.2426815974629974,
      "learning_rate": 3.7617728737039495e-06,
      "loss": 1.0091,
      "step": 2526
    },
    {
      "epoch": 0.18231665524331733,
      "grad_norm": 10.775920756002638,
      "learning_rate": 3.761551611272261e-06,
      "loss": 1.117,
      "step": 2527
    },
    {
      "epoch": 0.18238880271274485,
      "grad_norm": 0.6873052060991712,
      "learning_rate": 3.761330252648759e-06,
      "loss": 0.7898,
      "step": 2528
    },
    {
      "epoch": 0.18246095018217237,
      "grad_norm": 7.009615969076733,
      "learning_rate": 3.7611087978455328e-06,
      "loss": 0.9532,
      "step": 2529
    },
    {
      "epoch": 0.18253309765159986,
      "grad_norm": 8.351557932353755,
      "learning_rate": 3.7608872468746746e-06,
      "loss": 1.0433,
      "step": 2530
    },
    {
      "epoch": 0.18260524512102738,
      "grad_norm": 6.146287321575153,
      "learning_rate": 3.7606655997482826e-06,
      "loss": 1.0107,
      "step": 2531
    },
    {
      "epoch": 0.1826773925904549,
      "grad_norm": 7.195262399306438,
      "learning_rate": 3.7604438564784596e-06,
      "loss": 1.0357,
      "step": 2532
    },
    {
      "epoch": 0.1827495400598824,
      "grad_norm": 7.761964238301406,
      "learning_rate": 3.7602220170773144e-06,
      "loss": 0.9779,
      "step": 2533
    },
    {
      "epoch": 0.1828216875293099,
      "grad_norm": 9.80469800484995,
      "learning_rate": 3.760000081556961e-06,
      "loss": 0.9486,
      "step": 2534
    },
    {
      "epoch": 0.18289383499873743,
      "grad_norm": 0.7204818592686545,
      "learning_rate": 3.759778049929519e-06,
      "loss": 0.7982,
      "step": 2535
    },
    {
      "epoch": 0.18296598246816492,
      "grad_norm": 12.310433306284413,
      "learning_rate": 3.759555922207112e-06,
      "loss": 1.0905,
      "step": 2536
    },
    {
      "epoch": 0.18303812993759244,
      "grad_norm": 8.098506045617206,
      "learning_rate": 3.7593336984018695e-06,
      "loss": 1.0058,
      "step": 2537
    },
    {
      "epoch": 0.18311027740701996,
      "grad_norm": 4.755766931532249,
      "learning_rate": 3.7591113785259263e-06,
      "loss": 0.9763,
      "step": 2538
    },
    {
      "epoch": 0.18318242487644745,
      "grad_norm": 12.296072753174778,
      "learning_rate": 3.758888962591423e-06,
      "loss": 0.9571,
      "step": 2539
    },
    {
      "epoch": 0.18325457234587497,
      "grad_norm": 8.797796441666737,
      "learning_rate": 3.7586664506105047e-06,
      "loss": 0.9576,
      "step": 2540
    },
    {
      "epoch": 0.1833267198153025,
      "grad_norm": 7.603202305787633,
      "learning_rate": 3.758443842595322e-06,
      "loss": 0.9437,
      "step": 2541
    },
    {
      "epoch": 0.18339886728472998,
      "grad_norm": 11.296786190543896,
      "learning_rate": 3.7582211385580305e-06,
      "loss": 0.7634,
      "step": 2542
    },
    {
      "epoch": 0.1834710147541575,
      "grad_norm": 11.496697656581038,
      "learning_rate": 3.757998338510791e-06,
      "loss": 0.9117,
      "step": 2543
    },
    {
      "epoch": 0.18354316222358502,
      "grad_norm": 9.975843243821194,
      "learning_rate": 3.75777544246577e-06,
      "loss": 1.0144,
      "step": 2544
    },
    {
      "epoch": 0.1836153096930125,
      "grad_norm": 7.042257822116022,
      "learning_rate": 3.7575524504351402e-06,
      "loss": 0.9954,
      "step": 2545
    },
    {
      "epoch": 0.18368745716244003,
      "grad_norm": 4.827050740748409,
      "learning_rate": 3.7573293624310762e-06,
      "loss": 1.047,
      "step": 2546
    },
    {
      "epoch": 0.18375960463186752,
      "grad_norm": 7.932316807393747,
      "learning_rate": 3.757106178465762e-06,
      "loss": 0.9706,
      "step": 2547
    },
    {
      "epoch": 0.18383175210129504,
      "grad_norm": 0.7622667737637233,
      "learning_rate": 3.7568828985513833e-06,
      "loss": 0.878,
      "step": 2548
    },
    {
      "epoch": 0.18390389957072256,
      "grad_norm": 20.622848491001733,
      "learning_rate": 3.7566595227001336e-06,
      "loss": 1.024,
      "step": 2549
    },
    {
      "epoch": 0.18397604704015005,
      "grad_norm": 7.765686157722583,
      "learning_rate": 3.75643605092421e-06,
      "loss": 0.9818,
      "step": 2550
    },
    {
      "epoch": 0.18404819450957757,
      "grad_norm": 7.650638688297023,
      "learning_rate": 3.7562124832358163e-06,
      "loss": 1.0475,
      "step": 2551
    },
    {
      "epoch": 0.1841203419790051,
      "grad_norm": 6.7265409498057815,
      "learning_rate": 3.75598881964716e-06,
      "loss": 0.9815,
      "step": 2552
    },
    {
      "epoch": 0.18419248944843258,
      "grad_norm": 5.918446698418554,
      "learning_rate": 3.7557650601704544e-06,
      "loss": 1.0046,
      "step": 2553
    },
    {
      "epoch": 0.1842646369178601,
      "grad_norm": 7.5587845179414845,
      "learning_rate": 3.755541204817919e-06,
      "loss": 1.0464,
      "step": 2554
    },
    {
      "epoch": 0.18433678438728762,
      "grad_norm": 4.798413753668143,
      "learning_rate": 3.7553172536017763e-06,
      "loss": 1.0644,
      "step": 2555
    },
    {
      "epoch": 0.1844089318567151,
      "grad_norm": 13.350293308422046,
      "learning_rate": 3.755093206534257e-06,
      "loss": 1.0137,
      "step": 2556
    },
    {
      "epoch": 0.18448107932614263,
      "grad_norm": 7.2113699571534635,
      "learning_rate": 3.7548690636275946e-06,
      "loss": 0.9227,
      "step": 2557
    },
    {
      "epoch": 0.18455322679557015,
      "grad_norm": 15.057574341305617,
      "learning_rate": 3.7546448248940293e-06,
      "loss": 1.0029,
      "step": 2558
    },
    {
      "epoch": 0.18462537426499764,
      "grad_norm": 4.81741854509224,
      "learning_rate": 3.754420490345805e-06,
      "loss": 0.947,
      "step": 2559
    },
    {
      "epoch": 0.18469752173442516,
      "grad_norm": 6.599314289884949,
      "learning_rate": 3.754196059995173e-06,
      "loss": 1.0381,
      "step": 2560
    },
    {
      "epoch": 0.18476966920385268,
      "grad_norm": 11.927561512430463,
      "learning_rate": 3.7539715338543877e-06,
      "loss": 1.0387,
      "step": 2561
    },
    {
      "epoch": 0.18484181667328017,
      "grad_norm": 0.8639094970884962,
      "learning_rate": 3.75374691193571e-06,
      "loss": 0.9212,
      "step": 2562
    },
    {
      "epoch": 0.1849139641427077,
      "grad_norm": 4.8025141568945795,
      "learning_rate": 3.7535221942514053e-06,
      "loss": 1.034,
      "step": 2563
    },
    {
      "epoch": 0.1849861116121352,
      "grad_norm": 0.8102108178297227,
      "learning_rate": 3.7532973808137453e-06,
      "loss": 0.8687,
      "step": 2564
    },
    {
      "epoch": 0.1850582590815627,
      "grad_norm": 13.469342495306122,
      "learning_rate": 3.7530724716350057e-06,
      "loss": 1.0024,
      "step": 2565
    },
    {
      "epoch": 0.18513040655099022,
      "grad_norm": 8.793267223388249,
      "learning_rate": 3.752847466727468e-06,
      "loss": 0.959,
      "step": 2566
    },
    {
      "epoch": 0.18520255402041774,
      "grad_norm": 5.262259021596711,
      "learning_rate": 3.7526223661034184e-06,
      "loss": 1.0542,
      "step": 2567
    },
    {
      "epoch": 0.18527470148984523,
      "grad_norm": 6.503625005649644,
      "learning_rate": 3.75239716977515e-06,
      "loss": 1.0529,
      "step": 2568
    },
    {
      "epoch": 0.18534684895927275,
      "grad_norm": 9.084511590067907,
      "learning_rate": 3.7521718777549594e-06,
      "loss": 0.8874,
      "step": 2569
    },
    {
      "epoch": 0.18541899642870027,
      "grad_norm": 13.309160127000231,
      "learning_rate": 3.7519464900551485e-06,
      "loss": 1.086,
      "step": 2570
    },
    {
      "epoch": 0.18549114389812776,
      "grad_norm": 5.591565094360234,
      "learning_rate": 3.751721006688026e-06,
      "loss": 1.0711,
      "step": 2571
    },
    {
      "epoch": 0.18556329136755528,
      "grad_norm": 4.651550789108117,
      "learning_rate": 3.7514954276659026e-06,
      "loss": 1.0349,
      "step": 2572
    },
    {
      "epoch": 0.1856354388369828,
      "grad_norm": 8.117861854657825,
      "learning_rate": 3.751269753001099e-06,
      "loss": 1.0515,
      "step": 2573
    },
    {
      "epoch": 0.1857075863064103,
      "grad_norm": 6.0416255905684295,
      "learning_rate": 3.7510439827059366e-06,
      "loss": 1.1086,
      "step": 2574
    },
    {
      "epoch": 0.1857797337758378,
      "grad_norm": 21.990226655479816,
      "learning_rate": 3.7508181167927447e-06,
      "loss": 1.0005,
      "step": 2575
    },
    {
      "epoch": 0.18585188124526533,
      "grad_norm": 6.882883995975145,
      "learning_rate": 3.7505921552738565e-06,
      "loss": 1.1445,
      "step": 2576
    },
    {
      "epoch": 0.18592402871469282,
      "grad_norm": 10.110923872484442,
      "learning_rate": 3.750366098161611e-06,
      "loss": 1.018,
      "step": 2577
    },
    {
      "epoch": 0.18599617618412034,
      "grad_norm": 0.9978579288204167,
      "learning_rate": 3.750139945468353e-06,
      "loss": 0.8603,
      "step": 2578
    },
    {
      "epoch": 0.18606832365354786,
      "grad_norm": 9.661082222224891,
      "learning_rate": 3.749913697206431e-06,
      "loss": 0.9773,
      "step": 2579
    },
    {
      "epoch": 0.18614047112297535,
      "grad_norm": 14.434396649691493,
      "learning_rate": 3.7496873533882e-06,
      "loss": 0.9511,
      "step": 2580
    },
    {
      "epoch": 0.18621261859240287,
      "grad_norm": 7.109719203120247,
      "learning_rate": 3.74946091402602e-06,
      "loss": 1.0516,
      "step": 2581
    },
    {
      "epoch": 0.1862847660618304,
      "grad_norm": 15.968509679729088,
      "learning_rate": 3.7492343791322556e-06,
      "loss": 1.0278,
      "step": 2582
    },
    {
      "epoch": 0.18635691353125788,
      "grad_norm": 14.327675582510103,
      "learning_rate": 3.7490077487192773e-06,
      "loss": 1.0509,
      "step": 2583
    },
    {
      "epoch": 0.1864290610006854,
      "grad_norm": 6.861708120437514,
      "learning_rate": 3.7487810227994606e-06,
      "loss": 0.9447,
      "step": 2584
    },
    {
      "epoch": 0.18650120847011292,
      "grad_norm": 5.128819972617818,
      "learning_rate": 3.7485542013851855e-06,
      "loss": 1.0191,
      "step": 2585
    },
    {
      "epoch": 0.18657335593954041,
      "grad_norm": 8.260242461924781,
      "learning_rate": 3.748327284488839e-06,
      "loss": 0.9993,
      "step": 2586
    },
    {
      "epoch": 0.18664550340896793,
      "grad_norm": 12.66658106992477,
      "learning_rate": 3.748100272122811e-06,
      "loss": 1.0057,
      "step": 2587
    },
    {
      "epoch": 0.18671765087839545,
      "grad_norm": 6.745685187306048,
      "learning_rate": 3.7478731642994984e-06,
      "loss": 1.0543,
      "step": 2588
    },
    {
      "epoch": 0.18678979834782294,
      "grad_norm": 9.034096873015873,
      "learning_rate": 3.747645961031303e-06,
      "loss": 0.9178,
      "step": 2589
    },
    {
      "epoch": 0.18686194581725046,
      "grad_norm": 7.7810960624205885,
      "learning_rate": 3.7474186623306317e-06,
      "loss": 1.0088,
      "step": 2590
    },
    {
      "epoch": 0.18693409328667798,
      "grad_norm": 6.853741046093315,
      "learning_rate": 3.747191268209895e-06,
      "loss": 0.9252,
      "step": 2591
    },
    {
      "epoch": 0.18700624075610547,
      "grad_norm": 8.378362635337453,
      "learning_rate": 3.7469637786815117e-06,
      "loss": 0.9903,
      "step": 2592
    },
    {
      "epoch": 0.187078388225533,
      "grad_norm": 6.504401257438953,
      "learning_rate": 3.746736193757903e-06,
      "loss": 1.031,
      "step": 2593
    },
    {
      "epoch": 0.1871505356949605,
      "grad_norm": 10.221737859372185,
      "learning_rate": 3.7465085134514977e-06,
      "loss": 1.0342,
      "step": 2594
    },
    {
      "epoch": 0.187222683164388,
      "grad_norm": 5.472475537645813,
      "learning_rate": 3.746280737774727e-06,
      "loss": 0.971,
      "step": 2595
    },
    {
      "epoch": 0.18729483063381552,
      "grad_norm": 9.121383577434907,
      "learning_rate": 3.7460528667400306e-06,
      "loss": 1.0249,
      "step": 2596
    },
    {
      "epoch": 0.18736697810324304,
      "grad_norm": 8.821088785452726,
      "learning_rate": 3.7458249003598496e-06,
      "loss": 0.9568,
      "step": 2597
    },
    {
      "epoch": 0.18743912557267053,
      "grad_norm": 7.897430929308792,
      "learning_rate": 3.7455968386466346e-06,
      "loss": 0.9746,
      "step": 2598
    },
    {
      "epoch": 0.18751127304209805,
      "grad_norm": 5.195106903826667,
      "learning_rate": 3.745368681612838e-06,
      "loss": 1.008,
      "step": 2599
    },
    {
      "epoch": 0.18758342051152554,
      "grad_norm": 11.944389552106083,
      "learning_rate": 3.7451404292709184e-06,
      "loss": 0.9822,
      "step": 2600
    },
    {
      "epoch": 0.18765556798095306,
      "grad_norm": 79.59209362234103,
      "learning_rate": 3.7449120816333405e-06,
      "loss": 0.9091,
      "step": 2601
    },
    {
      "epoch": 0.18772771545038058,
      "grad_norm": 7.162002050855951,
      "learning_rate": 3.7446836387125737e-06,
      "loss": 0.9458,
      "step": 2602
    },
    {
      "epoch": 0.18779986291980807,
      "grad_norm": 12.27834481780194,
      "learning_rate": 3.7444551005210914e-06,
      "loss": 1.0521,
      "step": 2603
    },
    {
      "epoch": 0.1878720103892356,
      "grad_norm": 8.966908548299923,
      "learning_rate": 3.744226467071374e-06,
      "loss": 1.0034,
      "step": 2604
    },
    {
      "epoch": 0.1879441578586631,
      "grad_norm": 11.923141392232992,
      "learning_rate": 3.743997738375906e-06,
      "loss": 0.9416,
      "step": 2605
    },
    {
      "epoch": 0.1880163053280906,
      "grad_norm": 7.765905977688429,
      "learning_rate": 3.743768914447178e-06,
      "loss": 0.9884,
      "step": 2606
    },
    {
      "epoch": 0.18808845279751812,
      "grad_norm": 0.8821260264508131,
      "learning_rate": 3.7435399952976844e-06,
      "loss": 0.8134,
      "step": 2607
    },
    {
      "epoch": 0.18816060026694564,
      "grad_norm": 0.848519514134834,
      "learning_rate": 3.7433109809399264e-06,
      "loss": 0.9391,
      "step": 2608
    },
    {
      "epoch": 0.18823274773637313,
      "grad_norm": 11.431414877159781,
      "learning_rate": 3.743081871386409e-06,
      "loss": 1.1323,
      "step": 2609
    },
    {
      "epoch": 0.18830489520580065,
      "grad_norm": 8.919948603734312,
      "learning_rate": 3.7428526666496436e-06,
      "loss": 1.0636,
      "step": 2610
    },
    {
      "epoch": 0.18837704267522817,
      "grad_norm": 6.4629655858451205,
      "learning_rate": 3.742623366742145e-06,
      "loss": 1.1021,
      "step": 2611
    },
    {
      "epoch": 0.18844919014465566,
      "grad_norm": 15.165152729194807,
      "learning_rate": 3.742393971676436e-06,
      "loss": 1.0271,
      "step": 2612
    },
    {
      "epoch": 0.18852133761408318,
      "grad_norm": 10.236962190877767,
      "learning_rate": 3.7421644814650424e-06,
      "loss": 1.0097,
      "step": 2613
    },
    {
      "epoch": 0.1885934850835107,
      "grad_norm": 16.842188152310033,
      "learning_rate": 3.7419348961204956e-06,
      "loss": 0.9402,
      "step": 2614
    },
    {
      "epoch": 0.1886656325529382,
      "grad_norm": 9.247503407671843,
      "learning_rate": 3.7417052156553325e-06,
      "loss": 0.915,
      "step": 2615
    },
    {
      "epoch": 0.18873778002236571,
      "grad_norm": 17.836070958764818,
      "learning_rate": 3.7414754400820952e-06,
      "loss": 1.04,
      "step": 2616
    },
    {
      "epoch": 0.18880992749179323,
      "grad_norm": 6.815622716428434,
      "learning_rate": 3.7412455694133308e-06,
      "loss": 1.0179,
      "step": 2617
    },
    {
      "epoch": 0.18888207496122073,
      "grad_norm": 8.543781397105349,
      "learning_rate": 3.741015603661592e-06,
      "loss": 0.9719,
      "step": 2618
    },
    {
      "epoch": 0.18895422243064824,
      "grad_norm": 27.09297955219442,
      "learning_rate": 3.740785542839435e-06,
      "loss": 1.0625,
      "step": 2619
    },
    {
      "epoch": 0.18902636990007576,
      "grad_norm": 7.10435573591042,
      "learning_rate": 3.740555386959425e-06,
      "loss": 0.9865,
      "step": 2620
    },
    {
      "epoch": 0.18909851736950326,
      "grad_norm": 11.415660581786526,
      "learning_rate": 3.740325136034128e-06,
      "loss": 1.0163,
      "step": 2621
    },
    {
      "epoch": 0.18917066483893077,
      "grad_norm": 11.246017429295561,
      "learning_rate": 3.740094790076118e-06,
      "loss": 0.9308,
      "step": 2622
    },
    {
      "epoch": 0.1892428123083583,
      "grad_norm": 6.3590498707581355,
      "learning_rate": 3.7398643490979735e-06,
      "loss": 1.0625,
      "step": 2623
    },
    {
      "epoch": 0.18931495977778579,
      "grad_norm": 7.865100754312578,
      "learning_rate": 3.7396338131122762e-06,
      "loss": 1.0462,
      "step": 2624
    },
    {
      "epoch": 0.1893871072472133,
      "grad_norm": 10.392276581029831,
      "learning_rate": 3.7394031821316175e-06,
      "loss": 0.986,
      "step": 2625
    },
    {
      "epoch": 0.18945925471664082,
      "grad_norm": 14.03296105501899,
      "learning_rate": 3.7391724561685902e-06,
      "loss": 0.9826,
      "step": 2626
    },
    {
      "epoch": 0.18953140218606832,
      "grad_norm": 5.627594921475756,
      "learning_rate": 3.738941635235793e-06,
      "loss": 0.8876,
      "step": 2627
    },
    {
      "epoch": 0.18960354965549583,
      "grad_norm": 6.451490292153131,
      "learning_rate": 3.73871071934583e-06,
      "loss": 0.9434,
      "step": 2628
    },
    {
      "epoch": 0.18967569712492335,
      "grad_norm": 18.077156007954834,
      "learning_rate": 3.7384797085113114e-06,
      "loss": 1.0143,
      "step": 2629
    },
    {
      "epoch": 0.18974784459435085,
      "grad_norm": 0.8566998893947892,
      "learning_rate": 3.7382486027448515e-06,
      "loss": 0.8871,
      "step": 2630
    },
    {
      "epoch": 0.18981999206377836,
      "grad_norm": 9.33687397238347,
      "learning_rate": 3.7380174020590707e-06,
      "loss": 0.9653,
      "step": 2631
    },
    {
      "epoch": 0.18989213953320588,
      "grad_norm": 5.1662616981209615,
      "learning_rate": 3.737786106466593e-06,
      "loss": 0.9625,
      "step": 2632
    },
    {
      "epoch": 0.18996428700263338,
      "grad_norm": 6.237893548143376,
      "learning_rate": 3.737554715980049e-06,
      "loss": 1.0159,
      "step": 2633
    },
    {
      "epoch": 0.1900364344720609,
      "grad_norm": 7.646433239372438,
      "learning_rate": 3.737323230612074e-06,
      "loss": 1.005,
      "step": 2634
    },
    {
      "epoch": 0.1901085819414884,
      "grad_norm": 15.742640637864083,
      "learning_rate": 3.7370916503753095e-06,
      "loss": 1.0107,
      "step": 2635
    },
    {
      "epoch": 0.1901807294109159,
      "grad_norm": 8.745635769887244,
      "learning_rate": 3.7368599752824e-06,
      "loss": 1.0074,
      "step": 2636
    },
    {
      "epoch": 0.19025287688034342,
      "grad_norm": 0.8026072826155856,
      "learning_rate": 3.736628205345997e-06,
      "loss": 0.864,
      "step": 2637
    },
    {
      "epoch": 0.19032502434977094,
      "grad_norm": 14.014562390041831,
      "learning_rate": 3.736396340578756e-06,
      "loss": 0.9515,
      "step": 2638
    },
    {
      "epoch": 0.19039717181919844,
      "grad_norm": 5.777247347488465,
      "learning_rate": 3.7361643809933395e-06,
      "loss": 0.9939,
      "step": 2639
    },
    {
      "epoch": 0.19046931928862595,
      "grad_norm": 12.479990087651938,
      "learning_rate": 3.7359323266024126e-06,
      "loss": 0.9998,
      "step": 2640
    },
    {
      "epoch": 0.19054146675805347,
      "grad_norm": 9.526209764924955,
      "learning_rate": 3.735700177418648e-06,
      "loss": 0.982,
      "step": 2641
    },
    {
      "epoch": 0.19061361422748097,
      "grad_norm": 8.579355398297297,
      "learning_rate": 3.735467933454722e-06,
      "loss": 0.9922,
      "step": 2642
    },
    {
      "epoch": 0.19068576169690848,
      "grad_norm": 6.03206743620557,
      "learning_rate": 3.7352355947233166e-06,
      "loss": 0.9461,
      "step": 2643
    },
    {
      "epoch": 0.190757909166336,
      "grad_norm": 7.600366041503774,
      "learning_rate": 3.7350031612371186e-06,
      "loss": 0.9154,
      "step": 2644
    },
    {
      "epoch": 0.1908300566357635,
      "grad_norm": 6.1696748699172375,
      "learning_rate": 3.7347706330088213e-06,
      "loss": 1.0805,
      "step": 2645
    },
    {
      "epoch": 0.19090220410519101,
      "grad_norm": 4.449019832032704,
      "learning_rate": 3.7345380100511214e-06,
      "loss": 1.0232,
      "step": 2646
    },
    {
      "epoch": 0.19097435157461853,
      "grad_norm": 10.178889280629036,
      "learning_rate": 3.734305292376722e-06,
      "loss": 1.008,
      "step": 2647
    },
    {
      "epoch": 0.19104649904404603,
      "grad_norm": 7.495087604277937,
      "learning_rate": 3.73407247999833e-06,
      "loss": 0.9754,
      "step": 2648
    },
    {
      "epoch": 0.19111864651347354,
      "grad_norm": 6.7005533473835435,
      "learning_rate": 3.7338395729286595e-06,
      "loss": 1.016,
      "step": 2649
    },
    {
      "epoch": 0.19119079398290104,
      "grad_norm": 19.15067258604895,
      "learning_rate": 3.733606571180429e-06,
      "loss": 1.0805,
      "step": 2650
    },
    {
      "epoch": 0.19126294145232856,
      "grad_norm": 5.7778316805028735,
      "learning_rate": 3.7333734747663605e-06,
      "loss": 1.1564,
      "step": 2651
    },
    {
      "epoch": 0.19133508892175607,
      "grad_norm": 7.4068899401304105,
      "learning_rate": 3.7331402836991838e-06,
      "loss": 0.9737,
      "step": 2652
    },
    {
      "epoch": 0.19140723639118357,
      "grad_norm": 7.953137950474476,
      "learning_rate": 3.7329069979916322e-06,
      "loss": 0.9079,
      "step": 2653
    },
    {
      "epoch": 0.19147938386061109,
      "grad_norm": 11.283838202386487,
      "learning_rate": 3.732673617656444e-06,
      "loss": 0.9772,
      "step": 2654
    },
    {
      "epoch": 0.1915515313300386,
      "grad_norm": 9.231146904970819,
      "learning_rate": 3.7324401427063634e-06,
      "loss": 1.0572,
      "step": 2655
    },
    {
      "epoch": 0.1916236787994661,
      "grad_norm": 16.67043068661826,
      "learning_rate": 3.7322065731541406e-06,
      "loss": 0.8863,
      "step": 2656
    },
    {
      "epoch": 0.19169582626889362,
      "grad_norm": 8.744742421565729,
      "learning_rate": 3.7319729090125288e-06,
      "loss": 1.0522,
      "step": 2657
    },
    {
      "epoch": 0.19176797373832113,
      "grad_norm": 21.57055301387099,
      "learning_rate": 3.731739150294288e-06,
      "loss": 0.94,
      "step": 2658
    },
    {
      "epoch": 0.19184012120774863,
      "grad_norm": 5.05093532247602,
      "learning_rate": 3.731505297012183e-06,
      "loss": 1.0698,
      "step": 2659
    },
    {
      "epoch": 0.19191226867717615,
      "grad_norm": 12.543549625486051,
      "learning_rate": 3.731271349178984e-06,
      "loss": 0.9249,
      "step": 2660
    },
    {
      "epoch": 0.19198441614660366,
      "grad_norm": 7.420507429514144,
      "learning_rate": 3.7310373068074646e-06,
      "loss": 0.9748,
      "step": 2661
    },
    {
      "epoch": 0.19205656361603116,
      "grad_norm": 9.260324438355996,
      "learning_rate": 3.730803169910406e-06,
      "loss": 0.8437,
      "step": 2662
    },
    {
      "epoch": 0.19212871108545868,
      "grad_norm": 6.20116910679072,
      "learning_rate": 3.7305689385005945e-06,
      "loss": 0.9778,
      "step": 2663
    },
    {
      "epoch": 0.1922008585548862,
      "grad_norm": 9.469809494922123,
      "learning_rate": 3.730334612590819e-06,
      "loss": 0.9878,
      "step": 2664
    },
    {
      "epoch": 0.1922730060243137,
      "grad_norm": 0.7670740311690755,
      "learning_rate": 3.730100192193876e-06,
      "loss": 0.922,
      "step": 2665
    },
    {
      "epoch": 0.1923451534937412,
      "grad_norm": 0.8085024869711896,
      "learning_rate": 3.729865677322566e-06,
      "loss": 0.8516,
      "step": 2666
    },
    {
      "epoch": 0.19241730096316872,
      "grad_norm": 10.990095013944481,
      "learning_rate": 3.729631067989695e-06,
      "loss": 1.0528,
      "step": 2667
    },
    {
      "epoch": 0.19248944843259622,
      "grad_norm": 6.970352057126883,
      "learning_rate": 3.7293963642080747e-06,
      "loss": 1.1344,
      "step": 2668
    },
    {
      "epoch": 0.19256159590202374,
      "grad_norm": 16.499376689812976,
      "learning_rate": 3.72916156599052e-06,
      "loss": 1.0453,
      "step": 2669
    },
    {
      "epoch": 0.19263374337145125,
      "grad_norm": 22.48507330938718,
      "learning_rate": 3.7289266733498543e-06,
      "loss": 0.9961,
      "step": 2670
    },
    {
      "epoch": 0.19270589084087875,
      "grad_norm": 5.158070560761072,
      "learning_rate": 3.7286916862989024e-06,
      "loss": 1.0453,
      "step": 2671
    },
    {
      "epoch": 0.19277803831030627,
      "grad_norm": 8.10706827214612,
      "learning_rate": 3.7284566048504974e-06,
      "loss": 1.0221,
      "step": 2672
    },
    {
      "epoch": 0.19285018577973378,
      "grad_norm": 8.688532706155712,
      "learning_rate": 3.7282214290174763e-06,
      "loss": 0.895,
      "step": 2673
    },
    {
      "epoch": 0.19292233324916128,
      "grad_norm": 12.913957340391562,
      "learning_rate": 3.72798615881268e-06,
      "loss": 1.0333,
      "step": 2674
    },
    {
      "epoch": 0.1929944807185888,
      "grad_norm": 18.791199040427227,
      "learning_rate": 3.727750794248956e-06,
      "loss": 1.0051,
      "step": 2675
    },
    {
      "epoch": 0.19306662818801631,
      "grad_norm": 12.69824785776473,
      "learning_rate": 3.7275153353391576e-06,
      "loss": 0.9974,
      "step": 2676
    },
    {
      "epoch": 0.1931387756574438,
      "grad_norm": 5.741816128509627,
      "learning_rate": 3.7272797820961412e-06,
      "loss": 0.9047,
      "step": 2677
    },
    {
      "epoch": 0.19321092312687133,
      "grad_norm": 12.813154994690759,
      "learning_rate": 3.727044134532771e-06,
      "loss": 0.9071,
      "step": 2678
    },
    {
      "epoch": 0.19328307059629884,
      "grad_norm": 8.367396304875943,
      "learning_rate": 3.7268083926619126e-06,
      "loss": 0.8805,
      "step": 2679
    },
    {
      "epoch": 0.19335521806572634,
      "grad_norm": 12.883793762849233,
      "learning_rate": 3.7265725564964414e-06,
      "loss": 1.0168,
      "step": 2680
    },
    {
      "epoch": 0.19342736553515386,
      "grad_norm": 15.65815622751646,
      "learning_rate": 3.7263366260492345e-06,
      "loss": 0.9147,
      "step": 2681
    },
    {
      "epoch": 0.19349951300458137,
      "grad_norm": 5.784608927097244,
      "learning_rate": 3.726100601333174e-06,
      "loss": 1.0834,
      "step": 2682
    },
    {
      "epoch": 0.19357166047400887,
      "grad_norm": 5.857619854315063,
      "learning_rate": 3.7258644823611506e-06,
      "loss": 1.0493,
      "step": 2683
    },
    {
      "epoch": 0.19364380794343639,
      "grad_norm": 23.177091087311435,
      "learning_rate": 3.725628269146056e-06,
      "loss": 1.0256,
      "step": 2684
    },
    {
      "epoch": 0.1937159554128639,
      "grad_norm": 8.892264323957589,
      "learning_rate": 3.72539196170079e-06,
      "loss": 0.9164,
      "step": 2685
    },
    {
      "epoch": 0.1937881028822914,
      "grad_norm": 10.374301749023168,
      "learning_rate": 3.7251555600382563e-06,
      "loss": 0.9878,
      "step": 2686
    },
    {
      "epoch": 0.19386025035171892,
      "grad_norm": 7.118315120606139,
      "learning_rate": 3.724919064171364e-06,
      "loss": 0.9981,
      "step": 2687
    },
    {
      "epoch": 0.19393239782114643,
      "grad_norm": 5.626994648243617,
      "learning_rate": 3.7246824741130266e-06,
      "loss": 1.0812,
      "step": 2688
    },
    {
      "epoch": 0.19400454529057393,
      "grad_norm": 7.328693945298006,
      "learning_rate": 3.7244457898761635e-06,
      "loss": 0.9286,
      "step": 2689
    },
    {
      "epoch": 0.19407669276000145,
      "grad_norm": 9.385754736182564,
      "learning_rate": 3.7242090114737002e-06,
      "loss": 0.9931,
      "step": 2690
    },
    {
      "epoch": 0.19414884022942897,
      "grad_norm": 10.657431618070447,
      "learning_rate": 3.723972138918565e-06,
      "loss": 1.0324,
      "step": 2691
    },
    {
      "epoch": 0.19422098769885646,
      "grad_norm": 7.125004818563338,
      "learning_rate": 3.723735172223694e-06,
      "loss": 0.947,
      "step": 2692
    },
    {
      "epoch": 0.19429313516828398,
      "grad_norm": 10.33023746834233,
      "learning_rate": 3.723498111402026e-06,
      "loss": 0.9515,
      "step": 2693
    },
    {
      "epoch": 0.1943652826377115,
      "grad_norm": 10.116396038660959,
      "learning_rate": 3.723260956466506e-06,
      "loss": 1.0665,
      "step": 2694
    },
    {
      "epoch": 0.194437430107139,
      "grad_norm": 5.778282599932916,
      "learning_rate": 3.7230237074300847e-06,
      "loss": 0.9699,
      "step": 2695
    },
    {
      "epoch": 0.1945095775765665,
      "grad_norm": 6.834715850661046,
      "learning_rate": 3.7227863643057173e-06,
      "loss": 0.9833,
      "step": 2696
    },
    {
      "epoch": 0.19458172504599403,
      "grad_norm": 8.436564527858176,
      "learning_rate": 3.7225489271063637e-06,
      "loss": 0.9551,
      "step": 2697
    },
    {
      "epoch": 0.19465387251542152,
      "grad_norm": 11.147100548885915,
      "learning_rate": 3.7223113958449898e-06,
      "loss": 0.9666,
      "step": 2698
    },
    {
      "epoch": 0.19472601998484904,
      "grad_norm": 47.92291937238251,
      "learning_rate": 3.7220737705345666e-06,
      "loss": 0.9871,
      "step": 2699
    },
    {
      "epoch": 0.19479816745427656,
      "grad_norm": 9.775636664923152,
      "learning_rate": 3.7218360511880694e-06,
      "loss": 1.0033,
      "step": 2700
    },
    {
      "epoch": 0.19487031492370405,
      "grad_norm": 10.677374789594161,
      "learning_rate": 3.7215982378184794e-06,
      "loss": 1.0769,
      "step": 2701
    },
    {
      "epoch": 0.19494246239313157,
      "grad_norm": 17.274530190733028,
      "learning_rate": 3.7213603304387833e-06,
      "loss": 1.0484,
      "step": 2702
    },
    {
      "epoch": 0.19501460986255906,
      "grad_norm": 11.189436931335376,
      "learning_rate": 3.7211223290619717e-06,
      "loss": 0.8926,
      "step": 2703
    },
    {
      "epoch": 0.19508675733198658,
      "grad_norm": 11.6408366958519,
      "learning_rate": 3.720884233701041e-06,
      "loss": 0.9541,
      "step": 2704
    },
    {
      "epoch": 0.1951589048014141,
      "grad_norm": 0.7094828637078637,
      "learning_rate": 3.7206460443689927e-06,
      "loss": 0.9082,
      "step": 2705
    },
    {
      "epoch": 0.1952310522708416,
      "grad_norm": 13.623415233595177,
      "learning_rate": 3.7204077610788334e-06,
      "loss": 1.0911,
      "step": 2706
    },
    {
      "epoch": 0.1953031997402691,
      "grad_norm": 7.2955527567728184,
      "learning_rate": 3.7201693838435758e-06,
      "loss": 1.0195,
      "step": 2707
    },
    {
      "epoch": 0.19537534720969663,
      "grad_norm": 0.7464900137839546,
      "learning_rate": 3.719930912676235e-06,
      "loss": 0.8392,
      "step": 2708
    },
    {
      "epoch": 0.19544749467912412,
      "grad_norm": 15.600692029422124,
      "learning_rate": 3.7196923475898344e-06,
      "loss": 1.0029,
      "step": 2709
    },
    {
      "epoch": 0.19551964214855164,
      "grad_norm": 5.7553440636096935,
      "learning_rate": 3.7194536885974008e-06,
      "loss": 1.0763,
      "step": 2710
    },
    {
      "epoch": 0.19559178961797916,
      "grad_norm": 9.417049253900641,
      "learning_rate": 3.7192149357119663e-06,
      "loss": 1.0409,
      "step": 2711
    },
    {
      "epoch": 0.19566393708740665,
      "grad_norm": 5.911234694856548,
      "learning_rate": 3.7189760889465686e-06,
      "loss": 1.0689,
      "step": 2712
    },
    {
      "epoch": 0.19573608455683417,
      "grad_norm": 30.526074879659745,
      "learning_rate": 3.7187371483142506e-06,
      "loss": 0.9438,
      "step": 2713
    },
    {
      "epoch": 0.19580823202626169,
      "grad_norm": 16.847184830669782,
      "learning_rate": 3.7184981138280593e-06,
      "loss": 1.0049,
      "step": 2714
    },
    {
      "epoch": 0.19588037949568918,
      "grad_norm": 8.45821567196744,
      "learning_rate": 3.718258985501047e-06,
      "loss": 1.0938,
      "step": 2715
    },
    {
      "epoch": 0.1959525269651167,
      "grad_norm": 7.905158046660154,
      "learning_rate": 3.7180197633462726e-06,
      "loss": 0.9252,
      "step": 2716
    },
    {
      "epoch": 0.19602467443454422,
      "grad_norm": 9.990963095062185,
      "learning_rate": 3.717780447376799e-06,
      "loss": 1.0081,
      "step": 2717
    },
    {
      "epoch": 0.1960968219039717,
      "grad_norm": 6.08196884250657,
      "learning_rate": 3.7175410376056943e-06,
      "loss": 0.9722,
      "step": 2718
    },
    {
      "epoch": 0.19616896937339923,
      "grad_norm": 13.79480725321381,
      "learning_rate": 3.7173015340460314e-06,
      "loss": 0.8854,
      "step": 2719
    },
    {
      "epoch": 0.19624111684282675,
      "grad_norm": 8.608138256933858,
      "learning_rate": 3.7170619367108884e-06,
      "loss": 1.0148,
      "step": 2720
    },
    {
      "epoch": 0.19631326431225424,
      "grad_norm": 13.880655321816976,
      "learning_rate": 3.71682224561335e-06,
      "loss": 1.0475,
      "step": 2721
    },
    {
      "epoch": 0.19638541178168176,
      "grad_norm": 6.228003896252657,
      "learning_rate": 3.7165824607665047e-06,
      "loss": 0.9651,
      "step": 2722
    },
    {
      "epoch": 0.19645755925110928,
      "grad_norm": 7.519157356775893,
      "learning_rate": 3.716342582183445e-06,
      "loss": 0.9459,
      "step": 2723
    },
    {
      "epoch": 0.19652970672053677,
      "grad_norm": 8.262289892344418,
      "learning_rate": 3.7161026098772707e-06,
      "loss": 1.0293,
      "step": 2724
    },
    {
      "epoch": 0.1966018541899643,
      "grad_norm": 15.425210980374386,
      "learning_rate": 3.7158625438610857e-06,
      "loss": 1.0617,
      "step": 2725
    },
    {
      "epoch": 0.1966740016593918,
      "grad_norm": 5.54824786322733,
      "learning_rate": 3.715622384147999e-06,
      "loss": 0.9321,
      "step": 2726
    },
    {
      "epoch": 0.1967461491288193,
      "grad_norm": 9.70067030960059,
      "learning_rate": 3.7153821307511253e-06,
      "loss": 1.0574,
      "step": 2727
    },
    {
      "epoch": 0.19681829659824682,
      "grad_norm": 12.280597698564366,
      "learning_rate": 3.715141783683584e-06,
      "loss": 0.9567,
      "step": 2728
    },
    {
      "epoch": 0.19689044406767434,
      "grad_norm": 6.552348104766348,
      "learning_rate": 3.714901342958498e-06,
      "loss": 1.0164,
      "step": 2729
    },
    {
      "epoch": 0.19696259153710183,
      "grad_norm": 8.189723986770995,
      "learning_rate": 3.7146608085889987e-06,
      "loss": 0.9878,
      "step": 2730
    },
    {
      "epoch": 0.19703473900652935,
      "grad_norm": 64.33713616445598,
      "learning_rate": 3.7144201805882194e-06,
      "loss": 1.0756,
      "step": 2731
    },
    {
      "epoch": 0.19710688647595687,
      "grad_norm": 7.216834668794814,
      "learning_rate": 3.714179458969301e-06,
      "loss": 1.0465,
      "step": 2732
    },
    {
      "epoch": 0.19717903394538436,
      "grad_norm": 9.327302982782172,
      "learning_rate": 3.713938643745389e-06,
      "loss": 0.991,
      "step": 2733
    },
    {
      "epoch": 0.19725118141481188,
      "grad_norm": 19.32456578486144,
      "learning_rate": 3.7136977349296315e-06,
      "loss": 0.9621,
      "step": 2734
    },
    {
      "epoch": 0.1973233288842394,
      "grad_norm": 7.777364840974667,
      "learning_rate": 3.713456732535184e-06,
      "loss": 0.979,
      "step": 2735
    },
    {
      "epoch": 0.1973954763536669,
      "grad_norm": 10.75490653266532,
      "learning_rate": 3.7132156365752085e-06,
      "loss": 0.9769,
      "step": 2736
    },
    {
      "epoch": 0.1974676238230944,
      "grad_norm": 40.51848265327118,
      "learning_rate": 3.7129744470628685e-06,
      "loss": 1.0346,
      "step": 2737
    },
    {
      "epoch": 0.19753977129252193,
      "grad_norm": 10.099595966378162,
      "learning_rate": 3.7127331640113355e-06,
      "loss": 1.0759,
      "step": 2738
    },
    {
      "epoch": 0.19761191876194942,
      "grad_norm": 10.535811254318345,
      "learning_rate": 3.7124917874337846e-06,
      "loss": 0.9606,
      "step": 2739
    },
    {
      "epoch": 0.19768406623137694,
      "grad_norm": 10.21222406058895,
      "learning_rate": 3.7122503173433966e-06,
      "loss": 1.0223,
      "step": 2740
    },
    {
      "epoch": 0.19775621370080446,
      "grad_norm": 10.47887620972079,
      "learning_rate": 3.712008753753357e-06,
      "loss": 1.0185,
      "step": 2741
    },
    {
      "epoch": 0.19782836117023195,
      "grad_norm": 17.482181742185293,
      "learning_rate": 3.7117670966768575e-06,
      "loss": 1.0314,
      "step": 2742
    },
    {
      "epoch": 0.19790050863965947,
      "grad_norm": 7.638163476273809,
      "learning_rate": 3.7115253461270933e-06,
      "loss": 0.9532,
      "step": 2743
    },
    {
      "epoch": 0.197972656109087,
      "grad_norm": 12.95027451389937,
      "learning_rate": 3.7112835021172662e-06,
      "loss": 1.0053,
      "step": 2744
    },
    {
      "epoch": 0.19804480357851448,
      "grad_norm": 14.211597295190433,
      "learning_rate": 3.711041564660582e-06,
      "loss": 1.0544,
      "step": 2745
    },
    {
      "epoch": 0.198116951047942,
      "grad_norm": 8.151742910297523,
      "learning_rate": 3.7107995337702515e-06,
      "loss": 1.0956,
      "step": 2746
    },
    {
      "epoch": 0.19818909851736952,
      "grad_norm": 8.060000483131867,
      "learning_rate": 3.7105574094594917e-06,
      "loss": 1.0797,
      "step": 2747
    },
    {
      "epoch": 0.198261245986797,
      "grad_norm": 14.75182990668287,
      "learning_rate": 3.7103151917415244e-06,
      "loss": 0.9492,
      "step": 2748
    },
    {
      "epoch": 0.19833339345622453,
      "grad_norm": 10.716164502312616,
      "learning_rate": 3.7100728806295756e-06,
      "loss": 0.8775,
      "step": 2749
    },
    {
      "epoch": 0.19840554092565205,
      "grad_norm": 7.86824181622327,
      "learning_rate": 3.7098304761368776e-06,
      "loss": 1.0008,
      "step": 2750
    },
    {
      "epoch": 0.19847768839507954,
      "grad_norm": 5.452940492323145,
      "learning_rate": 3.709587978276666e-06,
      "loss": 1.0251,
      "step": 2751
    },
    {
      "epoch": 0.19854983586450706,
      "grad_norm": 7.515414355968308,
      "learning_rate": 3.7093453870621847e-06,
      "loss": 1.1501,
      "step": 2752
    },
    {
      "epoch": 0.19862198333393455,
      "grad_norm": 6.496410258938502,
      "learning_rate": 3.709102702506679e-06,
      "loss": 1.0396,
      "step": 2753
    },
    {
      "epoch": 0.19869413080336207,
      "grad_norm": 5.02848199557377,
      "learning_rate": 3.708859924623401e-06,
      "loss": 0.9752,
      "step": 2754
    },
    {
      "epoch": 0.1987662782727896,
      "grad_norm": 5.706516208685492,
      "learning_rate": 3.7086170534256098e-06,
      "loss": 1.0174,
      "step": 2755
    },
    {
      "epoch": 0.19883842574221708,
      "grad_norm": 12.95694469276937,
      "learning_rate": 3.7083740889265655e-06,
      "loss": 0.9345,
      "step": 2756
    },
    {
      "epoch": 0.1989105732116446,
      "grad_norm": 7.055982164650172,
      "learning_rate": 3.708131031139537e-06,
      "loss": 1.0502,
      "step": 2757
    },
    {
      "epoch": 0.19898272068107212,
      "grad_norm": 14.296909152964671,
      "learning_rate": 3.707887880077796e-06,
      "loss": 1.1159,
      "step": 2758
    },
    {
      "epoch": 0.1990548681504996,
      "grad_norm": 8.20052739750121,
      "learning_rate": 3.7076446357546196e-06,
      "loss": 0.9438,
      "step": 2759
    },
    {
      "epoch": 0.19912701561992713,
      "grad_norm": 11.051654909239167,
      "learning_rate": 3.7074012981832917e-06,
      "loss": 0.9808,
      "step": 2760
    },
    {
      "epoch": 0.19919916308935465,
      "grad_norm": 8.389969841012867,
      "learning_rate": 3.7071578673770995e-06,
      "loss": 1.1053,
      "step": 2761
    },
    {
      "epoch": 0.19927131055878214,
      "grad_norm": 6.3952112879084915,
      "learning_rate": 3.706914343349335e-06,
      "loss": 0.9873,
      "step": 2762
    },
    {
      "epoch": 0.19934345802820966,
      "grad_norm": 6.993317002434052,
      "learning_rate": 3.706670726113298e-06,
      "loss": 0.9369,
      "step": 2763
    },
    {
      "epoch": 0.19941560549763718,
      "grad_norm": 8.429867605524993,
      "learning_rate": 3.7064270156822906e-06,
      "loss": 1.0323,
      "step": 2764
    },
    {
      "epoch": 0.19948775296706467,
      "grad_norm": 4.594019460235818,
      "learning_rate": 3.70618321206962e-06,
      "loss": 0.968,
      "step": 2765
    },
    {
      "epoch": 0.1995599004364922,
      "grad_norm": 6.205823796476977,
      "learning_rate": 3.705939315288601e-06,
      "loss": 1.0087,
      "step": 2766
    },
    {
      "epoch": 0.1996320479059197,
      "grad_norm": 9.420111597133094,
      "learning_rate": 3.705695325352551e-06,
      "loss": 1.028,
      "step": 2767
    },
    {
      "epoch": 0.1997041953753472,
      "grad_norm": 0.7245206958560638,
      "learning_rate": 3.705451242274793e-06,
      "loss": 0.8085,
      "step": 2768
    },
    {
      "epoch": 0.19977634284477472,
      "grad_norm": 6.125522513387059,
      "learning_rate": 3.705207066068657e-06,
      "loss": 1.0427,
      "step": 2769
    },
    {
      "epoch": 0.19984849031420224,
      "grad_norm": 6.0740532659331015,
      "learning_rate": 3.704962796747475e-06,
      "loss": 0.9721,
      "step": 2770
    },
    {
      "epoch": 0.19992063778362973,
      "grad_norm": 0.7956409997522471,
      "learning_rate": 3.704718434324586e-06,
      "loss": 0.8051,
      "step": 2771
    },
    {
      "epoch": 0.19999278525305725,
      "grad_norm": 6.042627227542268,
      "learning_rate": 3.7044739788133343e-06,
      "loss": 1.0647,
      "step": 2772
    },
    {
      "epoch": 0.20006493272248477,
      "grad_norm": 13.648241850701297,
      "learning_rate": 3.7042294302270686e-06,
      "loss": 1.0052,
      "step": 2773
    },
    {
      "epoch": 0.20013708019191226,
      "grad_norm": 5.519197943923905,
      "learning_rate": 3.7039847885791426e-06,
      "loss": 1.0387,
      "step": 2774
    },
    {
      "epoch": 0.20020922766133978,
      "grad_norm": 5.90371795148064,
      "learning_rate": 3.7037400538829147e-06,
      "loss": 0.9375,
      "step": 2775
    },
    {
      "epoch": 0.2002813751307673,
      "grad_norm": 4.570558512008607,
      "learning_rate": 3.7034952261517494e-06,
      "loss": 0.9903,
      "step": 2776
    },
    {
      "epoch": 0.2003535226001948,
      "grad_norm": 6.537289979672028,
      "learning_rate": 3.703250305399016e-06,
      "loss": 0.9697,
      "step": 2777
    },
    {
      "epoch": 0.2004256700696223,
      "grad_norm": 5.232352797054752,
      "learning_rate": 3.703005291638089e-06,
      "loss": 1.0322,
      "step": 2778
    },
    {
      "epoch": 0.20049781753904983,
      "grad_norm": 7.4776139910091075,
      "learning_rate": 3.702760184882347e-06,
      "loss": 0.9706,
      "step": 2779
    },
    {
      "epoch": 0.20056996500847732,
      "grad_norm": 4.46462630362339,
      "learning_rate": 3.702514985145175e-06,
      "loss": 1.0785,
      "step": 2780
    },
    {
      "epoch": 0.20064211247790484,
      "grad_norm": 6.1039943562252805,
      "learning_rate": 3.7022696924399623e-06,
      "loss": 0.9615,
      "step": 2781
    },
    {
      "epoch": 0.20071425994733236,
      "grad_norm": 5.733874390082175,
      "learning_rate": 3.702024306780103e-06,
      "loss": 1.0305,
      "step": 2782
    },
    {
      "epoch": 0.20078640741675985,
      "grad_norm": 6.058058222523966,
      "learning_rate": 3.701778828178997e-06,
      "loss": 0.9756,
      "step": 2783
    },
    {
      "epoch": 0.20085855488618737,
      "grad_norm": 3.9152006828154735,
      "learning_rate": 3.701533256650049e-06,
      "loss": 0.9914,
      "step": 2784
    },
    {
      "epoch": 0.2009307023556149,
      "grad_norm": 6.2282467506249235,
      "learning_rate": 3.7012875922066688e-06,
      "loss": 1.0171,
      "step": 2785
    },
    {
      "epoch": 0.20100284982504238,
      "grad_norm": 9.338810766668555,
      "learning_rate": 3.701041834862271e-06,
      "loss": 0.944,
      "step": 2786
    },
    {
      "epoch": 0.2010749972944699,
      "grad_norm": 5.036816094784548,
      "learning_rate": 3.700795984630275e-06,
      "loss": 0.9696,
      "step": 2787
    },
    {
      "epoch": 0.20114714476389742,
      "grad_norm": 3.291157180051003,
      "learning_rate": 3.7005500415241073e-06,
      "loss": 1.0029,
      "step": 2788
    },
    {
      "epoch": 0.2012192922333249,
      "grad_norm": 9.608502405757378,
      "learning_rate": 3.7003040055571973e-06,
      "loss": 0.9468,
      "step": 2789
    },
    {
      "epoch": 0.20129143970275243,
      "grad_norm": 7.574039954858502,
      "learning_rate": 3.7000578767429792e-06,
      "loss": 1.0216,
      "step": 2790
    },
    {
      "epoch": 0.20136358717217995,
      "grad_norm": 7.960793026857003,
      "learning_rate": 3.699811655094894e-06,
      "loss": 0.8379,
      "step": 2791
    },
    {
      "epoch": 0.20143573464160744,
      "grad_norm": 4.564090190360733,
      "learning_rate": 3.699565340626387e-06,
      "loss": 1.0343,
      "step": 2792
    },
    {
      "epoch": 0.20150788211103496,
      "grad_norm": 3.479866292511441,
      "learning_rate": 3.6993189333509082e-06,
      "loss": 0.9553,
      "step": 2793
    },
    {
      "epoch": 0.20158002958046248,
      "grad_norm": 7.862337134392276,
      "learning_rate": 3.699072433281913e-06,
      "loss": 1.0622,
      "step": 2794
    },
    {
      "epoch": 0.20165217704988997,
      "grad_norm": 4.772390897052292,
      "learning_rate": 3.698825840432862e-06,
      "loss": 0.9941,
      "step": 2795
    },
    {
      "epoch": 0.2017243245193175,
      "grad_norm": 4.563218569672026,
      "learning_rate": 3.698579154817221e-06,
      "loss": 1.0014,
      "step": 2796
    },
    {
      "epoch": 0.201796471988745,
      "grad_norm": 4.512505004541892,
      "learning_rate": 3.69833237644846e-06,
      "loss": 1.0499,
      "step": 2797
    },
    {
      "epoch": 0.2018686194581725,
      "grad_norm": 7.3946323619605785,
      "learning_rate": 3.698085505340055e-06,
      "loss": 1.0419,
      "step": 2798
    },
    {
      "epoch": 0.20194076692760002,
      "grad_norm": 6.823715636863353,
      "learning_rate": 3.6978385415054863e-06,
      "loss": 1.1053,
      "step": 2799
    },
    {
      "epoch": 0.20201291439702754,
      "grad_norm": 5.83924961703881,
      "learning_rate": 3.6975914849582407e-06,
      "loss": 0.9815,
      "step": 2800
    },
    {
      "epoch": 0.20208506186645503,
      "grad_norm": 5.178571255688594,
      "learning_rate": 3.6973443357118077e-06,
      "loss": 1.0554,
      "step": 2801
    },
    {
      "epoch": 0.20215720933588255,
      "grad_norm": 0.724794431021224,
      "learning_rate": 3.697097093779684e-06,
      "loss": 0.9056,
      "step": 2802
    },
    {
      "epoch": 0.20222935680531004,
      "grad_norm": 15.903682718899969,
      "learning_rate": 3.6968497591753706e-06,
      "loss": 0.9235,
      "step": 2803
    },
    {
      "epoch": 0.20230150427473756,
      "grad_norm": 4.751133030728943,
      "learning_rate": 3.696602331912374e-06,
      "loss": 1.0932,
      "step": 2804
    },
    {
      "epoch": 0.20237365174416508,
      "grad_norm": 7.081483966029494,
      "learning_rate": 3.6963548120042037e-06,
      "loss": 0.99,
      "step": 2805
    },
    {
      "epoch": 0.20244579921359257,
      "grad_norm": 20.551225626610677,
      "learning_rate": 3.6961071994643774e-06,
      "loss": 0.9042,
      "step": 2806
    },
    {
      "epoch": 0.2025179466830201,
      "grad_norm": 4.732620770363827,
      "learning_rate": 3.6958594943064157e-06,
      "loss": 0.9537,
      "step": 2807
    },
    {
      "epoch": 0.2025900941524476,
      "grad_norm": 3.772707661246628,
      "learning_rate": 3.695611696543844e-06,
      "loss": 0.9469,
      "step": 2808
    },
    {
      "epoch": 0.2026622416218751,
      "grad_norm": 5.2528028726384575,
      "learning_rate": 3.6953638061901958e-06,
      "loss": 1.0588,
      "step": 2809
    },
    {
      "epoch": 0.20273438909130262,
      "grad_norm": 8.077673067477948,
      "learning_rate": 3.695115823259005e-06,
      "loss": 0.9359,
      "step": 2810
    },
    {
      "epoch": 0.20280653656073014,
      "grad_norm": 5.422901595049898,
      "learning_rate": 3.694867747763815e-06,
      "loss": 0.9945,
      "step": 2811
    },
    {
      "epoch": 0.20287868403015763,
      "grad_norm": 5.994174354336512,
      "learning_rate": 3.6946195797181714e-06,
      "loss": 0.9499,
      "step": 2812
    },
    {
      "epoch": 0.20295083149958515,
      "grad_norm": 5.661316346672354,
      "learning_rate": 3.6943713191356256e-06,
      "loss": 0.9327,
      "step": 2813
    },
    {
      "epoch": 0.20302297896901267,
      "grad_norm": 0.7586091563490106,
      "learning_rate": 3.6941229660297343e-06,
      "loss": 0.8891,
      "step": 2814
    },
    {
      "epoch": 0.20309512643844016,
      "grad_norm": 7.749332030028207,
      "learning_rate": 3.693874520414059e-06,
      "loss": 1.1184,
      "step": 2815
    },
    {
      "epoch": 0.20316727390786768,
      "grad_norm": 6.106890317050276,
      "learning_rate": 3.6936259823021673e-06,
      "loss": 0.8823,
      "step": 2816
    },
    {
      "epoch": 0.2032394213772952,
      "grad_norm": 4.915669916703162,
      "learning_rate": 3.6933773517076297e-06,
      "loss": 1.0454,
      "step": 2817
    },
    {
      "epoch": 0.2033115688467227,
      "grad_norm": 11.572487636319748,
      "learning_rate": 3.6931286286440245e-06,
      "loss": 0.9769,
      "step": 2818
    },
    {
      "epoch": 0.2033837163161502,
      "grad_norm": 6.732021832290761,
      "learning_rate": 3.692879813124932e-06,
      "loss": 1.0528,
      "step": 2819
    },
    {
      "epoch": 0.20345586378557773,
      "grad_norm": 5.481856330210893,
      "learning_rate": 3.6926309051639396e-06,
      "loss": 1.0725,
      "step": 2820
    },
    {
      "epoch": 0.20352801125500522,
      "grad_norm": 0.6913633387121337,
      "learning_rate": 3.69238190477464e-06,
      "loss": 0.8876,
      "step": 2821
    },
    {
      "epoch": 0.20360015872443274,
      "grad_norm": 11.471056797129407,
      "learning_rate": 3.6921328119706293e-06,
      "loss": 1.0999,
      "step": 2822
    },
    {
      "epoch": 0.20367230619386026,
      "grad_norm": 0.793809754833259,
      "learning_rate": 3.6918836267655093e-06,
      "loss": 0.917,
      "step": 2823
    },
    {
      "epoch": 0.20374445366328775,
      "grad_norm": 4.786398913900541,
      "learning_rate": 3.691634349172888e-06,
      "loss": 1.039,
      "step": 2824
    },
    {
      "epoch": 0.20381660113271527,
      "grad_norm": 11.60813024324506,
      "learning_rate": 3.691384979206377e-06,
      "loss": 0.8788,
      "step": 2825
    },
    {
      "epoch": 0.2038887486021428,
      "grad_norm": 0.8079143009934818,
      "learning_rate": 3.691135516879593e-06,
      "loss": 0.8363,
      "step": 2826
    },
    {
      "epoch": 0.20396089607157028,
      "grad_norm": 4.336749026674397,
      "learning_rate": 3.6908859622061596e-06,
      "loss": 1.0449,
      "step": 2827
    },
    {
      "epoch": 0.2040330435409978,
      "grad_norm": 0.6795915996376988,
      "learning_rate": 3.6906363151997033e-06,
      "loss": 0.8497,
      "step": 2828
    },
    {
      "epoch": 0.20410519101042532,
      "grad_norm": 8.332092141860244,
      "learning_rate": 3.6903865758738563e-06,
      "loss": 1.0726,
      "step": 2829
    },
    {
      "epoch": 0.2041773384798528,
      "grad_norm": 4.566209304256466,
      "learning_rate": 3.6901367442422556e-06,
      "loss": 0.9649,
      "step": 2830
    },
    {
      "epoch": 0.20424948594928033,
      "grad_norm": 3.007637316073303,
      "learning_rate": 3.689886820318544e-06,
      "loss": 1.0854,
      "step": 2831
    },
    {
      "epoch": 0.20432163341870785,
      "grad_norm": 4.216440204273721,
      "learning_rate": 3.689636804116369e-06,
      "loss": 1.0061,
      "step": 2832
    },
    {
      "epoch": 0.20439378088813534,
      "grad_norm": 10.68611144922574,
      "learning_rate": 3.6893866956493832e-06,
      "loss": 0.9351,
      "step": 2833
    },
    {
      "epoch": 0.20446592835756286,
      "grad_norm": 5.739970540891542,
      "learning_rate": 3.6891364949312437e-06,
      "loss": 1.0347,
      "step": 2834
    },
    {
      "epoch": 0.20453807582699038,
      "grad_norm": 6.130402654021957,
      "learning_rate": 3.688886201975613e-06,
      "loss": 0.9945,
      "step": 2835
    },
    {
      "epoch": 0.20461022329641787,
      "grad_norm": 6.77319617100004,
      "learning_rate": 3.688635816796159e-06,
      "loss": 0.9735,
      "step": 2836
    },
    {
      "epoch": 0.2046823707658454,
      "grad_norm": 5.7644246823898495,
      "learning_rate": 3.6883853394065546e-06,
      "loss": 0.9303,
      "step": 2837
    },
    {
      "epoch": 0.2047545182352729,
      "grad_norm": 4.852536605861976,
      "learning_rate": 3.6881347698204765e-06,
      "loss": 0.9698,
      "step": 2838
    },
    {
      "epoch": 0.2048266657047004,
      "grad_norm": 6.957806534917721,
      "learning_rate": 3.6878841080516074e-06,
      "loss": 0.9348,
      "step": 2839
    },
    {
      "epoch": 0.20489881317412792,
      "grad_norm": 5.772840827802593,
      "learning_rate": 3.687633354113636e-06,
      "loss": 0.995,
      "step": 2840
    },
    {
      "epoch": 0.20497096064355544,
      "grad_norm": 5.27336787425101,
      "learning_rate": 3.6873825080202544e-06,
      "loss": 0.8556,
      "step": 2841
    },
    {
      "epoch": 0.20504310811298293,
      "grad_norm": 27.779947153786804,
      "learning_rate": 3.687131569785161e-06,
      "loss": 0.9801,
      "step": 2842
    },
    {
      "epoch": 0.20511525558241045,
      "grad_norm": 4.8663987795811785,
      "learning_rate": 3.6868805394220575e-06,
      "loss": 0.9674,
      "step": 2843
    },
    {
      "epoch": 0.20518740305183797,
      "grad_norm": 6.937290222845396,
      "learning_rate": 3.6866294169446526e-06,
      "loss": 0.963,
      "step": 2844
    },
    {
      "epoch": 0.20525955052126546,
      "grad_norm": 3.1024145690320126,
      "learning_rate": 3.686378202366659e-06,
      "loss": 0.8934,
      "step": 2845
    },
    {
      "epoch": 0.20533169799069298,
      "grad_norm": 5.495379848246495,
      "learning_rate": 3.6861268957017947e-06,
      "loss": 1.0315,
      "step": 2846
    },
    {
      "epoch": 0.2054038454601205,
      "grad_norm": 5.498422222978031,
      "learning_rate": 3.6858754969637822e-06,
      "loss": 0.9664,
      "step": 2847
    },
    {
      "epoch": 0.205475992929548,
      "grad_norm": 5.738632575719701,
      "learning_rate": 3.6856240061663496e-06,
      "loss": 1.0238,
      "step": 2848
    },
    {
      "epoch": 0.2055481403989755,
      "grad_norm": 4.214359287423322,
      "learning_rate": 3.6853724233232307e-06,
      "loss": 1.0267,
      "step": 2849
    },
    {
      "epoch": 0.20562028786840303,
      "grad_norm": 6.042971275047953,
      "learning_rate": 3.6851207484481617e-06,
      "loss": 1.067,
      "step": 2850
    },
    {
      "epoch": 0.20569243533783052,
      "grad_norm": 2.925908532830901,
      "learning_rate": 3.6848689815548876e-06,
      "loss": 1.0109,
      "step": 2851
    },
    {
      "epoch": 0.20576458280725804,
      "grad_norm": 6.941781861935822,
      "learning_rate": 3.6846171226571547e-06,
      "loss": 0.9112,
      "step": 2852
    },
    {
      "epoch": 0.20583673027668556,
      "grad_norm": 9.366930120525748,
      "learning_rate": 3.6843651717687178e-06,
      "loss": 0.9928,
      "step": 2853
    },
    {
      "epoch": 0.20590887774611305,
      "grad_norm": 3.224717321501044,
      "learning_rate": 3.684113128903334e-06,
      "loss": 1.0578,
      "step": 2854
    },
    {
      "epoch": 0.20598102521554057,
      "grad_norm": 8.288828710769959,
      "learning_rate": 3.6838609940747665e-06,
      "loss": 0.9938,
      "step": 2855
    },
    {
      "epoch": 0.20605317268496806,
      "grad_norm": 5.670528367601509,
      "learning_rate": 3.6836087672967837e-06,
      "loss": 1.039,
      "step": 2856
    },
    {
      "epoch": 0.20612532015439558,
      "grad_norm": 6.979280051453809,
      "learning_rate": 3.683356448583158e-06,
      "loss": 0.9935,
      "step": 2857
    },
    {
      "epoch": 0.2061974676238231,
      "grad_norm": 7.41851435016299,
      "learning_rate": 3.683104037947669e-06,
      "loss": 1.0036,
      "step": 2858
    },
    {
      "epoch": 0.2062696150932506,
      "grad_norm": 3.9980750220369723,
      "learning_rate": 3.682851535404099e-06,
      "loss": 0.9775,
      "step": 2859
    },
    {
      "epoch": 0.2063417625626781,
      "grad_norm": 5.646296558403522,
      "learning_rate": 3.682598940966236e-06,
      "loss": 1.0852,
      "step": 2860
    },
    {
      "epoch": 0.20641391003210563,
      "grad_norm": 4.907469190985679,
      "learning_rate": 3.682346254647874e-06,
      "loss": 0.9348,
      "step": 2861
    },
    {
      "epoch": 0.20648605750153312,
      "grad_norm": 9.332604334381125,
      "learning_rate": 3.6820934764628102e-06,
      "loss": 1.0044,
      "step": 2862
    },
    {
      "epoch": 0.20655820497096064,
      "grad_norm": 0.7522186603122163,
      "learning_rate": 3.681840606424849e-06,
      "loss": 0.8197,
      "step": 2863
    },
    {
      "epoch": 0.20663035244038816,
      "grad_norm": 6.046085749728199,
      "learning_rate": 3.6815876445477983e-06,
      "loss": 0.9834,
      "step": 2864
    },
    {
      "epoch": 0.20670249990981565,
      "grad_norm": 4.2560119741601214,
      "learning_rate": 3.6813345908454716e-06,
      "loss": 1.0456,
      "step": 2865
    },
    {
      "epoch": 0.20677464737924317,
      "grad_norm": 4.2844981102816275,
      "learning_rate": 3.681081445331687e-06,
      "loss": 1.0412,
      "step": 2866
    },
    {
      "epoch": 0.2068467948486707,
      "grad_norm": 4.475368906510974,
      "learning_rate": 3.680828208020267e-06,
      "loss": 0.9673,
      "step": 2867
    },
    {
      "epoch": 0.20691894231809818,
      "grad_norm": 4.009396483556286,
      "learning_rate": 3.6805748789250414e-06,
      "loss": 0.9923,
      "step": 2868
    },
    {
      "epoch": 0.2069910897875257,
      "grad_norm": 3.843341588598441,
      "learning_rate": 3.680321458059843e-06,
      "loss": 1.0424,
      "step": 2869
    },
    {
      "epoch": 0.20706323725695322,
      "grad_norm": 3.8190851092607696,
      "learning_rate": 3.6800679454385096e-06,
      "loss": 0.8592,
      "step": 2870
    },
    {
      "epoch": 0.2071353847263807,
      "grad_norm": 11.680451088124563,
      "learning_rate": 3.679814341074886e-06,
      "loss": 0.9981,
      "step": 2871
    },
    {
      "epoch": 0.20720753219580823,
      "grad_norm": 4.5905488986257215,
      "learning_rate": 3.6795606449828185e-06,
      "loss": 0.9413,
      "step": 2872
    },
    {
      "epoch": 0.20727967966523575,
      "grad_norm": 0.8245376765362339,
      "learning_rate": 3.6793068571761624e-06,
      "loss": 0.8531,
      "step": 2873
    },
    {
      "epoch": 0.20735182713466324,
      "grad_norm": 4.855716223329356,
      "learning_rate": 3.6790529776687753e-06,
      "loss": 1.0135,
      "step": 2874
    },
    {
      "epoch": 0.20742397460409076,
      "grad_norm": 7.885747448237409,
      "learning_rate": 3.678799006474521e-06,
      "loss": 0.9844,
      "step": 2875
    },
    {
      "epoch": 0.20749612207351828,
      "grad_norm": 2.779899612233409,
      "learning_rate": 3.6785449436072672e-06,
      "loss": 0.9974,
      "step": 2876
    },
    {
      "epoch": 0.20756826954294577,
      "grad_norm": 4.2361732408869495,
      "learning_rate": 3.678290789080888e-06,
      "loss": 1.0124,
      "step": 2877
    },
    {
      "epoch": 0.2076404170123733,
      "grad_norm": 0.7116237304772666,
      "learning_rate": 3.678036542909262e-06,
      "loss": 0.8164,
      "step": 2878
    },
    {
      "epoch": 0.2077125644818008,
      "grad_norm": 21.274957339897373,
      "learning_rate": 3.677782205106272e-06,
      "loss": 0.9841,
      "step": 2879
    },
    {
      "epoch": 0.2077847119512283,
      "grad_norm": 3.4556746930119653,
      "learning_rate": 3.6775277756858066e-06,
      "loss": 1.0213,
      "step": 2880
    },
    {
      "epoch": 0.20785685942065582,
      "grad_norm": 4.313632982115485,
      "learning_rate": 3.677273254661759e-06,
      "loss": 0.9615,
      "step": 2881
    },
    {
      "epoch": 0.20792900689008334,
      "grad_norm": 5.5901808620394835,
      "learning_rate": 3.6770186420480282e-06,
      "loss": 1.0138,
      "step": 2882
    },
    {
      "epoch": 0.20800115435951083,
      "grad_norm": 3.7762343504403697,
      "learning_rate": 3.676763937858518e-06,
      "loss": 1.066,
      "step": 2883
    },
    {
      "epoch": 0.20807330182893835,
      "grad_norm": 3.1720433683761917,
      "learning_rate": 3.676509142107136e-06,
      "loss": 1.0508,
      "step": 2884
    },
    {
      "epoch": 0.20814544929836587,
      "grad_norm": 3.764258172256832,
      "learning_rate": 3.676254254807796e-06,
      "loss": 1.0144,
      "step": 2885
    },
    {
      "epoch": 0.20821759676779336,
      "grad_norm": 7.1361216695702545,
      "learning_rate": 3.6759992759744166e-06,
      "loss": 1.0679,
      "step": 2886
    },
    {
      "epoch": 0.20828974423722088,
      "grad_norm": 5.294923856221964,
      "learning_rate": 3.6757442056209204e-06,
      "loss": 1.0888,
      "step": 2887
    },
    {
      "epoch": 0.2083618917066484,
      "grad_norm": 4.793266206323473,
      "learning_rate": 3.675489043761237e-06,
      "loss": 0.9425,
      "step": 2888
    },
    {
      "epoch": 0.2084340391760759,
      "grad_norm": 3.5287284287736913,
      "learning_rate": 3.675233790409299e-06,
      "loss": 1.0054,
      "step": 2889
    },
    {
      "epoch": 0.2085061866455034,
      "grad_norm": 5.3976129766163226,
      "learning_rate": 3.6749784455790457e-06,
      "loss": 0.9307,
      "step": 2890
    },
    {
      "epoch": 0.20857833411493093,
      "grad_norm": 4.52872791475475,
      "learning_rate": 3.67472300928442e-06,
      "loss": 1.0133,
      "step": 2891
    },
    {
      "epoch": 0.20865048158435842,
      "grad_norm": 5.264958823424312,
      "learning_rate": 3.674467481539371e-06,
      "loss": 0.9514,
      "step": 2892
    },
    {
      "epoch": 0.20872262905378594,
      "grad_norm": 3.9248780687187064,
      "learning_rate": 3.6742118623578506e-06,
      "loss": 1.0738,
      "step": 2893
    },
    {
      "epoch": 0.20879477652321346,
      "grad_norm": 5.256458306465705,
      "learning_rate": 3.673956151753818e-06,
      "loss": 1.0048,
      "step": 2894
    },
    {
      "epoch": 0.20886692399264095,
      "grad_norm": 4.518573366039576,
      "learning_rate": 3.673700349741237e-06,
      "loss": 0.9608,
      "step": 2895
    },
    {
      "epoch": 0.20893907146206847,
      "grad_norm": 0.7293791779150858,
      "learning_rate": 3.673444456334076e-06,
      "loss": 0.8111,
      "step": 2896
    },
    {
      "epoch": 0.209011218931496,
      "grad_norm": 6.8266812332315805,
      "learning_rate": 3.673188471546308e-06,
      "loss": 1.0642,
      "step": 2897
    },
    {
      "epoch": 0.20908336640092348,
      "grad_norm": 2.5494859102424274,
      "learning_rate": 3.6729323953919115e-06,
      "loss": 1.1076,
      "step": 2898
    },
    {
      "epoch": 0.209155513870351,
      "grad_norm": 5.974383027851207,
      "learning_rate": 3.672676227884869e-06,
      "loss": 1.0065,
      "step": 2899
    },
    {
      "epoch": 0.20922766133977852,
      "grad_norm": 3.3470106537270103,
      "learning_rate": 3.672419969039171e-06,
      "loss": 1.0481,
      "step": 2900
    },
    {
      "epoch": 0.209299808809206,
      "grad_norm": 6.95730348683032,
      "learning_rate": 3.6721636188688086e-06,
      "loss": 0.9767,
      "step": 2901
    },
    {
      "epoch": 0.20937195627863353,
      "grad_norm": 4.456405555610286,
      "learning_rate": 3.6719071773877815e-06,
      "loss": 1.1001,
      "step": 2902
    },
    {
      "epoch": 0.20944410374806105,
      "grad_norm": 5.137989980531206,
      "learning_rate": 3.671650644610093e-06,
      "loss": 1.0009,
      "step": 2903
    },
    {
      "epoch": 0.20951625121748854,
      "grad_norm": 0.8554919514057294,
      "learning_rate": 3.671394020549751e-06,
      "loss": 0.8993,
      "step": 2904
    },
    {
      "epoch": 0.20958839868691606,
      "grad_norm": 4.309413524120771,
      "learning_rate": 3.6711373052207686e-06,
      "loss": 1.0085,
      "step": 2905
    },
    {
      "epoch": 0.20966054615634355,
      "grad_norm": 4.821304392350806,
      "learning_rate": 3.6708804986371637e-06,
      "loss": 0.9737,
      "step": 2906
    },
    {
      "epoch": 0.20973269362577107,
      "grad_norm": 0.7664777716532141,
      "learning_rate": 3.670623600812961e-06,
      "loss": 0.8708,
      "step": 2907
    },
    {
      "epoch": 0.2098048410951986,
      "grad_norm": 8.257003585764755,
      "learning_rate": 3.670366611762188e-06,
      "loss": 1.0286,
      "step": 2908
    },
    {
      "epoch": 0.20987698856462608,
      "grad_norm": 4.938608165916292,
      "learning_rate": 3.670109531498877e-06,
      "loss": 1.0046,
      "step": 2909
    },
    {
      "epoch": 0.2099491360340536,
      "grad_norm": 4.290520203179517,
      "learning_rate": 3.669852360037067e-06,
      "loss": 1.0214,
      "step": 2910
    },
    {
      "epoch": 0.21002128350348112,
      "grad_norm": 4.371302104548872,
      "learning_rate": 3.6695950973908022e-06,
      "loss": 1.0234,
      "step": 2911
    },
    {
      "epoch": 0.2100934309729086,
      "grad_norm": 5.209033440901271,
      "learning_rate": 3.6693377435741295e-06,
      "loss": 0.9352,
      "step": 2912
    },
    {
      "epoch": 0.21016557844233613,
      "grad_norm": 4.233647921029603,
      "learning_rate": 3.669080298601102e-06,
      "loss": 0.9596,
      "step": 2913
    },
    {
      "epoch": 0.21023772591176365,
      "grad_norm": 6.329244029627256,
      "learning_rate": 3.6688227624857777e-06,
      "loss": 1.0978,
      "step": 2914
    },
    {
      "epoch": 0.21030987338119114,
      "grad_norm": 16.290523765391377,
      "learning_rate": 3.6685651352422207e-06,
      "loss": 0.9293,
      "step": 2915
    },
    {
      "epoch": 0.21038202085061866,
      "grad_norm": 6.40095531963052,
      "learning_rate": 3.6683074168844984e-06,
      "loss": 0.9582,
      "step": 2916
    },
    {
      "epoch": 0.21045416832004618,
      "grad_norm": 0.9077538147964102,
      "learning_rate": 3.668049607426684e-06,
      "loss": 0.9195,
      "step": 2917
    },
    {
      "epoch": 0.21052631578947367,
      "grad_norm": 5.652033799326185,
      "learning_rate": 3.667791706882855e-06,
      "loss": 0.9719,
      "step": 2918
    },
    {
      "epoch": 0.2105984632589012,
      "grad_norm": 3.8585539033156553,
      "learning_rate": 3.6675337152670954e-06,
      "loss": 0.963,
      "step": 2919
    },
    {
      "epoch": 0.2106706107283287,
      "grad_norm": 6.034084464397754,
      "learning_rate": 3.667275632593492e-06,
      "loss": 0.952,
      "step": 2920
    },
    {
      "epoch": 0.2107427581977562,
      "grad_norm": 3.9923627184135486,
      "learning_rate": 3.667017458876139e-06,
      "loss": 0.9514,
      "step": 2921
    },
    {
      "epoch": 0.21081490566718372,
      "grad_norm": 4.060227448054067,
      "learning_rate": 3.666759194129133e-06,
      "loss": 1.0279,
      "step": 2922
    },
    {
      "epoch": 0.21088705313661124,
      "grad_norm": 4.300134594607276,
      "learning_rate": 3.666500838366578e-06,
      "loss": 0.9354,
      "step": 2923
    },
    {
      "epoch": 0.21095920060603873,
      "grad_norm": 13.094636217958255,
      "learning_rate": 3.6662423916025812e-06,
      "loss": 1.1103,
      "step": 2924
    },
    {
      "epoch": 0.21103134807546625,
      "grad_norm": 4.0390993334273135,
      "learning_rate": 3.665983853851256e-06,
      "loss": 1.0019,
      "step": 2925
    },
    {
      "epoch": 0.21110349554489377,
      "grad_norm": 0.9512199687352794,
      "learning_rate": 3.6657252251267193e-06,
      "loss": 0.9299,
      "step": 2926
    },
    {
      "epoch": 0.21117564301432126,
      "grad_norm": 5.845000597298676,
      "learning_rate": 3.6654665054430946e-06,
      "loss": 1.0415,
      "step": 2927
    },
    {
      "epoch": 0.21124779048374878,
      "grad_norm": 0.7665924266100723,
      "learning_rate": 3.6652076948145092e-06,
      "loss": 0.8528,
      "step": 2928
    },
    {
      "epoch": 0.2113199379531763,
      "grad_norm": 5.700405026150177,
      "learning_rate": 3.664948793255096e-06,
      "loss": 0.9418,
      "step": 2929
    },
    {
      "epoch": 0.2113920854226038,
      "grad_norm": 4.25646010453052,
      "learning_rate": 3.6646898007789934e-06,
      "loss": 1.0011,
      "step": 2930
    },
    {
      "epoch": 0.2114642328920313,
      "grad_norm": 8.014199053842459,
      "learning_rate": 3.664430717400342e-06,
      "loss": 1.0751,
      "step": 2931
    },
    {
      "epoch": 0.21153638036145883,
      "grad_norm": 10.541397955765468,
      "learning_rate": 3.6641715431332914e-06,
      "loss": 1.0271,
      "step": 2932
    },
    {
      "epoch": 0.21160852783088632,
      "grad_norm": 5.315281397923475,
      "learning_rate": 3.6639122779919935e-06,
      "loss": 0.8791,
      "step": 2933
    },
    {
      "epoch": 0.21168067530031384,
      "grad_norm": 3.9997143643437703,
      "learning_rate": 3.6636529219906054e-06,
      "loss": 0.9192,
      "step": 2934
    },
    {
      "epoch": 0.21175282276974136,
      "grad_norm": 4.51143274763144,
      "learning_rate": 3.6633934751432897e-06,
      "loss": 1.0323,
      "step": 2935
    },
    {
      "epoch": 0.21182497023916885,
      "grad_norm": 5.12037319581578,
      "learning_rate": 3.6631339374642138e-06,
      "loss": 1.0598,
      "step": 2936
    },
    {
      "epoch": 0.21189711770859637,
      "grad_norm": 4.012163265751398,
      "learning_rate": 3.662874308967551e-06,
      "loss": 1.0199,
      "step": 2937
    },
    {
      "epoch": 0.2119692651780239,
      "grad_norm": 4.574935479047893,
      "learning_rate": 3.662614589667478e-06,
      "loss": 0.9618,
      "step": 2938
    },
    {
      "epoch": 0.21204141264745138,
      "grad_norm": 3.668412674273042,
      "learning_rate": 3.662354779578176e-06,
      "loss": 1.2024,
      "step": 2939
    },
    {
      "epoch": 0.2121135601168789,
      "grad_norm": 5.54838038688792,
      "learning_rate": 3.662094878713834e-06,
      "loss": 1.0381,
      "step": 2940
    },
    {
      "epoch": 0.21218570758630642,
      "grad_norm": 6.030219904241126,
      "learning_rate": 3.6618348870886435e-06,
      "loss": 0.8768,
      "step": 2941
    },
    {
      "epoch": 0.2122578550557339,
      "grad_norm": 4.188871927561488,
      "learning_rate": 3.661574804716802e-06,
      "loss": 0.99,
      "step": 2942
    },
    {
      "epoch": 0.21233000252516143,
      "grad_norm": 6.4459691308513625,
      "learning_rate": 3.6613146316125105e-06,
      "loss": 1.0009,
      "step": 2943
    },
    {
      "epoch": 0.21240214999458895,
      "grad_norm": 3.9370252534978345,
      "learning_rate": 3.661054367789978e-06,
      "loss": 0.9168,
      "step": 2944
    },
    {
      "epoch": 0.21247429746401644,
      "grad_norm": 5.757253632668022,
      "learning_rate": 3.660794013263415e-06,
      "loss": 1.0704,
      "step": 2945
    },
    {
      "epoch": 0.21254644493344396,
      "grad_norm": 13.676344135623951,
      "learning_rate": 3.6605335680470384e-06,
      "loss": 1.0418,
      "step": 2946
    },
    {
      "epoch": 0.21261859240287148,
      "grad_norm": 4.454555134594649,
      "learning_rate": 3.6602730321550713e-06,
      "loss": 1.0246,
      "step": 2947
    },
    {
      "epoch": 0.21269073987229897,
      "grad_norm": 6.712319779063439,
      "learning_rate": 3.6600124056017404e-06,
      "loss": 1.0347,
      "step": 2948
    },
    {
      "epoch": 0.2127628873417265,
      "grad_norm": 3.920259059208791,
      "learning_rate": 3.659751688401277e-06,
      "loss": 0.9676,
      "step": 2949
    },
    {
      "epoch": 0.212835034811154,
      "grad_norm": 4.981434304382809,
      "learning_rate": 3.659490880567918e-06,
      "loss": 0.9057,
      "step": 2950
    },
    {
      "epoch": 0.2129071822805815,
      "grad_norm": 4.241000126977877,
      "learning_rate": 3.6592299821159048e-06,
      "loss": 1.0394,
      "step": 2951
    },
    {
      "epoch": 0.21297932975000902,
      "grad_norm": 6.217176382050564,
      "learning_rate": 3.658968993059485e-06,
      "loss": 1.0335,
      "step": 2952
    },
    {
      "epoch": 0.21305147721943654,
      "grad_norm": 6.843593090470784,
      "learning_rate": 3.65870791341291e-06,
      "loss": 1.049,
      "step": 2953
    },
    {
      "epoch": 0.21312362468886403,
      "grad_norm": 3.8955311921537694,
      "learning_rate": 3.6584467431904357e-06,
      "loss": 1.009,
      "step": 2954
    },
    {
      "epoch": 0.21319577215829155,
      "grad_norm": 6.939807405025564,
      "learning_rate": 3.6581854824063243e-06,
      "loss": 1.0064,
      "step": 2955
    },
    {
      "epoch": 0.21326791962771907,
      "grad_norm": 5.401137387620875,
      "learning_rate": 3.657924131074842e-06,
      "loss": 1.0305,
      "step": 2956
    },
    {
      "epoch": 0.21334006709714656,
      "grad_norm": 8.129984104567939,
      "learning_rate": 3.6576626892102605e-06,
      "loss": 0.9237,
      "step": 2957
    },
    {
      "epoch": 0.21341221456657408,
      "grad_norm": 4.652183446676754,
      "learning_rate": 3.6574011568268556e-06,
      "loss": 1.0785,
      "step": 2958
    },
    {
      "epoch": 0.21348436203600157,
      "grad_norm": 3.7520336994373955,
      "learning_rate": 3.657139533938909e-06,
      "loss": 0.9592,
      "step": 2959
    },
    {
      "epoch": 0.2135565095054291,
      "grad_norm": 9.14039337647238,
      "learning_rate": 3.656877820560708e-06,
      "loss": 1.06,
      "step": 2960
    },
    {
      "epoch": 0.2136286569748566,
      "grad_norm": 4.253845718835863,
      "learning_rate": 3.6566160167065415e-06,
      "loss": 0.9394,
      "step": 2961
    },
    {
      "epoch": 0.2137008044442841,
      "grad_norm": 7.199699331999317,
      "learning_rate": 3.6563541223907076e-06,
      "loss": 0.9743,
      "step": 2962
    },
    {
      "epoch": 0.21377295191371162,
      "grad_norm": 4.236120335858972,
      "learning_rate": 3.6560921376275068e-06,
      "loss": 1.0336,
      "step": 2963
    },
    {
      "epoch": 0.21384509938313914,
      "grad_norm": 4.956212955447053,
      "learning_rate": 3.655830062431244e-06,
      "loss": 0.9985,
      "step": 2964
    },
    {
      "epoch": 0.21391724685256663,
      "grad_norm": 5.455703257011368,
      "learning_rate": 3.655567896816232e-06,
      "loss": 1.0368,
      "step": 2965
    },
    {
      "epoch": 0.21398939432199415,
      "grad_norm": 3.8610513729776956,
      "learning_rate": 3.6553056407967858e-06,
      "loss": 1.0619,
      "step": 2966
    },
    {
      "epoch": 0.21406154179142167,
      "grad_norm": 5.14175894153133,
      "learning_rate": 3.6550432943872265e-06,
      "loss": 1.0015,
      "step": 2967
    },
    {
      "epoch": 0.21413368926084916,
      "grad_norm": 7.550472310431693,
      "learning_rate": 3.6547808576018796e-06,
      "loss": 1.0712,
      "step": 2968
    },
    {
      "epoch": 0.21420583673027668,
      "grad_norm": 3.9755902553863804,
      "learning_rate": 3.654518330455075e-06,
      "loss": 0.8274,
      "step": 2969
    },
    {
      "epoch": 0.2142779841997042,
      "grad_norm": 5.018719155488771,
      "learning_rate": 3.6542557129611505e-06,
      "loss": 1.0458,
      "step": 2970
    },
    {
      "epoch": 0.2143501316691317,
      "grad_norm": 4.057689694561257,
      "learning_rate": 3.6539930051344445e-06,
      "loss": 1.0662,
      "step": 2971
    },
    {
      "epoch": 0.2144222791385592,
      "grad_norm": 6.703131544559014,
      "learning_rate": 3.6537302069893035e-06,
      "loss": 1.1108,
      "step": 2972
    },
    {
      "epoch": 0.21449442660798673,
      "grad_norm": 7.452472138351165,
      "learning_rate": 3.6534673185400776e-06,
      "loss": 0.9403,
      "step": 2973
    },
    {
      "epoch": 0.21456657407741422,
      "grad_norm": 5.088383193527186,
      "learning_rate": 3.6532043398011227e-06,
      "loss": 1.0041,
      "step": 2974
    },
    {
      "epoch": 0.21463872154684174,
      "grad_norm": 5.772995122704295,
      "learning_rate": 3.6529412707867994e-06,
      "loss": 1.0281,
      "step": 2975
    },
    {
      "epoch": 0.21471086901626926,
      "grad_norm": 9.650906661534368,
      "learning_rate": 3.6526781115114708e-06,
      "loss": 0.9391,
      "step": 2976
    },
    {
      "epoch": 0.21478301648569675,
      "grad_norm": 5.1609878505245215,
      "learning_rate": 3.65241486198951e-06,
      "loss": 0.9414,
      "step": 2977
    },
    {
      "epoch": 0.21485516395512427,
      "grad_norm": 6.570967089400938,
      "learning_rate": 3.6521515222352893e-06,
      "loss": 1.0529,
      "step": 2978
    },
    {
      "epoch": 0.2149273114245518,
      "grad_norm": 3.913938478496072,
      "learning_rate": 3.651888092263191e-06,
      "loss": 0.9822,
      "step": 2979
    },
    {
      "epoch": 0.21499945889397928,
      "grad_norm": 12.985634716269166,
      "learning_rate": 3.6516245720875986e-06,
      "loss": 1.0105,
      "step": 2980
    },
    {
      "epoch": 0.2150716063634068,
      "grad_norm": 4.551003381113042,
      "learning_rate": 3.6513609617229024e-06,
      "loss": 1.0572,
      "step": 2981
    },
    {
      "epoch": 0.21514375383283432,
      "grad_norm": 5.137858379700476,
      "learning_rate": 3.6510972611834976e-06,
      "loss": 1.0024,
      "step": 2982
    },
    {
      "epoch": 0.21521590130226181,
      "grad_norm": 11.867748094066487,
      "learning_rate": 3.650833470483783e-06,
      "loss": 0.9813,
      "step": 2983
    },
    {
      "epoch": 0.21528804877168933,
      "grad_norm": 7.135416547863442,
      "learning_rate": 3.6505695896381643e-06,
      "loss": 0.9841,
      "step": 2984
    },
    {
      "epoch": 0.21536019624111685,
      "grad_norm": 6.491606648485575,
      "learning_rate": 3.65030561866105e-06,
      "loss": 0.9739,
      "step": 2985
    },
    {
      "epoch": 0.21543234371054434,
      "grad_norm": 4.959277642778147,
      "learning_rate": 3.6500415575668557e-06,
      "loss": 1.0968,
      "step": 2986
    },
    {
      "epoch": 0.21550449117997186,
      "grad_norm": 21.73865291966636,
      "learning_rate": 3.6497774063699994e-06,
      "loss": 0.9977,
      "step": 2987
    },
    {
      "epoch": 0.21557663864939938,
      "grad_norm": 6.605918323054649,
      "learning_rate": 3.6495131650849068e-06,
      "loss": 0.9609,
      "step": 2988
    },
    {
      "epoch": 0.21564878611882687,
      "grad_norm": 4.309636589005722,
      "learning_rate": 3.6492488337260065e-06,
      "loss": 1.0821,
      "step": 2989
    },
    {
      "epoch": 0.2157209335882544,
      "grad_norm": 5.611530707925096,
      "learning_rate": 3.648984412307732e-06,
      "loss": 1.1686,
      "step": 2990
    },
    {
      "epoch": 0.2157930810576819,
      "grad_norm": 7.1252021091640305,
      "learning_rate": 3.648719900844524e-06,
      "loss": 1.083,
      "step": 2991
    },
    {
      "epoch": 0.2158652285271094,
      "grad_norm": 0.7754298279506415,
      "learning_rate": 3.648455299350825e-06,
      "loss": 0.8342,
      "step": 2992
    },
    {
      "epoch": 0.21593737599653692,
      "grad_norm": 4.339615864783637,
      "learning_rate": 3.648190607841085e-06,
      "loss": 0.981,
      "step": 2993
    },
    {
      "epoch": 0.21600952346596444,
      "grad_norm": 0.7487009639954402,
      "learning_rate": 3.647925826329757e-06,
      "loss": 0.8251,
      "step": 2994
    },
    {
      "epoch": 0.21608167093539193,
      "grad_norm": 3.542037959364441,
      "learning_rate": 3.6476609548312996e-06,
      "loss": 1.0155,
      "step": 2995
    },
    {
      "epoch": 0.21615381840481945,
      "grad_norm": 3.8262522573629534,
      "learning_rate": 3.6473959933601777e-06,
      "loss": 0.9643,
      "step": 2996
    },
    {
      "epoch": 0.21622596587424697,
      "grad_norm": 8.754739839968252,
      "learning_rate": 3.6471309419308584e-06,
      "loss": 1.0677,
      "step": 2997
    },
    {
      "epoch": 0.21629811334367446,
      "grad_norm": 3.5433084310360523,
      "learning_rate": 3.6468658005578163e-06,
      "loss": 1.0695,
      "step": 2998
    },
    {
      "epoch": 0.21637026081310198,
      "grad_norm": 2.9878310718404046,
      "learning_rate": 3.646600569255529e-06,
      "loss": 1.0036,
      "step": 2999
    },
    {
      "epoch": 0.2164424082825295,
      "grad_norm": 4.463004520223364,
      "learning_rate": 3.64633524803848e-06,
      "loss": 1.0643,
      "step": 3000
    },
    {
      "epoch": 0.216514555751957,
      "grad_norm": 3.9700775329167826,
      "learning_rate": 3.6460698369211576e-06,
      "loss": 0.991,
      "step": 3001
    },
    {
      "epoch": 0.2165867032213845,
      "grad_norm": 13.480537230882273,
      "learning_rate": 3.645804335918056e-06,
      "loss": 1.0689,
      "step": 3002
    },
    {
      "epoch": 0.21665885069081203,
      "grad_norm": 6.802942413557367,
      "learning_rate": 3.6455387450436712e-06,
      "loss": 0.9623,
      "step": 3003
    },
    {
      "epoch": 0.21673099816023952,
      "grad_norm": 6.318034550189844,
      "learning_rate": 3.645273064312507e-06,
      "loss": 1.0717,
      "step": 3004
    },
    {
      "epoch": 0.21680314562966704,
      "grad_norm": 3.7773895734277816,
      "learning_rate": 3.6450072937390717e-06,
      "loss": 1.0062,
      "step": 3005
    },
    {
      "epoch": 0.21687529309909456,
      "grad_norm": 3.869510700333772,
      "learning_rate": 3.6447414333378775e-06,
      "loss": 0.9725,
      "step": 3006
    },
    {
      "epoch": 0.21694744056852205,
      "grad_norm": 4.037773118362533,
      "learning_rate": 3.6444754831234425e-06,
      "loss": 1.0198,
      "step": 3007
    },
    {
      "epoch": 0.21701958803794957,
      "grad_norm": 5.111165610773812,
      "learning_rate": 3.6442094431102886e-06,
      "loss": 0.9238,
      "step": 3008
    },
    {
      "epoch": 0.21709173550737706,
      "grad_norm": 8.293233701019558,
      "learning_rate": 3.6439433133129442e-06,
      "loss": 1.0159,
      "step": 3009
    },
    {
      "epoch": 0.21716388297680458,
      "grad_norm": 3.358187159409021,
      "learning_rate": 3.643677093745941e-06,
      "loss": 1.0843,
      "step": 3010
    },
    {
      "epoch": 0.2172360304462321,
      "grad_norm": 8.494322732081576,
      "learning_rate": 3.643410784423816e-06,
      "loss": 1.077,
      "step": 3011
    },
    {
      "epoch": 0.2173081779156596,
      "grad_norm": 6.1675772467183805,
      "learning_rate": 3.643144385361112e-06,
      "loss": 1.0861,
      "step": 3012
    },
    {
      "epoch": 0.21738032538508711,
      "grad_norm": 7.134737555122035,
      "learning_rate": 3.6428778965723755e-06,
      "loss": 1.0013,
      "step": 3013
    },
    {
      "epoch": 0.21745247285451463,
      "grad_norm": 4.793228204570877,
      "learning_rate": 3.6426113180721594e-06,
      "loss": 1.0664,
      "step": 3014
    },
    {
      "epoch": 0.21752462032394212,
      "grad_norm": 5.540558305006523,
      "learning_rate": 3.6423446498750197e-06,
      "loss": 0.9453,
      "step": 3015
    },
    {
      "epoch": 0.21759676779336964,
      "grad_norm": 0.7306570534572603,
      "learning_rate": 3.6420778919955184e-06,
      "loss": 0.7884,
      "step": 3016
    },
    {
      "epoch": 0.21766891526279716,
      "grad_norm": 4.984007436180384,
      "learning_rate": 3.6418110444482226e-06,
      "loss": 1.0281,
      "step": 3017
    },
    {
      "epoch": 0.21774106273222466,
      "grad_norm": 4.887030599936612,
      "learning_rate": 3.6415441072477027e-06,
      "loss": 0.9477,
      "step": 3018
    },
    {
      "epoch": 0.21781321020165217,
      "grad_norm": 4.086681757697348,
      "learning_rate": 3.6412770804085368e-06,
      "loss": 1.0351,
      "step": 3019
    },
    {
      "epoch": 0.2178853576710797,
      "grad_norm": 14.593484945565447,
      "learning_rate": 3.6410099639453042e-06,
      "loss": 0.9378,
      "step": 3020
    },
    {
      "epoch": 0.21795750514050719,
      "grad_norm": 5.833590401934221,
      "learning_rate": 3.640742757872593e-06,
      "loss": 0.9449,
      "step": 3021
    },
    {
      "epoch": 0.2180296526099347,
      "grad_norm": 5.8977270241605515,
      "learning_rate": 3.6404754622049937e-06,
      "loss": 1.0343,
      "step": 3022
    },
    {
      "epoch": 0.21810180007936222,
      "grad_norm": 3.529024080881277,
      "learning_rate": 3.6402080769571023e-06,
      "loss": 1.0171,
      "step": 3023
    },
    {
      "epoch": 0.21817394754878972,
      "grad_norm": 4.331611193284221,
      "learning_rate": 3.639940602143519e-06,
      "loss": 1.0902,
      "step": 3024
    },
    {
      "epoch": 0.21824609501821723,
      "grad_norm": 6.0971685748490625,
      "learning_rate": 3.6396730377788506e-06,
      "loss": 1.0309,
      "step": 3025
    },
    {
      "epoch": 0.21831824248764475,
      "grad_norm": 6.048810466756596,
      "learning_rate": 3.639405383877708e-06,
      "loss": 0.9866,
      "step": 3026
    },
    {
      "epoch": 0.21839038995707225,
      "grad_norm": 3.5537409937070548,
      "learning_rate": 3.6391376404547053e-06,
      "loss": 1.0001,
      "step": 3027
    },
    {
      "epoch": 0.21846253742649976,
      "grad_norm": 8.030805404765998,
      "learning_rate": 3.6388698075244644e-06,
      "loss": 1.0216,
      "step": 3028
    },
    {
      "epoch": 0.21853468489592728,
      "grad_norm": 5.081261238564313,
      "learning_rate": 3.6386018851016102e-06,
      "loss": 1.0334,
      "step": 3029
    },
    {
      "epoch": 0.21860683236535478,
      "grad_norm": 12.02458025859979,
      "learning_rate": 3.638333873200773e-06,
      "loss": 0.8108,
      "step": 3030
    },
    {
      "epoch": 0.2186789798347823,
      "grad_norm": 11.463259845870429,
      "learning_rate": 3.6380657718365877e-06,
      "loss": 0.9551,
      "step": 3031
    },
    {
      "epoch": 0.2187511273042098,
      "grad_norm": 4.469831429189594,
      "learning_rate": 3.6377975810236946e-06,
      "loss": 0.9617,
      "step": 3032
    },
    {
      "epoch": 0.2188232747736373,
      "grad_norm": 9.027061467892926,
      "learning_rate": 3.637529300776739e-06,
      "loss": 1.0181,
      "step": 3033
    },
    {
      "epoch": 0.21889542224306482,
      "grad_norm": 0.9143057605448869,
      "learning_rate": 3.6372609311103694e-06,
      "loss": 0.9194,
      "step": 3034
    },
    {
      "epoch": 0.21896756971249234,
      "grad_norm": 3.9921999220085604,
      "learning_rate": 3.636992472039242e-06,
      "loss": 1.0324,
      "step": 3035
    },
    {
      "epoch": 0.21903971718191984,
      "grad_norm": 4.843119131043595,
      "learning_rate": 3.6367239235780156e-06,
      "loss": 1.034,
      "step": 3036
    },
    {
      "epoch": 0.21911186465134735,
      "grad_norm": 7.5672907511700735,
      "learning_rate": 3.636455285741354e-06,
      "loss": 1.0161,
      "step": 3037
    },
    {
      "epoch": 0.21918401212077487,
      "grad_norm": 3.268938345798276,
      "learning_rate": 3.636186558543928e-06,
      "loss": 0.9053,
      "step": 3038
    },
    {
      "epoch": 0.21925615959020237,
      "grad_norm": 5.13413039290564,
      "learning_rate": 3.6359177420004104e-06,
      "loss": 1.0234,
      "step": 3039
    },
    {
      "epoch": 0.21932830705962988,
      "grad_norm": 5.298426578097616,
      "learning_rate": 3.635648836125481e-06,
      "loss": 0.9784,
      "step": 3040
    },
    {
      "epoch": 0.2194004545290574,
      "grad_norm": 5.211532627419707,
      "learning_rate": 3.635379840933824e-06,
      "loss": 0.9494,
      "step": 3041
    },
    {
      "epoch": 0.2194726019984849,
      "grad_norm": 5.681696015865264,
      "learning_rate": 3.6351107564401274e-06,
      "loss": 1.0099,
      "step": 3042
    },
    {
      "epoch": 0.21954474946791241,
      "grad_norm": 5.228040272447414,
      "learning_rate": 3.634841582659085e-06,
      "loss": 0.9641,
      "step": 3043
    },
    {
      "epoch": 0.21961689693733993,
      "grad_norm": 3.1435616651056453,
      "learning_rate": 3.6345723196053964e-06,
      "loss": 0.9404,
      "step": 3044
    },
    {
      "epoch": 0.21968904440676743,
      "grad_norm": 5.366964854768414,
      "learning_rate": 3.6343029672937645e-06,
      "loss": 1.0387,
      "step": 3045
    },
    {
      "epoch": 0.21976119187619494,
      "grad_norm": 15.63056931994561,
      "learning_rate": 3.634033525738897e-06,
      "loss": 0.9217,
      "step": 3046
    },
    {
      "epoch": 0.21983333934562246,
      "grad_norm": 6.902054422787968,
      "learning_rate": 3.633763994955508e-06,
      "loss": 0.9322,
      "step": 3047
    },
    {
      "epoch": 0.21990548681504996,
      "grad_norm": 4.479344327493483,
      "learning_rate": 3.633494374958315e-06,
      "loss": 1.0021,
      "step": 3048
    },
    {
      "epoch": 0.21997763428447747,
      "grad_norm": 4.803522033761582,
      "learning_rate": 3.6332246657620415e-06,
      "loss": 0.9538,
      "step": 3049
    },
    {
      "epoch": 0.220049781753905,
      "grad_norm": 8.085548280940381,
      "learning_rate": 3.6329548673814148e-06,
      "loss": 1.0959,
      "step": 3050
    },
    {
      "epoch": 0.22012192922333249,
      "grad_norm": 7.140080070153494,
      "learning_rate": 3.632684979831168e-06,
      "loss": 0.9176,
      "step": 3051
    },
    {
      "epoch": 0.22019407669276,
      "grad_norm": 8.817595611650916,
      "learning_rate": 3.6324150031260377e-06,
      "loss": 0.9833,
      "step": 3052
    },
    {
      "epoch": 0.22026622416218752,
      "grad_norm": 4.654918096131172,
      "learning_rate": 3.632144937280768e-06,
      "loss": 1.0003,
      "step": 3053
    },
    {
      "epoch": 0.22033837163161502,
      "grad_norm": 5.346863670493184,
      "learning_rate": 3.631874782310105e-06,
      "loss": 1.1255,
      "step": 3054
    },
    {
      "epoch": 0.22041051910104253,
      "grad_norm": 5.607465243595352,
      "learning_rate": 3.6316045382288012e-06,
      "loss": 1.0581,
      "step": 3055
    },
    {
      "epoch": 0.22048266657047005,
      "grad_norm": 6.843016981470337,
      "learning_rate": 3.6313342050516133e-06,
      "loss": 0.9819,
      "step": 3056
    },
    {
      "epoch": 0.22055481403989755,
      "grad_norm": 4.082583507798038,
      "learning_rate": 3.6310637827933035e-06,
      "loss": 0.9687,
      "step": 3057
    },
    {
      "epoch": 0.22062696150932506,
      "grad_norm": 8.624524089564849,
      "learning_rate": 3.6307932714686382e-06,
      "loss": 0.9374,
      "step": 3058
    },
    {
      "epoch": 0.22069910897875258,
      "grad_norm": 5.126414220189196,
      "learning_rate": 3.6305226710923897e-06,
      "loss": 1.0415,
      "step": 3059
    },
    {
      "epoch": 0.22077125644818008,
      "grad_norm": 5.558640416177969,
      "learning_rate": 3.6302519816793344e-06,
      "loss": 1.0331,
      "step": 3060
    },
    {
      "epoch": 0.2208434039176076,
      "grad_norm": 4.068178170097424,
      "learning_rate": 3.629981203244253e-06,
      "loss": 1.0092,
      "step": 3061
    },
    {
      "epoch": 0.2209155513870351,
      "grad_norm": 7.079767234553618,
      "learning_rate": 3.629710335801932e-06,
      "loss": 0.9295,
      "step": 3062
    },
    {
      "epoch": 0.2209876988564626,
      "grad_norm": 5.93093284024265,
      "learning_rate": 3.629439379367163e-06,
      "loss": 1.0344,
      "step": 3063
    },
    {
      "epoch": 0.22105984632589012,
      "grad_norm": 4.350048082458646,
      "learning_rate": 3.629168333954741e-06,
      "loss": 0.9977,
      "step": 3064
    },
    {
      "epoch": 0.22113199379531762,
      "grad_norm": 5.872292401966988,
      "learning_rate": 3.6288971995794673e-06,
      "loss": 0.9975,
      "step": 3065
    },
    {
      "epoch": 0.22120414126474514,
      "grad_norm": 2.926802941917243,
      "learning_rate": 3.6286259762561475e-06,
      "loss": 0.9535,
      "step": 3066
    },
    {
      "epoch": 0.22127628873417265,
      "grad_norm": 10.529428567913687,
      "learning_rate": 3.6283546639995927e-06,
      "loss": 1.0436,
      "step": 3067
    },
    {
      "epoch": 0.22134843620360015,
      "grad_norm": 3.3644492926731338,
      "learning_rate": 3.628083262824617e-06,
      "loss": 1.1102,
      "step": 3068
    },
    {
      "epoch": 0.22142058367302767,
      "grad_norm": 5.061740394401169,
      "learning_rate": 3.6278117727460414e-06,
      "loss": 1.0617,
      "step": 3069
    },
    {
      "epoch": 0.22149273114245518,
      "grad_norm": 4.045513616331291,
      "learning_rate": 3.627540193778691e-06,
      "loss": 1.0214,
      "step": 3070
    },
    {
      "epoch": 0.22156487861188268,
      "grad_norm": 8.237514411323225,
      "learning_rate": 3.6272685259373957e-06,
      "loss": 0.9186,
      "step": 3071
    },
    {
      "epoch": 0.2216370260813102,
      "grad_norm": 3.104333826125715,
      "learning_rate": 3.6269967692369894e-06,
      "loss": 0.9729,
      "step": 3072
    },
    {
      "epoch": 0.22170917355073771,
      "grad_norm": 11.82466008219984,
      "learning_rate": 3.626724923692313e-06,
      "loss": 1.0009,
      "step": 3073
    },
    {
      "epoch": 0.2217813210201652,
      "grad_norm": 7.071607540667717,
      "learning_rate": 3.6264529893182108e-06,
      "loss": 1.0819,
      "step": 3074
    },
    {
      "epoch": 0.22185346848959273,
      "grad_norm": 4.991836941489079,
      "learning_rate": 3.6261809661295316e-06,
      "loss": 1.0163,
      "step": 3075
    },
    {
      "epoch": 0.22192561595902024,
      "grad_norm": 5.359529987907911,
      "learning_rate": 3.6259088541411293e-06,
      "loss": 1.037,
      "step": 3076
    },
    {
      "epoch": 0.22199776342844774,
      "grad_norm": 0.6989344439756532,
      "learning_rate": 3.625636653367864e-06,
      "loss": 0.8074,
      "step": 3077
    },
    {
      "epoch": 0.22206991089787526,
      "grad_norm": 4.387830177338354,
      "learning_rate": 3.625364363824598e-06,
      "loss": 1.0623,
      "step": 3078
    },
    {
      "epoch": 0.22214205836730277,
      "grad_norm": 3.6154000333466625,
      "learning_rate": 3.6250919855262017e-06,
      "loss": 0.9599,
      "step": 3079
    },
    {
      "epoch": 0.22221420583673027,
      "grad_norm": 6.9455228587027,
      "learning_rate": 3.624819518487548e-06,
      "loss": 0.9736,
      "step": 3080
    },
    {
      "epoch": 0.22228635330615779,
      "grad_norm": 5.5689486723973225,
      "learning_rate": 3.624546962723515e-06,
      "loss": 0.9825,
      "step": 3081
    },
    {
      "epoch": 0.2223585007755853,
      "grad_norm": 3.5634155435478663,
      "learning_rate": 3.624274318248986e-06,
      "loss": 1.0333,
      "step": 3082
    },
    {
      "epoch": 0.2224306482450128,
      "grad_norm": 3.4989054194442453,
      "learning_rate": 3.62400158507885e-06,
      "loss": 0.9299,
      "step": 3083
    },
    {
      "epoch": 0.22250279571444032,
      "grad_norm": 15.845994779031999,
      "learning_rate": 3.6237287632279992e-06,
      "loss": 1.1389,
      "step": 3084
    },
    {
      "epoch": 0.22257494318386783,
      "grad_norm": 4.4001483198656395,
      "learning_rate": 3.623455852711331e-06,
      "loss": 0.9455,
      "step": 3085
    },
    {
      "epoch": 0.22264709065329533,
      "grad_norm": 4.978284792302124,
      "learning_rate": 3.623182853543749e-06,
      "loss": 0.8785,
      "step": 3086
    },
    {
      "epoch": 0.22271923812272285,
      "grad_norm": 6.263998605350422,
      "learning_rate": 3.62290976574016e-06,
      "loss": 1.0132,
      "step": 3087
    },
    {
      "epoch": 0.22279138559215036,
      "grad_norm": 0.9187763314139317,
      "learning_rate": 3.622636589315476e-06,
      "loss": 0.9099,
      "step": 3088
    },
    {
      "epoch": 0.22286353306157786,
      "grad_norm": 0.9079719823577226,
      "learning_rate": 3.622363324284615e-06,
      "loss": 0.8761,
      "step": 3089
    },
    {
      "epoch": 0.22293568053100538,
      "grad_norm": 4.526239821731513,
      "learning_rate": 3.622089970662499e-06,
      "loss": 0.9809,
      "step": 3090
    },
    {
      "epoch": 0.2230078280004329,
      "grad_norm": 10.799603306584453,
      "learning_rate": 3.6218165284640546e-06,
      "loss": 0.8536,
      "step": 3091
    },
    {
      "epoch": 0.2230799754698604,
      "grad_norm": 0.7408345388811026,
      "learning_rate": 3.621542997704212e-06,
      "loss": 0.8116,
      "step": 3092
    },
    {
      "epoch": 0.2231521229392879,
      "grad_norm": 4.278461968559058,
      "learning_rate": 3.6212693783979103e-06,
      "loss": 0.9631,
      "step": 3093
    },
    {
      "epoch": 0.22322427040871543,
      "grad_norm": 8.188624181082186,
      "learning_rate": 3.62099567056009e-06,
      "loss": 0.9058,
      "step": 3094
    },
    {
      "epoch": 0.22329641787814292,
      "grad_norm": 0.8067677099874647,
      "learning_rate": 3.620721874205696e-06,
      "loss": 0.8769,
      "step": 3095
    },
    {
      "epoch": 0.22336856534757044,
      "grad_norm": 6.062618608395974,
      "learning_rate": 3.6204479893496798e-06,
      "loss": 0.8732,
      "step": 3096
    },
    {
      "epoch": 0.22344071281699796,
      "grad_norm": 3.7950882337764478,
      "learning_rate": 3.620174016006998e-06,
      "loss": 1.0167,
      "step": 3097
    },
    {
      "epoch": 0.22351286028642545,
      "grad_norm": 5.866753167179388,
      "learning_rate": 3.6198999541926114e-06,
      "loss": 1.0709,
      "step": 3098
    },
    {
      "epoch": 0.22358500775585297,
      "grad_norm": 3.8346278174885793,
      "learning_rate": 3.619625803921484e-06,
      "loss": 0.9422,
      "step": 3099
    },
    {
      "epoch": 0.22365715522528049,
      "grad_norm": 4.404851380351373,
      "learning_rate": 3.619351565208588e-06,
      "loss": 0.8352,
      "step": 3100
    },
    {
      "epoch": 0.22372930269470798,
      "grad_norm": 6.380648037038134,
      "learning_rate": 3.6190772380688973e-06,
      "loss": 0.9833,
      "step": 3101
    },
    {
      "epoch": 0.2238014501641355,
      "grad_norm": 2.9280164505126427,
      "learning_rate": 3.618802822517392e-06,
      "loss": 0.9747,
      "step": 3102
    },
    {
      "epoch": 0.22387359763356302,
      "grad_norm": 9.70654865500566,
      "learning_rate": 3.6185283185690577e-06,
      "loss": 1.0659,
      "step": 3103
    },
    {
      "epoch": 0.2239457451029905,
      "grad_norm": 4.312205207811018,
      "learning_rate": 3.618253726238883e-06,
      "loss": 1.0056,
      "step": 3104
    },
    {
      "epoch": 0.22401789257241803,
      "grad_norm": 30.334485825104032,
      "learning_rate": 3.617979045541863e-06,
      "loss": 0.9683,
      "step": 3105
    },
    {
      "epoch": 0.22409004004184555,
      "grad_norm": 3.6663744694429545,
      "learning_rate": 3.617704276492997e-06,
      "loss": 1.0585,
      "step": 3106
    },
    {
      "epoch": 0.22416218751127304,
      "grad_norm": 7.95344299999687,
      "learning_rate": 3.6174294191072886e-06,
      "loss": 1.0238,
      "step": 3107
    },
    {
      "epoch": 0.22423433498070056,
      "grad_norm": 14.080063901236054,
      "learning_rate": 3.617154473399748e-06,
      "loss": 1.1188,
      "step": 3108
    },
    {
      "epoch": 0.22430648245012808,
      "grad_norm": 20.69577036229074,
      "learning_rate": 3.6168794393853874e-06,
      "loss": 0.9991,
      "step": 3109
    },
    {
      "epoch": 0.22437862991955557,
      "grad_norm": 6.494378226386517,
      "learning_rate": 3.6166043170792265e-06,
      "loss": 1.0186,
      "step": 3110
    },
    {
      "epoch": 0.22445077738898309,
      "grad_norm": 4.361072059559094,
      "learning_rate": 3.616329106496288e-06,
      "loss": 1.0628,
      "step": 3111
    },
    {
      "epoch": 0.22452292485841058,
      "grad_norm": 4.513068191388434,
      "learning_rate": 3.6160538076516007e-06,
      "loss": 0.9586,
      "step": 3112
    },
    {
      "epoch": 0.2245950723278381,
      "grad_norm": 3.2990696549290575,
      "learning_rate": 3.615778420560197e-06,
      "loss": 0.9677,
      "step": 3113
    },
    {
      "epoch": 0.22466721979726562,
      "grad_norm": 3.542893918896901,
      "learning_rate": 3.6155029452371154e-06,
      "loss": 0.9651,
      "step": 3114
    },
    {
      "epoch": 0.2247393672666931,
      "grad_norm": 2.961553585002443,
      "learning_rate": 3.615227381697398e-06,
      "loss": 0.9169,
      "step": 3115
    },
    {
      "epoch": 0.22481151473612063,
      "grad_norm": 6.086404765338015,
      "learning_rate": 3.6149517299560933e-06,
      "loss": 0.946,
      "step": 3116
    },
    {
      "epoch": 0.22488366220554815,
      "grad_norm": 4.5007491018070525,
      "learning_rate": 3.6146759900282526e-06,
      "loss": 1.0132,
      "step": 3117
    },
    {
      "epoch": 0.22495580967497564,
      "grad_norm": 4.928325192794467,
      "learning_rate": 3.6144001619289335e-06,
      "loss": 0.9931,
      "step": 3118
    },
    {
      "epoch": 0.22502795714440316,
      "grad_norm": 4.741715484707517,
      "learning_rate": 3.6141242456731977e-06,
      "loss": 0.9813,
      "step": 3119
    },
    {
      "epoch": 0.22510010461383068,
      "grad_norm": 2.961911969635347,
      "learning_rate": 3.613848241276113e-06,
      "loss": 1.0058,
      "step": 3120
    },
    {
      "epoch": 0.22517225208325817,
      "grad_norm": 5.638243618998538,
      "learning_rate": 3.6135721487527486e-06,
      "loss": 0.9324,
      "step": 3121
    },
    {
      "epoch": 0.2252443995526857,
      "grad_norm": 5.973783437823279,
      "learning_rate": 3.6132959681181835e-06,
      "loss": 0.9899,
      "step": 3122
    },
    {
      "epoch": 0.2253165470221132,
      "grad_norm": 66.3960359791908,
      "learning_rate": 3.6130196993874976e-06,
      "loss": 1.0416,
      "step": 3123
    },
    {
      "epoch": 0.2253886944915407,
      "grad_norm": 5.914016380785334,
      "learning_rate": 3.6127433425757766e-06,
      "loss": 0.9113,
      "step": 3124
    },
    {
      "epoch": 0.22546084196096822,
      "grad_norm": 0.7962046402807254,
      "learning_rate": 3.612466897698112e-06,
      "loss": 0.8333,
      "step": 3125
    },
    {
      "epoch": 0.22553298943039574,
      "grad_norm": 5.221855655727934,
      "learning_rate": 3.6121903647696e-06,
      "loss": 0.9303,
      "step": 3126
    },
    {
      "epoch": 0.22560513689982323,
      "grad_norm": 3.776913386605448,
      "learning_rate": 3.611913743805339e-06,
      "loss": 1.0213,
      "step": 3127
    },
    {
      "epoch": 0.22567728436925075,
      "grad_norm": 5.180179072050103,
      "learning_rate": 3.611637034820437e-06,
      "loss": 1.0246,
      "step": 3128
    },
    {
      "epoch": 0.22574943183867827,
      "grad_norm": 3.1570258367313153,
      "learning_rate": 3.611360237830002e-06,
      "loss": 1.0777,
      "step": 3129
    },
    {
      "epoch": 0.22582157930810576,
      "grad_norm": 4.191424452430186,
      "learning_rate": 3.611083352849149e-06,
      "loss": 1.0717,
      "step": 3130
    },
    {
      "epoch": 0.22589372677753328,
      "grad_norm": 8.664273396187745,
      "learning_rate": 3.610806379892999e-06,
      "loss": 0.9476,
      "step": 3131
    },
    {
      "epoch": 0.2259658742469608,
      "grad_norm": 7.145242145725009,
      "learning_rate": 3.610529318976675e-06,
      "loss": 0.9279,
      "step": 3132
    },
    {
      "epoch": 0.2260380217163883,
      "grad_norm": 5.408152162939286,
      "learning_rate": 3.6102521701153065e-06,
      "loss": 1.032,
      "step": 3133
    },
    {
      "epoch": 0.2261101691858158,
      "grad_norm": 5.432615344266495,
      "learning_rate": 3.609974933324029e-06,
      "loss": 1.0247,
      "step": 3134
    },
    {
      "epoch": 0.22618231665524333,
      "grad_norm": 5.5501355541274355,
      "learning_rate": 3.6096976086179797e-06,
      "loss": 0.9499,
      "step": 3135
    },
    {
      "epoch": 0.22625446412467082,
      "grad_norm": 3.888207967878911,
      "learning_rate": 3.609420196012303e-06,
      "loss": 0.9709,
      "step": 3136
    },
    {
      "epoch": 0.22632661159409834,
      "grad_norm": 3.2590318040118214,
      "learning_rate": 3.6091426955221465e-06,
      "loss": 0.9842,
      "step": 3137
    },
    {
      "epoch": 0.22639875906352586,
      "grad_norm": 4.557720659682243,
      "learning_rate": 3.608865107162665e-06,
      "loss": 0.9651,
      "step": 3138
    },
    {
      "epoch": 0.22647090653295335,
      "grad_norm": 4.604752010001016,
      "learning_rate": 3.6085874309490157e-06,
      "loss": 0.9561,
      "step": 3139
    },
    {
      "epoch": 0.22654305400238087,
      "grad_norm": 6.541766738936079,
      "learning_rate": 3.608309666896362e-06,
      "loss": 0.9144,
      "step": 3140
    },
    {
      "epoch": 0.2266152014718084,
      "grad_norm": 10.713910937339463,
      "learning_rate": 3.6080318150198705e-06,
      "loss": 1.0031,
      "step": 3141
    },
    {
      "epoch": 0.22668734894123588,
      "grad_norm": 2.8613111267626494,
      "learning_rate": 3.6077538753347145e-06,
      "loss": 1.0873,
      "step": 3142
    },
    {
      "epoch": 0.2267594964106634,
      "grad_norm": 0.8436623987004263,
      "learning_rate": 3.6074758478560716e-06,
      "loss": 0.9218,
      "step": 3143
    },
    {
      "epoch": 0.22683164388009092,
      "grad_norm": 4.004028437541084,
      "learning_rate": 3.6071977325991227e-06,
      "loss": 0.9521,
      "step": 3144
    },
    {
      "epoch": 0.2269037913495184,
      "grad_norm": 5.032475861189702,
      "learning_rate": 3.6069195295790556e-06,
      "loss": 0.8923,
      "step": 3145
    },
    {
      "epoch": 0.22697593881894593,
      "grad_norm": 3.684144772753102,
      "learning_rate": 3.6066412388110614e-06,
      "loss": 1.0537,
      "step": 3146
    },
    {
      "epoch": 0.22704808628837345,
      "grad_norm": 3.331627806805348,
      "learning_rate": 3.606362860310337e-06,
      "loss": 1.0854,
      "step": 3147
    },
    {
      "epoch": 0.22712023375780094,
      "grad_norm": 5.857254499109116,
      "learning_rate": 3.606084394092083e-06,
      "loss": 0.9097,
      "step": 3148
    },
    {
      "epoch": 0.22719238122722846,
      "grad_norm": 4.862454435237993,
      "learning_rate": 3.605805840171506e-06,
      "loss": 0.9748,
      "step": 3149
    },
    {
      "epoch": 0.22726452869665598,
      "grad_norm": 4.3999109866068284,
      "learning_rate": 3.6055271985638163e-06,
      "loss": 1.0102,
      "step": 3150
    },
    {
      "epoch": 0.22733667616608347,
      "grad_norm": 6.539877418045877,
      "learning_rate": 3.6052484692842297e-06,
      "loss": 0.9803,
      "step": 3151
    },
    {
      "epoch": 0.227408823635511,
      "grad_norm": 9.239283359364453,
      "learning_rate": 3.6049696523479673e-06,
      "loss": 0.982,
      "step": 3152
    },
    {
      "epoch": 0.2274809711049385,
      "grad_norm": 4.007398676932929,
      "learning_rate": 3.6046907477702524e-06,
      "loss": 1.0252,
      "step": 3153
    },
    {
      "epoch": 0.227553118574366,
      "grad_norm": 3.7458053016391664,
      "learning_rate": 3.6044117555663166e-06,
      "loss": 1.0495,
      "step": 3154
    },
    {
      "epoch": 0.22762526604379352,
      "grad_norm": 6.255450809129543,
      "learning_rate": 3.604132675751395e-06,
      "loss": 0.9571,
      "step": 3155
    },
    {
      "epoch": 0.22769741351322104,
      "grad_norm": 3.3706893236278033,
      "learning_rate": 3.6038535083407256e-06,
      "loss": 0.9607,
      "step": 3156
    },
    {
      "epoch": 0.22776956098264853,
      "grad_norm": 4.77253197604818,
      "learning_rate": 3.6035742533495533e-06,
      "loss": 1.0655,
      "step": 3157
    },
    {
      "epoch": 0.22784170845207605,
      "grad_norm": 3.909752092234173,
      "learning_rate": 3.6032949107931276e-06,
      "loss": 0.9221,
      "step": 3158
    },
    {
      "epoch": 0.22791385592150357,
      "grad_norm": 3.5450701049415363,
      "learning_rate": 3.6030154806867018e-06,
      "loss": 0.9303,
      "step": 3159
    },
    {
      "epoch": 0.22798600339093106,
      "grad_norm": 5.544254339386819,
      "learning_rate": 3.6027359630455347e-06,
      "loss": 1.0655,
      "step": 3160
    },
    {
      "epoch": 0.22805815086035858,
      "grad_norm": 3.7288405774861344,
      "learning_rate": 3.6024563578848896e-06,
      "loss": 1.036,
      "step": 3161
    },
    {
      "epoch": 0.2281302983297861,
      "grad_norm": 3.439731775073202,
      "learning_rate": 3.6021766652200354e-06,
      "loss": 1.0316,
      "step": 3162
    },
    {
      "epoch": 0.2282024457992136,
      "grad_norm": 8.074536232872855,
      "learning_rate": 3.601896885066244e-06,
      "loss": 1.0014,
      "step": 3163
    },
    {
      "epoch": 0.2282745932686411,
      "grad_norm": 3.5065492934054867,
      "learning_rate": 3.601617017438794e-06,
      "loss": 1.0048,
      "step": 3164
    },
    {
      "epoch": 0.2283467407380686,
      "grad_norm": 0.7894578075694758,
      "learning_rate": 3.601337062352968e-06,
      "loss": 0.8724,
      "step": 3165
    },
    {
      "epoch": 0.22841888820749612,
      "grad_norm": 3.2409269892947243,
      "learning_rate": 3.6010570198240523e-06,
      "loss": 0.9721,
      "step": 3166
    },
    {
      "epoch": 0.22849103567692364,
      "grad_norm": 3.085790703094697,
      "learning_rate": 3.6007768898673404e-06,
      "loss": 0.954,
      "step": 3167
    },
    {
      "epoch": 0.22856318314635113,
      "grad_norm": 4.394669485673031,
      "learning_rate": 3.6004966724981277e-06,
      "loss": 1.0599,
      "step": 3168
    },
    {
      "epoch": 0.22863533061577865,
      "grad_norm": 4.418647369962682,
      "learning_rate": 3.6002163677317166e-06,
      "loss": 1.0469,
      "step": 3169
    },
    {
      "epoch": 0.22870747808520617,
      "grad_norm": 3.2878250367515056,
      "learning_rate": 3.599935975583414e-06,
      "loss": 0.9995,
      "step": 3170
    },
    {
      "epoch": 0.22877962555463366,
      "grad_norm": 4.359999745220212,
      "learning_rate": 3.5996554960685304e-06,
      "loss": 0.9648,
      "step": 3171
    },
    {
      "epoch": 0.22885177302406118,
      "grad_norm": 5.532509353291931,
      "learning_rate": 3.599374929202382e-06,
      "loss": 1.1039,
      "step": 3172
    },
    {
      "epoch": 0.2289239204934887,
      "grad_norm": 3.856144468078369,
      "learning_rate": 3.599094275000289e-06,
      "loss": 1.1251,
      "step": 3173
    },
    {
      "epoch": 0.2289960679629162,
      "grad_norm": 5.691890824215197,
      "learning_rate": 3.5988135334775775e-06,
      "loss": 0.9207,
      "step": 3174
    },
    {
      "epoch": 0.2290682154323437,
      "grad_norm": 2.5776028190995186,
      "learning_rate": 3.5985327046495773e-06,
      "loss": 0.9248,
      "step": 3175
    },
    {
      "epoch": 0.22914036290177123,
      "grad_norm": 4.359311557978104,
      "learning_rate": 3.5982517885316232e-06,
      "loss": 0.9873,
      "step": 3176
    },
    {
      "epoch": 0.22921251037119872,
      "grad_norm": 4.626648480235771,
      "learning_rate": 3.597970785139057e-06,
      "loss": 1.0233,
      "step": 3177
    },
    {
      "epoch": 0.22928465784062624,
      "grad_norm": 4.325213503115676,
      "learning_rate": 3.5976896944872205e-06,
      "loss": 1.0219,
      "step": 3178
    },
    {
      "epoch": 0.22935680531005376,
      "grad_norm": 4.33862023742286,
      "learning_rate": 3.597408516591464e-06,
      "loss": 1.0634,
      "step": 3179
    },
    {
      "epoch": 0.22942895277948125,
      "grad_norm": 4.484609139208201,
      "learning_rate": 3.5971272514671426e-06,
      "loss": 0.88,
      "step": 3180
    },
    {
      "epoch": 0.22950110024890877,
      "grad_norm": 2.898860766822436,
      "learning_rate": 3.5968458991296144e-06,
      "loss": 1.0858,
      "step": 3181
    },
    {
      "epoch": 0.2295732477183363,
      "grad_norm": 2.701223597717472,
      "learning_rate": 3.5965644595942423e-06,
      "loss": 1.0073,
      "step": 3182
    },
    {
      "epoch": 0.22964539518776378,
      "grad_norm": 3.3780379574383903,
      "learning_rate": 3.596282932876396e-06,
      "loss": 1.0334,
      "step": 3183
    },
    {
      "epoch": 0.2297175426571913,
      "grad_norm": 4.402892653086106,
      "learning_rate": 3.5960013189914477e-06,
      "loss": 0.906,
      "step": 3184
    },
    {
      "epoch": 0.22978969012661882,
      "grad_norm": 3.0397167139070156,
      "learning_rate": 3.5957196179547756e-06,
      "loss": 0.9944,
      "step": 3185
    },
    {
      "epoch": 0.2298618375960463,
      "grad_norm": 2.983980961695423,
      "learning_rate": 3.5954378297817617e-06,
      "loss": 0.9814,
      "step": 3186
    },
    {
      "epoch": 0.22993398506547383,
      "grad_norm": 3.3284025121333927,
      "learning_rate": 3.595155954487795e-06,
      "loss": 0.9612,
      "step": 3187
    },
    {
      "epoch": 0.23000613253490135,
      "grad_norm": 4.4020404592731905,
      "learning_rate": 3.594873992088266e-06,
      "loss": 1.0088,
      "step": 3188
    },
    {
      "epoch": 0.23007828000432884,
      "grad_norm": 3.6738930209247003,
      "learning_rate": 3.594591942598573e-06,
      "loss": 1.0119,
      "step": 3189
    },
    {
      "epoch": 0.23015042747375636,
      "grad_norm": 3.2966874055689024,
      "learning_rate": 3.5943098060341166e-06,
      "loss": 0.981,
      "step": 3190
    },
    {
      "epoch": 0.23022257494318388,
      "grad_norm": 3.435612385674439,
      "learning_rate": 3.5940275824103033e-06,
      "loss": 1.0444,
      "step": 3191
    },
    {
      "epoch": 0.23029472241261137,
      "grad_norm": 4.008677844643267,
      "learning_rate": 3.593745271742545e-06,
      "loss": 0.9925,
      "step": 3192
    },
    {
      "epoch": 0.2303668698820389,
      "grad_norm": 2.846630105164252,
      "learning_rate": 3.593462874046257e-06,
      "loss": 1.0051,
      "step": 3193
    },
    {
      "epoch": 0.2304390173514664,
      "grad_norm": 3.5103053557334185,
      "learning_rate": 3.5931803893368604e-06,
      "loss": 0.9899,
      "step": 3194
    },
    {
      "epoch": 0.2305111648208939,
      "grad_norm": 5.068652711586083,
      "learning_rate": 3.592897817629781e-06,
      "loss": 1.0057,
      "step": 3195
    },
    {
      "epoch": 0.23058331229032142,
      "grad_norm": 3.0255468142280413,
      "learning_rate": 3.592615158940448e-06,
      "loss": 0.9755,
      "step": 3196
    },
    {
      "epoch": 0.23065545975974894,
      "grad_norm": 0.8602238364349738,
      "learning_rate": 3.592332413284296e-06,
      "loss": 0.8742,
      "step": 3197
    },
    {
      "epoch": 0.23072760722917643,
      "grad_norm": 0.7713320424789776,
      "learning_rate": 3.592049580676767e-06,
      "loss": 0.8893,
      "step": 3198
    },
    {
      "epoch": 0.23079975469860395,
      "grad_norm": 5.273064222900014,
      "learning_rate": 3.5917666611333032e-06,
      "loss": 1.1153,
      "step": 3199
    },
    {
      "epoch": 0.23087190216803147,
      "grad_norm": 4.387069185412701,
      "learning_rate": 3.5914836546693547e-06,
      "loss": 1.0203,
      "step": 3200
    },
    {
      "epoch": 0.23094404963745896,
      "grad_norm": 3.9989712107395485,
      "learning_rate": 3.5912005613003755e-06,
      "loss": 0.8874,
      "step": 3201
    },
    {
      "epoch": 0.23101619710688648,
      "grad_norm": 5.395771963980712,
      "learning_rate": 3.5909173810418236e-06,
      "loss": 1.0752,
      "step": 3202
    },
    {
      "epoch": 0.231088344576314,
      "grad_norm": 3.1358080310258614,
      "learning_rate": 3.590634113909163e-06,
      "loss": 1.0405,
      "step": 3203
    },
    {
      "epoch": 0.2311604920457415,
      "grad_norm": 2.885402436112293,
      "learning_rate": 3.5903507599178616e-06,
      "loss": 1.0003,
      "step": 3204
    },
    {
      "epoch": 0.231232639515169,
      "grad_norm": 4.670435042447092,
      "learning_rate": 3.5900673190833932e-06,
      "loss": 0.9038,
      "step": 3205
    },
    {
      "epoch": 0.23130478698459653,
      "grad_norm": 3.8062214667091685,
      "learning_rate": 3.5897837914212342e-06,
      "loss": 1.1258,
      "step": 3206
    },
    {
      "epoch": 0.23137693445402402,
      "grad_norm": 4.037785636317093,
      "learning_rate": 3.589500176946868e-06,
      "loss": 1.0146,
      "step": 3207
    },
    {
      "epoch": 0.23144908192345154,
      "grad_norm": 4.8241562812251,
      "learning_rate": 3.5892164756757807e-06,
      "loss": 0.8843,
      "step": 3208
    },
    {
      "epoch": 0.23152122939287906,
      "grad_norm": 0.7739624784896835,
      "learning_rate": 3.5889326876234643e-06,
      "loss": 0.8853,
      "step": 3209
    },
    {
      "epoch": 0.23159337686230655,
      "grad_norm": 3.494206128857779,
      "learning_rate": 3.5886488128054172e-06,
      "loss": 0.9896,
      "step": 3210
    },
    {
      "epoch": 0.23166552433173407,
      "grad_norm": 5.803452536234963,
      "learning_rate": 3.5883648512371386e-06,
      "loss": 0.9778,
      "step": 3211
    },
    {
      "epoch": 0.2317376718011616,
      "grad_norm": 4.1531716326182995,
      "learning_rate": 3.5880808029341355e-06,
      "loss": 0.9141,
      "step": 3212
    },
    {
      "epoch": 0.23180981927058908,
      "grad_norm": 2.5935962125848206,
      "learning_rate": 3.5877966679119183e-06,
      "loss": 1.1398,
      "step": 3213
    },
    {
      "epoch": 0.2318819667400166,
      "grad_norm": 2.9712820398178366,
      "learning_rate": 3.5875124461860034e-06,
      "loss": 0.9056,
      "step": 3214
    },
    {
      "epoch": 0.2319541142094441,
      "grad_norm": 3.0688423344847453,
      "learning_rate": 3.5872281377719105e-06,
      "loss": 0.9706,
      "step": 3215
    },
    {
      "epoch": 0.2320262616788716,
      "grad_norm": 6.96030040355757,
      "learning_rate": 3.5869437426851645e-06,
      "loss": 1.0512,
      "step": 3216
    },
    {
      "epoch": 0.23209840914829913,
      "grad_norm": 3.10265894037762,
      "learning_rate": 3.586659260941295e-06,
      "loss": 0.8407,
      "step": 3217
    },
    {
      "epoch": 0.23217055661772662,
      "grad_norm": 4.415120603959734,
      "learning_rate": 3.5863746925558376e-06,
      "loss": 1.0181,
      "step": 3218
    },
    {
      "epoch": 0.23224270408715414,
      "grad_norm": 3.1411654638352635,
      "learning_rate": 3.5860900375443305e-06,
      "loss": 0.9872,
      "step": 3219
    },
    {
      "epoch": 0.23231485155658166,
      "grad_norm": 4.200658819206918,
      "learning_rate": 3.5858052959223184e-06,
      "loss": 1.0178,
      "step": 3220
    },
    {
      "epoch": 0.23238699902600915,
      "grad_norm": 4.39858536954483,
      "learning_rate": 3.5855204677053494e-06,
      "loss": 1.0156,
      "step": 3221
    },
    {
      "epoch": 0.23245914649543667,
      "grad_norm": 4.978005480005015,
      "learning_rate": 3.5852355529089763e-06,
      "loss": 1.0752,
      "step": 3222
    },
    {
      "epoch": 0.2325312939648642,
      "grad_norm": 3.6531143436325406,
      "learning_rate": 3.5849505515487586e-06,
      "loss": 1.0445,
      "step": 3223
    },
    {
      "epoch": 0.23260344143429168,
      "grad_norm": 2.900763470353538,
      "learning_rate": 3.5846654636402586e-06,
      "loss": 0.9609,
      "step": 3224
    },
    {
      "epoch": 0.2326755889037192,
      "grad_norm": 3.339377137543474,
      "learning_rate": 3.5843802891990438e-06,
      "loss": 0.9159,
      "step": 3225
    },
    {
      "epoch": 0.23274773637314672,
      "grad_norm": 5.981487324469807,
      "learning_rate": 3.584095028240686e-06,
      "loss": 1.0582,
      "step": 3226
    },
    {
      "epoch": 0.2328198838425742,
      "grad_norm": 4.48764078668539,
      "learning_rate": 3.5838096807807635e-06,
      "loss": 0.9726,
      "step": 3227
    },
    {
      "epoch": 0.23289203131200173,
      "grad_norm": 3.5847699294491018,
      "learning_rate": 3.5835242468348574e-06,
      "loss": 0.9702,
      "step": 3228
    },
    {
      "epoch": 0.23296417878142925,
      "grad_norm": 4.236247081693678,
      "learning_rate": 3.583238726418554e-06,
      "loss": 0.8508,
      "step": 3229
    },
    {
      "epoch": 0.23303632625085674,
      "grad_norm": 5.3593936841989365,
      "learning_rate": 3.5829531195474443e-06,
      "loss": 0.8777,
      "step": 3230
    },
    {
      "epoch": 0.23310847372028426,
      "grad_norm": 5.32932051058033,
      "learning_rate": 3.5826674262371254e-06,
      "loss": 0.9603,
      "step": 3231
    },
    {
      "epoch": 0.23318062118971178,
      "grad_norm": 4.116733475321901,
      "learning_rate": 3.5823816465031968e-06,
      "loss": 0.9624,
      "step": 3232
    },
    {
      "epoch": 0.23325276865913927,
      "grad_norm": 5.793247620036671,
      "learning_rate": 3.5820957803612645e-06,
      "loss": 0.9987,
      "step": 3233
    },
    {
      "epoch": 0.2333249161285668,
      "grad_norm": 21.440833719469435,
      "learning_rate": 3.5818098278269383e-06,
      "loss": 1.0407,
      "step": 3234
    },
    {
      "epoch": 0.2333970635979943,
      "grad_norm": 6.793622380859193,
      "learning_rate": 3.5815237889158326e-06,
      "loss": 1.0269,
      "step": 3235
    },
    {
      "epoch": 0.2334692110674218,
      "grad_norm": 10.424947717476062,
      "learning_rate": 3.5812376636435676e-06,
      "loss": 0.9518,
      "step": 3236
    },
    {
      "epoch": 0.23354135853684932,
      "grad_norm": 6.598149716503596,
      "learning_rate": 3.5809514520257673e-06,
      "loss": 0.9584,
      "step": 3237
    },
    {
      "epoch": 0.23361350600627684,
      "grad_norm": 6.493900665209576,
      "learning_rate": 3.580665154078061e-06,
      "loss": 0.982,
      "step": 3238
    },
    {
      "epoch": 0.23368565347570433,
      "grad_norm": 3.9797845463926267,
      "learning_rate": 3.5803787698160824e-06,
      "loss": 1.0284,
      "step": 3239
    },
    {
      "epoch": 0.23375780094513185,
      "grad_norm": 4.660342677209322,
      "learning_rate": 3.580092299255469e-06,
      "loss": 1.0129,
      "step": 3240
    },
    {
      "epoch": 0.23382994841455937,
      "grad_norm": 5.056785089308534,
      "learning_rate": 3.5798057424118637e-06,
      "loss": 1.0567,
      "step": 3241
    },
    {
      "epoch": 0.23390209588398686,
      "grad_norm": 5.230462550757156,
      "learning_rate": 3.579519099300916e-06,
      "loss": 1.0142,
      "step": 3242
    },
    {
      "epoch": 0.23397424335341438,
      "grad_norm": 6.097509857707796,
      "learning_rate": 3.5792323699382774e-06,
      "loss": 1.0732,
      "step": 3243
    },
    {
      "epoch": 0.2340463908228419,
      "grad_norm": 0.8397776023187337,
      "learning_rate": 3.578945554339605e-06,
      "loss": 0.8892,
      "step": 3244
    },
    {
      "epoch": 0.2341185382922694,
      "grad_norm": 7.014881120083665,
      "learning_rate": 3.5786586525205607e-06,
      "loss": 1.0289,
      "step": 3245
    },
    {
      "epoch": 0.2341906857616969,
      "grad_norm": 3.8673772591937805,
      "learning_rate": 3.578371664496812e-06,
      "loss": 0.9632,
      "step": 3246
    },
    {
      "epoch": 0.23426283323112443,
      "grad_norm": 5.197215163521,
      "learning_rate": 3.578084590284029e-06,
      "loss": 1.0138,
      "step": 3247
    },
    {
      "epoch": 0.23433498070055192,
      "grad_norm": 4.707505324933509,
      "learning_rate": 3.577797429897889e-06,
      "loss": 1.0566,
      "step": 3248
    },
    {
      "epoch": 0.23440712816997944,
      "grad_norm": 11.24686065952736,
      "learning_rate": 3.577510183354072e-06,
      "loss": 0.9083,
      "step": 3249
    },
    {
      "epoch": 0.23447927563940696,
      "grad_norm": 4.305522858416918,
      "learning_rate": 3.5772228506682636e-06,
      "loss": 0.9614,
      "step": 3250
    },
    {
      "epoch": 0.23455142310883445,
      "grad_norm": 3.7127508932807007,
      "learning_rate": 3.576935431856154e-06,
      "loss": 0.9612,
      "step": 3251
    },
    {
      "epoch": 0.23462357057826197,
      "grad_norm": 4.255525699330358,
      "learning_rate": 3.5766479269334377e-06,
      "loss": 0.8985,
      "step": 3252
    },
    {
      "epoch": 0.2346957180476895,
      "grad_norm": 3.2475990817113227,
      "learning_rate": 3.576360335915815e-06,
      "loss": 0.9725,
      "step": 3253
    },
    {
      "epoch": 0.23476786551711698,
      "grad_norm": 5.226233501942033,
      "learning_rate": 3.57607265881899e-06,
      "loss": 1.0056,
      "step": 3254
    },
    {
      "epoch": 0.2348400129865445,
      "grad_norm": 5.890987961815893,
      "learning_rate": 3.575784895658671e-06,
      "loss": 0.9631,
      "step": 3255
    },
    {
      "epoch": 0.23491216045597202,
      "grad_norm": 3.6303245638935624,
      "learning_rate": 3.5754970464505724e-06,
      "loss": 1.0392,
      "step": 3256
    },
    {
      "epoch": 0.2349843079253995,
      "grad_norm": 3.2323395594692648,
      "learning_rate": 3.575209111210413e-06,
      "loss": 1.0065,
      "step": 3257
    },
    {
      "epoch": 0.23505645539482703,
      "grad_norm": 5.067483403649651,
      "learning_rate": 3.574921089953914e-06,
      "loss": 0.9463,
      "step": 3258
    },
    {
      "epoch": 0.23512860286425455,
      "grad_norm": 3.4283006510075413,
      "learning_rate": 3.5746329826968058e-06,
      "loss": 1.0707,
      "step": 3259
    },
    {
      "epoch": 0.23520075033368204,
      "grad_norm": 0.7404184238439463,
      "learning_rate": 3.5743447894548185e-06,
      "loss": 0.8499,
      "step": 3260
    },
    {
      "epoch": 0.23527289780310956,
      "grad_norm": 4.985047871963378,
      "learning_rate": 3.5740565102436908e-06,
      "loss": 0.9552,
      "step": 3261
    },
    {
      "epoch": 0.23534504527253708,
      "grad_norm": 3.6572947517424743,
      "learning_rate": 3.573768145079164e-06,
      "loss": 0.946,
      "step": 3262
    },
    {
      "epoch": 0.23541719274196457,
      "grad_norm": 2.5461087624691654,
      "learning_rate": 3.573479693976985e-06,
      "loss": 1.0366,
      "step": 3263
    },
    {
      "epoch": 0.2354893402113921,
      "grad_norm": 2.7034738128167053,
      "learning_rate": 3.5731911569529036e-06,
      "loss": 0.9984,
      "step": 3264
    },
    {
      "epoch": 0.2355614876808196,
      "grad_norm": 4.65621190407549,
      "learning_rate": 3.5729025340226774e-06,
      "loss": 0.8497,
      "step": 3265
    },
    {
      "epoch": 0.2356336351502471,
      "grad_norm": 4.591558855086861,
      "learning_rate": 3.572613825202067e-06,
      "loss": 0.9403,
      "step": 3266
    },
    {
      "epoch": 0.23570578261967462,
      "grad_norm": 3.9651310558234965,
      "learning_rate": 3.572325030506837e-06,
      "loss": 0.9556,
      "step": 3267
    },
    {
      "epoch": 0.2357779300891021,
      "grad_norm": 5.523207685337931,
      "learning_rate": 3.5720361499527575e-06,
      "loss": 1.007,
      "step": 3268
    },
    {
      "epoch": 0.23585007755852963,
      "grad_norm": 8.506123749120487,
      "learning_rate": 3.571747183555604e-06,
      "loss": 0.8832,
      "step": 3269
    },
    {
      "epoch": 0.23592222502795715,
      "grad_norm": 6.247538577812065,
      "learning_rate": 3.571458131331154e-06,
      "loss": 1.0017,
      "step": 3270
    },
    {
      "epoch": 0.23599437249738464,
      "grad_norm": 3.635925009672828,
      "learning_rate": 3.571168993295194e-06,
      "loss": 0.9898,
      "step": 3271
    },
    {
      "epoch": 0.23606651996681216,
      "grad_norm": 5.14702469886559,
      "learning_rate": 3.5708797694635106e-06,
      "loss": 0.9695,
      "step": 3272
    },
    {
      "epoch": 0.23613866743623968,
      "grad_norm": 4.658296071530776,
      "learning_rate": 3.5705904598518983e-06,
      "loss": 1.0385,
      "step": 3273
    },
    {
      "epoch": 0.23621081490566717,
      "grad_norm": 10.876395300440551,
      "learning_rate": 3.5703010644761557e-06,
      "loss": 0.8121,
      "step": 3274
    },
    {
      "epoch": 0.2362829623750947,
      "grad_norm": 4.5451369486740605,
      "learning_rate": 3.570011583352085e-06,
      "loss": 1.0237,
      "step": 3275
    },
    {
      "epoch": 0.2363551098445222,
      "grad_norm": 5.354043961639095,
      "learning_rate": 3.5697220164954924e-06,
      "loss": 0.9275,
      "step": 3276
    },
    {
      "epoch": 0.2364272573139497,
      "grad_norm": 2.767204480197188,
      "learning_rate": 3.5694323639221926e-06,
      "loss": 1.0509,
      "step": 3277
    },
    {
      "epoch": 0.23649940478337722,
      "grad_norm": 4.294401143327154,
      "learning_rate": 3.569142625648001e-06,
      "loss": 1.0083,
      "step": 3278
    },
    {
      "epoch": 0.23657155225280474,
      "grad_norm": 4.670919365250786,
      "learning_rate": 3.56885280168874e-06,
      "loss": 1.0245,
      "step": 3279
    },
    {
      "epoch": 0.23664369972223223,
      "grad_norm": 0.8514906826707918,
      "learning_rate": 3.5685628920602343e-06,
      "loss": 0.8475,
      "step": 3280
    },
    {
      "epoch": 0.23671584719165975,
      "grad_norm": 7.66063654831891,
      "learning_rate": 3.568272896778316e-06,
      "loss": 0.8507,
      "step": 3281
    },
    {
      "epoch": 0.23678799466108727,
      "grad_norm": 5.718196988075314,
      "learning_rate": 3.5679828158588203e-06,
      "loss": 1.0252,
      "step": 3282
    },
    {
      "epoch": 0.23686014213051476,
      "grad_norm": 6.226619782417658,
      "learning_rate": 3.5676926493175875e-06,
      "loss": 1.0055,
      "step": 3283
    },
    {
      "epoch": 0.23693228959994228,
      "grad_norm": 4.821466984671506,
      "learning_rate": 3.5674023971704625e-06,
      "loss": 1.0157,
      "step": 3284
    },
    {
      "epoch": 0.2370044370693698,
      "grad_norm": 4.006666112467071,
      "learning_rate": 3.567112059433295e-06,
      "loss": 0.9963,
      "step": 3285
    },
    {
      "epoch": 0.2370765845387973,
      "grad_norm": 29.53811304180266,
      "learning_rate": 3.566821636121939e-06,
      "loss": 1.0643,
      "step": 3286
    },
    {
      "epoch": 0.2371487320082248,
      "grad_norm": 8.123860440669365,
      "learning_rate": 3.5665311272522536e-06,
      "loss": 0.9723,
      "step": 3287
    },
    {
      "epoch": 0.23722087947765233,
      "grad_norm": 3.1666296070004982,
      "learning_rate": 3.566240532840103e-06,
      "loss": 0.9249,
      "step": 3288
    },
    {
      "epoch": 0.23729302694707982,
      "grad_norm": 3.506916704896974,
      "learning_rate": 3.5659498529013545e-06,
      "loss": 0.9413,
      "step": 3289
    },
    {
      "epoch": 0.23736517441650734,
      "grad_norm": 3.191728517283315,
      "learning_rate": 3.5656590874518814e-06,
      "loss": 0.9513,
      "step": 3290
    },
    {
      "epoch": 0.23743732188593486,
      "grad_norm": 10.199687010506974,
      "learning_rate": 3.5653682365075614e-06,
      "loss": 1.0216,
      "step": 3291
    },
    {
      "epoch": 0.23750946935536235,
      "grad_norm": 6.071796329163372,
      "learning_rate": 3.565077300084277e-06,
      "loss": 0.9343,
      "step": 3292
    },
    {
      "epoch": 0.23758161682478987,
      "grad_norm": 0.7939141561816108,
      "learning_rate": 3.564786278197915e-06,
      "loss": 0.8642,
      "step": 3293
    },
    {
      "epoch": 0.2376537642942174,
      "grad_norm": 8.132555925497694,
      "learning_rate": 3.564495170864367e-06,
      "loss": 0.8394,
      "step": 3294
    },
    {
      "epoch": 0.23772591176364488,
      "grad_norm": 14.863260204485757,
      "learning_rate": 3.5642039780995293e-06,
      "loss": 0.9743,
      "step": 3295
    },
    {
      "epoch": 0.2377980592330724,
      "grad_norm": 3.7403974293358617,
      "learning_rate": 3.5639126999193025e-06,
      "loss": 1.0639,
      "step": 3296
    },
    {
      "epoch": 0.23787020670249992,
      "grad_norm": 4.62768229467086,
      "learning_rate": 3.563621336339593e-06,
      "loss": 1.0166,
      "step": 3297
    },
    {
      "epoch": 0.2379423541719274,
      "grad_norm": 4.134230026275298,
      "learning_rate": 3.56332988737631e-06,
      "loss": 1.0394,
      "step": 3298
    },
    {
      "epoch": 0.23801450164135493,
      "grad_norm": 5.500146863883908,
      "learning_rate": 3.5630383530453693e-06,
      "loss": 0.9075,
      "step": 3299
    },
    {
      "epoch": 0.23808664911078245,
      "grad_norm": 5.386503285835388,
      "learning_rate": 3.562746733362691e-06,
      "loss": 1.0183,
      "step": 3300
    },
    {
      "epoch": 0.23815879658020994,
      "grad_norm": 2.5747926387810236,
      "learning_rate": 3.562455028344198e-06,
      "loss": 0.9089,
      "step": 3301
    },
    {
      "epoch": 0.23823094404963746,
      "grad_norm": 3.6870093584968773,
      "learning_rate": 3.56216323800582e-06,
      "loss": 1.0131,
      "step": 3302
    },
    {
      "epoch": 0.23830309151906498,
      "grad_norm": 4.662447498015707,
      "learning_rate": 3.5618713623634905e-06,
      "loss": 0.9825,
      "step": 3303
    },
    {
      "epoch": 0.23837523898849247,
      "grad_norm": 10.090390050662647,
      "learning_rate": 3.5615794014331486e-06,
      "loss": 0.84,
      "step": 3304
    },
    {
      "epoch": 0.23844738645792,
      "grad_norm": 4.925382593465872,
      "learning_rate": 3.5612873552307352e-06,
      "loss": 1.0181,
      "step": 3305
    },
    {
      "epoch": 0.2385195339273475,
      "grad_norm": 9.403545409034065,
      "learning_rate": 3.5609952237721993e-06,
      "loss": 1.0312,
      "step": 3306
    },
    {
      "epoch": 0.238591681396775,
      "grad_norm": 0.7276693547473395,
      "learning_rate": 3.560703007073493e-06,
      "loss": 0.7911,
      "step": 3307
    },
    {
      "epoch": 0.23866382886620252,
      "grad_norm": 7.550875722863468,
      "learning_rate": 3.5604107051505732e-06,
      "loss": 1.0421,
      "step": 3308
    },
    {
      "epoch": 0.23873597633563004,
      "grad_norm": 6.899060127267224,
      "learning_rate": 3.560118318019401e-06,
      "loss": 0.9835,
      "step": 3309
    },
    {
      "epoch": 0.23880812380505753,
      "grad_norm": 5.919435345810648,
      "learning_rate": 3.559825845695943e-06,
      "loss": 1.1153,
      "step": 3310
    },
    {
      "epoch": 0.23888027127448505,
      "grad_norm": 4.030654508388335,
      "learning_rate": 3.5595332881961695e-06,
      "loss": 1.0069,
      "step": 3311
    },
    {
      "epoch": 0.23895241874391257,
      "grad_norm": 0.8518642625810716,
      "learning_rate": 3.5592406455360563e-06,
      "loss": 0.8723,
      "step": 3312
    },
    {
      "epoch": 0.23902456621334006,
      "grad_norm": 16.267717767327472,
      "learning_rate": 3.5589479177315836e-06,
      "loss": 0.963,
      "step": 3313
    },
    {
      "epoch": 0.23909671368276758,
      "grad_norm": 5.204509745585508,
      "learning_rate": 3.5586551047987365e-06,
      "loss": 0.9345,
      "step": 3314
    },
    {
      "epoch": 0.2391688611521951,
      "grad_norm": 4.450524357109245,
      "learning_rate": 3.558362206753504e-06,
      "loss": 1.0372,
      "step": 3315
    },
    {
      "epoch": 0.2392410086216226,
      "grad_norm": 5.2019329366559335,
      "learning_rate": 3.5580692236118806e-06,
      "loss": 1.0694,
      "step": 3316
    },
    {
      "epoch": 0.2393131560910501,
      "grad_norm": 2.943519754982737,
      "learning_rate": 3.557776155389864e-06,
      "loss": 1.0007,
      "step": 3317
    },
    {
      "epoch": 0.2393853035604776,
      "grad_norm": 3.039878912070155,
      "learning_rate": 3.557483002103458e-06,
      "loss": 1.0197,
      "step": 3318
    },
    {
      "epoch": 0.23945745102990512,
      "grad_norm": 7.616547198649008,
      "learning_rate": 3.5571897637686712e-06,
      "loss": 0.9965,
      "step": 3319
    },
    {
      "epoch": 0.23952959849933264,
      "grad_norm": 5.537535635874419,
      "learning_rate": 3.556896440401516e-06,
      "loss": 0.9298,
      "step": 3320
    },
    {
      "epoch": 0.23960174596876013,
      "grad_norm": 4.56571387400933,
      "learning_rate": 3.5566030320180104e-06,
      "loss": 1.0352,
      "step": 3321
    },
    {
      "epoch": 0.23967389343818765,
      "grad_norm": 3.196267579843022,
      "learning_rate": 3.5563095386341747e-06,
      "loss": 0.9904,
      "step": 3322
    },
    {
      "epoch": 0.23974604090761517,
      "grad_norm": 2.9024308244214465,
      "learning_rate": 3.5560159602660363e-06,
      "loss": 1.068,
      "step": 3323
    },
    {
      "epoch": 0.23981818837704266,
      "grad_norm": 8.72834102637803,
      "learning_rate": 3.555722296929627e-06,
      "loss": 0.9338,
      "step": 3324
    },
    {
      "epoch": 0.23989033584647018,
      "grad_norm": 0.8198434851304113,
      "learning_rate": 3.5554285486409813e-06,
      "loss": 0.8704,
      "step": 3325
    },
    {
      "epoch": 0.2399624833158977,
      "grad_norm": 5.137336769202521,
      "learning_rate": 3.5551347154161412e-06,
      "loss": 1.0124,
      "step": 3326
    },
    {
      "epoch": 0.2400346307853252,
      "grad_norm": 3.0731138694116296,
      "learning_rate": 3.554840797271151e-06,
      "loss": 1.0409,
      "step": 3327
    },
    {
      "epoch": 0.2401067782547527,
      "grad_norm": 3.1684603045160507,
      "learning_rate": 3.5545467942220604e-06,
      "loss": 1.1187,
      "step": 3328
    },
    {
      "epoch": 0.24017892572418023,
      "grad_norm": 4.150728157230908,
      "learning_rate": 3.5542527062849247e-06,
      "loss": 1.0547,
      "step": 3329
    },
    {
      "epoch": 0.24025107319360772,
      "grad_norm": 7.985624749305389,
      "learning_rate": 3.553958533475802e-06,
      "loss": 1.0747,
      "step": 3330
    },
    {
      "epoch": 0.24032322066303524,
      "grad_norm": 5.114450529385273,
      "learning_rate": 3.5536642758107567e-06,
      "loss": 1.0202,
      "step": 3331
    },
    {
      "epoch": 0.24039536813246276,
      "grad_norm": 34.118930821901316,
      "learning_rate": 3.5533699333058566e-06,
      "loss": 1.0589,
      "step": 3332
    },
    {
      "epoch": 0.24046751560189025,
      "grad_norm": 4.577391686150258,
      "learning_rate": 3.5530755059771744e-06,
      "loss": 1.0077,
      "step": 3333
    },
    {
      "epoch": 0.24053966307131777,
      "grad_norm": 3.198335816854679,
      "learning_rate": 3.5527809938407883e-06,
      "loss": 1.0638,
      "step": 3334
    },
    {
      "epoch": 0.2406118105407453,
      "grad_norm": 3.542726081520286,
      "learning_rate": 3.5524863969127806e-06,
      "loss": 0.9756,
      "step": 3335
    },
    {
      "epoch": 0.24068395801017278,
      "grad_norm": 4.203979405346341,
      "learning_rate": 3.552191715209238e-06,
      "loss": 1.0077,
      "step": 3336
    },
    {
      "epoch": 0.2407561054796003,
      "grad_norm": 2.569525229985442,
      "learning_rate": 3.551896948746252e-06,
      "loss": 0.9815,
      "step": 3337
    },
    {
      "epoch": 0.24082825294902782,
      "grad_norm": 5.082514253925652,
      "learning_rate": 3.551602097539918e-06,
      "loss": 0.9785,
      "step": 3338
    },
    {
      "epoch": 0.2409004004184553,
      "grad_norm": 3.587668406292011,
      "learning_rate": 3.5513071616063373e-06,
      "loss": 1.0258,
      "step": 3339
    },
    {
      "epoch": 0.24097254788788283,
      "grad_norm": 3.5733801686600533,
      "learning_rate": 3.5510121409616157e-06,
      "loss": 0.9978,
      "step": 3340
    },
    {
      "epoch": 0.24104469535731035,
      "grad_norm": 5.349134977767525,
      "learning_rate": 3.550717035621862e-06,
      "loss": 0.9638,
      "step": 3341
    },
    {
      "epoch": 0.24111684282673784,
      "grad_norm": 4.399040221526222,
      "learning_rate": 3.550421845603192e-06,
      "loss": 0.8941,
      "step": 3342
    },
    {
      "epoch": 0.24118899029616536,
      "grad_norm": 3.432180832195847,
      "learning_rate": 3.5501265709217245e-06,
      "loss": 0.94,
      "step": 3343
    },
    {
      "epoch": 0.24126113776559288,
      "grad_norm": 6.74929918960341,
      "learning_rate": 3.5498312115935834e-06,
      "loss": 0.9495,
      "step": 3344
    },
    {
      "epoch": 0.24133328523502037,
      "grad_norm": 3.1870836472200477,
      "learning_rate": 3.5495357676348964e-06,
      "loss": 0.9634,
      "step": 3345
    },
    {
      "epoch": 0.2414054327044479,
      "grad_norm": 2.437770339450637,
      "learning_rate": 3.5492402390617985e-06,
      "loss": 1.0557,
      "step": 3346
    },
    {
      "epoch": 0.2414775801738754,
      "grad_norm": 5.003979625059708,
      "learning_rate": 3.548944625890425e-06,
      "loss": 1.0199,
      "step": 3347
    },
    {
      "epoch": 0.2415497276433029,
      "grad_norm": 11.646435080025226,
      "learning_rate": 3.5486489281369206e-06,
      "loss": 1.0021,
      "step": 3348
    },
    {
      "epoch": 0.24162187511273042,
      "grad_norm": 6.4307476296060315,
      "learning_rate": 3.5483531458174304e-06,
      "loss": 1.0234,
      "step": 3349
    },
    {
      "epoch": 0.24169402258215794,
      "grad_norm": 10.378356023122148,
      "learning_rate": 3.548057278948107e-06,
      "loss": 0.9386,
      "step": 3350
    },
    {
      "epoch": 0.24176617005158543,
      "grad_norm": 3.067871521180636,
      "learning_rate": 3.5477613275451065e-06,
      "loss": 0.9301,
      "step": 3351
    },
    {
      "epoch": 0.24183831752101295,
      "grad_norm": 3.4690377055956367,
      "learning_rate": 3.547465291624589e-06,
      "loss": 0.9914,
      "step": 3352
    },
    {
      "epoch": 0.24191046499044047,
      "grad_norm": 4.024893310270176,
      "learning_rate": 3.547169171202721e-06,
      "loss": 0.9601,
      "step": 3353
    },
    {
      "epoch": 0.24198261245986796,
      "grad_norm": 3.679267616370287,
      "learning_rate": 3.546872966295672e-06,
      "loss": 0.8979,
      "step": 3354
    },
    {
      "epoch": 0.24205475992929548,
      "grad_norm": 3.761701637300445,
      "learning_rate": 3.546576676919616e-06,
      "loss": 0.9956,
      "step": 3355
    },
    {
      "epoch": 0.242126907398723,
      "grad_norm": 3.326656393622652,
      "learning_rate": 3.546280303090733e-06,
      "loss": 0.9964,
      "step": 3356
    },
    {
      "epoch": 0.2421990548681505,
      "grad_norm": 5.141249227451866,
      "learning_rate": 3.5459838448252073e-06,
      "loss": 1.0668,
      "step": 3357
    },
    {
      "epoch": 0.242271202337578,
      "grad_norm": 3.3465316468114197,
      "learning_rate": 3.5456873021392265e-06,
      "loss": 1.056,
      "step": 3358
    },
    {
      "epoch": 0.24234334980700553,
      "grad_norm": 5.6548871669423795,
      "learning_rate": 3.5453906750489845e-06,
      "loss": 1.114,
      "step": 3359
    },
    {
      "epoch": 0.24241549727643302,
      "grad_norm": 4.247311134203326,
      "learning_rate": 3.5450939635706782e-06,
      "loss": 0.9232,
      "step": 3360
    },
    {
      "epoch": 0.24248764474586054,
      "grad_norm": 3.827284989350737,
      "learning_rate": 3.5447971677205103e-06,
      "loss": 0.8246,
      "step": 3361
    },
    {
      "epoch": 0.24255979221528806,
      "grad_norm": 7.487785310993191,
      "learning_rate": 3.5445002875146877e-06,
      "loss": 0.9462,
      "step": 3362
    },
    {
      "epoch": 0.24263193968471555,
      "grad_norm": 4.5405029532510435,
      "learning_rate": 3.5442033229694223e-06,
      "loss": 1.0635,
      "step": 3363
    },
    {
      "epoch": 0.24270408715414307,
      "grad_norm": 3.7919664089791425,
      "learning_rate": 3.5439062741009292e-06,
      "loss": 1.0671,
      "step": 3364
    },
    {
      "epoch": 0.2427762346235706,
      "grad_norm": 8.062438018324011,
      "learning_rate": 3.5436091409254305e-06,
      "loss": 0.9022,
      "step": 3365
    },
    {
      "epoch": 0.24284838209299808,
      "grad_norm": 3.004301801614736,
      "learning_rate": 3.5433119234591505e-06,
      "loss": 0.9482,
      "step": 3366
    },
    {
      "epoch": 0.2429205295624256,
      "grad_norm": 4.07459442367056,
      "learning_rate": 3.5430146217183197e-06,
      "loss": 0.9886,
      "step": 3367
    },
    {
      "epoch": 0.24299267703185312,
      "grad_norm": 4.089270084229867,
      "learning_rate": 3.542717235719172e-06,
      "loss": 1.0127,
      "step": 3368
    },
    {
      "epoch": 0.2430648245012806,
      "grad_norm": 6.8056373677775,
      "learning_rate": 3.5424197654779474e-06,
      "loss": 0.9983,
      "step": 3369
    },
    {
      "epoch": 0.24313697197070813,
      "grad_norm": 0.603485699254263,
      "learning_rate": 3.5421222110108896e-06,
      "loss": 0.7906,
      "step": 3370
    },
    {
      "epoch": 0.24320911944013562,
      "grad_norm": 3.38180759648282,
      "learning_rate": 3.541824572334246e-06,
      "loss": 1.0734,
      "step": 3371
    },
    {
      "epoch": 0.24328126690956314,
      "grad_norm": 0.8703507132504882,
      "learning_rate": 3.54152684946427e-06,
      "loss": 0.8227,
      "step": 3372
    },
    {
      "epoch": 0.24335341437899066,
      "grad_norm": 3.239140929079045,
      "learning_rate": 3.54122904241722e-06,
      "loss": 1.0264,
      "step": 3373
    },
    {
      "epoch": 0.24342556184841815,
      "grad_norm": 3.8839953211668767,
      "learning_rate": 3.5409311512093563e-06,
      "loss": 0.9142,
      "step": 3374
    },
    {
      "epoch": 0.24349770931784567,
      "grad_norm": 3.279416980100292,
      "learning_rate": 3.540633175856948e-06,
      "loss": 0.9717,
      "step": 3375
    },
    {
      "epoch": 0.2435698567872732,
      "grad_norm": 20.501447673016976,
      "learning_rate": 3.540335116376265e-06,
      "loss": 0.9509,
      "step": 3376
    },
    {
      "epoch": 0.24364200425670068,
      "grad_norm": 4.342891251458764,
      "learning_rate": 3.5400369727835833e-06,
      "loss": 0.9999,
      "step": 3377
    },
    {
      "epoch": 0.2437141517261282,
      "grad_norm": 4.544116464814781,
      "learning_rate": 3.5397387450951824e-06,
      "loss": 1.0467,
      "step": 3378
    },
    {
      "epoch": 0.24378629919555572,
      "grad_norm": 6.199943591445762,
      "learning_rate": 3.53944043332735e-06,
      "loss": 1.0897,
      "step": 3379
    },
    {
      "epoch": 0.2438584466649832,
      "grad_norm": 4.400652420918553,
      "learning_rate": 3.539142037496374e-06,
      "loss": 1.0513,
      "step": 3380
    },
    {
      "epoch": 0.24393059413441073,
      "grad_norm": 2.9430445836248578,
      "learning_rate": 3.538843557618549e-06,
      "loss": 0.9887,
      "step": 3381
    },
    {
      "epoch": 0.24400274160383825,
      "grad_norm": 5.643379194175068,
      "learning_rate": 3.5385449937101733e-06,
      "loss": 0.9945,
      "step": 3382
    },
    {
      "epoch": 0.24407488907326574,
      "grad_norm": 0.745835464268487,
      "learning_rate": 3.538246345787552e-06,
      "loss": 0.8354,
      "step": 3383
    },
    {
      "epoch": 0.24414703654269326,
      "grad_norm": 3.37182453672814,
      "learning_rate": 3.5379476138669914e-06,
      "loss": 1.0606,
      "step": 3384
    },
    {
      "epoch": 0.24421918401212078,
      "grad_norm": 3.2359064134755906,
      "learning_rate": 3.537648797964805e-06,
      "loss": 0.9883,
      "step": 3385
    },
    {
      "epoch": 0.24429133148154827,
      "grad_norm": 3.364175321013329,
      "learning_rate": 3.53734989809731e-06,
      "loss": 1.0475,
      "step": 3386
    },
    {
      "epoch": 0.2443634789509758,
      "grad_norm": 4.6323151345584614,
      "learning_rate": 3.5370509142808283e-06,
      "loss": 1.0397,
      "step": 3387
    },
    {
      "epoch": 0.2444356264204033,
      "grad_norm": 6.232784823731327,
      "learning_rate": 3.536751846531686e-06,
      "loss": 1.0492,
      "step": 3388
    },
    {
      "epoch": 0.2445077738898308,
      "grad_norm": 3.305704596086425,
      "learning_rate": 3.5364526948662145e-06,
      "loss": 1.0085,
      "step": 3389
    },
    {
      "epoch": 0.24457992135925832,
      "grad_norm": 9.091968689635943,
      "learning_rate": 3.5361534593007488e-06,
      "loss": 1.0524,
      "step": 3390
    },
    {
      "epoch": 0.24465206882868584,
      "grad_norm": 0.7158186887798135,
      "learning_rate": 3.5358541398516295e-06,
      "loss": 0.8525,
      "step": 3391
    },
    {
      "epoch": 0.24472421629811333,
      "grad_norm": 6.366790871465511,
      "learning_rate": 3.535554736535201e-06,
      "loss": 0.9525,
      "step": 3392
    },
    {
      "epoch": 0.24479636376754085,
      "grad_norm": 2.4978961675426907,
      "learning_rate": 3.5352552493678127e-06,
      "loss": 0.9162,
      "step": 3393
    },
    {
      "epoch": 0.24486851123696837,
      "grad_norm": 4.514862946619511,
      "learning_rate": 3.5349556783658187e-06,
      "loss": 0.9061,
      "step": 3394
    },
    {
      "epoch": 0.24494065870639586,
      "grad_norm": 3.5086265248152597,
      "learning_rate": 3.534656023545577e-06,
      "loss": 0.9439,
      "step": 3395
    },
    {
      "epoch": 0.24501280617582338,
      "grad_norm": 4.058536415319278,
      "learning_rate": 3.5343562849234516e-06,
      "loss": 0.9818,
      "step": 3396
    },
    {
      "epoch": 0.2450849536452509,
      "grad_norm": 7.400971797333231,
      "learning_rate": 3.534056462515809e-06,
      "loss": 0.938,
      "step": 3397
    },
    {
      "epoch": 0.2451571011146784,
      "grad_norm": 2.679199441221856,
      "learning_rate": 3.5337565563390223e-06,
      "loss": 1.0953,
      "step": 3398
    },
    {
      "epoch": 0.2452292485841059,
      "grad_norm": 1.4980463975410745,
      "learning_rate": 3.5334565664094675e-06,
      "loss": 0.7505,
      "step": 3399
    },
    {
      "epoch": 0.24530139605353343,
      "grad_norm": 0.7846327597523215,
      "learning_rate": 3.533156492743526e-06,
      "loss": 0.8341,
      "step": 3400
    },
    {
      "epoch": 0.24537354352296092,
      "grad_norm": 7.081106606333764,
      "learning_rate": 3.532856335357584e-06,
      "loss": 1.111,
      "step": 3401
    },
    {
      "epoch": 0.24544569099238844,
      "grad_norm": 4.678518821229183,
      "learning_rate": 3.532556094268032e-06,
      "loss": 0.9865,
      "step": 3402
    },
    {
      "epoch": 0.24551783846181596,
      "grad_norm": 4.151585535711284,
      "learning_rate": 3.532255769491266e-06,
      "loss": 1.1934,
      "step": 3403
    },
    {
      "epoch": 0.24558998593124345,
      "grad_norm": 3.6338079585809417,
      "learning_rate": 3.5319553610436836e-06,
      "loss": 0.9934,
      "step": 3404
    },
    {
      "epoch": 0.24566213340067097,
      "grad_norm": 8.8123770630831,
      "learning_rate": 3.5316548689416903e-06,
      "loss": 1.0428,
      "step": 3405
    },
    {
      "epoch": 0.2457342808700985,
      "grad_norm": 4.736294700268657,
      "learning_rate": 3.5313542932016945e-06,
      "loss": 1.0051,
      "step": 3406
    },
    {
      "epoch": 0.24580642833952598,
      "grad_norm": 6.870478027237545,
      "learning_rate": 3.53105363384011e-06,
      "loss": 1.0839,
      "step": 3407
    },
    {
      "epoch": 0.2458785758089535,
      "grad_norm": 8.768617714110414,
      "learning_rate": 3.5307528908733542e-06,
      "loss": 0.8916,
      "step": 3408
    },
    {
      "epoch": 0.24595072327838102,
      "grad_norm": 4.184686655143178,
      "learning_rate": 3.5304520643178496e-06,
      "loss": 0.9128,
      "step": 3409
    },
    {
      "epoch": 0.24602287074780851,
      "grad_norm": 4.1113313724749645,
      "learning_rate": 3.530151154190023e-06,
      "loss": 1.019,
      "step": 3410
    },
    {
      "epoch": 0.24609501821723603,
      "grad_norm": 9.698428972577611,
      "learning_rate": 3.5298501605063067e-06,
      "loss": 1.0241,
      "step": 3411
    },
    {
      "epoch": 0.24616716568666355,
      "grad_norm": 0.7361267381084986,
      "learning_rate": 3.529549083283136e-06,
      "loss": 0.8024,
      "step": 3412
    },
    {
      "epoch": 0.24623931315609104,
      "grad_norm": 4.3643197622659,
      "learning_rate": 3.5292479225369524e-06,
      "loss": 1.0161,
      "step": 3413
    },
    {
      "epoch": 0.24631146062551856,
      "grad_norm": 4.063995790209054,
      "learning_rate": 3.5289466782842015e-06,
      "loss": 0.9732,
      "step": 3414
    },
    {
      "epoch": 0.24638360809494608,
      "grad_norm": 18.31578268892194,
      "learning_rate": 3.5286453505413315e-06,
      "loss": 0.8627,
      "step": 3415
    },
    {
      "epoch": 0.24645575556437357,
      "grad_norm": 4.007509811303687,
      "learning_rate": 3.5283439393247977e-06,
      "loss": 1.0486,
      "step": 3416
    },
    {
      "epoch": 0.2465279030338011,
      "grad_norm": 2.8684091070415474,
      "learning_rate": 3.5280424446510593e-06,
      "loss": 0.9681,
      "step": 3417
    },
    {
      "epoch": 0.2466000505032286,
      "grad_norm": 4.34847787118005,
      "learning_rate": 3.5277408665365796e-06,
      "loss": 1.0287,
      "step": 3418
    },
    {
      "epoch": 0.2466721979726561,
      "grad_norm": 3.7243268166370123,
      "learning_rate": 3.527439204997827e-06,
      "loss": 1.0565,
      "step": 3419
    },
    {
      "epoch": 0.24674434544208362,
      "grad_norm": 0.8397515534650329,
      "learning_rate": 3.527137460051274e-06,
      "loss": 0.8289,
      "step": 3420
    },
    {
      "epoch": 0.24681649291151112,
      "grad_norm": 4.513521225782684,
      "learning_rate": 3.526835631713397e-06,
      "loss": 0.9629,
      "step": 3421
    },
    {
      "epoch": 0.24688864038093863,
      "grad_norm": 2.968976463915828,
      "learning_rate": 3.5265337200006784e-06,
      "loss": 0.9231,
      "step": 3422
    },
    {
      "epoch": 0.24696078785036615,
      "grad_norm": 4.208822819912825,
      "learning_rate": 3.5262317249296044e-06,
      "loss": 0.9787,
      "step": 3423
    },
    {
      "epoch": 0.24703293531979365,
      "grad_norm": 3.7766602458938556,
      "learning_rate": 3.5259296465166665e-06,
      "loss": 0.9187,
      "step": 3424
    },
    {
      "epoch": 0.24710508278922116,
      "grad_norm": 4.3516545448636,
      "learning_rate": 3.525627484778359e-06,
      "loss": 1.0528,
      "step": 3425
    },
    {
      "epoch": 0.24717723025864868,
      "grad_norm": 6.032843977048969,
      "learning_rate": 3.525325239731182e-06,
      "loss": 0.9953,
      "step": 3426
    },
    {
      "epoch": 0.24724937772807618,
      "grad_norm": 3.4445280881864173,
      "learning_rate": 3.52502291139164e-06,
      "loss": 0.9353,
      "step": 3427
    },
    {
      "epoch": 0.2473215251975037,
      "grad_norm": 4.082147594783376,
      "learning_rate": 3.5247204997762424e-06,
      "loss": 0.8754,
      "step": 3428
    },
    {
      "epoch": 0.2473936726669312,
      "grad_norm": 4.154198161585778,
      "learning_rate": 3.524418004901503e-06,
      "loss": 1.0851,
      "step": 3429
    },
    {
      "epoch": 0.2474658201363587,
      "grad_norm": 2.5940911861442437,
      "learning_rate": 3.524115426783939e-06,
      "loss": 0.9901,
      "step": 3430
    },
    {
      "epoch": 0.24753796760578622,
      "grad_norm": 4.649389756060441,
      "learning_rate": 3.5238127654400744e-06,
      "loss": 1.1357,
      "step": 3431
    },
    {
      "epoch": 0.24761011507521374,
      "grad_norm": 4.799375596759897,
      "learning_rate": 3.5235100208864345e-06,
      "loss": 0.974,
      "step": 3432
    },
    {
      "epoch": 0.24768226254464124,
      "grad_norm": 10.958139357403223,
      "learning_rate": 3.5232071931395535e-06,
      "loss": 0.9367,
      "step": 3433
    },
    {
      "epoch": 0.24775441001406875,
      "grad_norm": 5.329172160146038,
      "learning_rate": 3.5229042822159657e-06,
      "loss": 0.9066,
      "step": 3434
    },
    {
      "epoch": 0.24782655748349627,
      "grad_norm": 2.967538847568573,
      "learning_rate": 3.522601288132212e-06,
      "loss": 1.0049,
      "step": 3435
    },
    {
      "epoch": 0.24789870495292377,
      "grad_norm": 3.6004729701986182,
      "learning_rate": 3.5222982109048398e-06,
      "loss": 0.9954,
      "step": 3436
    },
    {
      "epoch": 0.24797085242235128,
      "grad_norm": 4.089694977744224,
      "learning_rate": 3.521995050550397e-06,
      "loss": 1.0286,
      "step": 3437
    },
    {
      "epoch": 0.2480429998917788,
      "grad_norm": 4.8095602990060735,
      "learning_rate": 3.521691807085439e-06,
      "loss": 1.0118,
      "step": 3438
    },
    {
      "epoch": 0.2481151473612063,
      "grad_norm": 4.2025214119932475,
      "learning_rate": 3.521388480526524e-06,
      "loss": 0.9863,
      "step": 3439
    },
    {
      "epoch": 0.24818729483063381,
      "grad_norm": 6.906725284568233,
      "learning_rate": 3.5210850708902163e-06,
      "loss": 1.0659,
      "step": 3440
    },
    {
      "epoch": 0.24825944230006133,
      "grad_norm": 4.483385568383023,
      "learning_rate": 3.5207815781930834e-06,
      "loss": 1.0455,
      "step": 3441
    },
    {
      "epoch": 0.24833158976948883,
      "grad_norm": 7.1682287184635145,
      "learning_rate": 3.520478002451699e-06,
      "loss": 1.0453,
      "step": 3442
    },
    {
      "epoch": 0.24840373723891634,
      "grad_norm": 4.176369749588549,
      "learning_rate": 3.520174343682639e-06,
      "loss": 0.9499,
      "step": 3443
    },
    {
      "epoch": 0.24847588470834386,
      "grad_norm": 4.676368619122541,
      "learning_rate": 3.519870601902486e-06,
      "loss": 1.0749,
      "step": 3444
    },
    {
      "epoch": 0.24854803217777136,
      "grad_norm": 0.8720918761659812,
      "learning_rate": 3.519566777127825e-06,
      "loss": 0.9338,
      "step": 3445
    },
    {
      "epoch": 0.24862017964719887,
      "grad_norm": 8.744639471265918,
      "learning_rate": 3.5192628693752485e-06,
      "loss": 0.7614,
      "step": 3446
    },
    {
      "epoch": 0.2486923271166264,
      "grad_norm": 4.4049845291887895,
      "learning_rate": 3.5189588786613504e-06,
      "loss": 0.8012,
      "step": 3447
    },
    {
      "epoch": 0.24876447458605389,
      "grad_norm": 4.889737667231822,
      "learning_rate": 3.5186548050027306e-06,
      "loss": 0.9153,
      "step": 3448
    },
    {
      "epoch": 0.2488366220554814,
      "grad_norm": 21.485521276239,
      "learning_rate": 3.5183506484159946e-06,
      "loss": 1.002,
      "step": 3449
    },
    {
      "epoch": 0.24890876952490892,
      "grad_norm": 6.930974967597621,
      "learning_rate": 3.5180464089177494e-06,
      "loss": 1.0285,
      "step": 3450
    },
    {
      "epoch": 0.24898091699433642,
      "grad_norm": 6.225013837071249,
      "learning_rate": 3.5177420865246096e-06,
      "loss": 0.9111,
      "step": 3451
    },
    {
      "epoch": 0.24905306446376393,
      "grad_norm": 0.7711237583679107,
      "learning_rate": 3.5174376812531936e-06,
      "loss": 0.8456,
      "step": 3452
    },
    {
      "epoch": 0.24912521193319145,
      "grad_norm": 3.6764532887668877,
      "learning_rate": 3.517133193120123e-06,
      "loss": 1.0425,
      "step": 3453
    },
    {
      "epoch": 0.24919735940261895,
      "grad_norm": 6.993518280745053,
      "learning_rate": 3.516828622142025e-06,
      "loss": 1.0071,
      "step": 3454
    },
    {
      "epoch": 0.24926950687204646,
      "grad_norm": 3.3986857005719098,
      "learning_rate": 3.5165239683355308e-06,
      "loss": 1.0049,
      "step": 3455
    },
    {
      "epoch": 0.24934165434147398,
      "grad_norm": 7.693297394393641,
      "learning_rate": 3.5162192317172767e-06,
      "loss": 1.0581,
      "step": 3456
    },
    {
      "epoch": 0.24941380181090148,
      "grad_norm": 5.720763805584627,
      "learning_rate": 3.5159144123039034e-06,
      "loss": 0.8244,
      "step": 3457
    },
    {
      "epoch": 0.249485949280329,
      "grad_norm": 4.780132287720744,
      "learning_rate": 3.5156095101120555e-06,
      "loss": 1.1218,
      "step": 3458
    },
    {
      "epoch": 0.24955809674975651,
      "grad_norm": 16.91870876570751,
      "learning_rate": 3.5153045251583827e-06,
      "loss": 1.0162,
      "step": 3459
    },
    {
      "epoch": 0.249630244219184,
      "grad_norm": 3.8581880914252893,
      "learning_rate": 3.5149994574595398e-06,
      "loss": 1.1025,
      "step": 3460
    },
    {
      "epoch": 0.24970239168861152,
      "grad_norm": 5.0705709288676495,
      "learning_rate": 3.5146943070321845e-06,
      "loss": 0.9901,
      "step": 3461
    },
    {
      "epoch": 0.24977453915803904,
      "grad_norm": 12.038994058046855,
      "learning_rate": 3.5143890738929805e-06,
      "loss": 1.0375,
      "step": 3462
    },
    {
      "epoch": 0.24984668662746654,
      "grad_norm": 4.2835112677005505,
      "learning_rate": 3.514083758058595e-06,
      "loss": 1.0304,
      "step": 3463
    },
    {
      "epoch": 0.24991883409689405,
      "grad_norm": 0.8567304669333301,
      "learning_rate": 3.5137783595457006e-06,
      "loss": 0.8686,
      "step": 3464
    },
    {
      "epoch": 0.24999098156632157,
      "grad_norm": 2.5595178696995418,
      "learning_rate": 3.5134728783709737e-06,
      "loss": 1.0886,
      "step": 3465
    },
    {
      "epoch": 0.25006312903574907,
      "grad_norm": 12.49670488795509,
      "learning_rate": 3.513167314551095e-06,
      "loss": 1.0326,
      "step": 3466
    },
    {
      "epoch": 0.2501352765051766,
      "grad_norm": 4.432919999742957,
      "learning_rate": 3.512861668102752e-06,
      "loss": 1.0178,
      "step": 3467
    },
    {
      "epoch": 0.2502074239746041,
      "grad_norm": 5.105642839658131,
      "learning_rate": 3.5125559390426333e-06,
      "loss": 1.0753,
      "step": 3468
    },
    {
      "epoch": 0.2502795714440316,
      "grad_norm": 4.051038801931655,
      "learning_rate": 3.5122501273874336e-06,
      "loss": 0.994,
      "step": 3469
    },
    {
      "epoch": 0.2503517189134591,
      "grad_norm": 5.212279734543027,
      "learning_rate": 3.511944233153853e-06,
      "loss": 0.9106,
      "step": 3470
    },
    {
      "epoch": 0.2504238663828866,
      "grad_norm": 27.921706683180428,
      "learning_rate": 3.511638256358595e-06,
      "loss": 0.8771,
      "step": 3471
    },
    {
      "epoch": 0.2504960138523141,
      "grad_norm": 4.371406605342097,
      "learning_rate": 3.5113321970183674e-06,
      "loss": 1.0653,
      "step": 3472
    },
    {
      "epoch": 0.25056816132174164,
      "grad_norm": 4.929950205336724,
      "learning_rate": 3.5110260551498834e-06,
      "loss": 1.0385,
      "step": 3473
    },
    {
      "epoch": 0.25064030879116916,
      "grad_norm": 4.0489654444121825,
      "learning_rate": 3.51071983076986e-06,
      "loss": 0.9594,
      "step": 3474
    },
    {
      "epoch": 0.2507124562605967,
      "grad_norm": 6.466788321592803,
      "learning_rate": 3.5104135238950194e-06,
      "loss": 0.9539,
      "step": 3475
    },
    {
      "epoch": 0.25078460373002415,
      "grad_norm": 4.322190385345474,
      "learning_rate": 3.5101071345420874e-06,
      "loss": 0.9923,
      "step": 3476
    },
    {
      "epoch": 0.25085675119945167,
      "grad_norm": 5.238618775828873,
      "learning_rate": 3.5098006627277956e-06,
      "loss": 1.0072,
      "step": 3477
    },
    {
      "epoch": 0.2509288986688792,
      "grad_norm": 10.37776753844637,
      "learning_rate": 3.509494108468878e-06,
      "loss": 0.958,
      "step": 3478
    },
    {
      "epoch": 0.2510010461383067,
      "grad_norm": 0.7662230899808766,
      "learning_rate": 3.5091874717820754e-06,
      "loss": 0.8386,
      "step": 3479
    },
    {
      "epoch": 0.2510731936077342,
      "grad_norm": 4.011941014032141,
      "learning_rate": 3.508880752684131e-06,
      "loss": 1.0373,
      "step": 3480
    },
    {
      "epoch": 0.25114534107716174,
      "grad_norm": 5.347552814695431,
      "learning_rate": 3.5085739511917956e-06,
      "loss": 0.8738,
      "step": 3481
    },
    {
      "epoch": 0.2512174885465892,
      "grad_norm": 2.8908521408329917,
      "learning_rate": 3.508267067321821e-06,
      "loss": 0.999,
      "step": 3482
    },
    {
      "epoch": 0.2512896360160167,
      "grad_norm": 5.328099225560826,
      "learning_rate": 3.5079601010909647e-06,
      "loss": 1.027,
      "step": 3483
    },
    {
      "epoch": 0.25136178348544425,
      "grad_norm": 7.090344388746247,
      "learning_rate": 3.50765305251599e-06,
      "loss": 0.928,
      "step": 3484
    },
    {
      "epoch": 0.25143393095487176,
      "grad_norm": 3.688330217427074,
      "learning_rate": 3.507345921613663e-06,
      "loss": 0.9787,
      "step": 3485
    },
    {
      "epoch": 0.2515060784242993,
      "grad_norm": 3.3316020443176675,
      "learning_rate": 3.5070387084007553e-06,
      "loss": 0.9614,
      "step": 3486
    },
    {
      "epoch": 0.2515782258937268,
      "grad_norm": 3.965900870272679,
      "learning_rate": 3.5067314128940427e-06,
      "loss": 0.9465,
      "step": 3487
    },
    {
      "epoch": 0.25165037336315427,
      "grad_norm": 3.9172401549456417,
      "learning_rate": 3.5064240351103053e-06,
      "loss": 0.8824,
      "step": 3488
    },
    {
      "epoch": 0.2517225208325818,
      "grad_norm": 3.5870618865026707,
      "learning_rate": 3.506116575066328e-06,
      "loss": 1.0719,
      "step": 3489
    },
    {
      "epoch": 0.2517946683020093,
      "grad_norm": 4.851993166963526,
      "learning_rate": 3.5058090327788994e-06,
      "loss": 0.9345,
      "step": 3490
    },
    {
      "epoch": 0.2518668157714368,
      "grad_norm": 0.6512892188858671,
      "learning_rate": 3.505501408264815e-06,
      "loss": 0.799,
      "step": 3491
    },
    {
      "epoch": 0.25193896324086434,
      "grad_norm": 2.8289926789109128,
      "learning_rate": 3.5051937015408706e-06,
      "loss": 1.0397,
      "step": 3492
    },
    {
      "epoch": 0.25201111071029186,
      "grad_norm": 3.6369745510785294,
      "learning_rate": 3.5048859126238714e-06,
      "loss": 0.879,
      "step": 3493
    },
    {
      "epoch": 0.2520832581797193,
      "grad_norm": 6.707846519229335,
      "learning_rate": 3.5045780415306227e-06,
      "loss": 0.8386,
      "step": 3494
    },
    {
      "epoch": 0.25215540564914685,
      "grad_norm": 4.012875101495211,
      "learning_rate": 3.504270088277937e-06,
      "loss": 0.9219,
      "step": 3495
    },
    {
      "epoch": 0.25222755311857437,
      "grad_norm": 3.4788555683836355,
      "learning_rate": 3.50396205288263e-06,
      "loss": 0.9793,
      "step": 3496
    },
    {
      "epoch": 0.2522997005880019,
      "grad_norm": 5.848436730334073,
      "learning_rate": 3.503653935361524e-06,
      "loss": 0.9493,
      "step": 3497
    },
    {
      "epoch": 0.2523718480574294,
      "grad_norm": 4.714480862562366,
      "learning_rate": 3.5033457357314417e-06,
      "loss": 1.0568,
      "step": 3498
    },
    {
      "epoch": 0.25244399552685687,
      "grad_norm": 4.649413139472971,
      "learning_rate": 3.5030374540092143e-06,
      "loss": 0.914,
      "step": 3499
    },
    {
      "epoch": 0.2525161429962844,
      "grad_norm": 5.938843585131882,
      "learning_rate": 3.502729090211676e-06,
      "loss": 0.9635,
      "step": 3500
    },
    {
      "epoch": 0.2525882904657119,
      "grad_norm": 4.833622934719756,
      "learning_rate": 3.5024206443556645e-06,
      "loss": 1.046,
      "step": 3501
    },
    {
      "epoch": 0.2526604379351394,
      "grad_norm": 3.2734546888532896,
      "learning_rate": 3.502112116458024e-06,
      "loss": 0.9612,
      "step": 3502
    },
    {
      "epoch": 0.25273258540456695,
      "grad_norm": 0.7015205090366748,
      "learning_rate": 3.501803506535601e-06,
      "loss": 0.7999,
      "step": 3503
    },
    {
      "epoch": 0.25280473287399446,
      "grad_norm": 2.988090080180065,
      "learning_rate": 3.5014948146052477e-06,
      "loss": 1.0962,
      "step": 3504
    },
    {
      "epoch": 0.25287688034342193,
      "grad_norm": 7.668558495453586,
      "learning_rate": 3.5011860406838214e-06,
      "loss": 1.023,
      "step": 3505
    },
    {
      "epoch": 0.25294902781284945,
      "grad_norm": 0.8145309154949603,
      "learning_rate": 3.500877184788182e-06,
      "loss": 0.9045,
      "step": 3506
    },
    {
      "epoch": 0.25302117528227697,
      "grad_norm": 0.7438221648011212,
      "learning_rate": 3.500568246935196e-06,
      "loss": 0.8079,
      "step": 3507
    },
    {
      "epoch": 0.2530933227517045,
      "grad_norm": 3.2995981838776713,
      "learning_rate": 3.5002592271417334e-06,
      "loss": 0.9929,
      "step": 3508
    },
    {
      "epoch": 0.253165470221132,
      "grad_norm": 5.088563490176249,
      "learning_rate": 3.4999501254246678e-06,
      "loss": 0.9682,
      "step": 3509
    },
    {
      "epoch": 0.2532376176905595,
      "grad_norm": 5.307906554755703,
      "learning_rate": 3.4996409418008782e-06,
      "loss": 0.9953,
      "step": 3510
    },
    {
      "epoch": 0.253309765159987,
      "grad_norm": 3.5809979628040747,
      "learning_rate": 3.4993316762872486e-06,
      "loss": 0.9525,
      "step": 3511
    },
    {
      "epoch": 0.2533819126294145,
      "grad_norm": 3.980762235025066,
      "learning_rate": 3.499022328900666e-06,
      "loss": 0.9567,
      "step": 3512
    },
    {
      "epoch": 0.253454060098842,
      "grad_norm": 3.7238405151225087,
      "learning_rate": 3.4987128996580243e-06,
      "loss": 1.0485,
      "step": 3513
    },
    {
      "epoch": 0.25352620756826955,
      "grad_norm": 9.831263394355744,
      "learning_rate": 3.498403388576218e-06,
      "loss": 1.0277,
      "step": 3514
    },
    {
      "epoch": 0.25359835503769707,
      "grad_norm": 4.272035581504858,
      "learning_rate": 3.49809379567215e-06,
      "loss": 1.0644,
      "step": 3515
    },
    {
      "epoch": 0.2536705025071246,
      "grad_norm": 4.312743249197208,
      "learning_rate": 3.497784120962726e-06,
      "loss": 0.9472,
      "step": 3516
    },
    {
      "epoch": 0.25374264997655205,
      "grad_norm": 12.000161487764316,
      "learning_rate": 3.497474364464855e-06,
      "loss": 0.883,
      "step": 3517
    },
    {
      "epoch": 0.25381479744597957,
      "grad_norm": 4.877516268297807,
      "learning_rate": 3.497164526195453e-06,
      "loss": 1.016,
      "step": 3518
    },
    {
      "epoch": 0.2538869449154071,
      "grad_norm": 4.560427260208663,
      "learning_rate": 3.496854606171438e-06,
      "loss": 1.0743,
      "step": 3519
    },
    {
      "epoch": 0.2539590923848346,
      "grad_norm": 0.720325608868587,
      "learning_rate": 3.4965446044097347e-06,
      "loss": 0.868,
      "step": 3520
    },
    {
      "epoch": 0.2540312398542621,
      "grad_norm": 9.153783680875877,
      "learning_rate": 3.49623452092727e-06,
      "loss": 1.0124,
      "step": 3521
    },
    {
      "epoch": 0.25410338732368964,
      "grad_norm": 4.465485133763004,
      "learning_rate": 3.4959243557409774e-06,
      "loss": 1.0831,
      "step": 3522
    },
    {
      "epoch": 0.2541755347931171,
      "grad_norm": 4.631414347215203,
      "learning_rate": 3.4956141088677937e-06,
      "loss": 1.0888,
      "step": 3523
    },
    {
      "epoch": 0.2542476822625446,
      "grad_norm": 4.382396657871004,
      "learning_rate": 3.4953037803246596e-06,
      "loss": 0.997,
      "step": 3524
    },
    {
      "epoch": 0.25431982973197215,
      "grad_norm": 7.922622795850629,
      "learning_rate": 3.494993370128522e-06,
      "loss": 0.9443,
      "step": 3525
    },
    {
      "epoch": 0.25439197720139967,
      "grad_norm": 3.7657065481639385,
      "learning_rate": 3.49468287829633e-06,
      "loss": 1.0075,
      "step": 3526
    },
    {
      "epoch": 0.2544641246708272,
      "grad_norm": 11.572237110988077,
      "learning_rate": 3.4943723048450397e-06,
      "loss": 0.9324,
      "step": 3527
    },
    {
      "epoch": 0.2545362721402547,
      "grad_norm": 3.7708740372682628,
      "learning_rate": 3.4940616497916097e-06,
      "loss": 1.0578,
      "step": 3528
    },
    {
      "epoch": 0.25460841960968217,
      "grad_norm": 6.687843670437062,
      "learning_rate": 3.493750913153003e-06,
      "loss": 0.9134,
      "step": 3529
    },
    {
      "epoch": 0.2546805670791097,
      "grad_norm": 3.34342170601072,
      "learning_rate": 3.49344009494619e-06,
      "loss": 1.0009,
      "step": 3530
    },
    {
      "epoch": 0.2547527145485372,
      "grad_norm": 0.7289040683006354,
      "learning_rate": 3.4931291951881415e-06,
      "loss": 0.8396,
      "step": 3531
    },
    {
      "epoch": 0.2548248620179647,
      "grad_norm": 7.198024309147485,
      "learning_rate": 3.492818213895835e-06,
      "loss": 1.0537,
      "step": 3532
    },
    {
      "epoch": 0.25489700948739225,
      "grad_norm": 5.786979187174684,
      "learning_rate": 3.4925071510862515e-06,
      "loss": 0.9206,
      "step": 3533
    },
    {
      "epoch": 0.25496915695681976,
      "grad_norm": 5.345725960172272,
      "learning_rate": 3.492196006776378e-06,
      "loss": 1.0715,
      "step": 3534
    },
    {
      "epoch": 0.25504130442624723,
      "grad_norm": 6.665580692850428,
      "learning_rate": 3.4918847809832043e-06,
      "loss": 0.9646,
      "step": 3535
    },
    {
      "epoch": 0.25511345189567475,
      "grad_norm": 5.7085948058153875,
      "learning_rate": 3.4915734737237257e-06,
      "loss": 1.0805,
      "step": 3536
    },
    {
      "epoch": 0.25518559936510227,
      "grad_norm": 6.665531761207651,
      "learning_rate": 3.4912620850149416e-06,
      "loss": 0.9695,
      "step": 3537
    },
    {
      "epoch": 0.2552577468345298,
      "grad_norm": 5.512565131462238,
      "learning_rate": 3.4909506148738546e-06,
      "loss": 0.9941,
      "step": 3538
    },
    {
      "epoch": 0.2553298943039573,
      "grad_norm": 4.8987569750409845,
      "learning_rate": 3.4906390633174743e-06,
      "loss": 0.9158,
      "step": 3539
    },
    {
      "epoch": 0.2554020417733848,
      "grad_norm": 12.713226543658799,
      "learning_rate": 3.4903274303628135e-06,
      "loss": 1.0774,
      "step": 3540
    },
    {
      "epoch": 0.2554741892428123,
      "grad_norm": 2.8783592832890563,
      "learning_rate": 3.490015716026888e-06,
      "loss": 1.0442,
      "step": 3541
    },
    {
      "epoch": 0.2555463367122398,
      "grad_norm": 4.71061945864916,
      "learning_rate": 3.48970392032672e-06,
      "loss": 1.022,
      "step": 3542
    },
    {
      "epoch": 0.2556184841816673,
      "grad_norm": 4.241946275010239,
      "learning_rate": 3.489392043279336e-06,
      "loss": 1.1552,
      "step": 3543
    },
    {
      "epoch": 0.25569063165109485,
      "grad_norm": 5.7673310854433435,
      "learning_rate": 3.489080084901766e-06,
      "loss": 0.9181,
      "step": 3544
    },
    {
      "epoch": 0.25576277912052237,
      "grad_norm": 3.5112711755936323,
      "learning_rate": 3.488768045211045e-06,
      "loss": 1.0142,
      "step": 3545
    },
    {
      "epoch": 0.2558349265899499,
      "grad_norm": 4.048664654653414,
      "learning_rate": 3.488455924224212e-06,
      "loss": 0.9806,
      "step": 3546
    },
    {
      "epoch": 0.25590707405937735,
      "grad_norm": 4.773517214644851,
      "learning_rate": 3.4881437219583117e-06,
      "loss": 0.9349,
      "step": 3547
    },
    {
      "epoch": 0.25597922152880487,
      "grad_norm": 3.0492968202058135,
      "learning_rate": 3.4878314384303908e-06,
      "loss": 0.9372,
      "step": 3548
    },
    {
      "epoch": 0.2560513689982324,
      "grad_norm": 4.50337220042238,
      "learning_rate": 3.4875190736575035e-06,
      "loss": 1.0373,
      "step": 3549
    },
    {
      "epoch": 0.2561235164676599,
      "grad_norm": 5.651705776399139,
      "learning_rate": 3.4872066276567058e-06,
      "loss": 0.9309,
      "step": 3550
    },
    {
      "epoch": 0.2561956639370874,
      "grad_norm": 3.7018980416014426,
      "learning_rate": 3.4868941004450597e-06,
      "loss": 0.9773,
      "step": 3551
    },
    {
      "epoch": 0.2562678114065149,
      "grad_norm": 5.169210525532442,
      "learning_rate": 3.4865814920396314e-06,
      "loss": 0.9813,
      "step": 3552
    },
    {
      "epoch": 0.2563399588759424,
      "grad_norm": 4.1523366301826306,
      "learning_rate": 3.4862688024574906e-06,
      "loss": 0.9903,
      "step": 3553
    },
    {
      "epoch": 0.25641210634536993,
      "grad_norm": 6.913703038413578,
      "learning_rate": 3.4859560317157124e-06,
      "loss": 1.046,
      "step": 3554
    },
    {
      "epoch": 0.25648425381479745,
      "grad_norm": 3.027003509106097,
      "learning_rate": 3.4856431798313757e-06,
      "loss": 1.0064,
      "step": 3555
    },
    {
      "epoch": 0.25655640128422497,
      "grad_norm": 5.211264169682422,
      "learning_rate": 3.4853302468215653e-06,
      "loss": 1.007,
      "step": 3556
    },
    {
      "epoch": 0.2566285487536525,
      "grad_norm": 7.162766334148922,
      "learning_rate": 3.485017232703369e-06,
      "loss": 0.9362,
      "step": 3557
    },
    {
      "epoch": 0.25670069622307995,
      "grad_norm": 9.010129102994958,
      "learning_rate": 3.484704137493878e-06,
      "loss": 0.8882,
      "step": 3558
    },
    {
      "epoch": 0.25677284369250747,
      "grad_norm": 0.7928873405572692,
      "learning_rate": 3.4843909612101906e-06,
      "loss": 0.9061,
      "step": 3559
    },
    {
      "epoch": 0.256844991161935,
      "grad_norm": 6.7350767969228995,
      "learning_rate": 3.484077703869408e-06,
      "loss": 0.9568,
      "step": 3560
    },
    {
      "epoch": 0.2569171386313625,
      "grad_norm": 12.745455661318996,
      "learning_rate": 3.4837643654886357e-06,
      "loss": 0.9922,
      "step": 3561
    },
    {
      "epoch": 0.25698928610079,
      "grad_norm": 5.964786192011098,
      "learning_rate": 3.4834509460849844e-06,
      "loss": 1.0389,
      "step": 3562
    },
    {
      "epoch": 0.25706143357021755,
      "grad_norm": 3.684901728361377,
      "learning_rate": 3.4831374456755684e-06,
      "loss": 0.9215,
      "step": 3563
    },
    {
      "epoch": 0.257133581039645,
      "grad_norm": 3.512781917892618,
      "learning_rate": 3.482823864277507e-06,
      "loss": 1.0551,
      "step": 3564
    },
    {
      "epoch": 0.25720572850907253,
      "grad_norm": 4.019560432884223,
      "learning_rate": 3.482510201907923e-06,
      "loss": 1.057,
      "step": 3565
    },
    {
      "epoch": 0.25727787597850005,
      "grad_norm": 5.018338714127119,
      "learning_rate": 3.4821964585839453e-06,
      "loss": 1.0441,
      "step": 3566
    },
    {
      "epoch": 0.25735002344792757,
      "grad_norm": 4.520277958990806,
      "learning_rate": 3.481882634322706e-06,
      "loss": 0.8635,
      "step": 3567
    },
    {
      "epoch": 0.2574221709173551,
      "grad_norm": 4.374004359713642,
      "learning_rate": 3.4815687291413416e-06,
      "loss": 0.8188,
      "step": 3568
    },
    {
      "epoch": 0.2574943183867826,
      "grad_norm": 5.877725496562378,
      "learning_rate": 3.4812547430569933e-06,
      "loss": 1.0162,
      "step": 3569
    },
    {
      "epoch": 0.25756646585621007,
      "grad_norm": 3.0224519449078664,
      "learning_rate": 3.480940676086807e-06,
      "loss": 0.9193,
      "step": 3570
    },
    {
      "epoch": 0.2576386133256376,
      "grad_norm": 5.000916969139416,
      "learning_rate": 3.480626528247933e-06,
      "loss": 0.957,
      "step": 3571
    },
    {
      "epoch": 0.2577107607950651,
      "grad_norm": 3.7885784665868214,
      "learning_rate": 3.4803122995575258e-06,
      "loss": 0.948,
      "step": 3572
    },
    {
      "epoch": 0.2577829082644926,
      "grad_norm": 3.8991045486275344,
      "learning_rate": 3.4799979900327427e-06,
      "loss": 1.0101,
      "step": 3573
    },
    {
      "epoch": 0.25785505573392015,
      "grad_norm": 6.2296684683917345,
      "learning_rate": 3.479683599690749e-06,
      "loss": 0.9701,
      "step": 3574
    },
    {
      "epoch": 0.25792720320334767,
      "grad_norm": 4.1963816724203005,
      "learning_rate": 3.4793691285487112e-06,
      "loss": 1.0224,
      "step": 3575
    },
    {
      "epoch": 0.25799935067277513,
      "grad_norm": 3.883928288301605,
      "learning_rate": 3.4790545766238015e-06,
      "loss": 1.0744,
      "step": 3576
    },
    {
      "epoch": 0.25807149814220265,
      "grad_norm": 5.101870937866852,
      "learning_rate": 3.478739943933197e-06,
      "loss": 0.9614,
      "step": 3577
    },
    {
      "epoch": 0.25814364561163017,
      "grad_norm": 4.572870252832805,
      "learning_rate": 3.478425230494078e-06,
      "loss": 0.9554,
      "step": 3578
    },
    {
      "epoch": 0.2582157930810577,
      "grad_norm": 3.6995193375433817,
      "learning_rate": 3.4781104363236305e-06,
      "loss": 1.156,
      "step": 3579
    },
    {
      "epoch": 0.2582879405504852,
      "grad_norm": 3.4124831817547876,
      "learning_rate": 3.477795561439043e-06,
      "loss": 1.0242,
      "step": 3580
    },
    {
      "epoch": 0.2583600880199127,
      "grad_norm": 4.613839163382434,
      "learning_rate": 3.4774806058575116e-06,
      "loss": 0.967,
      "step": 3581
    },
    {
      "epoch": 0.2584322354893402,
      "grad_norm": 8.160661578567165,
      "learning_rate": 3.477165569596233e-06,
      "loss": 1.0404,
      "step": 3582
    },
    {
      "epoch": 0.2585043829587677,
      "grad_norm": 2.9915568431648145,
      "learning_rate": 3.476850452672411e-06,
      "loss": 1.0205,
      "step": 3583
    },
    {
      "epoch": 0.25857653042819523,
      "grad_norm": 0.7557948673922958,
      "learning_rate": 3.476535255103253e-06,
      "loss": 0.8467,
      "step": 3584
    },
    {
      "epoch": 0.25864867789762275,
      "grad_norm": 4.150674392872731,
      "learning_rate": 3.476219976905971e-06,
      "loss": 0.9424,
      "step": 3585
    },
    {
      "epoch": 0.25872082536705027,
      "grad_norm": 4.666756061424341,
      "learning_rate": 3.4759046180977807e-06,
      "loss": 1.0463,
      "step": 3586
    },
    {
      "epoch": 0.2587929728364778,
      "grad_norm": 0.6754388106958107,
      "learning_rate": 3.4755891786959025e-06,
      "loss": 0.877,
      "step": 3587
    },
    {
      "epoch": 0.25886512030590525,
      "grad_norm": 4.666499997750338,
      "learning_rate": 3.475273658717562e-06,
      "loss": 0.9795,
      "step": 3588
    },
    {
      "epoch": 0.25893726777533277,
      "grad_norm": 3.8727925073271834,
      "learning_rate": 3.4749580581799876e-06,
      "loss": 0.9829,
      "step": 3589
    },
    {
      "epoch": 0.2590094152447603,
      "grad_norm": 3.4725720767621375,
      "learning_rate": 3.474642377100415e-06,
      "loss": 1.0293,
      "step": 3590
    },
    {
      "epoch": 0.2590815627141878,
      "grad_norm": 5.789408767372747,
      "learning_rate": 3.47432661549608e-06,
      "loss": 0.9163,
      "step": 3591
    },
    {
      "epoch": 0.2591537101836153,
      "grad_norm": 6.3699248812811025,
      "learning_rate": 3.4740107733842265e-06,
      "loss": 0.8521,
      "step": 3592
    },
    {
      "epoch": 0.25922585765304285,
      "grad_norm": 3.872122280546596,
      "learning_rate": 3.4736948507821016e-06,
      "loss": 0.9309,
      "step": 3593
    },
    {
      "epoch": 0.2592980051224703,
      "grad_norm": 4.863020434145833,
      "learning_rate": 3.4733788477069562e-06,
      "loss": 1.0282,
      "step": 3594
    },
    {
      "epoch": 0.25937015259189783,
      "grad_norm": 4.9396237984550675,
      "learning_rate": 3.4730627641760463e-06,
      "loss": 1.0026,
      "step": 3595
    },
    {
      "epoch": 0.25944230006132535,
      "grad_norm": 6.636211768819995,
      "learning_rate": 3.4727466002066323e-06,
      "loss": 1.0102,
      "step": 3596
    },
    {
      "epoch": 0.25951444753075287,
      "grad_norm": 2.997257250452591,
      "learning_rate": 3.4724303558159774e-06,
      "loss": 1.0922,
      "step": 3597
    },
    {
      "epoch": 0.2595865950001804,
      "grad_norm": 3.751760451032767,
      "learning_rate": 3.472114031021352e-06,
      "loss": 0.9785,
      "step": 3598
    },
    {
      "epoch": 0.2596587424696079,
      "grad_norm": 4.962532039510079,
      "learning_rate": 3.471797625840029e-06,
      "loss": 1.0329,
      "step": 3599
    },
    {
      "epoch": 0.25973088993903537,
      "grad_norm": 3.833558504429633,
      "learning_rate": 3.4714811402892866e-06,
      "loss": 1.0076,
      "step": 3600
    },
    {
      "epoch": 0.2598030374084629,
      "grad_norm": 3.9317534156231937,
      "learning_rate": 3.4711645743864058e-06,
      "loss": 0.9008,
      "step": 3601
    },
    {
      "epoch": 0.2598751848778904,
      "grad_norm": 5.7033106107424505,
      "learning_rate": 3.4708479281486734e-06,
      "loss": 0.9574,
      "step": 3602
    },
    {
      "epoch": 0.2599473323473179,
      "grad_norm": 3.1838660375314616,
      "learning_rate": 3.4705312015933814e-06,
      "loss": 1.0051,
      "step": 3603
    },
    {
      "epoch": 0.26001947981674545,
      "grad_norm": 3.0627455320997488,
      "learning_rate": 3.470214394737823e-06,
      "loss": 0.9451,
      "step": 3604
    },
    {
      "epoch": 0.2600916272861729,
      "grad_norm": 3.782537950324106,
      "learning_rate": 3.4698975075993e-06,
      "loss": 1.0709,
      "step": 3605
    },
    {
      "epoch": 0.26016377475560043,
      "grad_norm": 4.224969419109117,
      "learning_rate": 3.469580540195115e-06,
      "loss": 1.0531,
      "step": 3606
    },
    {
      "epoch": 0.26023592222502795,
      "grad_norm": 3.9033892112226716,
      "learning_rate": 3.469263492542577e-06,
      "loss": 0.9956,
      "step": 3607
    },
    {
      "epoch": 0.26030806969445547,
      "grad_norm": 7.911868545381333,
      "learning_rate": 3.468946364658999e-06,
      "loss": 1.0126,
      "step": 3608
    },
    {
      "epoch": 0.260380217163883,
      "grad_norm": 3.8292798849863448,
      "learning_rate": 3.4686291565616976e-06,
      "loss": 1.0218,
      "step": 3609
    },
    {
      "epoch": 0.2604523646333105,
      "grad_norm": 32.77064853300034,
      "learning_rate": 3.468311868267994e-06,
      "loss": 0.9379,
      "step": 3610
    },
    {
      "epoch": 0.26052451210273797,
      "grad_norm": 2.8708494328978014,
      "learning_rate": 3.467994499795216e-06,
      "loss": 0.9576,
      "step": 3611
    },
    {
      "epoch": 0.2605966595721655,
      "grad_norm": 4.328888422983168,
      "learning_rate": 3.467677051160692e-06,
      "loss": 1.0518,
      "step": 3612
    },
    {
      "epoch": 0.260668807041593,
      "grad_norm": 4.179813903608488,
      "learning_rate": 3.4673595223817573e-06,
      "loss": 1.098,
      "step": 3613
    },
    {
      "epoch": 0.26074095451102053,
      "grad_norm": 8.660682617489162,
      "learning_rate": 3.4670419134757516e-06,
      "loss": 1.0938,
      "step": 3614
    },
    {
      "epoch": 0.26081310198044805,
      "grad_norm": 9.771901108255816,
      "learning_rate": 3.4667242244600176e-06,
      "loss": 1.0734,
      "step": 3615
    },
    {
      "epoch": 0.26088524944987557,
      "grad_norm": 0.9634241324557465,
      "learning_rate": 3.466406455351903e-06,
      "loss": 0.859,
      "step": 3616
    },
    {
      "epoch": 0.26095739691930303,
      "grad_norm": 7.203356691552378,
      "learning_rate": 3.4660886061687604e-06,
      "loss": 0.949,
      "step": 3617
    },
    {
      "epoch": 0.26102954438873055,
      "grad_norm": 4.250331416952172,
      "learning_rate": 3.4657706769279465e-06,
      "loss": 1.0091,
      "step": 3618
    },
    {
      "epoch": 0.26110169185815807,
      "grad_norm": 4.367483603278019,
      "learning_rate": 3.465452667646822e-06,
      "loss": 0.9238,
      "step": 3619
    },
    {
      "epoch": 0.2611738393275856,
      "grad_norm": 2.621341244515207,
      "learning_rate": 3.4651345783427522e-06,
      "loss": 0.9293,
      "step": 3620
    },
    {
      "epoch": 0.2612459867970131,
      "grad_norm": 3.1124726014674016,
      "learning_rate": 3.4648164090331067e-06,
      "loss": 1.0331,
      "step": 3621
    },
    {
      "epoch": 0.2613181342664406,
      "grad_norm": 5.280575680993984,
      "learning_rate": 3.4644981597352604e-06,
      "loss": 0.96,
      "step": 3622
    },
    {
      "epoch": 0.2613902817358681,
      "grad_norm": 4.142026038028615,
      "learning_rate": 3.4641798304665903e-06,
      "loss": 0.9189,
      "step": 3623
    },
    {
      "epoch": 0.2614624292052956,
      "grad_norm": 4.9119977388532075,
      "learning_rate": 3.4638614212444797e-06,
      "loss": 0.9459,
      "step": 3624
    },
    {
      "epoch": 0.26153457667472313,
      "grad_norm": 2.9951616053215204,
      "learning_rate": 3.4635429320863163e-06,
      "loss": 0.9882,
      "step": 3625
    },
    {
      "epoch": 0.26160672414415065,
      "grad_norm": 3.5252986483846636,
      "learning_rate": 3.463224363009491e-06,
      "loss": 1.0406,
      "step": 3626
    },
    {
      "epoch": 0.26167887161357817,
      "grad_norm": 2.8011693964830062,
      "learning_rate": 3.4629057140314e-06,
      "loss": 1.0512,
      "step": 3627
    },
    {
      "epoch": 0.2617510190830057,
      "grad_norm": 4.080180267764655,
      "learning_rate": 3.4625869851694437e-06,
      "loss": 0.9462,
      "step": 3628
    },
    {
      "epoch": 0.26182316655243315,
      "grad_norm": 2.793480118041179,
      "learning_rate": 3.462268176441026e-06,
      "loss": 0.958,
      "step": 3629
    },
    {
      "epoch": 0.26189531402186067,
      "grad_norm": 2.4588210920519873,
      "learning_rate": 3.4619492878635565e-06,
      "loss": 0.9691,
      "step": 3630
    },
    {
      "epoch": 0.2619674614912882,
      "grad_norm": 5.169531345929518,
      "learning_rate": 3.461630319454448e-06,
      "loss": 0.9454,
      "step": 3631
    },
    {
      "epoch": 0.2620396089607157,
      "grad_norm": 5.044676688716211,
      "learning_rate": 3.4613112712311194e-06,
      "loss": 0.9292,
      "step": 3632
    },
    {
      "epoch": 0.26211175643014323,
      "grad_norm": 13.302305231157892,
      "learning_rate": 3.460992143210991e-06,
      "loss": 1.017,
      "step": 3633
    },
    {
      "epoch": 0.26218390389957075,
      "grad_norm": 3.502057832938597,
      "learning_rate": 3.460672935411491e-06,
      "loss": 0.9429,
      "step": 3634
    },
    {
      "epoch": 0.2622560513689982,
      "grad_norm": 10.634013223785852,
      "learning_rate": 3.4603536478500484e-06,
      "loss": 1.0566,
      "step": 3635
    },
    {
      "epoch": 0.26232819883842573,
      "grad_norm": 2.5116515439832723,
      "learning_rate": 3.460034280544099e-06,
      "loss": 1.017,
      "step": 3636
    },
    {
      "epoch": 0.26240034630785325,
      "grad_norm": 4.138139316918364,
      "learning_rate": 3.459714833511083e-06,
      "loss": 0.8902,
      "step": 3637
    },
    {
      "epoch": 0.26247249377728077,
      "grad_norm": 3.66364642222973,
      "learning_rate": 3.4593953067684435e-06,
      "loss": 0.988,
      "step": 3638
    },
    {
      "epoch": 0.2625446412467083,
      "grad_norm": 6.563124563378622,
      "learning_rate": 3.4590757003336285e-06,
      "loss": 0.9095,
      "step": 3639
    },
    {
      "epoch": 0.2626167887161358,
      "grad_norm": 5.6244911811427825,
      "learning_rate": 3.458756014224091e-06,
      "loss": 1.0357,
      "step": 3640
    },
    {
      "epoch": 0.26268893618556327,
      "grad_norm": 5.111591756491727,
      "learning_rate": 3.458436248457288e-06,
      "loss": 0.9008,
      "step": 3641
    },
    {
      "epoch": 0.2627610836549908,
      "grad_norm": 3.0726443067639106,
      "learning_rate": 3.4581164030506804e-06,
      "loss": 1.0738,
      "step": 3642
    },
    {
      "epoch": 0.2628332311244183,
      "grad_norm": 8.87928947699144,
      "learning_rate": 3.457796478021733e-06,
      "loss": 0.9141,
      "step": 3643
    },
    {
      "epoch": 0.26290537859384583,
      "grad_norm": 4.872883263786465,
      "learning_rate": 3.457476473387917e-06,
      "loss": 0.9892,
      "step": 3644
    },
    {
      "epoch": 0.26297752606327335,
      "grad_norm": 3.0223910470009026,
      "learning_rate": 3.4571563891667065e-06,
      "loss": 0.9778,
      "step": 3645
    },
    {
      "epoch": 0.26304967353270087,
      "grad_norm": 0.7220751513813511,
      "learning_rate": 3.4568362253755796e-06,
      "loss": 0.8419,
      "step": 3646
    },
    {
      "epoch": 0.26312182100212833,
      "grad_norm": 4.296422872164699,
      "learning_rate": 3.4565159820320196e-06,
      "loss": 0.9521,
      "step": 3647
    },
    {
      "epoch": 0.26319396847155585,
      "grad_norm": 2.954134006794976,
      "learning_rate": 3.456195659153513e-06,
      "loss": 0.987,
      "step": 3648
    },
    {
      "epoch": 0.26326611594098337,
      "grad_norm": 3.243869574957652,
      "learning_rate": 3.4558752567575526e-06,
      "loss": 0.9932,
      "step": 3649
    },
    {
      "epoch": 0.2633382634104109,
      "grad_norm": 2.4607622356674486,
      "learning_rate": 3.4555547748616337e-06,
      "loss": 1.0494,
      "step": 3650
    },
    {
      "epoch": 0.2634104108798384,
      "grad_norm": 4.570868981546065,
      "learning_rate": 3.455234213483257e-06,
      "loss": 0.9376,
      "step": 3651
    },
    {
      "epoch": 0.26348255834926587,
      "grad_norm": 4.5812998104119815,
      "learning_rate": 3.4549135726399273e-06,
      "loss": 1.0116,
      "step": 3652
    },
    {
      "epoch": 0.2635547058186934,
      "grad_norm": 4.489631575865855,
      "learning_rate": 3.4545928523491527e-06,
      "loss": 0.955,
      "step": 3653
    },
    {
      "epoch": 0.2636268532881209,
      "grad_norm": 4.246870122574117,
      "learning_rate": 3.454272052628447e-06,
      "loss": 0.9115,
      "step": 3654
    },
    {
      "epoch": 0.26369900075754843,
      "grad_norm": 3.6318702049926133,
      "learning_rate": 3.4539511734953285e-06,
      "loss": 1.0312,
      "step": 3655
    },
    {
      "epoch": 0.26377114822697595,
      "grad_norm": 8.010784033266452,
      "learning_rate": 3.453630214967319e-06,
      "loss": 1.0748,
      "step": 3656
    },
    {
      "epoch": 0.26384329569640347,
      "grad_norm": 4.189986672027439,
      "learning_rate": 3.4533091770619446e-06,
      "loss": 1.0274,
      "step": 3657
    },
    {
      "epoch": 0.26391544316583093,
      "grad_norm": 5.486680983335064,
      "learning_rate": 3.452988059796736e-06,
      "loss": 0.9811,
      "step": 3658
    },
    {
      "epoch": 0.26398759063525845,
      "grad_norm": 4.159293071762499,
      "learning_rate": 3.4526668631892273e-06,
      "loss": 1.0466,
      "step": 3659
    },
    {
      "epoch": 0.26405973810468597,
      "grad_norm": 3.993070799572729,
      "learning_rate": 3.4523455872569596e-06,
      "loss": 0.9624,
      "step": 3660
    },
    {
      "epoch": 0.2641318855741135,
      "grad_norm": 4.054113329033064,
      "learning_rate": 3.4520242320174755e-06,
      "loss": 0.9995,
      "step": 3661
    },
    {
      "epoch": 0.264204033043541,
      "grad_norm": 3.450413947219308,
      "learning_rate": 3.451702797488323e-06,
      "loss": 1.0156,
      "step": 3662
    },
    {
      "epoch": 0.26427618051296853,
      "grad_norm": 3.0273813946752206,
      "learning_rate": 3.4513812836870556e-06,
      "loss": 0.9979,
      "step": 3663
    },
    {
      "epoch": 0.264348327982396,
      "grad_norm": 4.002415166333484,
      "learning_rate": 3.451059690631229e-06,
      "loss": 0.8899,
      "step": 3664
    },
    {
      "epoch": 0.2644204754518235,
      "grad_norm": 3.4241083251359945,
      "learning_rate": 3.450738018338404e-06,
      "loss": 1.0026,
      "step": 3665
    },
    {
      "epoch": 0.26449262292125103,
      "grad_norm": 6.942242625118746,
      "learning_rate": 3.450416266826146e-06,
      "loss": 0.9631,
      "step": 3666
    },
    {
      "epoch": 0.26456477039067855,
      "grad_norm": 3.42025690708839,
      "learning_rate": 3.4500944361120256e-06,
      "loss": 0.9765,
      "step": 3667
    },
    {
      "epoch": 0.26463691786010607,
      "grad_norm": 2.6670694245727318,
      "learning_rate": 3.4497725262136154e-06,
      "loss": 0.9975,
      "step": 3668
    },
    {
      "epoch": 0.2647090653295336,
      "grad_norm": 3.264607248094932,
      "learning_rate": 3.449450537148495e-06,
      "loss": 0.9762,
      "step": 3669
    },
    {
      "epoch": 0.26478121279896105,
      "grad_norm": 6.9770775548258515,
      "learning_rate": 3.449128468934246e-06,
      "loss": 0.9276,
      "step": 3670
    },
    {
      "epoch": 0.26485336026838857,
      "grad_norm": 4.956532939517055,
      "learning_rate": 3.448806321588456e-06,
      "loss": 0.9221,
      "step": 3671
    },
    {
      "epoch": 0.2649255077378161,
      "grad_norm": 4.389774335131467,
      "learning_rate": 3.448484095128716e-06,
      "loss": 0.9353,
      "step": 3672
    },
    {
      "epoch": 0.2649976552072436,
      "grad_norm": 2.8417336158659348,
      "learning_rate": 3.448161789572622e-06,
      "loss": 1.0021,
      "step": 3673
    },
    {
      "epoch": 0.26506980267667113,
      "grad_norm": 2.9164584130916333,
      "learning_rate": 3.447839404937773e-06,
      "loss": 1.0684,
      "step": 3674
    },
    {
      "epoch": 0.26514195014609865,
      "grad_norm": 2.1498215534764724,
      "learning_rate": 3.4475169412417745e-06,
      "loss": 1.0003,
      "step": 3675
    },
    {
      "epoch": 0.2652140976155261,
      "grad_norm": 8.993231029263239,
      "learning_rate": 3.447194398502234e-06,
      "loss": 1.0793,
      "step": 3676
    },
    {
      "epoch": 0.26528624508495363,
      "grad_norm": 4.3554133527166465,
      "learning_rate": 3.446871776736764e-06,
      "loss": 0.9325,
      "step": 3677
    },
    {
      "epoch": 0.26535839255438115,
      "grad_norm": 6.51025139764188,
      "learning_rate": 3.4465490759629835e-06,
      "loss": 0.9068,
      "step": 3678
    },
    {
      "epoch": 0.26543054002380867,
      "grad_norm": 4.013138175435368,
      "learning_rate": 3.446226296198513e-06,
      "loss": 1.0983,
      "step": 3679
    },
    {
      "epoch": 0.2655026874932362,
      "grad_norm": 0.9003315076638837,
      "learning_rate": 3.4459034374609773e-06,
      "loss": 0.855,
      "step": 3680
    },
    {
      "epoch": 0.2655748349626637,
      "grad_norm": 5.142180882753642,
      "learning_rate": 3.445580499768008e-06,
      "loss": 1.0319,
      "step": 3681
    },
    {
      "epoch": 0.26564698243209117,
      "grad_norm": 3.4190087703426104,
      "learning_rate": 3.445257483137239e-06,
      "loss": 1.0683,
      "step": 3682
    },
    {
      "epoch": 0.2657191299015187,
      "grad_norm": 3.64114520240676,
      "learning_rate": 3.4449343875863087e-06,
      "loss": 0.9252,
      "step": 3683
    },
    {
      "epoch": 0.2657912773709462,
      "grad_norm": 2.7602346012365553,
      "learning_rate": 3.4446112131328604e-06,
      "loss": 1.0295,
      "step": 3684
    },
    {
      "epoch": 0.26586342484037373,
      "grad_norm": 3.06497294950761,
      "learning_rate": 3.444287959794542e-06,
      "loss": 0.973,
      "step": 3685
    },
    {
      "epoch": 0.26593557230980125,
      "grad_norm": 3.4246389992356057,
      "learning_rate": 3.4439646275890044e-06,
      "loss": 1.0957,
      "step": 3686
    },
    {
      "epoch": 0.26600771977922877,
      "grad_norm": 4.96214383093898,
      "learning_rate": 3.443641216533904e-06,
      "loss": 0.945,
      "step": 3687
    },
    {
      "epoch": 0.26607986724865623,
      "grad_norm": 3.692319920409568,
      "learning_rate": 3.4433177266469005e-06,
      "loss": 1.0056,
      "step": 3688
    },
    {
      "epoch": 0.26615201471808375,
      "grad_norm": 0.6320753631474878,
      "learning_rate": 3.4429941579456595e-06,
      "loss": 0.8104,
      "step": 3689
    },
    {
      "epoch": 0.26622416218751127,
      "grad_norm": 6.131499909295874,
      "learning_rate": 3.442670510447849e-06,
      "loss": 1.0429,
      "step": 3690
    },
    {
      "epoch": 0.2662963096569388,
      "grad_norm": 7.592906010197117,
      "learning_rate": 3.4423467841711425e-06,
      "loss": 1.0473,
      "step": 3691
    },
    {
      "epoch": 0.2663684571263663,
      "grad_norm": 3.570723962759527,
      "learning_rate": 3.442022979133218e-06,
      "loss": 1.04,
      "step": 3692
    },
    {
      "epoch": 0.26644060459579383,
      "grad_norm": 0.8989040987548337,
      "learning_rate": 3.441699095351756e-06,
      "loss": 0.9059,
      "step": 3693
    },
    {
      "epoch": 0.2665127520652213,
      "grad_norm": 4.481784404561939,
      "learning_rate": 3.4413751328444438e-06,
      "loss": 0.7992,
      "step": 3694
    },
    {
      "epoch": 0.2665848995346488,
      "grad_norm": 3.307097726176745,
      "learning_rate": 3.4410510916289716e-06,
      "loss": 1.0485,
      "step": 3695
    },
    {
      "epoch": 0.26665704700407633,
      "grad_norm": 3.6120698454937976,
      "learning_rate": 3.4407269717230333e-06,
      "loss": 0.9767,
      "step": 3696
    },
    {
      "epoch": 0.26672919447350385,
      "grad_norm": 2.2406241785364496,
      "learning_rate": 3.4404027731443284e-06,
      "loss": 0.9644,
      "step": 3697
    },
    {
      "epoch": 0.26680134194293137,
      "grad_norm": 4.677732540122292,
      "learning_rate": 3.440078495910561e-06,
      "loss": 0.9821,
      "step": 3698
    },
    {
      "epoch": 0.2668734894123589,
      "grad_norm": 3.0275784309238807,
      "learning_rate": 3.4397541400394376e-06,
      "loss": 0.9132,
      "step": 3699
    },
    {
      "epoch": 0.26694563688178635,
      "grad_norm": 3.004319260558097,
      "learning_rate": 3.4394297055486704e-06,
      "loss": 0.958,
      "step": 3700
    },
    {
      "epoch": 0.26701778435121387,
      "grad_norm": 3.0742297626768282,
      "learning_rate": 3.4391051924559756e-06,
      "loss": 1.0773,
      "step": 3701
    },
    {
      "epoch": 0.2670899318206414,
      "grad_norm": 4.212232061631937,
      "learning_rate": 3.4387806007790733e-06,
      "loss": 0.9776,
      "step": 3702
    },
    {
      "epoch": 0.2671620792900689,
      "grad_norm": 3.2536791004336356,
      "learning_rate": 3.4384559305356893e-06,
      "loss": 1.0942,
      "step": 3703
    },
    {
      "epoch": 0.26723422675949643,
      "grad_norm": 2.8386342071473925,
      "learning_rate": 3.438131181743551e-06,
      "loss": 1.0251,
      "step": 3704
    },
    {
      "epoch": 0.2673063742289239,
      "grad_norm": 0.7889489951503202,
      "learning_rate": 3.437806354420393e-06,
      "loss": 0.9334,
      "step": 3705
    },
    {
      "epoch": 0.2673785216983514,
      "grad_norm": 3.0539148626745236,
      "learning_rate": 3.437481448583953e-06,
      "loss": 0.97,
      "step": 3706
    },
    {
      "epoch": 0.26745066916777893,
      "grad_norm": 4.030531471741927,
      "learning_rate": 3.437156464251972e-06,
      "loss": 1.0336,
      "step": 3707
    },
    {
      "epoch": 0.26752281663720645,
      "grad_norm": 31.468418335516365,
      "learning_rate": 3.4368314014421962e-06,
      "loss": 0.9065,
      "step": 3708
    },
    {
      "epoch": 0.26759496410663397,
      "grad_norm": 12.152444827408315,
      "learning_rate": 3.4365062601723774e-06,
      "loss": 0.9346,
      "step": 3709
    },
    {
      "epoch": 0.2676671115760615,
      "grad_norm": 3.96359988089213,
      "learning_rate": 3.4361810404602693e-06,
      "loss": 1.0291,
      "step": 3710
    },
    {
      "epoch": 0.26773925904548895,
      "grad_norm": 4.0017210119058655,
      "learning_rate": 3.435855742323631e-06,
      "loss": 1.0553,
      "step": 3711
    },
    {
      "epoch": 0.2678114065149165,
      "grad_norm": 3.2196034901694235,
      "learning_rate": 3.4355303657802265e-06,
      "loss": 0.9703,
      "step": 3712
    },
    {
      "epoch": 0.267883553984344,
      "grad_norm": 5.329532202541308,
      "learning_rate": 3.435204910847822e-06,
      "loss": 1.0031,
      "step": 3713
    },
    {
      "epoch": 0.2679557014537715,
      "grad_norm": 3.884295236101547,
      "learning_rate": 3.4348793775441907e-06,
      "loss": 0.9782,
      "step": 3714
    },
    {
      "epoch": 0.26802784892319903,
      "grad_norm": 3.6873123638453653,
      "learning_rate": 3.4345537658871086e-06,
      "loss": 0.991,
      "step": 3715
    },
    {
      "epoch": 0.26809999639262655,
      "grad_norm": 3.5861096974501487,
      "learning_rate": 3.434228075894355e-06,
      "loss": 0.8852,
      "step": 3716
    },
    {
      "epoch": 0.268172143862054,
      "grad_norm": 2.762730090885505,
      "learning_rate": 3.4339023075837165e-06,
      "loss": 0.9345,
      "step": 3717
    },
    {
      "epoch": 0.26824429133148153,
      "grad_norm": 8.798863458932939,
      "learning_rate": 3.4335764609729806e-06,
      "loss": 0.9959,
      "step": 3718
    },
    {
      "epoch": 0.26831643880090905,
      "grad_norm": 3.6397385725524134,
      "learning_rate": 3.4332505360799412e-06,
      "loss": 0.9708,
      "step": 3719
    },
    {
      "epoch": 0.26838858627033657,
      "grad_norm": 2.748295776113654,
      "learning_rate": 3.432924532922396e-06,
      "loss": 0.8963,
      "step": 3720
    },
    {
      "epoch": 0.2684607337397641,
      "grad_norm": 3.724693357403996,
      "learning_rate": 3.4325984515181457e-06,
      "loss": 0.9623,
      "step": 3721
    },
    {
      "epoch": 0.2685328812091916,
      "grad_norm": 3.3043758164187094,
      "learning_rate": 3.432272291884998e-06,
      "loss": 0.9664,
      "step": 3722
    },
    {
      "epoch": 0.2686050286786191,
      "grad_norm": 4.761362639857673,
      "learning_rate": 3.4319460540407623e-06,
      "loss": 1.0103,
      "step": 3723
    },
    {
      "epoch": 0.2686771761480466,
      "grad_norm": 6.167460038295717,
      "learning_rate": 3.4316197380032536e-06,
      "loss": 1.0132,
      "step": 3724
    },
    {
      "epoch": 0.2687493236174741,
      "grad_norm": 2.965144166062944,
      "learning_rate": 3.4312933437902904e-06,
      "loss": 1.0491,
      "step": 3725
    },
    {
      "epoch": 0.26882147108690163,
      "grad_norm": 8.060829166897792,
      "learning_rate": 3.4309668714196964e-06,
      "loss": 0.888,
      "step": 3726
    },
    {
      "epoch": 0.26889361855632915,
      "grad_norm": 5.233257299628151,
      "learning_rate": 3.4306403209092983e-06,
      "loss": 0.9246,
      "step": 3727
    },
    {
      "epoch": 0.26896576602575667,
      "grad_norm": 3.4411524008958696,
      "learning_rate": 3.4303136922769295e-06,
      "loss": 0.9798,
      "step": 3728
    },
    {
      "epoch": 0.26903791349518413,
      "grad_norm": 1.9726690008676704,
      "learning_rate": 3.429986985540424e-06,
      "loss": 1.0643,
      "step": 3729
    },
    {
      "epoch": 0.26911006096461165,
      "grad_norm": 5.074984568304644,
      "learning_rate": 3.429660200717623e-06,
      "loss": 0.9605,
      "step": 3730
    },
    {
      "epoch": 0.26918220843403917,
      "grad_norm": 3.838419622019217,
      "learning_rate": 3.429333337826371e-06,
      "loss": 1.0363,
      "step": 3731
    },
    {
      "epoch": 0.2692543559034667,
      "grad_norm": 6.206990380509723,
      "learning_rate": 3.429006396884516e-06,
      "loss": 1.0114,
      "step": 3732
    },
    {
      "epoch": 0.2693265033728942,
      "grad_norm": 3.819070625892724,
      "learning_rate": 3.4286793779099124e-06,
      "loss": 1.0358,
      "step": 3733
    },
    {
      "epoch": 0.26939865084232173,
      "grad_norm": 5.167840496152794,
      "learning_rate": 3.4283522809204163e-06,
      "loss": 1.0571,
      "step": 3734
    },
    {
      "epoch": 0.2694707983117492,
      "grad_norm": 2.7705425788706894,
      "learning_rate": 3.4280251059338898e-06,
      "loss": 0.9773,
      "step": 3735
    },
    {
      "epoch": 0.2695429457811767,
      "grad_norm": 3.027272239578795,
      "learning_rate": 3.427697852968199e-06,
      "loss": 1.0148,
      "step": 3736
    },
    {
      "epoch": 0.26961509325060423,
      "grad_norm": 4.712496826818395,
      "learning_rate": 3.4273705220412137e-06,
      "loss": 0.9245,
      "step": 3737
    },
    {
      "epoch": 0.26968724072003175,
      "grad_norm": 3.490714016208796,
      "learning_rate": 3.427043113170807e-06,
      "loss": 1.0271,
      "step": 3738
    },
    {
      "epoch": 0.26975938818945927,
      "grad_norm": 5.627546793515626,
      "learning_rate": 3.42671562637486e-06,
      "loss": 0.9664,
      "step": 3739
    },
    {
      "epoch": 0.2698315356588868,
      "grad_norm": 7.7275876251356985,
      "learning_rate": 3.426388061671253e-06,
      "loss": 0.9141,
      "step": 3740
    },
    {
      "epoch": 0.26990368312831425,
      "grad_norm": 3.5258445215074006,
      "learning_rate": 3.426060419077875e-06,
      "loss": 1.0521,
      "step": 3741
    },
    {
      "epoch": 0.2699758305977418,
      "grad_norm": 3.143569401128582,
      "learning_rate": 3.4257326986126165e-06,
      "loss": 0.9455,
      "step": 3742
    },
    {
      "epoch": 0.2700479780671693,
      "grad_norm": 12.20964409536768,
      "learning_rate": 3.425404900293373e-06,
      "loss": 0.949,
      "step": 3743
    },
    {
      "epoch": 0.2701201255365968,
      "grad_norm": 4.481595018371406,
      "learning_rate": 3.4250770241380445e-06,
      "loss": 0.9788,
      "step": 3744
    },
    {
      "epoch": 0.27019227300602433,
      "grad_norm": 3.1944844635009306,
      "learning_rate": 3.4247490701645356e-06,
      "loss": 0.9949,
      "step": 3745
    },
    {
      "epoch": 0.27026442047545185,
      "grad_norm": 5.096569568082504,
      "learning_rate": 3.424421038390754e-06,
      "loss": 1.049,
      "step": 3746
    },
    {
      "epoch": 0.2703365679448793,
      "grad_norm": 4.24470324589976,
      "learning_rate": 3.4240929288346126e-06,
      "loss": 1.0663,
      "step": 3747
    },
    {
      "epoch": 0.27040871541430683,
      "grad_norm": 5.069367448521068,
      "learning_rate": 3.423764741514028e-06,
      "loss": 1.0538,
      "step": 3748
    },
    {
      "epoch": 0.27048086288373435,
      "grad_norm": 4.218741409858152,
      "learning_rate": 3.4234364764469216e-06,
      "loss": 1.0008,
      "step": 3749
    },
    {
      "epoch": 0.27055301035316187,
      "grad_norm": 3.706666050117313,
      "learning_rate": 3.423108133651219e-06,
      "loss": 0.9452,
      "step": 3750
    },
    {
      "epoch": 0.2706251578225894,
      "grad_norm": 5.608288332950856,
      "learning_rate": 3.4227797131448486e-06,
      "loss": 0.9478,
      "step": 3751
    },
    {
      "epoch": 0.2706973052920169,
      "grad_norm": 5.456587864223861,
      "learning_rate": 3.422451214945745e-06,
      "loss": 0.9615,
      "step": 3752
    },
    {
      "epoch": 0.2707694527614444,
      "grad_norm": 3.394579431611093,
      "learning_rate": 3.422122639071847e-06,
      "loss": 1.0404,
      "step": 3753
    },
    {
      "epoch": 0.2708416002308719,
      "grad_norm": 3.4556862838755946,
      "learning_rate": 3.4217939855410958e-06,
      "loss": 0.9354,
      "step": 3754
    },
    {
      "epoch": 0.2709137477002994,
      "grad_norm": 0.7937742379760641,
      "learning_rate": 3.421465254371438e-06,
      "loss": 0.8283,
      "step": 3755
    },
    {
      "epoch": 0.27098589516972693,
      "grad_norm": 9.26379602917277,
      "learning_rate": 3.421136445580825e-06,
      "loss": 0.9841,
      "step": 3756
    },
    {
      "epoch": 0.27105804263915445,
      "grad_norm": 3.6529686704263833,
      "learning_rate": 3.420807559187212e-06,
      "loss": 0.993,
      "step": 3757
    },
    {
      "epoch": 0.2711301901085819,
      "grad_norm": 3.096887866011761,
      "learning_rate": 3.4204785952085574e-06,
      "loss": 1.0479,
      "step": 3758
    },
    {
      "epoch": 0.27120233757800943,
      "grad_norm": 3.838148299635531,
      "learning_rate": 3.420149553662825e-06,
      "loss": 0.9444,
      "step": 3759
    },
    {
      "epoch": 0.27127448504743695,
      "grad_norm": 19.01711666679151,
      "learning_rate": 3.419820434567983e-06,
      "loss": 0.9403,
      "step": 3760
    },
    {
      "epoch": 0.27134663251686447,
      "grad_norm": 2.873592197885945,
      "learning_rate": 3.4194912379420035e-06,
      "loss": 0.8951,
      "step": 3761
    },
    {
      "epoch": 0.271418779986292,
      "grad_norm": 3.543105891390346,
      "learning_rate": 3.4191619638028623e-06,
      "loss": 1.0467,
      "step": 3762
    },
    {
      "epoch": 0.2714909274557195,
      "grad_norm": 9.29376794522162,
      "learning_rate": 3.4188326121685387e-06,
      "loss": 1.0374,
      "step": 3763
    },
    {
      "epoch": 0.271563074925147,
      "grad_norm": 3.4265422606490166,
      "learning_rate": 3.4185031830570195e-06,
      "loss": 1.0381,
      "step": 3764
    },
    {
      "epoch": 0.2716352223945745,
      "grad_norm": 3.013323602435624,
      "learning_rate": 3.418173676486292e-06,
      "loss": 0.9703,
      "step": 3765
    },
    {
      "epoch": 0.271707369864002,
      "grad_norm": 5.711538301481231,
      "learning_rate": 3.41784409247435e-06,
      "loss": 1.0615,
      "step": 3766
    },
    {
      "epoch": 0.27177951733342953,
      "grad_norm": 4.031704899161259,
      "learning_rate": 3.417514431039191e-06,
      "loss": 0.9411,
      "step": 3767
    },
    {
      "epoch": 0.27185166480285705,
      "grad_norm": 3.370899959576906,
      "learning_rate": 3.417184692198816e-06,
      "loss": 0.8402,
      "step": 3768
    },
    {
      "epoch": 0.27192381227228457,
      "grad_norm": 11.095774806965691,
      "learning_rate": 3.416854875971232e-06,
      "loss": 0.9678,
      "step": 3769
    },
    {
      "epoch": 0.27199595974171203,
      "grad_norm": 6.261699559877799,
      "learning_rate": 3.4165249823744474e-06,
      "loss": 0.9545,
      "step": 3770
    },
    {
      "epoch": 0.27206810721113955,
      "grad_norm": 4.429323211707352,
      "learning_rate": 3.4161950114264775e-06,
      "loss": 1.0069,
      "step": 3771
    },
    {
      "epoch": 0.2721402546805671,
      "grad_norm": 3.8243269783268015,
      "learning_rate": 3.415864963145341e-06,
      "loss": 0.9969,
      "step": 3772
    },
    {
      "epoch": 0.2722124021499946,
      "grad_norm": 0.7480161974321068,
      "learning_rate": 3.41553483754906e-06,
      "loss": 0.8333,
      "step": 3773
    },
    {
      "epoch": 0.2722845496194221,
      "grad_norm": 6.616993510567808,
      "learning_rate": 3.4152046346556615e-06,
      "loss": 0.9491,
      "step": 3774
    },
    {
      "epoch": 0.27235669708884963,
      "grad_norm": 3.615420608216478,
      "learning_rate": 3.414874354483177e-06,
      "loss": 0.8676,
      "step": 3775
    },
    {
      "epoch": 0.2724288445582771,
      "grad_norm": 0.7609059619046831,
      "learning_rate": 3.4145439970496416e-06,
      "loss": 0.8634,
      "step": 3776
    },
    {
      "epoch": 0.2725009920277046,
      "grad_norm": 3.09282678257594,
      "learning_rate": 3.414213562373095e-06,
      "loss": 1.1009,
      "step": 3777
    },
    {
      "epoch": 0.27257313949713213,
      "grad_norm": 3.2120856763783587,
      "learning_rate": 3.4138830504715807e-06,
      "loss": 1.1064,
      "step": 3778
    },
    {
      "epoch": 0.27264528696655965,
      "grad_norm": 2.892172734136824,
      "learning_rate": 3.4135524613631476e-06,
      "loss": 0.9828,
      "step": 3779
    },
    {
      "epoch": 0.27271743443598717,
      "grad_norm": 3.79564932674839,
      "learning_rate": 3.413221795065847e-06,
      "loss": 1.0555,
      "step": 3780
    },
    {
      "epoch": 0.2727895819054147,
      "grad_norm": 9.863406171857426,
      "learning_rate": 3.4128910515977357e-06,
      "loss": 0.8943,
      "step": 3781
    },
    {
      "epoch": 0.27286172937484215,
      "grad_norm": 0.7937283565544107,
      "learning_rate": 3.4125602309768747e-06,
      "loss": 0.8567,
      "step": 3782
    },
    {
      "epoch": 0.2729338768442697,
      "grad_norm": 5.104452112221917,
      "learning_rate": 3.4122293332213286e-06,
      "loss": 0.9609,
      "step": 3783
    },
    {
      "epoch": 0.2730060243136972,
      "grad_norm": 3.96511433998138,
      "learning_rate": 3.411898358349166e-06,
      "loss": 1.0049,
      "step": 3784
    },
    {
      "epoch": 0.2730781717831247,
      "grad_norm": 2.562601785266504,
      "learning_rate": 3.4115673063784614e-06,
      "loss": 0.986,
      "step": 3785
    },
    {
      "epoch": 0.27315031925255223,
      "grad_norm": 4.049010666938462,
      "learning_rate": 3.411236177327291e-06,
      "loss": 0.9103,
      "step": 3786
    },
    {
      "epoch": 0.27322246672197975,
      "grad_norm": 2.6193650026524296,
      "learning_rate": 3.4109049712137375e-06,
      "loss": 0.9978,
      "step": 3787
    },
    {
      "epoch": 0.2732946141914072,
      "grad_norm": 5.478363481064767,
      "learning_rate": 3.4105736880558867e-06,
      "loss": 0.9912,
      "step": 3788
    },
    {
      "epoch": 0.27336676166083473,
      "grad_norm": 6.55295646279756,
      "learning_rate": 3.4102423278718285e-06,
      "loss": 0.9364,
      "step": 3789
    },
    {
      "epoch": 0.27343890913026225,
      "grad_norm": 12.532091982896288,
      "learning_rate": 3.409910890679657e-06,
      "loss": 1.0853,
      "step": 3790
    },
    {
      "epoch": 0.2735110565996898,
      "grad_norm": 3.814300487177424,
      "learning_rate": 3.409579376497471e-06,
      "loss": 1.1273,
      "step": 3791
    },
    {
      "epoch": 0.2735832040691173,
      "grad_norm": 3.048535329870839,
      "learning_rate": 3.409247785343374e-06,
      "loss": 1.0711,
      "step": 3792
    },
    {
      "epoch": 0.2736553515385448,
      "grad_norm": 3.155876213073195,
      "learning_rate": 3.4089161172354718e-06,
      "loss": 0.9574,
      "step": 3793
    },
    {
      "epoch": 0.2737274990079723,
      "grad_norm": 3.25215942828153,
      "learning_rate": 3.4085843721918766e-06,
      "loss": 1.0432,
      "step": 3794
    },
    {
      "epoch": 0.2737996464773998,
      "grad_norm": 3.6770461011231768,
      "learning_rate": 3.4082525502307026e-06,
      "loss": 1.1002,
      "step": 3795
    },
    {
      "epoch": 0.2738717939468273,
      "grad_norm": 2.971129578046615,
      "learning_rate": 3.40792065137007e-06,
      "loss": 0.954,
      "step": 3796
    },
    {
      "epoch": 0.27394394141625483,
      "grad_norm": 4.657780747687709,
      "learning_rate": 3.4075886756281028e-06,
      "loss": 1.0155,
      "step": 3797
    },
    {
      "epoch": 0.27401608888568235,
      "grad_norm": 2.4424386489019336,
      "learning_rate": 3.4072566230229286e-06,
      "loss": 1.0518,
      "step": 3798
    },
    {
      "epoch": 0.27408823635510987,
      "grad_norm": 2.8129461146569485,
      "learning_rate": 3.40692449357268e-06,
      "loss": 0.9529,
      "step": 3799
    },
    {
      "epoch": 0.27416038382453733,
      "grad_norm": 4.7881220879027415,
      "learning_rate": 3.4065922872954927e-06,
      "loss": 0.9962,
      "step": 3800
    },
    {
      "epoch": 0.27423253129396485,
      "grad_norm": 2.9059405418451454,
      "learning_rate": 3.4062600042095077e-06,
      "loss": 1.002,
      "step": 3801
    },
    {
      "epoch": 0.2743046787633924,
      "grad_norm": 3.464266517147758,
      "learning_rate": 3.4059276443328697e-06,
      "loss": 0.9776,
      "step": 3802
    },
    {
      "epoch": 0.2743768262328199,
      "grad_norm": 3.3739044389290136,
      "learning_rate": 3.4055952076837277e-06,
      "loss": 0.9214,
      "step": 3803
    },
    {
      "epoch": 0.2744489737022474,
      "grad_norm": 2.574847918713724,
      "learning_rate": 3.4052626942802343e-06,
      "loss": 1.032,
      "step": 3804
    },
    {
      "epoch": 0.27452112117167493,
      "grad_norm": 5.419181748451874,
      "learning_rate": 3.4049301041405487e-06,
      "loss": 1.015,
      "step": 3805
    },
    {
      "epoch": 0.2745932686411024,
      "grad_norm": 4.971833238120589,
      "learning_rate": 3.4045974372828297e-06,
      "loss": 1.0024,
      "step": 3806
    },
    {
      "epoch": 0.2746654161105299,
      "grad_norm": 5.967392372595891,
      "learning_rate": 3.4042646937252447e-06,
      "loss": 0.9698,
      "step": 3807
    },
    {
      "epoch": 0.27473756357995743,
      "grad_norm": 2.7974465068649215,
      "learning_rate": 3.4039318734859627e-06,
      "loss": 0.9398,
      "step": 3808
    },
    {
      "epoch": 0.27480971104938495,
      "grad_norm": 2.9101524813838853,
      "learning_rate": 3.403598976583159e-06,
      "loss": 1.0138,
      "step": 3809
    },
    {
      "epoch": 0.27488185851881247,
      "grad_norm": 53.751117357572205,
      "learning_rate": 3.40326600303501e-06,
      "loss": 0.9198,
      "step": 3810
    },
    {
      "epoch": 0.27495400598823994,
      "grad_norm": 5.135913858338974,
      "learning_rate": 3.4029329528597005e-06,
      "loss": 0.9432,
      "step": 3811
    },
    {
      "epoch": 0.27502615345766745,
      "grad_norm": 5.7814971046219465,
      "learning_rate": 3.4025998260754154e-06,
      "loss": 0.9359,
      "step": 3812
    },
    {
      "epoch": 0.275098300927095,
      "grad_norm": 3.6320462639084576,
      "learning_rate": 3.402266622700346e-06,
      "loss": 0.9333,
      "step": 3813
    },
    {
      "epoch": 0.2751704483965225,
      "grad_norm": 12.252247506875078,
      "learning_rate": 3.4019333427526875e-06,
      "loss": 0.9918,
      "step": 3814
    },
    {
      "epoch": 0.27524259586595,
      "grad_norm": 4.296039513944337,
      "learning_rate": 3.4015999862506387e-06,
      "loss": 0.9869,
      "step": 3815
    },
    {
      "epoch": 0.27531474333537753,
      "grad_norm": 5.674126296489674,
      "learning_rate": 3.4012665532124037e-06,
      "loss": 1.0106,
      "step": 3816
    },
    {
      "epoch": 0.275386890804805,
      "grad_norm": 5.93539519899163,
      "learning_rate": 3.400933043656189e-06,
      "loss": 0.954,
      "step": 3817
    },
    {
      "epoch": 0.2754590382742325,
      "grad_norm": 8.64951720902495,
      "learning_rate": 3.400599457600207e-06,
      "loss": 0.8708,
      "step": 3818
    },
    {
      "epoch": 0.27553118574366003,
      "grad_norm": 4.0795783599198785,
      "learning_rate": 3.4002657950626736e-06,
      "loss": 0.8903,
      "step": 3819
    },
    {
      "epoch": 0.27560333321308755,
      "grad_norm": 3.2735948183408357,
      "learning_rate": 3.399932056061809e-06,
      "loss": 1.0116,
      "step": 3820
    },
    {
      "epoch": 0.2756754806825151,
      "grad_norm": 0.628553089947988,
      "learning_rate": 3.3995982406158366e-06,
      "loss": 0.8104,
      "step": 3821
    },
    {
      "epoch": 0.2757476281519426,
      "grad_norm": 2.742491007036402,
      "learning_rate": 3.399264348742986e-06,
      "loss": 1.0868,
      "step": 3822
    },
    {
      "epoch": 0.27581977562137006,
      "grad_norm": 3.9486248954621503,
      "learning_rate": 3.398930380461489e-06,
      "loss": 1.0384,
      "step": 3823
    },
    {
      "epoch": 0.2758919230907976,
      "grad_norm": 4.111592785795556,
      "learning_rate": 3.398596335789582e-06,
      "loss": 0.936,
      "step": 3824
    },
    {
      "epoch": 0.2759640705602251,
      "grad_norm": 3.353593544340836,
      "learning_rate": 3.3982622147455072e-06,
      "loss": 1.0377,
      "step": 3825
    },
    {
      "epoch": 0.2760362180296526,
      "grad_norm": 7.9773658997954175,
      "learning_rate": 3.397928017347509e-06,
      "loss": 0.9965,
      "step": 3826
    },
    {
      "epoch": 0.27610836549908013,
      "grad_norm": 5.201946686444727,
      "learning_rate": 3.3975937436138365e-06,
      "loss": 1.0992,
      "step": 3827
    },
    {
      "epoch": 0.27618051296850765,
      "grad_norm": 3.5508741779987805,
      "learning_rate": 3.3972593935627433e-06,
      "loss": 1.0475,
      "step": 3828
    },
    {
      "epoch": 0.2762526604379351,
      "grad_norm": 5.1346612443032695,
      "learning_rate": 3.396924967212487e-06,
      "loss": 0.9609,
      "step": 3829
    },
    {
      "epoch": 0.27632480790736264,
      "grad_norm": 2.686140293101768,
      "learning_rate": 3.39659046458133e-06,
      "loss": 0.9724,
      "step": 3830
    },
    {
      "epoch": 0.27639695537679015,
      "grad_norm": 2.7291377019922516,
      "learning_rate": 3.3962558856875374e-06,
      "loss": 0.946,
      "step": 3831
    },
    {
      "epoch": 0.2764691028462177,
      "grad_norm": 3.359095788042465,
      "learning_rate": 3.3959212305493797e-06,
      "loss": 0.9072,
      "step": 3832
    },
    {
      "epoch": 0.2765412503156452,
      "grad_norm": 2.985977461630812,
      "learning_rate": 3.395586499185131e-06,
      "loss": 0.987,
      "step": 3833
    },
    {
      "epoch": 0.2766133977850727,
      "grad_norm": 4.507217976257421,
      "learning_rate": 3.3952516916130703e-06,
      "loss": 1.0497,
      "step": 3834
    },
    {
      "epoch": 0.2766855452545002,
      "grad_norm": 2.8757572006219005,
      "learning_rate": 3.394916807851479e-06,
      "loss": 0.8599,
      "step": 3835
    },
    {
      "epoch": 0.2767576927239277,
      "grad_norm": 6.873525010453948,
      "learning_rate": 3.3945818479186453e-06,
      "loss": 1.0677,
      "step": 3836
    },
    {
      "epoch": 0.2768298401933552,
      "grad_norm": 0.9227840338326547,
      "learning_rate": 3.3942468118328592e-06,
      "loss": 0.9058,
      "step": 3837
    },
    {
      "epoch": 0.27690198766278273,
      "grad_norm": 4.285940677476151,
      "learning_rate": 3.3939116996124163e-06,
      "loss": 0.9804,
      "step": 3838
    },
    {
      "epoch": 0.27697413513221025,
      "grad_norm": 5.422934129149255,
      "learning_rate": 3.3935765112756152e-06,
      "loss": 1.0544,
      "step": 3839
    },
    {
      "epoch": 0.27704628260163777,
      "grad_norm": 4.622254458848911,
      "learning_rate": 3.3932412468407597e-06,
      "loss": 1.0619,
      "step": 3840
    },
    {
      "epoch": 0.27711843007106524,
      "grad_norm": 3.3376877635473323,
      "learning_rate": 3.392905906326158e-06,
      "loss": 0.9901,
      "step": 3841
    },
    {
      "epoch": 0.27719057754049276,
      "grad_norm": 3.1900389787929333,
      "learning_rate": 3.3925704897501204e-06,
      "loss": 1.0212,
      "step": 3842
    },
    {
      "epoch": 0.2772627250099203,
      "grad_norm": 4.143225662866146,
      "learning_rate": 3.3922349971309637e-06,
      "loss": 1.0552,
      "step": 3843
    },
    {
      "epoch": 0.2773348724793478,
      "grad_norm": 3.2670069056713213,
      "learning_rate": 3.3918994284870072e-06,
      "loss": 0.982,
      "step": 3844
    },
    {
      "epoch": 0.2774070199487753,
      "grad_norm": 4.851123341511793,
      "learning_rate": 3.391563783836576e-06,
      "loss": 1.0188,
      "step": 3845
    },
    {
      "epoch": 0.27747916741820283,
      "grad_norm": 6.45785947312338,
      "learning_rate": 3.391228063197998e-06,
      "loss": 0.7824,
      "step": 3846
    },
    {
      "epoch": 0.2775513148876303,
      "grad_norm": 6.150688641457975,
      "learning_rate": 3.3908922665896054e-06,
      "loss": 0.811,
      "step": 3847
    },
    {
      "epoch": 0.2776234623570578,
      "grad_norm": 4.322745274589329,
      "learning_rate": 3.390556394029735e-06,
      "loss": 0.8165,
      "step": 3848
    },
    {
      "epoch": 0.27769560982648533,
      "grad_norm": 3.90626928706176,
      "learning_rate": 3.390220445536728e-06,
      "loss": 0.9841,
      "step": 3849
    },
    {
      "epoch": 0.27776775729591285,
      "grad_norm": 3.54646641734804,
      "learning_rate": 3.3898844211289282e-06,
      "loss": 0.9687,
      "step": 3850
    },
    {
      "epoch": 0.2778399047653404,
      "grad_norm": 2.7683033877901795,
      "learning_rate": 3.3895483208246864e-06,
      "loss": 0.8949,
      "step": 3851
    },
    {
      "epoch": 0.2779120522347679,
      "grad_norm": 6.602543901256362,
      "learning_rate": 3.3892121446423535e-06,
      "loss": 0.9003,
      "step": 3852
    },
    {
      "epoch": 0.27798419970419536,
      "grad_norm": 2.859827724950236,
      "learning_rate": 3.3888758926002894e-06,
      "loss": 1.0035,
      "step": 3853
    },
    {
      "epoch": 0.2780563471736229,
      "grad_norm": 2.177331888955629,
      "learning_rate": 3.388539564716853e-06,
      "loss": 0.9973,
      "step": 3854
    },
    {
      "epoch": 0.2781284946430504,
      "grad_norm": 5.550639162117479,
      "learning_rate": 3.3882031610104117e-06,
      "loss": 1.0168,
      "step": 3855
    },
    {
      "epoch": 0.2782006421124779,
      "grad_norm": 4.152699036349615,
      "learning_rate": 3.3878666814993348e-06,
      "loss": 1.0795,
      "step": 3856
    },
    {
      "epoch": 0.27827278958190543,
      "grad_norm": 4.312397388951345,
      "learning_rate": 3.3875301262019957e-06,
      "loss": 1.0513,
      "step": 3857
    },
    {
      "epoch": 0.2783449370513329,
      "grad_norm": 4.613164449661861,
      "learning_rate": 3.387193495136773e-06,
      "loss": 0.9994,
      "step": 3858
    },
    {
      "epoch": 0.2784170845207604,
      "grad_norm": 3.329562883985387,
      "learning_rate": 3.386856788322049e-06,
      "loss": 0.953,
      "step": 3859
    },
    {
      "epoch": 0.27848923199018794,
      "grad_norm": 3.258625468811027,
      "learning_rate": 3.38652000577621e-06,
      "loss": 1.0872,
      "step": 3860
    },
    {
      "epoch": 0.27856137945961545,
      "grad_norm": 2.658151641504103,
      "learning_rate": 3.386183147517646e-06,
      "loss": 0.9581,
      "step": 3861
    },
    {
      "epoch": 0.278633526929043,
      "grad_norm": 12.881876766350372,
      "learning_rate": 3.3858462135647513e-06,
      "loss": 0.9218,
      "step": 3862
    },
    {
      "epoch": 0.2787056743984705,
      "grad_norm": 6.858710671755838,
      "learning_rate": 3.3855092039359257e-06,
      "loss": 1.0287,
      "step": 3863
    },
    {
      "epoch": 0.27877782186789796,
      "grad_norm": 2.6998853729782084,
      "learning_rate": 3.3851721186495714e-06,
      "loss": 0.9762,
      "step": 3864
    },
    {
      "epoch": 0.2788499693373255,
      "grad_norm": 8.630688546694236,
      "learning_rate": 3.3848349577240953e-06,
      "loss": 0.9741,
      "step": 3865
    },
    {
      "epoch": 0.278922116806753,
      "grad_norm": 2.4305255585287764,
      "learning_rate": 3.3844977211779087e-06,
      "loss": 1.0368,
      "step": 3866
    },
    {
      "epoch": 0.2789942642761805,
      "grad_norm": 4.718047639225464,
      "learning_rate": 3.3841604090294268e-06,
      "loss": 1.0388,
      "step": 3867
    },
    {
      "epoch": 0.27906641174560803,
      "grad_norm": 6.5968828758724,
      "learning_rate": 3.3838230212970685e-06,
      "loss": 1.0711,
      "step": 3868
    },
    {
      "epoch": 0.27913855921503555,
      "grad_norm": 2.751443657374621,
      "learning_rate": 3.3834855579992582e-06,
      "loss": 1.0633,
      "step": 3869
    },
    {
      "epoch": 0.279210706684463,
      "grad_norm": 4.0586186574368055,
      "learning_rate": 3.383148019154423e-06,
      "loss": 1.0436,
      "step": 3870
    },
    {
      "epoch": 0.27928285415389054,
      "grad_norm": 3.812433711241719,
      "learning_rate": 3.3828104047809945e-06,
      "loss": 0.9465,
      "step": 3871
    },
    {
      "epoch": 0.27935500162331806,
      "grad_norm": 2.589204044651342,
      "learning_rate": 3.382472714897409e-06,
      "loss": 1.0491,
      "step": 3872
    },
    {
      "epoch": 0.2794271490927456,
      "grad_norm": 3.8268446669123235,
      "learning_rate": 3.3821349495221067e-06,
      "loss": 0.9983,
      "step": 3873
    },
    {
      "epoch": 0.2794992965621731,
      "grad_norm": 12.458184452932187,
      "learning_rate": 3.38179710867353e-06,
      "loss": 0.9538,
      "step": 3874
    },
    {
      "epoch": 0.2795714440316006,
      "grad_norm": 7.955064218622758,
      "learning_rate": 3.38145919237013e-06,
      "loss": 0.948,
      "step": 3875
    },
    {
      "epoch": 0.2796435915010281,
      "grad_norm": 3.2259736713013774,
      "learning_rate": 3.3811212006303564e-06,
      "loss": 1.02,
      "step": 3876
    },
    {
      "epoch": 0.2797157389704556,
      "grad_norm": 3.7689049076868515,
      "learning_rate": 3.3807831334726677e-06,
      "loss": 1.0377,
      "step": 3877
    },
    {
      "epoch": 0.2797878864398831,
      "grad_norm": 2.9897545068807196,
      "learning_rate": 3.3804449909155226e-06,
      "loss": 1.0346,
      "step": 3878
    },
    {
      "epoch": 0.27986003390931063,
      "grad_norm": 3.299532140453587,
      "learning_rate": 3.380106772977387e-06,
      "loss": 1.0145,
      "step": 3879
    },
    {
      "epoch": 0.27993218137873815,
      "grad_norm": 3.384322677306474,
      "learning_rate": 3.3797684796767297e-06,
      "loss": 1.0323,
      "step": 3880
    },
    {
      "epoch": 0.2800043288481657,
      "grad_norm": 3.0181493125974277,
      "learning_rate": 3.3794301110320236e-06,
      "loss": 1.0474,
      "step": 3881
    },
    {
      "epoch": 0.28007647631759314,
      "grad_norm": 10.59428687971711,
      "learning_rate": 3.3790916670617457e-06,
      "loss": 1.0358,
      "step": 3882
    },
    {
      "epoch": 0.28014862378702066,
      "grad_norm": 9.2680759629852,
      "learning_rate": 3.3787531477843768e-06,
      "loss": 1.0664,
      "step": 3883
    },
    {
      "epoch": 0.2802207712564482,
      "grad_norm": 3.5211649330380332,
      "learning_rate": 3.3784145532184022e-06,
      "loss": 0.9427,
      "step": 3884
    },
    {
      "epoch": 0.2802929187258757,
      "grad_norm": 4.203630948724706,
      "learning_rate": 3.378075883382312e-06,
      "loss": 1.0885,
      "step": 3885
    },
    {
      "epoch": 0.2803650661953032,
      "grad_norm": 2.5781934324487894,
      "learning_rate": 3.377737138294599e-06,
      "loss": 1.0082,
      "step": 3886
    },
    {
      "epoch": 0.28043721366473073,
      "grad_norm": 3.8603544034978796,
      "learning_rate": 3.377398317973761e-06,
      "loss": 0.9903,
      "step": 3887
    },
    {
      "epoch": 0.2805093611341582,
      "grad_norm": 4.526733883979936,
      "learning_rate": 3.3770594224383004e-06,
      "loss": 1.0496,
      "step": 3888
    },
    {
      "epoch": 0.2805815086035857,
      "grad_norm": 5.90966472107162,
      "learning_rate": 3.3767204517067223e-06,
      "loss": 0.9064,
      "step": 3889
    },
    {
      "epoch": 0.28065365607301324,
      "grad_norm": 0.6906758561547859,
      "learning_rate": 3.3763814057975363e-06,
      "loss": 0.8451,
      "step": 3890
    },
    {
      "epoch": 0.28072580354244075,
      "grad_norm": 3.1786758350596385,
      "learning_rate": 3.376042284729257e-06,
      "loss": 0.9412,
      "step": 3891
    },
    {
      "epoch": 0.2807979510118683,
      "grad_norm": 3.51227229999178,
      "learning_rate": 3.375703088520403e-06,
      "loss": 0.9534,
      "step": 3892
    },
    {
      "epoch": 0.2808700984812958,
      "grad_norm": 15.426755061532404,
      "learning_rate": 3.3753638171894955e-06,
      "loss": 1.0146,
      "step": 3893
    },
    {
      "epoch": 0.28094224595072326,
      "grad_norm": 3.4455380874098593,
      "learning_rate": 3.375024470755061e-06,
      "loss": 0.9682,
      "step": 3894
    },
    {
      "epoch": 0.2810143934201508,
      "grad_norm": 4.182605673030205,
      "learning_rate": 3.374685049235631e-06,
      "loss": 0.8591,
      "step": 3895
    },
    {
      "epoch": 0.2810865408895783,
      "grad_norm": 3.72068954024751,
      "learning_rate": 3.374345552649739e-06,
      "loss": 0.9372,
      "step": 3896
    },
    {
      "epoch": 0.2811586883590058,
      "grad_norm": 6.3140377777680445,
      "learning_rate": 3.374005981015925e-06,
      "loss": 1.0624,
      "step": 3897
    },
    {
      "epoch": 0.28123083582843333,
      "grad_norm": 3.041154391553339,
      "learning_rate": 3.3736663343527296e-06,
      "loss": 0.9238,
      "step": 3898
    },
    {
      "epoch": 0.28130298329786085,
      "grad_norm": 2.934749045617767,
      "learning_rate": 3.3733266126787016e-06,
      "loss": 1.0487,
      "step": 3899
    },
    {
      "epoch": 0.2813751307672883,
      "grad_norm": 3.7243292492653755,
      "learning_rate": 3.3729868160123906e-06,
      "loss": 1.0034,
      "step": 3900
    },
    {
      "epoch": 0.28144727823671584,
      "grad_norm": 4.038984818145032,
      "learning_rate": 3.372646944372352e-06,
      "loss": 1.0638,
      "step": 3901
    },
    {
      "epoch": 0.28151942570614336,
      "grad_norm": 3.2304392286287142,
      "learning_rate": 3.3723069977771457e-06,
      "loss": 0.977,
      "step": 3902
    },
    {
      "epoch": 0.2815915731755709,
      "grad_norm": 5.449261627575483,
      "learning_rate": 3.3719669762453343e-06,
      "loss": 1.039,
      "step": 3903
    },
    {
      "epoch": 0.2816637206449984,
      "grad_norm": 3.2815249373281685,
      "learning_rate": 3.3716268797954853e-06,
      "loss": 0.9693,
      "step": 3904
    },
    {
      "epoch": 0.2817358681144259,
      "grad_norm": 11.9267862385669,
      "learning_rate": 3.3712867084461694e-06,
      "loss": 1.0463,
      "step": 3905
    },
    {
      "epoch": 0.2818080155838534,
      "grad_norm": 7.846455506384515,
      "learning_rate": 3.3709464622159633e-06,
      "loss": 0.9393,
      "step": 3906
    },
    {
      "epoch": 0.2818801630532809,
      "grad_norm": 2.716283358892647,
      "learning_rate": 3.370606141123446e-06,
      "loss": 1.0056,
      "step": 3907
    },
    {
      "epoch": 0.2819523105227084,
      "grad_norm": 4.7541570040944245,
      "learning_rate": 3.3702657451872e-06,
      "loss": 0.979,
      "step": 3908
    },
    {
      "epoch": 0.28202445799213594,
      "grad_norm": 5.08980796256845,
      "learning_rate": 3.369925274425815e-06,
      "loss": 0.9777,
      "step": 3909
    },
    {
      "epoch": 0.28209660546156345,
      "grad_norm": 4.089809005866628,
      "learning_rate": 3.369584728857882e-06,
      "loss": 0.9631,
      "step": 3910
    },
    {
      "epoch": 0.2821687529309909,
      "grad_norm": 3.1970796431953463,
      "learning_rate": 3.3692441085019977e-06,
      "loss": 1.0577,
      "step": 3911
    },
    {
      "epoch": 0.28224090040041844,
      "grad_norm": 5.576653465411941,
      "learning_rate": 3.3689034133767607e-06,
      "loss": 0.9788,
      "step": 3912
    },
    {
      "epoch": 0.28231304786984596,
      "grad_norm": 4.910039907941992,
      "learning_rate": 3.368562643500776e-06,
      "loss": 0.9836,
      "step": 3913
    },
    {
      "epoch": 0.2823851953392735,
      "grad_norm": 13.141575981864175,
      "learning_rate": 3.3682217988926506e-06,
      "loss": 0.9619,
      "step": 3914
    },
    {
      "epoch": 0.282457342808701,
      "grad_norm": 5.5925006856564075,
      "learning_rate": 3.3678808795709995e-06,
      "loss": 1.0082,
      "step": 3915
    },
    {
      "epoch": 0.2825294902781285,
      "grad_norm": 5.283576949286186,
      "learning_rate": 3.3675398855544357e-06,
      "loss": 0.9478,
      "step": 3916
    },
    {
      "epoch": 0.282601637747556,
      "grad_norm": 7.230250853517128,
      "learning_rate": 3.3671988168615825e-06,
      "loss": 1.0219,
      "step": 3917
    },
    {
      "epoch": 0.2826737852169835,
      "grad_norm": 9.544131120367817,
      "learning_rate": 3.3668576735110624e-06,
      "loss": 1.0542,
      "step": 3918
    },
    {
      "epoch": 0.282745932686411,
      "grad_norm": 4.331108084331212,
      "learning_rate": 3.3665164555215046e-06,
      "loss": 0.8718,
      "step": 3919
    },
    {
      "epoch": 0.28281808015583854,
      "grad_norm": 1.0418326944916672,
      "learning_rate": 3.366175162911542e-06,
      "loss": 0.8884,
      "step": 3920
    },
    {
      "epoch": 0.28289022762526606,
      "grad_norm": 2.810621269890344,
      "learning_rate": 3.365833795699811e-06,
      "loss": 1.0587,
      "step": 3921
    },
    {
      "epoch": 0.2829623750946936,
      "grad_norm": 11.522756865492303,
      "learning_rate": 3.365492353904953e-06,
      "loss": 0.884,
      "step": 3922
    },
    {
      "epoch": 0.28303452256412104,
      "grad_norm": 6.616272557946442,
      "learning_rate": 3.3651508375456123e-06,
      "loss": 0.9685,
      "step": 3923
    },
    {
      "epoch": 0.28310667003354856,
      "grad_norm": 8.947016844322897,
      "learning_rate": 3.364809246640438e-06,
      "loss": 0.948,
      "step": 3924
    },
    {
      "epoch": 0.2831788175029761,
      "grad_norm": 3.2357698095941796,
      "learning_rate": 3.3644675812080827e-06,
      "loss": 0.9333,
      "step": 3925
    },
    {
      "epoch": 0.2832509649724036,
      "grad_norm": 3.7814302007439125,
      "learning_rate": 3.3641258412672043e-06,
      "loss": 1.0375,
      "step": 3926
    },
    {
      "epoch": 0.2833231124418311,
      "grad_norm": 2.817911156049145,
      "learning_rate": 3.3637840268364634e-06,
      "loss": 0.8903,
      "step": 3927
    },
    {
      "epoch": 0.28339525991125863,
      "grad_norm": 10.010563802018213,
      "learning_rate": 3.3634421379345252e-06,
      "loss": 1.0259,
      "step": 3928
    },
    {
      "epoch": 0.2834674073806861,
      "grad_norm": 3.9568061425666934,
      "learning_rate": 3.3631001745800588e-06,
      "loss": 0.9017,
      "step": 3929
    },
    {
      "epoch": 0.2835395548501136,
      "grad_norm": 5.922149732351638,
      "learning_rate": 3.3627581367917386e-06,
      "loss": 0.9705,
      "step": 3930
    },
    {
      "epoch": 0.28361170231954114,
      "grad_norm": 5.757169814454055,
      "learning_rate": 3.362416024588241e-06,
      "loss": 0.9582,
      "step": 3931
    },
    {
      "epoch": 0.28368384978896866,
      "grad_norm": 5.00560713127644,
      "learning_rate": 3.362073837988248e-06,
      "loss": 0.9686,
      "step": 3932
    },
    {
      "epoch": 0.2837559972583962,
      "grad_norm": 6.330429448751106,
      "learning_rate": 3.3617315770104444e-06,
      "loss": 1.0225,
      "step": 3933
    },
    {
      "epoch": 0.2838281447278237,
      "grad_norm": 2.871380393322901,
      "learning_rate": 3.361389241673521e-06,
      "loss": 1.02,
      "step": 3934
    },
    {
      "epoch": 0.28390029219725116,
      "grad_norm": 3.4764094286877576,
      "learning_rate": 3.3610468319961703e-06,
      "loss": 1.0055,
      "step": 3935
    },
    {
      "epoch": 0.2839724396666787,
      "grad_norm": 3.593251732823891,
      "learning_rate": 3.360704347997091e-06,
      "loss": 1.0991,
      "step": 3936
    },
    {
      "epoch": 0.2840445871361062,
      "grad_norm": 3.579202414805212,
      "learning_rate": 3.360361789694984e-06,
      "loss": 1.0445,
      "step": 3937
    },
    {
      "epoch": 0.2841167346055337,
      "grad_norm": 2.6837244349036244,
      "learning_rate": 3.3600191571085557e-06,
      "loss": 1.0264,
      "step": 3938
    },
    {
      "epoch": 0.28418888207496124,
      "grad_norm": 5.503282607776887,
      "learning_rate": 3.359676450256516e-06,
      "loss": 1.0139,
      "step": 3939
    },
    {
      "epoch": 0.28426102954438875,
      "grad_norm": 3.6945452871558113,
      "learning_rate": 3.3593336691575784e-06,
      "loss": 1.0827,
      "step": 3940
    },
    {
      "epoch": 0.2843331770138162,
      "grad_norm": 6.3498586924291756,
      "learning_rate": 3.3589908138304616e-06,
      "loss": 1.0963,
      "step": 3941
    },
    {
      "epoch": 0.28440532448324374,
      "grad_norm": 5.330991807166026,
      "learning_rate": 3.358647884293887e-06,
      "loss": 1.0155,
      "step": 3942
    },
    {
      "epoch": 0.28447747195267126,
      "grad_norm": 3.0030708490703533,
      "learning_rate": 3.3583048805665808e-06,
      "loss": 0.8762,
      "step": 3943
    },
    {
      "epoch": 0.2845496194220988,
      "grad_norm": 4.077208434251676,
      "learning_rate": 3.3579618026672735e-06,
      "loss": 1.061,
      "step": 3944
    },
    {
      "epoch": 0.2846217668915263,
      "grad_norm": 4.276848305353131,
      "learning_rate": 3.3576186506146994e-06,
      "loss": 0.9462,
      "step": 3945
    },
    {
      "epoch": 0.2846939143609538,
      "grad_norm": 5.999812759020947,
      "learning_rate": 3.357275424427596e-06,
      "loss": 0.964,
      "step": 3946
    },
    {
      "epoch": 0.2847660618303813,
      "grad_norm": 5.870360795066842,
      "learning_rate": 3.356932124124706e-06,
      "loss": 0.9333,
      "step": 3947
    },
    {
      "epoch": 0.2848382092998088,
      "grad_norm": 3.894387505575055,
      "learning_rate": 3.3565887497247772e-06,
      "loss": 1.0382,
      "step": 3948
    },
    {
      "epoch": 0.2849103567692363,
      "grad_norm": 4.696012981039429,
      "learning_rate": 3.356245301246558e-06,
      "loss": 0.9576,
      "step": 3949
    },
    {
      "epoch": 0.28498250423866384,
      "grad_norm": 3.3862504294315263,
      "learning_rate": 3.3559017787088033e-06,
      "loss": 1.0194,
      "step": 3950
    },
    {
      "epoch": 0.28505465170809136,
      "grad_norm": 2.653336037374042,
      "learning_rate": 3.355558182130272e-06,
      "loss": 0.9879,
      "step": 3951
    },
    {
      "epoch": 0.2851267991775189,
      "grad_norm": 3.750449598699176,
      "learning_rate": 3.3552145115297262e-06,
      "loss": 0.931,
      "step": 3952
    },
    {
      "epoch": 0.28519894664694634,
      "grad_norm": 2.830066836211526,
      "learning_rate": 3.3548707669259327e-06,
      "loss": 1.017,
      "step": 3953
    },
    {
      "epoch": 0.28527109411637386,
      "grad_norm": 4.763923116004308,
      "learning_rate": 3.3545269483376622e-06,
      "loss": 0.9112,
      "step": 3954
    },
    {
      "epoch": 0.2853432415858014,
      "grad_norm": 4.1917748338053915,
      "learning_rate": 3.35418305578369e-06,
      "loss": 0.9933,
      "step": 3955
    },
    {
      "epoch": 0.2854153890552289,
      "grad_norm": 5.046847789206831,
      "learning_rate": 3.353839089282794e-06,
      "loss": 1.0261,
      "step": 3956
    },
    {
      "epoch": 0.2854875365246564,
      "grad_norm": 3.6003785411180393,
      "learning_rate": 3.3534950488537567e-06,
      "loss": 0.8598,
      "step": 3957
    },
    {
      "epoch": 0.28555968399408393,
      "grad_norm": 3.109156922020352,
      "learning_rate": 3.353150934515365e-06,
      "loss": 1.1002,
      "step": 3958
    },
    {
      "epoch": 0.2856318314635114,
      "grad_norm": 4.55803576979551,
      "learning_rate": 3.3528067462864107e-06,
      "loss": 0.9937,
      "step": 3959
    },
    {
      "epoch": 0.2857039789329389,
      "grad_norm": 4.473012958278431,
      "learning_rate": 3.3524624841856874e-06,
      "loss": 0.9865,
      "step": 3960
    },
    {
      "epoch": 0.28577612640236644,
      "grad_norm": 4.53567835704532,
      "learning_rate": 3.3521181482319945e-06,
      "loss": 1.0076,
      "step": 3961
    },
    {
      "epoch": 0.28584827387179396,
      "grad_norm": 4.263946538335947,
      "learning_rate": 3.351773738444135e-06,
      "loss": 0.9019,
      "step": 3962
    },
    {
      "epoch": 0.2859204213412215,
      "grad_norm": 4.090930462791026,
      "learning_rate": 3.351429254840916e-06,
      "loss": 0.8858,
      "step": 3963
    },
    {
      "epoch": 0.28599256881064894,
      "grad_norm": 4.557945800338486,
      "learning_rate": 3.3510846974411477e-06,
      "loss": 1.0261,
      "step": 3964
    },
    {
      "epoch": 0.28606471628007646,
      "grad_norm": 4.232917788277884,
      "learning_rate": 3.350740066263646e-06,
      "loss": 0.9104,
      "step": 3965
    },
    {
      "epoch": 0.286136863749504,
      "grad_norm": 2.418397437134854,
      "learning_rate": 3.3503953613272293e-06,
      "loss": 0.978,
      "step": 3966
    },
    {
      "epoch": 0.2862090112189315,
      "grad_norm": 4.996529137413341,
      "learning_rate": 3.3500505826507208e-06,
      "loss": 1.0451,
      "step": 3967
    },
    {
      "epoch": 0.286281158688359,
      "grad_norm": 4.231180147076193,
      "learning_rate": 3.349705730252948e-06,
      "loss": 1.0654,
      "step": 3968
    },
    {
      "epoch": 0.28635330615778654,
      "grad_norm": 5.083860380439047,
      "learning_rate": 3.3493608041527414e-06,
      "loss": 1.0243,
      "step": 3969
    },
    {
      "epoch": 0.286425453627214,
      "grad_norm": 3.272367534710391,
      "learning_rate": 3.3490158043689373e-06,
      "loss": 0.9965,
      "step": 3970
    },
    {
      "epoch": 0.2864976010966415,
      "grad_norm": 3.4322330699466708,
      "learning_rate": 3.348670730920373e-06,
      "loss": 1.131,
      "step": 3971
    },
    {
      "epoch": 0.28656974856606904,
      "grad_norm": 3.2604077291315368,
      "learning_rate": 3.348325583825893e-06,
      "loss": 0.9872,
      "step": 3972
    },
    {
      "epoch": 0.28664189603549656,
      "grad_norm": 33.150023714405165,
      "learning_rate": 3.3479803631043443e-06,
      "loss": 0.9355,
      "step": 3973
    },
    {
      "epoch": 0.2867140435049241,
      "grad_norm": 2.8641577572129515,
      "learning_rate": 3.3476350687745783e-06,
      "loss": 1.0011,
      "step": 3974
    },
    {
      "epoch": 0.2867861909743516,
      "grad_norm": 6.3511165448263345,
      "learning_rate": 3.3472897008554495e-06,
      "loss": 0.9644,
      "step": 3975
    },
    {
      "epoch": 0.28685833844377906,
      "grad_norm": 4.062767137400831,
      "learning_rate": 3.346944259365818e-06,
      "loss": 0.9313,
      "step": 3976
    },
    {
      "epoch": 0.2869304859132066,
      "grad_norm": 4.207910472410653,
      "learning_rate": 3.3465987443245466e-06,
      "loss": 1.0875,
      "step": 3977
    },
    {
      "epoch": 0.2870026333826341,
      "grad_norm": 4.990569185196848,
      "learning_rate": 3.3462531557505026e-06,
      "loss": 0.8739,
      "step": 3978
    },
    {
      "epoch": 0.2870747808520616,
      "grad_norm": 2.9112035722908227,
      "learning_rate": 3.345907493662557e-06,
      "loss": 0.964,
      "step": 3979
    },
    {
      "epoch": 0.28714692832148914,
      "grad_norm": 0.9746370789406984,
      "learning_rate": 3.3455617580795866e-06,
      "loss": 0.8662,
      "step": 3980
    },
    {
      "epoch": 0.28721907579091666,
      "grad_norm": 3.5068266912820185,
      "learning_rate": 3.345215949020469e-06,
      "loss": 1.0139,
      "step": 3981
    },
    {
      "epoch": 0.2872912232603441,
      "grad_norm": 5.809241909659247,
      "learning_rate": 3.344870066504088e-06,
      "loss": 0.9402,
      "step": 3982
    },
    {
      "epoch": 0.28736337072977164,
      "grad_norm": 4.546167902022955,
      "learning_rate": 3.344524110549332e-06,
      "loss": 1.1576,
      "step": 3983
    },
    {
      "epoch": 0.28743551819919916,
      "grad_norm": 0.8336149375675451,
      "learning_rate": 3.3441780811750907e-06,
      "loss": 0.8848,
      "step": 3984
    },
    {
      "epoch": 0.2875076656686267,
      "grad_norm": 4.431613940532267,
      "learning_rate": 3.343831978400261e-06,
      "loss": 0.9615,
      "step": 3985
    },
    {
      "epoch": 0.2875798131380542,
      "grad_norm": 9.157058628035815,
      "learning_rate": 3.343485802243741e-06,
      "loss": 0.9939,
      "step": 3986
    },
    {
      "epoch": 0.2876519606074817,
      "grad_norm": 2.9876368407508824,
      "learning_rate": 3.3431395527244356e-06,
      "loss": 1.0117,
      "step": 3987
    },
    {
      "epoch": 0.2877241080769092,
      "grad_norm": 5.277098157972056,
      "learning_rate": 3.342793229861251e-06,
      "loss": 0.9732,
      "step": 3988
    },
    {
      "epoch": 0.2877962555463367,
      "grad_norm": 5.777136416580558,
      "learning_rate": 3.3424468336730993e-06,
      "loss": 1.0864,
      "step": 3989
    },
    {
      "epoch": 0.2878684030157642,
      "grad_norm": 5.093786695851332,
      "learning_rate": 3.3421003641788944e-06,
      "loss": 1.0851,
      "step": 3990
    },
    {
      "epoch": 0.28794055048519174,
      "grad_norm": 3.6926498659542224,
      "learning_rate": 3.3417538213975575e-06,
      "loss": 1.0523,
      "step": 3991
    },
    {
      "epoch": 0.28801269795461926,
      "grad_norm": 3.9087226061079114,
      "learning_rate": 3.3414072053480125e-06,
      "loss": 0.897,
      "step": 3992
    },
    {
      "epoch": 0.2880848454240468,
      "grad_norm": 0.7268451428041891,
      "learning_rate": 3.3410605160491842e-06,
      "loss": 0.9403,
      "step": 3993
    },
    {
      "epoch": 0.28815699289347424,
      "grad_norm": 3.960819521017863,
      "learning_rate": 3.340713753520006e-06,
      "loss": 0.9338,
      "step": 3994
    },
    {
      "epoch": 0.28822914036290176,
      "grad_norm": 3.9567699892157737,
      "learning_rate": 3.3403669177794134e-06,
      "loss": 0.9851,
      "step": 3995
    },
    {
      "epoch": 0.2883012878323293,
      "grad_norm": 0.9036501720371947,
      "learning_rate": 3.3400200088463447e-06,
      "loss": 0.8431,
      "step": 3996
    },
    {
      "epoch": 0.2883734353017568,
      "grad_norm": 3.2742066856677194,
      "learning_rate": 3.3396730267397445e-06,
      "loss": 0.9948,
      "step": 3997
    },
    {
      "epoch": 0.2884455827711843,
      "grad_norm": 0.6541825607418023,
      "learning_rate": 3.3393259714785595e-06,
      "loss": 0.7778,
      "step": 3998
    },
    {
      "epoch": 0.28851773024061184,
      "grad_norm": 4.763723926017391,
      "learning_rate": 3.3389788430817414e-06,
      "loss": 0.9432,
      "step": 3999
    },
    {
      "epoch": 0.2885898777100393,
      "grad_norm": 5.255099227070079,
      "learning_rate": 3.338631641568244e-06,
      "loss": 1.0482,
      "step": 4000
    },
    {
      "epoch": 0.2886620251794668,
      "grad_norm": 4.328949446945412,
      "learning_rate": 3.3382843669570296e-06,
      "loss": 1.1278,
      "step": 4001
    },
    {
      "epoch": 0.28873417264889434,
      "grad_norm": 4.0180388442839625,
      "learning_rate": 3.3379370192670593e-06,
      "loss": 0.8605,
      "step": 4002
    },
    {
      "epoch": 0.28880632011832186,
      "grad_norm": 7.482375226215196,
      "learning_rate": 3.337589598517302e-06,
      "loss": 1.0,
      "step": 4003
    },
    {
      "epoch": 0.2888784675877494,
      "grad_norm": 9.881455255122313,
      "learning_rate": 3.3372421047267274e-06,
      "loss": 0.9227,
      "step": 4004
    },
    {
      "epoch": 0.2889506150571769,
      "grad_norm": 7.474534097396608,
      "learning_rate": 3.336894537914312e-06,
      "loss": 1.0419,
      "step": 4005
    },
    {
      "epoch": 0.28902276252660436,
      "grad_norm": 3.462071218003794,
      "learning_rate": 3.3365468980990353e-06,
      "loss": 0.9987,
      "step": 4006
    },
    {
      "epoch": 0.2890949099960319,
      "grad_norm": 4.182354398884807,
      "learning_rate": 3.33619918529988e-06,
      "loss": 0.9657,
      "step": 4007
    },
    {
      "epoch": 0.2891670574654594,
      "grad_norm": 6.974120666422701,
      "learning_rate": 3.3358513995358336e-06,
      "loss": 1.0584,
      "step": 4008
    },
    {
      "epoch": 0.2892392049348869,
      "grad_norm": 7.285941646672099,
      "learning_rate": 3.335503540825888e-06,
      "loss": 1.0381,
      "step": 4009
    },
    {
      "epoch": 0.28931135240431444,
      "grad_norm": 3.952168224484072,
      "learning_rate": 3.335155609189037e-06,
      "loss": 0.9985,
      "step": 4010
    },
    {
      "epoch": 0.28938349987374196,
      "grad_norm": 3.7820117277773493,
      "learning_rate": 3.334807604644281e-06,
      "loss": 0.9104,
      "step": 4011
    },
    {
      "epoch": 0.2894556473431694,
      "grad_norm": 3.0823432948120417,
      "learning_rate": 3.3344595272106235e-06,
      "loss": 1.0926,
      "step": 4012
    },
    {
      "epoch": 0.28952779481259694,
      "grad_norm": 7.9006112840013,
      "learning_rate": 3.334111376907071e-06,
      "loss": 0.9654,
      "step": 4013
    },
    {
      "epoch": 0.28959994228202446,
      "grad_norm": 5.083707305826475,
      "learning_rate": 3.3337631537526347e-06,
      "loss": 0.9879,
      "step": 4014
    },
    {
      "epoch": 0.289672089751452,
      "grad_norm": 14.140732907046845,
      "learning_rate": 3.3334148577663303e-06,
      "loss": 0.941,
      "step": 4015
    },
    {
      "epoch": 0.2897442372208795,
      "grad_norm": 23.610246199180928,
      "learning_rate": 3.333066488967176e-06,
      "loss": 1.0132,
      "step": 4016
    },
    {
      "epoch": 0.28981638469030696,
      "grad_norm": 1.0225631240162933,
      "learning_rate": 3.3327180473741957e-06,
      "loss": 0.841,
      "step": 4017
    },
    {
      "epoch": 0.2898885321597345,
      "grad_norm": 9.21708017517296,
      "learning_rate": 3.3323695330064167e-06,
      "loss": 1.0957,
      "step": 4018
    },
    {
      "epoch": 0.289960679629162,
      "grad_norm": 5.022969320499353,
      "learning_rate": 3.33202094588287e-06,
      "loss": 1.0682,
      "step": 4019
    },
    {
      "epoch": 0.2900328270985895,
      "grad_norm": 4.506249432821675,
      "learning_rate": 3.3316722860225906e-06,
      "loss": 0.9445,
      "step": 4020
    },
    {
      "epoch": 0.29010497456801704,
      "grad_norm": 5.061601723932727,
      "learning_rate": 3.3313235534446165e-06,
      "loss": 0.9482,
      "step": 4021
    },
    {
      "epoch": 0.29017712203744456,
      "grad_norm": 5.965002671118855,
      "learning_rate": 3.330974748167992e-06,
      "loss": 1.0268,
      "step": 4022
    },
    {
      "epoch": 0.290249269506872,
      "grad_norm": 3.622193400990578,
      "learning_rate": 3.3306258702117633e-06,
      "loss": 1.0153,
      "step": 4023
    },
    {
      "epoch": 0.29032141697629954,
      "grad_norm": 4.961441903491237,
      "learning_rate": 3.3302769195949817e-06,
      "loss": 0.9012,
      "step": 4024
    },
    {
      "epoch": 0.29039356444572706,
      "grad_norm": 4.345002137985845,
      "learning_rate": 3.329927896336703e-06,
      "loss": 0.9737,
      "step": 4025
    },
    {
      "epoch": 0.2904657119151546,
      "grad_norm": 5.167718329096685,
      "learning_rate": 3.3295788004559837e-06,
      "loss": 0.8863,
      "step": 4026
    },
    {
      "epoch": 0.2905378593845821,
      "grad_norm": 5.643887999950162,
      "learning_rate": 3.3292296319718893e-06,
      "loss": 1.0823,
      "step": 4027
    },
    {
      "epoch": 0.2906100068540096,
      "grad_norm": 3.2620627216177294,
      "learning_rate": 3.3288803909034845e-06,
      "loss": 1.0106,
      "step": 4028
    },
    {
      "epoch": 0.2906821543234371,
      "grad_norm": 6.699561534597572,
      "learning_rate": 3.328531077269841e-06,
      "loss": 0.954,
      "step": 4029
    },
    {
      "epoch": 0.2907543017928646,
      "grad_norm": 7.715479849171956,
      "learning_rate": 3.328181691090033e-06,
      "loss": 1.0049,
      "step": 4030
    },
    {
      "epoch": 0.2908264492622921,
      "grad_norm": 4.414691819422238,
      "learning_rate": 3.327832232383141e-06,
      "loss": 0.9275,
      "step": 4031
    },
    {
      "epoch": 0.29089859673171964,
      "grad_norm": 3.568424184593789,
      "learning_rate": 3.327482701168245e-06,
      "loss": 0.9971,
      "step": 4032
    },
    {
      "epoch": 0.29097074420114716,
      "grad_norm": 5.340945216043809,
      "learning_rate": 3.3271330974644336e-06,
      "loss": 0.9274,
      "step": 4033
    },
    {
      "epoch": 0.2910428916705747,
      "grad_norm": 4.871727456335532,
      "learning_rate": 3.3267834212907957e-06,
      "loss": 1.0213,
      "step": 4034
    },
    {
      "epoch": 0.29111503914000214,
      "grad_norm": 12.54121197016716,
      "learning_rate": 3.3264336726664275e-06,
      "loss": 0.8487,
      "step": 4035
    },
    {
      "epoch": 0.29118718660942966,
      "grad_norm": 6.835289589831251,
      "learning_rate": 3.326083851610427e-06,
      "loss": 1.007,
      "step": 4036
    },
    {
      "epoch": 0.2912593340788572,
      "grad_norm": 5.227411266797945,
      "learning_rate": 3.325733958141896e-06,
      "loss": 1.0118,
      "step": 4037
    },
    {
      "epoch": 0.2913314815482847,
      "grad_norm": 5.831463468677776,
      "learning_rate": 3.325383992279941e-06,
      "loss": 1.0053,
      "step": 4038
    },
    {
      "epoch": 0.2914036290177122,
      "grad_norm": 4.365244424624898,
      "learning_rate": 3.3250339540436727e-06,
      "loss": 1.0704,
      "step": 4039
    },
    {
      "epoch": 0.29147577648713974,
      "grad_norm": 7.347386275609022,
      "learning_rate": 3.324683843452205e-06,
      "loss": 0.8721,
      "step": 4040
    },
    {
      "epoch": 0.2915479239565672,
      "grad_norm": 4.968785699680124,
      "learning_rate": 3.3243336605246567e-06,
      "loss": 1.0384,
      "step": 4041
    },
    {
      "epoch": 0.2916200714259947,
      "grad_norm": 4.989772254165264,
      "learning_rate": 3.32398340528015e-06,
      "loss": 1.0803,
      "step": 4042
    },
    {
      "epoch": 0.29169221889542224,
      "grad_norm": 23.751685795433552,
      "learning_rate": 3.323633077737811e-06,
      "loss": 0.9612,
      "step": 4043
    },
    {
      "epoch": 0.29176436636484976,
      "grad_norm": 2.7737583888143216,
      "learning_rate": 3.3232826779167685e-06,
      "loss": 0.9584,
      "step": 4044
    },
    {
      "epoch": 0.2918365138342773,
      "grad_norm": 4.163648173204539,
      "learning_rate": 3.3229322058361585e-06,
      "loss": 0.9605,
      "step": 4045
    },
    {
      "epoch": 0.2919086613037048,
      "grad_norm": 8.22751350376223,
      "learning_rate": 3.3225816615151176e-06,
      "loss": 1.0172,
      "step": 4046
    },
    {
      "epoch": 0.29198080877313226,
      "grad_norm": 4.232759399756909,
      "learning_rate": 3.3222310449727882e-06,
      "loss": 0.8592,
      "step": 4047
    },
    {
      "epoch": 0.2920529562425598,
      "grad_norm": 4.292636403764227,
      "learning_rate": 3.3218803562283163e-06,
      "loss": 0.9087,
      "step": 4048
    },
    {
      "epoch": 0.2921251037119873,
      "grad_norm": 3.4646909867148112,
      "learning_rate": 3.3215295953008515e-06,
      "loss": 0.8488,
      "step": 4049
    },
    {
      "epoch": 0.2921972511814148,
      "grad_norm": 3.8061499320971355,
      "learning_rate": 3.3211787622095474e-06,
      "loss": 0.9792,
      "step": 4050
    },
    {
      "epoch": 0.29226939865084234,
      "grad_norm": 5.190394869209416,
      "learning_rate": 3.3208278569735626e-06,
      "loss": 0.9576,
      "step": 4051
    },
    {
      "epoch": 0.29234154612026986,
      "grad_norm": 0.8312211727220469,
      "learning_rate": 3.3204768796120574e-06,
      "loss": 0.8392,
      "step": 4052
    },
    {
      "epoch": 0.2924136935896973,
      "grad_norm": 2.926047380513558,
      "learning_rate": 3.3201258301441986e-06,
      "loss": 0.993,
      "step": 4053
    },
    {
      "epoch": 0.29248584105912484,
      "grad_norm": 3.200152697496243,
      "learning_rate": 3.319774708589155e-06,
      "loss": 1.0008,
      "step": 4054
    },
    {
      "epoch": 0.29255798852855236,
      "grad_norm": 5.269780090202691,
      "learning_rate": 3.3194235149661e-06,
      "loss": 0.8802,
      "step": 4055
    },
    {
      "epoch": 0.2926301359979799,
      "grad_norm": 5.0862570104131155,
      "learning_rate": 3.319072249294211e-06,
      "loss": 0.8621,
      "step": 4056
    },
    {
      "epoch": 0.2927022834674074,
      "grad_norm": 3.092295449229376,
      "learning_rate": 3.3187209115926696e-06,
      "loss": 0.9581,
      "step": 4057
    },
    {
      "epoch": 0.2927744309368349,
      "grad_norm": 6.212519759639885,
      "learning_rate": 3.3183695018806615e-06,
      "loss": 1.0568,
      "step": 4058
    },
    {
      "epoch": 0.2928465784062624,
      "grad_norm": 3.3512995690134195,
      "learning_rate": 3.3180180201773746e-06,
      "loss": 1.0362,
      "step": 4059
    },
    {
      "epoch": 0.2929187258756899,
      "grad_norm": 14.441228740118136,
      "learning_rate": 3.3176664665020035e-06,
      "loss": 1.0895,
      "step": 4060
    },
    {
      "epoch": 0.2929908733451174,
      "grad_norm": 4.211795301949911,
      "learning_rate": 3.3173148408737437e-06,
      "loss": 0.9823,
      "step": 4061
    },
    {
      "epoch": 0.29306302081454494,
      "grad_norm": 4.707395522082074,
      "learning_rate": 3.316963143311798e-06,
      "loss": 1.0229,
      "step": 4062
    },
    {
      "epoch": 0.29313516828397246,
      "grad_norm": 4.395431548403876,
      "learning_rate": 3.3166113738353695e-06,
      "loss": 0.9639,
      "step": 4063
    },
    {
      "epoch": 0.2932073157533999,
      "grad_norm": 3.606640593269387,
      "learning_rate": 3.316259532463668e-06,
      "loss": 1.0486,
      "step": 4064
    },
    {
      "epoch": 0.29327946322282744,
      "grad_norm": 4.680789467500381,
      "learning_rate": 3.3159076192159063e-06,
      "loss": 0.8806,
      "step": 4065
    },
    {
      "epoch": 0.29335161069225496,
      "grad_norm": 3.691331844488221,
      "learning_rate": 3.3155556341113e-06,
      "loss": 0.9905,
      "step": 4066
    },
    {
      "epoch": 0.2934237581616825,
      "grad_norm": 2.1801447781775174,
      "learning_rate": 3.315203577169071e-06,
      "loss": 1.0411,
      "step": 4067
    },
    {
      "epoch": 0.29349590563111,
      "grad_norm": 3.2666317788680956,
      "learning_rate": 3.3148514484084434e-06,
      "loss": 0.9997,
      "step": 4068
    },
    {
      "epoch": 0.2935680531005375,
      "grad_norm": 3.858857401992664,
      "learning_rate": 3.314499247848646e-06,
      "loss": 0.8008,
      "step": 4069
    },
    {
      "epoch": 0.293640200569965,
      "grad_norm": 5.029129055149085,
      "learning_rate": 3.314146975508911e-06,
      "loss": 0.9795,
      "step": 4070
    },
    {
      "epoch": 0.2937123480393925,
      "grad_norm": 3.158043380217094,
      "learning_rate": 3.3137946314084736e-06,
      "loss": 0.9559,
      "step": 4071
    },
    {
      "epoch": 0.29378449550882,
      "grad_norm": 3.4160004097985195,
      "learning_rate": 3.3134422155665755e-06,
      "loss": 0.9747,
      "step": 4072
    },
    {
      "epoch": 0.29385664297824754,
      "grad_norm": 4.891409814038491,
      "learning_rate": 3.3130897280024595e-06,
      "loss": 0.9428,
      "step": 4073
    },
    {
      "epoch": 0.29392879044767506,
      "grad_norm": 0.8247176554183574,
      "learning_rate": 3.3127371687353747e-06,
      "loss": 0.8735,
      "step": 4074
    },
    {
      "epoch": 0.2940009379171026,
      "grad_norm": 2.82410309318637,
      "learning_rate": 3.312384537784573e-06,
      "loss": 1.0667,
      "step": 4075
    },
    {
      "epoch": 0.29407308538653004,
      "grad_norm": 5.569773173071134,
      "learning_rate": 3.31203183516931e-06,
      "loss": 0.9413,
      "step": 4076
    },
    {
      "epoch": 0.29414523285595756,
      "grad_norm": 3.9713868520256725,
      "learning_rate": 3.3116790609088446e-06,
      "loss": 0.9969,
      "step": 4077
    },
    {
      "epoch": 0.2942173803253851,
      "grad_norm": 3.267753090195752,
      "learning_rate": 3.3113262150224424e-06,
      "loss": 0.968,
      "step": 4078
    },
    {
      "epoch": 0.2942895277948126,
      "grad_norm": 5.920471186057863,
      "learning_rate": 3.3109732975293695e-06,
      "loss": 0.9749,
      "step": 4079
    },
    {
      "epoch": 0.2943616752642401,
      "grad_norm": 8.35500807738356,
      "learning_rate": 3.310620308448898e-06,
      "loss": 0.9528,
      "step": 4080
    },
    {
      "epoch": 0.29443382273366764,
      "grad_norm": 5.680409465591133,
      "learning_rate": 3.310267247800303e-06,
      "loss": 0.9538,
      "step": 4081
    },
    {
      "epoch": 0.2945059702030951,
      "grad_norm": 0.9243500346229321,
      "learning_rate": 3.309914115602864e-06,
      "loss": 0.9027,
      "step": 4082
    },
    {
      "epoch": 0.2945781176725226,
      "grad_norm": 3.695477020278934,
      "learning_rate": 3.309560911875865e-06,
      "loss": 0.9722,
      "step": 4083
    },
    {
      "epoch": 0.29465026514195014,
      "grad_norm": 2.4153733135932747,
      "learning_rate": 3.3092076366385915e-06,
      "loss": 0.9572,
      "step": 4084
    },
    {
      "epoch": 0.29472241261137766,
      "grad_norm": 3.0494029977321997,
      "learning_rate": 3.308854289910336e-06,
      "loss": 0.9857,
      "step": 4085
    },
    {
      "epoch": 0.2947945600808052,
      "grad_norm": 5.040340385750117,
      "learning_rate": 3.3085008717103926e-06,
      "loss": 0.9443,
      "step": 4086
    },
    {
      "epoch": 0.2948667075502327,
      "grad_norm": 6.590627789847234,
      "learning_rate": 3.3081473820580606e-06,
      "loss": 0.8611,
      "step": 4087
    },
    {
      "epoch": 0.29493885501966016,
      "grad_norm": 3.4807005525712666,
      "learning_rate": 3.3077938209726428e-06,
      "loss": 0.9731,
      "step": 4088
    },
    {
      "epoch": 0.2950110024890877,
      "grad_norm": 4.8270359230704205,
      "learning_rate": 3.3074401884734456e-06,
      "loss": 1.0545,
      "step": 4089
    },
    {
      "epoch": 0.2950831499585152,
      "grad_norm": 11.799096308872755,
      "learning_rate": 3.30708648457978e-06,
      "loss": 1.0333,
      "step": 4090
    },
    {
      "epoch": 0.2951552974279427,
      "grad_norm": 5.433305723124056,
      "learning_rate": 3.3067327093109598e-06,
      "loss": 1.0425,
      "step": 4091
    },
    {
      "epoch": 0.29522744489737024,
      "grad_norm": 5.793359888599027,
      "learning_rate": 3.306378862686304e-06,
      "loss": 1.034,
      "step": 4092
    },
    {
      "epoch": 0.29529959236679776,
      "grad_norm": 3.7349789182994866,
      "learning_rate": 3.3060249447251344e-06,
      "loss": 0.9259,
      "step": 4093
    },
    {
      "epoch": 0.2953717398362252,
      "grad_norm": 3.4697019628226613,
      "learning_rate": 3.3056709554467766e-06,
      "loss": 0.9706,
      "step": 4094
    },
    {
      "epoch": 0.29544388730565274,
      "grad_norm": 2.947286172217558,
      "learning_rate": 3.3053168948705623e-06,
      "loss": 1.0244,
      "step": 4095
    },
    {
      "epoch": 0.29551603477508026,
      "grad_norm": 5.330408585796057,
      "learning_rate": 3.3049627630158245e-06,
      "loss": 0.9258,
      "step": 4096
    },
    {
      "epoch": 0.2955881822445078,
      "grad_norm": 12.26279326388619,
      "learning_rate": 3.3046085599019007e-06,
      "loss": 0.9584,
      "step": 4097
    },
    {
      "epoch": 0.2956603297139353,
      "grad_norm": 3.16508275289962,
      "learning_rate": 3.3042542855481327e-06,
      "loss": 1.0459,
      "step": 4098
    },
    {
      "epoch": 0.2957324771833628,
      "grad_norm": 16.139886652703275,
      "learning_rate": 3.303899939973867e-06,
      "loss": 1.0601,
      "step": 4099
    },
    {
      "epoch": 0.2958046246527903,
      "grad_norm": 5.8352445831947035,
      "learning_rate": 3.303545523198452e-06,
      "loss": 1.0465,
      "step": 4100
    },
    {
      "epoch": 0.2958767721222178,
      "grad_norm": 5.415475371990779,
      "learning_rate": 3.303191035241242e-06,
      "loss": 0.9203,
      "step": 4101
    },
    {
      "epoch": 0.2959489195916453,
      "grad_norm": 4.624828850955975,
      "learning_rate": 3.3028364761215934e-06,
      "loss": 0.9143,
      "step": 4102
    },
    {
      "epoch": 0.29602106706107284,
      "grad_norm": 2.9758844679690135,
      "learning_rate": 3.3024818458588675e-06,
      "loss": 1.0492,
      "step": 4103
    },
    {
      "epoch": 0.29609321453050036,
      "grad_norm": 7.685191234227072,
      "learning_rate": 3.30212714447243e-06,
      "loss": 0.9169,
      "step": 4104
    },
    {
      "epoch": 0.2961653619999279,
      "grad_norm": 4.426096919250485,
      "learning_rate": 3.3017723719816495e-06,
      "loss": 0.8959,
      "step": 4105
    },
    {
      "epoch": 0.29623750946935534,
      "grad_norm": 3.8857269708642748,
      "learning_rate": 3.301417528405898e-06,
      "loss": 1.0082,
      "step": 4106
    },
    {
      "epoch": 0.29630965693878286,
      "grad_norm": 18.623317975379933,
      "learning_rate": 3.301062613764554e-06,
      "loss": 0.9198,
      "step": 4107
    },
    {
      "epoch": 0.2963818044082104,
      "grad_norm": 3.7000253315651275,
      "learning_rate": 3.3007076280769967e-06,
      "loss": 0.9238,
      "step": 4108
    },
    {
      "epoch": 0.2964539518776379,
      "grad_norm": 6.441873194586102,
      "learning_rate": 3.3003525713626108e-06,
      "loss": 1.0777,
      "step": 4109
    },
    {
      "epoch": 0.2965260993470654,
      "grad_norm": 0.6984485454667341,
      "learning_rate": 3.2999974436407846e-06,
      "loss": 0.8558,
      "step": 4110
    },
    {
      "epoch": 0.29659824681649294,
      "grad_norm": 3.3302168741714273,
      "learning_rate": 3.2996422449309104e-06,
      "loss": 0.9926,
      "step": 4111
    },
    {
      "epoch": 0.2966703942859204,
      "grad_norm": 4.020052002148179,
      "learning_rate": 3.2992869752523847e-06,
      "loss": 0.8439,
      "step": 4112
    },
    {
      "epoch": 0.2967425417553479,
      "grad_norm": 7.560339500837243,
      "learning_rate": 3.298931634624607e-06,
      "loss": 0.9248,
      "step": 4113
    },
    {
      "epoch": 0.29681468922477544,
      "grad_norm": 5.186513967905825,
      "learning_rate": 3.2985762230669807e-06,
      "loss": 0.9112,
      "step": 4114
    },
    {
      "epoch": 0.29688683669420296,
      "grad_norm": 6.231088025160206,
      "learning_rate": 3.2982207405989144e-06,
      "loss": 0.9431,
      "step": 4115
    },
    {
      "epoch": 0.2969589841636305,
      "grad_norm": 5.212209657774864,
      "learning_rate": 3.2978651872398194e-06,
      "loss": 1.007,
      "step": 4116
    },
    {
      "epoch": 0.29703113163305794,
      "grad_norm": 3.6466164465403694,
      "learning_rate": 3.297509563009111e-06,
      "loss": 0.9012,
      "step": 4117
    },
    {
      "epoch": 0.29710327910248546,
      "grad_norm": 5.240728456662272,
      "learning_rate": 3.297153867926209e-06,
      "loss": 0.855,
      "step": 4118
    },
    {
      "epoch": 0.297175426571913,
      "grad_norm": 6.656943414952752,
      "learning_rate": 3.296798102010536e-06,
      "loss": 0.95,
      "step": 4119
    },
    {
      "epoch": 0.2972475740413405,
      "grad_norm": 5.826841320385276,
      "learning_rate": 3.2964422652815184e-06,
      "loss": 0.9398,
      "step": 4120
    },
    {
      "epoch": 0.297319721510768,
      "grad_norm": 3.6393869284071245,
      "learning_rate": 3.2960863577585884e-06,
      "loss": 1.0432,
      "step": 4121
    },
    {
      "epoch": 0.29739186898019554,
      "grad_norm": 8.858799722283559,
      "learning_rate": 3.2957303794611805e-06,
      "loss": 0.9614,
      "step": 4122
    },
    {
      "epoch": 0.297464016449623,
      "grad_norm": 10.245379267049298,
      "learning_rate": 3.295374330408733e-06,
      "loss": 0.9684,
      "step": 4123
    },
    {
      "epoch": 0.2975361639190505,
      "grad_norm": 0.6112411729686235,
      "learning_rate": 3.2950182106206884e-06,
      "loss": 0.8014,
      "step": 4124
    },
    {
      "epoch": 0.29760831138847804,
      "grad_norm": 3.491514409887303,
      "learning_rate": 3.2946620201164944e-06,
      "loss": 0.9418,
      "step": 4125
    },
    {
      "epoch": 0.29768045885790556,
      "grad_norm": 3.4674478097080734,
      "learning_rate": 3.2943057589155993e-06,
      "loss": 0.9878,
      "step": 4126
    },
    {
      "epoch": 0.2977526063273331,
      "grad_norm": 6.781197683949982,
      "learning_rate": 3.293949427037458e-06,
      "loss": 0.9804,
      "step": 4127
    },
    {
      "epoch": 0.2978247537967606,
      "grad_norm": 4.543923380192341,
      "learning_rate": 3.293593024501529e-06,
      "loss": 0.998,
      "step": 4128
    },
    {
      "epoch": 0.29789690126618806,
      "grad_norm": 8.63205817193625,
      "learning_rate": 3.293236551327273e-06,
      "loss": 1.0514,
      "step": 4129
    },
    {
      "epoch": 0.2979690487356156,
      "grad_norm": 5.870434874456438,
      "learning_rate": 3.292880007534157e-06,
      "loss": 1.0058,
      "step": 4130
    },
    {
      "epoch": 0.2980411962050431,
      "grad_norm": 3.220812979005074,
      "learning_rate": 3.29252339314165e-06,
      "loss": 0.9414,
      "step": 4131
    },
    {
      "epoch": 0.2981133436744706,
      "grad_norm": 9.92419810094384,
      "learning_rate": 3.2921667081692244e-06,
      "loss": 0.9466,
      "step": 4132
    },
    {
      "epoch": 0.29818549114389814,
      "grad_norm": 3.5907324022078195,
      "learning_rate": 3.2918099526363586e-06,
      "loss": 1.0026,
      "step": 4133
    },
    {
      "epoch": 0.29825763861332566,
      "grad_norm": 5.283550777032105,
      "learning_rate": 3.2914531265625334e-06,
      "loss": 0.9906,
      "step": 4134
    },
    {
      "epoch": 0.2983297860827531,
      "grad_norm": 4.414220864484932,
      "learning_rate": 3.291096229967234e-06,
      "loss": 1.0165,
      "step": 4135
    },
    {
      "epoch": 0.29840193355218064,
      "grad_norm": 3.2420985198162664,
      "learning_rate": 3.290739262869949e-06,
      "loss": 0.8676,
      "step": 4136
    },
    {
      "epoch": 0.29847408102160816,
      "grad_norm": 10.576545721405106,
      "learning_rate": 3.290382225290171e-06,
      "loss": 0.9798,
      "step": 4137
    },
    {
      "epoch": 0.2985462284910357,
      "grad_norm": 4.71263544915018,
      "learning_rate": 3.290025117247396e-06,
      "loss": 1.0425,
      "step": 4138
    },
    {
      "epoch": 0.2986183759604632,
      "grad_norm": 2.9900522606406574,
      "learning_rate": 3.289667938761125e-06,
      "loss": 1.0077,
      "step": 4139
    },
    {
      "epoch": 0.2986905234298907,
      "grad_norm": 3.9335336235474236,
      "learning_rate": 3.2893106898508625e-06,
      "loss": 0.9544,
      "step": 4140
    },
    {
      "epoch": 0.2987626708993182,
      "grad_norm": 4.567645998637818,
      "learning_rate": 3.288953370536116e-06,
      "loss": 0.9722,
      "step": 4141
    },
    {
      "epoch": 0.2988348183687457,
      "grad_norm": 2.7662102893542277,
      "learning_rate": 3.2885959808363976e-06,
      "loss": 0.9924,
      "step": 4142
    },
    {
      "epoch": 0.2989069658381732,
      "grad_norm": 3.3622827055536884,
      "learning_rate": 3.2882385207712223e-06,
      "loss": 1.0229,
      "step": 4143
    },
    {
      "epoch": 0.29897911330760074,
      "grad_norm": 3.682116312314774,
      "learning_rate": 3.2878809903601107e-06,
      "loss": 0.9684,
      "step": 4144
    },
    {
      "epoch": 0.29905126077702826,
      "grad_norm": 5.960504878632609,
      "learning_rate": 3.2875233896225856e-06,
      "loss": 0.9273,
      "step": 4145
    },
    {
      "epoch": 0.2991234082464558,
      "grad_norm": 3.473670699733639,
      "learning_rate": 3.287165718578175e-06,
      "loss": 0.9679,
      "step": 4146
    },
    {
      "epoch": 0.29919555571588324,
      "grad_norm": 2.468239067418937,
      "learning_rate": 3.2868079772464087e-06,
      "loss": 0.9137,
      "step": 4147
    },
    {
      "epoch": 0.29926770318531076,
      "grad_norm": 4.507093349061379,
      "learning_rate": 3.2864501656468225e-06,
      "loss": 1.0187,
      "step": 4148
    },
    {
      "epoch": 0.2993398506547383,
      "grad_norm": 4.3875513144424,
      "learning_rate": 3.2860922837989556e-06,
      "loss": 0.787,
      "step": 4149
    },
    {
      "epoch": 0.2994119981241658,
      "grad_norm": 4.193905841416214,
      "learning_rate": 3.2857343317223496e-06,
      "loss": 0.9949,
      "step": 4150
    },
    {
      "epoch": 0.2994841455935933,
      "grad_norm": 3.822704354331818,
      "learning_rate": 3.2853763094365513e-06,
      "loss": 1.0421,
      "step": 4151
    },
    {
      "epoch": 0.29955629306302084,
      "grad_norm": 3.151832283815288,
      "learning_rate": 3.2850182169611112e-06,
      "loss": 1.0904,
      "step": 4152
    },
    {
      "epoch": 0.2996284405324483,
      "grad_norm": 7.059277094499396,
      "learning_rate": 3.284660054315583e-06,
      "loss": 0.8885,
      "step": 4153
    },
    {
      "epoch": 0.2997005880018758,
      "grad_norm": 7.373316685874499,
      "learning_rate": 3.2843018215195253e-06,
      "loss": 0.9246,
      "step": 4154
    },
    {
      "epoch": 0.29977273547130334,
      "grad_norm": 4.595394540811243,
      "learning_rate": 3.2839435185924992e-06,
      "loss": 0.9891,
      "step": 4155
    },
    {
      "epoch": 0.29984488294073086,
      "grad_norm": 2.5096087811217602,
      "learning_rate": 3.2835851455540705e-06,
      "loss": 1.0153,
      "step": 4156
    },
    {
      "epoch": 0.2999170304101584,
      "grad_norm": 4.080710808039764,
      "learning_rate": 3.2832267024238085e-06,
      "loss": 1.1136,
      "step": 4157
    },
    {
      "epoch": 0.2999891778795859,
      "grad_norm": 4.931639271348824,
      "learning_rate": 3.282868189221287e-06,
      "loss": 1.0713,
      "step": 4158
    },
    {
      "epoch": 0.30006132534901336,
      "grad_norm": 5.894310564747281,
      "learning_rate": 3.2825096059660825e-06,
      "loss": 0.9823,
      "step": 4159
    },
    {
      "epoch": 0.3001334728184409,
      "grad_norm": 6.271811002472286,
      "learning_rate": 3.2821509526777763e-06,
      "loss": 1.0371,
      "step": 4160
    },
    {
      "epoch": 0.3002056202878684,
      "grad_norm": 10.235838993572719,
      "learning_rate": 3.2817922293759528e-06,
      "loss": 0.956,
      "step": 4161
    },
    {
      "epoch": 0.3002777677572959,
      "grad_norm": 5.145689536285007,
      "learning_rate": 3.2814334360802002e-06,
      "loss": 0.9435,
      "step": 4162
    },
    {
      "epoch": 0.30034991522672344,
      "grad_norm": 0.7602909106079407,
      "learning_rate": 3.2810745728101124e-06,
      "loss": 0.8394,
      "step": 4163
    },
    {
      "epoch": 0.30042206269615096,
      "grad_norm": 5.136241400408143,
      "learning_rate": 3.280715639585284e-06,
      "loss": 0.9644,
      "step": 4164
    },
    {
      "epoch": 0.3004942101655784,
      "grad_norm": 4.4660630050704375,
      "learning_rate": 3.2803566364253154e-06,
      "loss": 1.0211,
      "step": 4165
    },
    {
      "epoch": 0.30056635763500594,
      "grad_norm": 2.9882793351709553,
      "learning_rate": 3.2799975633498103e-06,
      "loss": 0.9464,
      "step": 4166
    },
    {
      "epoch": 0.30063850510443346,
      "grad_norm": 11.454790231090277,
      "learning_rate": 3.2796384203783776e-06,
      "loss": 1.116,
      "step": 4167
    },
    {
      "epoch": 0.300710652573861,
      "grad_norm": 6.717516976701688,
      "learning_rate": 3.279279207530627e-06,
      "loss": 0.9363,
      "step": 4168
    },
    {
      "epoch": 0.3007828000432885,
      "grad_norm": 33.40194700128366,
      "learning_rate": 3.278919924826175e-06,
      "loss": 0.9852,
      "step": 4169
    },
    {
      "epoch": 0.30085494751271596,
      "grad_norm": 4.229895217355234,
      "learning_rate": 3.2785605722846407e-06,
      "loss": 0.8668,
      "step": 4170
    },
    {
      "epoch": 0.3009270949821435,
      "grad_norm": 4.630490729606599,
      "learning_rate": 3.278201149925646e-06,
      "loss": 0.9252,
      "step": 4171
    },
    {
      "epoch": 0.300999242451571,
      "grad_norm": 4.41343870903441,
      "learning_rate": 3.2778416577688184e-06,
      "loss": 0.9347,
      "step": 4172
    },
    {
      "epoch": 0.3010713899209985,
      "grad_norm": 3.316990813193424,
      "learning_rate": 3.277482095833788e-06,
      "loss": 1.0305,
      "step": 4173
    },
    {
      "epoch": 0.30114353739042604,
      "grad_norm": 9.276343284870904,
      "learning_rate": 3.27712246414019e-06,
      "loss": 1.029,
      "step": 4174
    },
    {
      "epoch": 0.30121568485985356,
      "grad_norm": 3.9819994258525746,
      "learning_rate": 3.2767627627076617e-06,
      "loss": 1.0178,
      "step": 4175
    },
    {
      "epoch": 0.301287832329281,
      "grad_norm": 3.7817968808504583,
      "learning_rate": 3.2764029915558455e-06,
      "loss": 0.947,
      "step": 4176
    },
    {
      "epoch": 0.30135997979870854,
      "grad_norm": 4.509234384312487,
      "learning_rate": 3.2760431507043863e-06,
      "loss": 0.9995,
      "step": 4177
    },
    {
      "epoch": 0.30143212726813606,
      "grad_norm": 0.7858224955270232,
      "learning_rate": 3.275683240172935e-06,
      "loss": 0.8632,
      "step": 4178
    },
    {
      "epoch": 0.3015042747375636,
      "grad_norm": 6.400226124106653,
      "learning_rate": 3.2753232599811445e-06,
      "loss": 1.0544,
      "step": 4179
    },
    {
      "epoch": 0.3015764222069911,
      "grad_norm": 4.288302439343487,
      "learning_rate": 3.274963210148672e-06,
      "loss": 1.0432,
      "step": 4180
    },
    {
      "epoch": 0.3016485696764186,
      "grad_norm": 10.688179931428083,
      "learning_rate": 3.2746030906951772e-06,
      "loss": 1.0825,
      "step": 4181
    },
    {
      "epoch": 0.3017207171458461,
      "grad_norm": 5.405225359005915,
      "learning_rate": 3.2742429016403273e-06,
      "loss": 1.0353,
      "step": 4182
    },
    {
      "epoch": 0.3017928646152736,
      "grad_norm": 4.4188370799936,
      "learning_rate": 3.273882643003789e-06,
      "loss": 0.9624,
      "step": 4183
    },
    {
      "epoch": 0.3018650120847011,
      "grad_norm": 5.679746268071993,
      "learning_rate": 3.2735223148052356e-06,
      "loss": 0.9152,
      "step": 4184
    },
    {
      "epoch": 0.30193715955412864,
      "grad_norm": 3.5636839656003794,
      "learning_rate": 3.273161917064343e-06,
      "loss": 0.9753,
      "step": 4185
    },
    {
      "epoch": 0.30200930702355616,
      "grad_norm": 3.489144384001566,
      "learning_rate": 3.272801449800791e-06,
      "loss": 0.9166,
      "step": 4186
    },
    {
      "epoch": 0.3020814544929837,
      "grad_norm": 6.120782451102204,
      "learning_rate": 3.2724409130342636e-06,
      "loss": 1.0123,
      "step": 4187
    },
    {
      "epoch": 0.30215360196241114,
      "grad_norm": 4.317696577852979,
      "learning_rate": 3.272080306784448e-06,
      "loss": 0.9313,
      "step": 4188
    },
    {
      "epoch": 0.30222574943183866,
      "grad_norm": 4.567340947698073,
      "learning_rate": 3.271719631071037e-06,
      "loss": 1.0339,
      "step": 4189
    },
    {
      "epoch": 0.3022978969012662,
      "grad_norm": 6.765007008438258,
      "learning_rate": 3.2713588859137236e-06,
      "loss": 1.0115,
      "step": 4190
    },
    {
      "epoch": 0.3023700443706937,
      "grad_norm": 5.377960631692117,
      "learning_rate": 3.270998071332209e-06,
      "loss": 1.0094,
      "step": 4191
    },
    {
      "epoch": 0.3024421918401212,
      "grad_norm": 8.676954276845127,
      "learning_rate": 3.2706371873461938e-06,
      "loss": 1.0619,
      "step": 4192
    },
    {
      "epoch": 0.30251433930954874,
      "grad_norm": 5.400535076616903,
      "learning_rate": 3.2702762339753855e-06,
      "loss": 0.928,
      "step": 4193
    },
    {
      "epoch": 0.3025864867789762,
      "grad_norm": 3.8703799855181926,
      "learning_rate": 3.2699152112394952e-06,
      "loss": 0.954,
      "step": 4194
    },
    {
      "epoch": 0.3026586342484037,
      "grad_norm": 7.713826080031151,
      "learning_rate": 3.2695541191582356e-06,
      "loss": 0.9809,
      "step": 4195
    },
    {
      "epoch": 0.30273078171783124,
      "grad_norm": 49.37010952837506,
      "learning_rate": 3.2691929577513256e-06,
      "loss": 1.0218,
      "step": 4196
    },
    {
      "epoch": 0.30280292918725876,
      "grad_norm": 5.281174698693475,
      "learning_rate": 3.2688317270384866e-06,
      "loss": 0.99,
      "step": 4197
    },
    {
      "epoch": 0.3028750766566863,
      "grad_norm": 3.8440665758224672,
      "learning_rate": 3.268470427039444e-06,
      "loss": 0.9976,
      "step": 4198
    },
    {
      "epoch": 0.3029472241261138,
      "grad_norm": 4.800648152777662,
      "learning_rate": 3.2681090577739266e-06,
      "loss": 0.9812,
      "step": 4199
    },
    {
      "epoch": 0.30301937159554126,
      "grad_norm": 4.386933318925893,
      "learning_rate": 3.2677476192616685e-06,
      "loss": 0.9959,
      "step": 4200
    },
    {
      "epoch": 0.3030915190649688,
      "grad_norm": 3.0363507957893026,
      "learning_rate": 3.2673861115224063e-06,
      "loss": 0.9747,
      "step": 4201
    },
    {
      "epoch": 0.3031636665343963,
      "grad_norm": 12.977179818624764,
      "learning_rate": 3.2670245345758796e-06,
      "loss": 0.9151,
      "step": 4202
    },
    {
      "epoch": 0.3032358140038238,
      "grad_norm": 0.7066830995024076,
      "learning_rate": 3.266662888441834e-06,
      "loss": 0.8501,
      "step": 4203
    },
    {
      "epoch": 0.30330796147325134,
      "grad_norm": 4.515332275195251,
      "learning_rate": 3.266301173140016e-06,
      "loss": 0.993,
      "step": 4204
    },
    {
      "epoch": 0.30338010894267886,
      "grad_norm": 7.089999245023486,
      "learning_rate": 3.26593938869018e-06,
      "loss": 1.0133,
      "step": 4205
    },
    {
      "epoch": 0.3034522564121063,
      "grad_norm": 7.974186976016226,
      "learning_rate": 3.2655775351120795e-06,
      "loss": 0.8952,
      "step": 4206
    },
    {
      "epoch": 0.30352440388153384,
      "grad_norm": 42.39845693407515,
      "learning_rate": 3.2652156124254752e-06,
      "loss": 0.9844,
      "step": 4207
    },
    {
      "epoch": 0.30359655135096136,
      "grad_norm": 3.7659220598110217,
      "learning_rate": 3.2648536206501293e-06,
      "loss": 0.9771,
      "step": 4208
    },
    {
      "epoch": 0.3036686988203889,
      "grad_norm": 4.4223628229977265,
      "learning_rate": 3.26449155980581e-06,
      "loss": 1.0184,
      "step": 4209
    },
    {
      "epoch": 0.3037408462898164,
      "grad_norm": 0.7971715188452664,
      "learning_rate": 3.2641294299122877e-06,
      "loss": 0.9019,
      "step": 4210
    },
    {
      "epoch": 0.3038129937592439,
      "grad_norm": 2.974088025758642,
      "learning_rate": 3.2637672309893365e-06,
      "loss": 0.9996,
      "step": 4211
    },
    {
      "epoch": 0.3038851412286714,
      "grad_norm": 5.340850400325516,
      "learning_rate": 3.263404963056736e-06,
      "loss": 0.9353,
      "step": 4212
    },
    {
      "epoch": 0.3039572886980989,
      "grad_norm": 3.655130198569355,
      "learning_rate": 3.2630426261342663e-06,
      "loss": 0.912,
      "step": 4213
    },
    {
      "epoch": 0.3040294361675264,
      "grad_norm": 4.68890197133373,
      "learning_rate": 3.2626802202417148e-06,
      "loss": 1.02,
      "step": 4214
    },
    {
      "epoch": 0.30410158363695394,
      "grad_norm": 4.230298096093409,
      "learning_rate": 3.2623177453988708e-06,
      "loss": 1.0114,
      "step": 4215
    },
    {
      "epoch": 0.30417373110638146,
      "grad_norm": 4.525849379598434,
      "learning_rate": 3.261955201625528e-06,
      "loss": 0.9916,
      "step": 4216
    },
    {
      "epoch": 0.304245878575809,
      "grad_norm": 6.794175727932085,
      "learning_rate": 3.2615925889414827e-06,
      "loss": 0.9956,
      "step": 4217
    },
    {
      "epoch": 0.30431802604523644,
      "grad_norm": 4.135974042534307,
      "learning_rate": 3.261229907366537e-06,
      "loss": 0.9998,
      "step": 4218
    },
    {
      "epoch": 0.30439017351466396,
      "grad_norm": 6.789050981272593,
      "learning_rate": 3.2608671569204943e-06,
      "loss": 1.008,
      "step": 4219
    },
    {
      "epoch": 0.3044623209840915,
      "grad_norm": 2.43317695097748,
      "learning_rate": 3.260504337623164e-06,
      "loss": 1.099,
      "step": 4220
    },
    {
      "epoch": 0.304534468453519,
      "grad_norm": 5.0656304747112415,
      "learning_rate": 3.260141449494358e-06,
      "loss": 1.0738,
      "step": 4221
    },
    {
      "epoch": 0.3046066159229465,
      "grad_norm": 5.23822353709585,
      "learning_rate": 3.2597784925538927e-06,
      "loss": 1.031,
      "step": 4222
    },
    {
      "epoch": 0.304678763392374,
      "grad_norm": 5.3144750177562665,
      "learning_rate": 3.2594154668215873e-06,
      "loss": 0.9809,
      "step": 4223
    },
    {
      "epoch": 0.3047509108618015,
      "grad_norm": 6.500827443202063,
      "learning_rate": 3.2590523723172654e-06,
      "loss": 0.9842,
      "step": 4224
    },
    {
      "epoch": 0.304823058331229,
      "grad_norm": 4.035584950454393,
      "learning_rate": 3.258689209060754e-06,
      "loss": 1.029,
      "step": 4225
    },
    {
      "epoch": 0.30489520580065654,
      "grad_norm": 3.020688564295535,
      "learning_rate": 3.2583259770718846e-06,
      "loss": 1.0194,
      "step": 4226
    },
    {
      "epoch": 0.30496735327008406,
      "grad_norm": 3.86098097773255,
      "learning_rate": 3.257962676370492e-06,
      "loss": 0.9072,
      "step": 4227
    },
    {
      "epoch": 0.3050395007395116,
      "grad_norm": 3.558703624984598,
      "learning_rate": 3.257599306976414e-06,
      "loss": 0.9408,
      "step": 4228
    },
    {
      "epoch": 0.30511164820893905,
      "grad_norm": 3.57344435343817,
      "learning_rate": 3.2572358689094936e-06,
      "loss": 0.8658,
      "step": 4229
    },
    {
      "epoch": 0.30518379567836657,
      "grad_norm": 6.565664345915245,
      "learning_rate": 3.2568723621895766e-06,
      "loss": 1.0012,
      "step": 4230
    },
    {
      "epoch": 0.3052559431477941,
      "grad_norm": 4.095682168669563,
      "learning_rate": 3.2565087868365123e-06,
      "loss": 1.0071,
      "step": 4231
    },
    {
      "epoch": 0.3053280906172216,
      "grad_norm": 2.7475619345740085,
      "learning_rate": 3.256145142870155e-06,
      "loss": 1.0711,
      "step": 4232
    },
    {
      "epoch": 0.3054002380866491,
      "grad_norm": 4.447437817469603,
      "learning_rate": 3.2557814303103613e-06,
      "loss": 0.9123,
      "step": 4233
    },
    {
      "epoch": 0.30547238555607664,
      "grad_norm": 13.297759729306263,
      "learning_rate": 3.2554176491769928e-06,
      "loss": 0.9493,
      "step": 4234
    },
    {
      "epoch": 0.3055445330255041,
      "grad_norm": 3.5755169954809536,
      "learning_rate": 3.2550537994899135e-06,
      "loss": 1.077,
      "step": 4235
    },
    {
      "epoch": 0.3056166804949316,
      "grad_norm": 2.9694352814648286,
      "learning_rate": 3.2546898812689927e-06,
      "loss": 0.9006,
      "step": 4236
    },
    {
      "epoch": 0.30568882796435914,
      "grad_norm": 5.216705287185076,
      "learning_rate": 3.2543258945341013e-06,
      "loss": 1.0559,
      "step": 4237
    },
    {
      "epoch": 0.30576097543378666,
      "grad_norm": 0.7641875338296593,
      "learning_rate": 3.253961839305117e-06,
      "loss": 0.8704,
      "step": 4238
    },
    {
      "epoch": 0.3058331229032142,
      "grad_norm": 10.186446995521948,
      "learning_rate": 3.2535977156019185e-06,
      "loss": 0.9412,
      "step": 4239
    },
    {
      "epoch": 0.3059052703726417,
      "grad_norm": 3.582718877821021,
      "learning_rate": 3.2532335234443893e-06,
      "loss": 1.0678,
      "step": 4240
    },
    {
      "epoch": 0.30597741784206917,
      "grad_norm": 3.7917382928699355,
      "learning_rate": 3.2528692628524166e-06,
      "loss": 1.0409,
      "step": 4241
    },
    {
      "epoch": 0.3060495653114967,
      "grad_norm": 3.058957756666128,
      "learning_rate": 3.2525049338458918e-06,
      "loss": 1.0647,
      "step": 4242
    },
    {
      "epoch": 0.3061217127809242,
      "grad_norm": 4.870804375050364,
      "learning_rate": 3.2521405364447086e-06,
      "loss": 1.0474,
      "step": 4243
    },
    {
      "epoch": 0.3061938602503517,
      "grad_norm": 9.1437436253075,
      "learning_rate": 3.251776070668766e-06,
      "loss": 0.9445,
      "step": 4244
    },
    {
      "epoch": 0.30626600771977924,
      "grad_norm": 4.47448874169013,
      "learning_rate": 3.251411536537967e-06,
      "loss": 1.0115,
      "step": 4245
    },
    {
      "epoch": 0.30633815518920676,
      "grad_norm": 8.64945811079235,
      "learning_rate": 3.2510469340722156e-06,
      "loss": 0.9973,
      "step": 4246
    },
    {
      "epoch": 0.3064103026586342,
      "grad_norm": 25.221576669342067,
      "learning_rate": 3.2506822632914227e-06,
      "loss": 0.8977,
      "step": 4247
    },
    {
      "epoch": 0.30648245012806175,
      "grad_norm": 2.779032219451892,
      "learning_rate": 3.250317524215501e-06,
      "loss": 1.065,
      "step": 4248
    },
    {
      "epoch": 0.30655459759748926,
      "grad_norm": 3.470349604559842,
      "learning_rate": 3.249952716864368e-06,
      "loss": 0.9774,
      "step": 4249
    },
    {
      "epoch": 0.3066267450669168,
      "grad_norm": 4.51244202468276,
      "learning_rate": 3.2495878412579444e-06,
      "loss": 1.0088,
      "step": 4250
    },
    {
      "epoch": 0.3066988925363443,
      "grad_norm": 4.041403824167428,
      "learning_rate": 3.249222897416155e-06,
      "loss": 0.9054,
      "step": 4251
    },
    {
      "epoch": 0.3067710400057718,
      "grad_norm": 4.695753638542669,
      "learning_rate": 3.2488578853589268e-06,
      "loss": 0.9423,
      "step": 4252
    },
    {
      "epoch": 0.3068431874751993,
      "grad_norm": 4.145790189330053,
      "learning_rate": 3.248492805106193e-06,
      "loss": 1.0807,
      "step": 4253
    },
    {
      "epoch": 0.3069153349446268,
      "grad_norm": 7.912798796642221,
      "learning_rate": 3.248127656677889e-06,
      "loss": 0.8403,
      "step": 4254
    },
    {
      "epoch": 0.3069874824140543,
      "grad_norm": 3.649783553930226,
      "learning_rate": 3.247762440093954e-06,
      "loss": 1.0354,
      "step": 4255
    },
    {
      "epoch": 0.30705962988348184,
      "grad_norm": 7.7487796930198645,
      "learning_rate": 3.2473971553743314e-06,
      "loss": 1.0591,
      "step": 4256
    },
    {
      "epoch": 0.30713177735290936,
      "grad_norm": 5.771178344255285,
      "learning_rate": 3.2470318025389674e-06,
      "loss": 1.0148,
      "step": 4257
    },
    {
      "epoch": 0.3072039248223369,
      "grad_norm": 6.806772606772794,
      "learning_rate": 3.246666381607813e-06,
      "loss": 0.9546,
      "step": 4258
    },
    {
      "epoch": 0.30727607229176435,
      "grad_norm": 3.7761199451970717,
      "learning_rate": 3.2463008926008225e-06,
      "loss": 0.9799,
      "step": 4259
    },
    {
      "epoch": 0.30734821976119187,
      "grad_norm": 3.201583416909084,
      "learning_rate": 3.2459353355379545e-06,
      "loss": 1.0617,
      "step": 4260
    },
    {
      "epoch": 0.3074203672306194,
      "grad_norm": 4.339036978130402,
      "learning_rate": 3.245569710439169e-06,
      "loss": 1.0133,
      "step": 4261
    },
    {
      "epoch": 0.3074925147000469,
      "grad_norm": 4.776987606819085,
      "learning_rate": 3.245204017324434e-06,
      "loss": 0.922,
      "step": 4262
    },
    {
      "epoch": 0.3075646621694744,
      "grad_norm": 3.4236332804744674,
      "learning_rate": 3.2448382562137163e-06,
      "loss": 1.045,
      "step": 4263
    },
    {
      "epoch": 0.30763680963890194,
      "grad_norm": 5.173174830168253,
      "learning_rate": 3.2444724271269898e-06,
      "loss": 0.9735,
      "step": 4264
    },
    {
      "epoch": 0.3077089571083294,
      "grad_norm": 3.2050592304535233,
      "learning_rate": 3.244106530084231e-06,
      "loss": 1.0348,
      "step": 4265
    },
    {
      "epoch": 0.3077811045777569,
      "grad_norm": 4.378152855642225,
      "learning_rate": 3.24374056510542e-06,
      "loss": 1.0573,
      "step": 4266
    },
    {
      "epoch": 0.30785325204718444,
      "grad_norm": 6.755944636640358,
      "learning_rate": 3.243374532210541e-06,
      "loss": 1.0012,
      "step": 4267
    },
    {
      "epoch": 0.30792539951661196,
      "grad_norm": 15.416854829756842,
      "learning_rate": 3.2430084314195823e-06,
      "loss": 0.9484,
      "step": 4268
    },
    {
      "epoch": 0.3079975469860395,
      "grad_norm": 3.2750540226564753,
      "learning_rate": 3.2426422627525336e-06,
      "loss": 0.9797,
      "step": 4269
    },
    {
      "epoch": 0.30806969445546695,
      "grad_norm": 9.531534159442392,
      "learning_rate": 3.2422760262293913e-06,
      "loss": 0.8745,
      "step": 4270
    },
    {
      "epoch": 0.30814184192489447,
      "grad_norm": 2.2387532906478818,
      "learning_rate": 3.241909721870154e-06,
      "loss": 0.9543,
      "step": 4271
    },
    {
      "epoch": 0.308213989394322,
      "grad_norm": 3.7956652813813405,
      "learning_rate": 3.2415433496948243e-06,
      "loss": 0.8002,
      "step": 4272
    },
    {
      "epoch": 0.3082861368637495,
      "grad_norm": 3.2971622287684577,
      "learning_rate": 3.241176909723408e-06,
      "loss": 1.0055,
      "step": 4273
    },
    {
      "epoch": 0.308358284333177,
      "grad_norm": 2.484344002392362,
      "learning_rate": 3.2408104019759157e-06,
      "loss": 0.9612,
      "step": 4274
    },
    {
      "epoch": 0.30843043180260454,
      "grad_norm": 5.293492860640261,
      "learning_rate": 3.2404438264723605e-06,
      "loss": 1.0123,
      "step": 4275
    },
    {
      "epoch": 0.308502579272032,
      "grad_norm": 4.570477970094821,
      "learning_rate": 3.2400771832327596e-06,
      "loss": 1.0755,
      "step": 4276
    },
    {
      "epoch": 0.3085747267414595,
      "grad_norm": 4.773082264170303,
      "learning_rate": 3.239710472277135e-06,
      "loss": 0.9868,
      "step": 4277
    },
    {
      "epoch": 0.30864687421088705,
      "grad_norm": 0.7084607589396871,
      "learning_rate": 3.2393436936255106e-06,
      "loss": 0.7793,
      "step": 4278
    },
    {
      "epoch": 0.30871902168031456,
      "grad_norm": 6.3937622428172745,
      "learning_rate": 3.2389768472979144e-06,
      "loss": 1.0004,
      "step": 4279
    },
    {
      "epoch": 0.3087911691497421,
      "grad_norm": 9.084626225375766,
      "learning_rate": 3.23860993331438e-06,
      "loss": 0.9759,
      "step": 4280
    },
    {
      "epoch": 0.3088633166191696,
      "grad_norm": 5.8529704581172926,
      "learning_rate": 3.238242951694942e-06,
      "loss": 1.0549,
      "step": 4281
    },
    {
      "epoch": 0.30893546408859707,
      "grad_norm": 4.210873180457295,
      "learning_rate": 3.2378759024596395e-06,
      "loss": 1.0112,
      "step": 4282
    },
    {
      "epoch": 0.3090076115580246,
      "grad_norm": 6.226475809563551,
      "learning_rate": 3.237508785628517e-06,
      "loss": 0.9062,
      "step": 4283
    },
    {
      "epoch": 0.3090797590274521,
      "grad_norm": 8.277824413102767,
      "learning_rate": 3.237141601221621e-06,
      "loss": 1.0513,
      "step": 4284
    },
    {
      "epoch": 0.3091519064968796,
      "grad_norm": 4.9910054366996075,
      "learning_rate": 3.2367743492590016e-06,
      "loss": 1.0536,
      "step": 4285
    },
    {
      "epoch": 0.30922405396630714,
      "grad_norm": 5.884870961864242,
      "learning_rate": 3.2364070297607135e-06,
      "loss": 0.9806,
      "step": 4286
    },
    {
      "epoch": 0.30929620143573466,
      "grad_norm": 2.292000093190456,
      "learning_rate": 3.2360396427468146e-06,
      "loss": 0.943,
      "step": 4287
    },
    {
      "epoch": 0.3093683489051621,
      "grad_norm": 3.590998649577793,
      "learning_rate": 3.2356721882373663e-06,
      "loss": 1.0573,
      "step": 4288
    },
    {
      "epoch": 0.30944049637458965,
      "grad_norm": 6.163584101959962,
      "learning_rate": 3.2353046662524336e-06,
      "loss": 0.9227,
      "step": 4289
    },
    {
      "epoch": 0.30951264384401717,
      "grad_norm": 3.8305709809664363,
      "learning_rate": 3.2349370768120877e-06,
      "loss": 0.991,
      "step": 4290
    },
    {
      "epoch": 0.3095847913134447,
      "grad_norm": 5.6333346184422455,
      "learning_rate": 3.234569419936398e-06,
      "loss": 0.9656,
      "step": 4291
    },
    {
      "epoch": 0.3096569387828722,
      "grad_norm": 6.559572275220229,
      "learning_rate": 3.234201695645443e-06,
      "loss": 0.9634,
      "step": 4292
    },
    {
      "epoch": 0.3097290862522997,
      "grad_norm": 2.9303277295239507,
      "learning_rate": 3.2338339039593026e-06,
      "loss": 0.9237,
      "step": 4293
    },
    {
      "epoch": 0.3098012337217272,
      "grad_norm": 8.037043637717186,
      "learning_rate": 3.233466044898059e-06,
      "loss": 1.0029,
      "step": 4294
    },
    {
      "epoch": 0.3098733811911547,
      "grad_norm": 2.951293700434707,
      "learning_rate": 3.2330981184818027e-06,
      "loss": 0.8836,
      "step": 4295
    },
    {
      "epoch": 0.3099455286605822,
      "grad_norm": 3.536938971254233,
      "learning_rate": 3.2327301247306214e-06,
      "loss": 0.8504,
      "step": 4296
    },
    {
      "epoch": 0.31001767613000974,
      "grad_norm": 0.8960092652727369,
      "learning_rate": 3.232362063664612e-06,
      "loss": 0.905,
      "step": 4297
    },
    {
      "epoch": 0.31008982359943726,
      "grad_norm": 4.385751646512011,
      "learning_rate": 3.2319939353038725e-06,
      "loss": 1.0934,
      "step": 4298
    },
    {
      "epoch": 0.3101619710688648,
      "grad_norm": 20.533454950806426,
      "learning_rate": 3.2316257396685042e-06,
      "loss": 0.8802,
      "step": 4299
    },
    {
      "epoch": 0.31023411853829225,
      "grad_norm": 4.22955092515061,
      "learning_rate": 3.231257476778614e-06,
      "loss": 0.9889,
      "step": 4300
    },
    {
      "epoch": 0.31030626600771977,
      "grad_norm": 21.002569132281174,
      "learning_rate": 3.230889146654311e-06,
      "loss": 0.823,
      "step": 4301
    },
    {
      "epoch": 0.3103784134771473,
      "grad_norm": 5.3458385290099315,
      "learning_rate": 3.2305207493157083e-06,
      "loss": 0.8632,
      "step": 4302
    },
    {
      "epoch": 0.3104505609465748,
      "grad_norm": 6.561309706506565,
      "learning_rate": 3.230152284782923e-06,
      "loss": 0.8257,
      "step": 4303
    },
    {
      "epoch": 0.3105227084160023,
      "grad_norm": 4.301145117228435,
      "learning_rate": 3.2297837530760744e-06,
      "loss": 1.1446,
      "step": 4304
    },
    {
      "epoch": 0.31059485588542984,
      "grad_norm": 5.752559258173602,
      "learning_rate": 3.2294151542152887e-06,
      "loss": 1.0422,
      "step": 4305
    },
    {
      "epoch": 0.3106670033548573,
      "grad_norm": 4.189494099449977,
      "learning_rate": 3.2290464882206915e-06,
      "loss": 0.9461,
      "step": 4306
    },
    {
      "epoch": 0.3107391508242848,
      "grad_norm": 6.207538562398312,
      "learning_rate": 3.228677755112416e-06,
      "loss": 0.9329,
      "step": 4307
    },
    {
      "epoch": 0.31081129829371235,
      "grad_norm": 4.3920398753845316,
      "learning_rate": 3.228308954910597e-06,
      "loss": 1.0364,
      "step": 4308
    },
    {
      "epoch": 0.31088344576313987,
      "grad_norm": 2.6739325381081085,
      "learning_rate": 3.227940087635372e-06,
      "loss": 1.0358,
      "step": 4309
    },
    {
      "epoch": 0.3109555932325674,
      "grad_norm": 5.1269244790866395,
      "learning_rate": 3.2275711533068856e-06,
      "loss": 0.9497,
      "step": 4310
    },
    {
      "epoch": 0.3110277407019949,
      "grad_norm": 5.121144775322648,
      "learning_rate": 3.227202151945282e-06,
      "loss": 1.0087,
      "step": 4311
    },
    {
      "epoch": 0.31109988817142237,
      "grad_norm": 4.622722580514267,
      "learning_rate": 3.226833083570712e-06,
      "loss": 0.9756,
      "step": 4312
    },
    {
      "epoch": 0.3111720356408499,
      "grad_norm": 2.9882309853309357,
      "learning_rate": 3.2264639482033294e-06,
      "loss": 0.969,
      "step": 4313
    },
    {
      "epoch": 0.3112441831102774,
      "grad_norm": 4.719411841455719,
      "learning_rate": 3.22609474586329e-06,
      "loss": 1.0369,
      "step": 4314
    },
    {
      "epoch": 0.3113163305797049,
      "grad_norm": 6.942126681627866,
      "learning_rate": 3.2257254765707558e-06,
      "loss": 1.0064,
      "step": 4315
    },
    {
      "epoch": 0.31138847804913244,
      "grad_norm": 5.723939302625135,
      "learning_rate": 3.2253561403458908e-06,
      "loss": 0.9853,
      "step": 4316
    },
    {
      "epoch": 0.31146062551855996,
      "grad_norm": 4.213666551748746,
      "learning_rate": 3.224986737208863e-06,
      "loss": 1.0639,
      "step": 4317
    },
    {
      "epoch": 0.3115327729879874,
      "grad_norm": 0.6979180497302875,
      "learning_rate": 3.2246172671798446e-06,
      "loss": 0.8179,
      "step": 4318
    },
    {
      "epoch": 0.31160492045741495,
      "grad_norm": 8.888203430237471,
      "learning_rate": 3.2242477302790106e-06,
      "loss": 1.0826,
      "step": 4319
    },
    {
      "epoch": 0.31167706792684247,
      "grad_norm": 5.600453372041598,
      "learning_rate": 3.223878126526539e-06,
      "loss": 0.8733,
      "step": 4320
    },
    {
      "epoch": 0.31174921539627,
      "grad_norm": 4.974710979405063,
      "learning_rate": 3.2235084559426147e-06,
      "loss": 0.9623,
      "step": 4321
    },
    {
      "epoch": 0.3118213628656975,
      "grad_norm": 5.143182831245659,
      "learning_rate": 3.2231387185474227e-06,
      "loss": 1.0351,
      "step": 4322
    },
    {
      "epoch": 0.31189351033512497,
      "grad_norm": 6.522176325974299,
      "learning_rate": 3.2227689143611532e-06,
      "loss": 0.9118,
      "step": 4323
    },
    {
      "epoch": 0.3119656578045525,
      "grad_norm": 4.473585167252014,
      "learning_rate": 3.2223990434039994e-06,
      "loss": 0.9864,
      "step": 4324
    },
    {
      "epoch": 0.31203780527398,
      "grad_norm": 5.095827018544653,
      "learning_rate": 3.2220291056961593e-06,
      "loss": 0.9968,
      "step": 4325
    },
    {
      "epoch": 0.3121099527434075,
      "grad_norm": 6.300197586109775,
      "learning_rate": 3.221659101257833e-06,
      "loss": 0.9757,
      "step": 4326
    },
    {
      "epoch": 0.31218210021283505,
      "grad_norm": 2.8171444162299797,
      "learning_rate": 3.221289030109227e-06,
      "loss": 0.9797,
      "step": 4327
    },
    {
      "epoch": 0.31225424768226256,
      "grad_norm": 5.028407650454119,
      "learning_rate": 3.220918892270547e-06,
      "loss": 1.0645,
      "step": 4328
    },
    {
      "epoch": 0.31232639515169003,
      "grad_norm": 3.3799131447732407,
      "learning_rate": 3.2205486877620057e-06,
      "loss": 0.9883,
      "step": 4329
    },
    {
      "epoch": 0.31239854262111755,
      "grad_norm": 3.4571738693215486,
      "learning_rate": 3.2201784166038195e-06,
      "loss": 1.0389,
      "step": 4330
    },
    {
      "epoch": 0.31247069009054507,
      "grad_norm": 7.713819403910269,
      "learning_rate": 3.219808078816207e-06,
      "loss": 1.0341,
      "step": 4331
    },
    {
      "epoch": 0.3125428375599726,
      "grad_norm": 6.7493203845332586,
      "learning_rate": 3.2194376744193907e-06,
      "loss": 0.9674,
      "step": 4332
    },
    {
      "epoch": 0.3126149850294001,
      "grad_norm": 3.118407658381524,
      "learning_rate": 3.219067203433597e-06,
      "loss": 0.9678,
      "step": 4333
    },
    {
      "epoch": 0.3126871324988276,
      "grad_norm": 6.366596142975767,
      "learning_rate": 3.218696665879056e-06,
      "loss": 0.9773,
      "step": 4334
    },
    {
      "epoch": 0.3127592799682551,
      "grad_norm": 11.211522061197174,
      "learning_rate": 3.218326061776002e-06,
      "loss": 0.8592,
      "step": 4335
    },
    {
      "epoch": 0.3128314274376826,
      "grad_norm": 3.2107787452675733,
      "learning_rate": 3.217955391144672e-06,
      "loss": 0.9176,
      "step": 4336
    },
    {
      "epoch": 0.3129035749071101,
      "grad_norm": 6.356106154653079,
      "learning_rate": 3.217584654005306e-06,
      "loss": 0.9673,
      "step": 4337
    },
    {
      "epoch": 0.31297572237653765,
      "grad_norm": 4.417343999399775,
      "learning_rate": 3.21721385037815e-06,
      "loss": 0.9272,
      "step": 4338
    },
    {
      "epoch": 0.31304786984596517,
      "grad_norm": 17.7653750600822,
      "learning_rate": 3.216842980283452e-06,
      "loss": 0.9958,
      "step": 4339
    },
    {
      "epoch": 0.3131200173153927,
      "grad_norm": 3.3306920713377672,
      "learning_rate": 3.216472043741463e-06,
      "loss": 0.9472,
      "step": 4340
    },
    {
      "epoch": 0.31319216478482015,
      "grad_norm": 9.01293736003735,
      "learning_rate": 3.2161010407724385e-06,
      "loss": 0.9774,
      "step": 4341
    },
    {
      "epoch": 0.31326431225424767,
      "grad_norm": 5.582419913811233,
      "learning_rate": 3.215729971396638e-06,
      "loss": 0.9628,
      "step": 4342
    },
    {
      "epoch": 0.3133364597236752,
      "grad_norm": 6.626664726262157,
      "learning_rate": 3.215358835634325e-06,
      "loss": 1.042,
      "step": 4343
    },
    {
      "epoch": 0.3134086071931027,
      "grad_norm": 6.714250628255058,
      "learning_rate": 3.2149876335057647e-06,
      "loss": 0.9864,
      "step": 4344
    },
    {
      "epoch": 0.3134807546625302,
      "grad_norm": 0.8623490284355398,
      "learning_rate": 3.2146163650312276e-06,
      "loss": 0.8526,
      "step": 4345
    },
    {
      "epoch": 0.31355290213195774,
      "grad_norm": 4.776505852969527,
      "learning_rate": 3.2142450302309873e-06,
      "loss": 0.8361,
      "step": 4346
    },
    {
      "epoch": 0.3136250496013852,
      "grad_norm": 4.764185353437582,
      "learning_rate": 3.21387362912532e-06,
      "loss": 1.0677,
      "step": 4347
    },
    {
      "epoch": 0.3136971970708127,
      "grad_norm": 4.503561094347181,
      "learning_rate": 3.2135021617345084e-06,
      "loss": 1.0079,
      "step": 4348
    },
    {
      "epoch": 0.31376934454024025,
      "grad_norm": 6.581987492899804,
      "learning_rate": 3.2131306280788354e-06,
      "loss": 0.9225,
      "step": 4349
    },
    {
      "epoch": 0.31384149200966777,
      "grad_norm": 4.086418050657716,
      "learning_rate": 3.21275902817859e-06,
      "loss": 0.93,
      "step": 4350
    },
    {
      "epoch": 0.3139136394790953,
      "grad_norm": 4.289603666479984,
      "learning_rate": 3.2123873620540636e-06,
      "loss": 0.9767,
      "step": 4351
    },
    {
      "epoch": 0.3139857869485228,
      "grad_norm": 6.155807866519386,
      "learning_rate": 3.2120156297255517e-06,
      "loss": 0.9432,
      "step": 4352
    },
    {
      "epoch": 0.31405793441795027,
      "grad_norm": 5.229908965213814,
      "learning_rate": 3.2116438312133517e-06,
      "loss": 0.9432,
      "step": 4353
    },
    {
      "epoch": 0.3141300818873778,
      "grad_norm": 5.051680128649515,
      "learning_rate": 3.2112719665377687e-06,
      "loss": 1.0226,
      "step": 4354
    },
    {
      "epoch": 0.3142022293568053,
      "grad_norm": 9.632770522769446,
      "learning_rate": 3.210900035719107e-06,
      "loss": 0.9848,
      "step": 4355
    },
    {
      "epoch": 0.3142743768262328,
      "grad_norm": 2.8725788038581292,
      "learning_rate": 3.2105280387776773e-06,
      "loss": 1.0388,
      "step": 4356
    },
    {
      "epoch": 0.31434652429566035,
      "grad_norm": 9.581432474666904,
      "learning_rate": 3.2101559757337925e-06,
      "loss": 1.0407,
      "step": 4357
    },
    {
      "epoch": 0.31441867176508786,
      "grad_norm": 7.261051567992537,
      "learning_rate": 3.2097838466077697e-06,
      "loss": 0.9503,
      "step": 4358
    },
    {
      "epoch": 0.31449081923451533,
      "grad_norm": 3.9414431491784283,
      "learning_rate": 3.2094116514199294e-06,
      "loss": 1.034,
      "step": 4359
    },
    {
      "epoch": 0.31456296670394285,
      "grad_norm": 4.038398732406127,
      "learning_rate": 3.209039390190596e-06,
      "loss": 0.9907,
      "step": 4360
    },
    {
      "epoch": 0.31463511417337037,
      "grad_norm": 4.173868110404515,
      "learning_rate": 3.2086670629400978e-06,
      "loss": 1.0483,
      "step": 4361
    },
    {
      "epoch": 0.3147072616427979,
      "grad_norm": 4.091597129298221,
      "learning_rate": 3.2082946696887643e-06,
      "loss": 0.8999,
      "step": 4362
    },
    {
      "epoch": 0.3147794091122254,
      "grad_norm": 4.524862272579734,
      "learning_rate": 3.2079222104569325e-06,
      "loss": 0.8868,
      "step": 4363
    },
    {
      "epoch": 0.3148515565816529,
      "grad_norm": 4.637706725250487,
      "learning_rate": 3.207549685264941e-06,
      "loss": 0.9788,
      "step": 4364
    },
    {
      "epoch": 0.3149237040510804,
      "grad_norm": 3.073827696846377,
      "learning_rate": 3.2071770941331304e-06,
      "loss": 1.0406,
      "step": 4365
    },
    {
      "epoch": 0.3149958515205079,
      "grad_norm": 5.4223623798752625,
      "learning_rate": 3.2068044370818478e-06,
      "loss": 1.0411,
      "step": 4366
    },
    {
      "epoch": 0.3150679989899354,
      "grad_norm": 6.660207798554177,
      "learning_rate": 3.206431714131443e-06,
      "loss": 1.0609,
      "step": 4367
    },
    {
      "epoch": 0.31514014645936295,
      "grad_norm": 7.7034684234267115,
      "learning_rate": 3.2060589253022676e-06,
      "loss": 1.0109,
      "step": 4368
    },
    {
      "epoch": 0.31521229392879047,
      "grad_norm": 7.6917375500003855,
      "learning_rate": 3.2056860706146793e-06,
      "loss": 1.0785,
      "step": 4369
    },
    {
      "epoch": 0.315284441398218,
      "grad_norm": 3.251970720673117,
      "learning_rate": 3.2053131500890372e-06,
      "loss": 1.03,
      "step": 4370
    },
    {
      "epoch": 0.31535658886764545,
      "grad_norm": 8.202209305290433,
      "learning_rate": 3.204940163745707e-06,
      "loss": 0.9461,
      "step": 4371
    },
    {
      "epoch": 0.31542873633707297,
      "grad_norm": 3.3854965591785016,
      "learning_rate": 3.2045671116050547e-06,
      "loss": 0.9978,
      "step": 4372
    },
    {
      "epoch": 0.3155008838065005,
      "grad_norm": 0.762280418484682,
      "learning_rate": 3.2041939936874514e-06,
      "loss": 0.8689,
      "step": 4373
    },
    {
      "epoch": 0.315573031275928,
      "grad_norm": 5.09495839020544,
      "learning_rate": 3.2038208100132718e-06,
      "loss": 1.007,
      "step": 4374
    },
    {
      "epoch": 0.3156451787453555,
      "grad_norm": 3.3150032229860957,
      "learning_rate": 3.203447560602894e-06,
      "loss": 0.9919,
      "step": 4375
    },
    {
      "epoch": 0.315717326214783,
      "grad_norm": 4.360062083580997,
      "learning_rate": 3.2030742454767003e-06,
      "loss": 1.0455,
      "step": 4376
    },
    {
      "epoch": 0.3157894736842105,
      "grad_norm": 0.8115987914868378,
      "learning_rate": 3.2027008646550754e-06,
      "loss": 0.9075,
      "step": 4377
    },
    {
      "epoch": 0.31586162115363803,
      "grad_norm": 0.7098706803281931,
      "learning_rate": 3.202327418158409e-06,
      "loss": 0.8588,
      "step": 4378
    },
    {
      "epoch": 0.31593376862306555,
      "grad_norm": 3.8496294413004413,
      "learning_rate": 3.201953906007093e-06,
      "loss": 0.9306,
      "step": 4379
    },
    {
      "epoch": 0.31600591609249307,
      "grad_norm": 4.514885336939676,
      "learning_rate": 3.2015803282215234e-06,
      "loss": 0.9888,
      "step": 4380
    },
    {
      "epoch": 0.3160780635619206,
      "grad_norm": 5.3327217744832565,
      "learning_rate": 3.2012066848221e-06,
      "loss": 1.0104,
      "step": 4381
    },
    {
      "epoch": 0.31615021103134805,
      "grad_norm": 3.402212427278695,
      "learning_rate": 3.200832975829227e-06,
      "loss": 0.8786,
      "step": 4382
    },
    {
      "epoch": 0.31622235850077557,
      "grad_norm": 2.9297008463237666,
      "learning_rate": 3.2004592012633097e-06,
      "loss": 0.9693,
      "step": 4383
    },
    {
      "epoch": 0.3162945059702031,
      "grad_norm": 0.7458936412943755,
      "learning_rate": 3.20008536114476e-06,
      "loss": 0.9086,
      "step": 4384
    },
    {
      "epoch": 0.3163666534396306,
      "grad_norm": 6.305925023275326,
      "learning_rate": 3.1997114554939907e-06,
      "loss": 0.9682,
      "step": 4385
    },
    {
      "epoch": 0.3164388009090581,
      "grad_norm": 3.1383078458388853,
      "learning_rate": 3.1993374843314203e-06,
      "loss": 1.0117,
      "step": 4386
    },
    {
      "epoch": 0.31651094837848565,
      "grad_norm": 3.7063470014001116,
      "learning_rate": 3.19896344767747e-06,
      "loss": 0.8826,
      "step": 4387
    },
    {
      "epoch": 0.3165830958479131,
      "grad_norm": 3.1966828856257297,
      "learning_rate": 3.198589345552564e-06,
      "loss": 0.9574,
      "step": 4388
    },
    {
      "epoch": 0.31665524331734063,
      "grad_norm": 3.577956016581506,
      "learning_rate": 3.198215177977131e-06,
      "loss": 1.0488,
      "step": 4389
    },
    {
      "epoch": 0.31672739078676815,
      "grad_norm": 3.5115228074245777,
      "learning_rate": 3.1978409449716023e-06,
      "loss": 1.0048,
      "step": 4390
    },
    {
      "epoch": 0.31679953825619567,
      "grad_norm": 32.81573086772276,
      "learning_rate": 3.197466646556414e-06,
      "loss": 1.0089,
      "step": 4391
    },
    {
      "epoch": 0.3168716857256232,
      "grad_norm": 9.386709402955189,
      "learning_rate": 3.197092282752005e-06,
      "loss": 0.9874,
      "step": 4392
    },
    {
      "epoch": 0.3169438331950507,
      "grad_norm": 5.869704761781338,
      "learning_rate": 3.196717853578818e-06,
      "loss": 1.0256,
      "step": 4393
    },
    {
      "epoch": 0.31701598066447817,
      "grad_norm": 9.214863620686502,
      "learning_rate": 3.1963433590572993e-06,
      "loss": 0.9353,
      "step": 4394
    },
    {
      "epoch": 0.3170881281339057,
      "grad_norm": 4.125216391696357,
      "learning_rate": 3.1959687992078987e-06,
      "loss": 1.0005,
      "step": 4395
    },
    {
      "epoch": 0.3171602756033332,
      "grad_norm": 4.554695037783214,
      "learning_rate": 3.1955941740510687e-06,
      "loss": 0.9483,
      "step": 4396
    },
    {
      "epoch": 0.3172324230727607,
      "grad_norm": 0.6776472359959997,
      "learning_rate": 3.1952194836072667e-06,
      "loss": 0.8723,
      "step": 4397
    },
    {
      "epoch": 0.31730457054218825,
      "grad_norm": 2.686329875200508,
      "learning_rate": 3.1948447278969537e-06,
      "loss": 1.083,
      "step": 4398
    },
    {
      "epoch": 0.31737671801161577,
      "grad_norm": 4.851798968610666,
      "learning_rate": 3.194469906940593e-06,
      "loss": 1.0232,
      "step": 4399
    },
    {
      "epoch": 0.31744886548104323,
      "grad_norm": 7.531447032076416,
      "learning_rate": 3.1940950207586533e-06,
      "loss": 0.9976,
      "step": 4400
    },
    {
      "epoch": 0.31752101295047075,
      "grad_norm": 4.460669621376295,
      "learning_rate": 3.193720069371604e-06,
      "loss": 1.0522,
      "step": 4401
    },
    {
      "epoch": 0.31759316041989827,
      "grad_norm": 3.9843079804412476,
      "learning_rate": 3.1933450527999205e-06,
      "loss": 0.9777,
      "step": 4402
    },
    {
      "epoch": 0.3176653078893258,
      "grad_norm": 6.342552335084908,
      "learning_rate": 3.192969971064082e-06,
      "loss": 1.086,
      "step": 4403
    },
    {
      "epoch": 0.3177374553587533,
      "grad_norm": 2.668073928569403,
      "learning_rate": 3.192594824184569e-06,
      "loss": 1.0428,
      "step": 4404
    },
    {
      "epoch": 0.3178096028281808,
      "grad_norm": 5.180107088197157,
      "learning_rate": 3.1922196121818682e-06,
      "loss": 0.9944,
      "step": 4405
    },
    {
      "epoch": 0.3178817502976083,
      "grad_norm": 5.17602391087995,
      "learning_rate": 3.1918443350764673e-06,
      "loss": 0.8545,
      "step": 4406
    },
    {
      "epoch": 0.3179538977670358,
      "grad_norm": 6.843102829524719,
      "learning_rate": 3.1914689928888595e-06,
      "loss": 0.8998,
      "step": 4407
    },
    {
      "epoch": 0.31802604523646333,
      "grad_norm": 3.0775079253874806,
      "learning_rate": 3.1910935856395405e-06,
      "loss": 0.9116,
      "step": 4408
    },
    {
      "epoch": 0.31809819270589085,
      "grad_norm": 3.4269242331881524,
      "learning_rate": 3.1907181133490104e-06,
      "loss": 0.9433,
      "step": 4409
    },
    {
      "epoch": 0.31817034017531837,
      "grad_norm": 4.681046990549697,
      "learning_rate": 3.1903425760377714e-06,
      "loss": 1.0703,
      "step": 4410
    },
    {
      "epoch": 0.3182424876447459,
      "grad_norm": 3.258482646854372,
      "learning_rate": 3.1899669737263315e-06,
      "loss": 1.0248,
      "step": 4411
    },
    {
      "epoch": 0.31831463511417335,
      "grad_norm": 5.875380483440653,
      "learning_rate": 3.1895913064352e-06,
      "loss": 0.9563,
      "step": 4412
    },
    {
      "epoch": 0.31838678258360087,
      "grad_norm": 5.119713825036372,
      "learning_rate": 3.1892155741848906e-06,
      "loss": 0.9145,
      "step": 4413
    },
    {
      "epoch": 0.3184589300530284,
      "grad_norm": 5.449872027653297,
      "learning_rate": 3.1888397769959216e-06,
      "loss": 0.8499,
      "step": 4414
    },
    {
      "epoch": 0.3185310775224559,
      "grad_norm": 6.817373789677634,
      "learning_rate": 3.188463914888813e-06,
      "loss": 1.0275,
      "step": 4415
    },
    {
      "epoch": 0.3186032249918834,
      "grad_norm": 4.1349863489887415,
      "learning_rate": 3.18808798788409e-06,
      "loss": 0.9773,
      "step": 4416
    },
    {
      "epoch": 0.31867537246131095,
      "grad_norm": 0.9354112244060427,
      "learning_rate": 3.1877119960022797e-06,
      "loss": 0.9329,
      "step": 4417
    },
    {
      "epoch": 0.3187475199307384,
      "grad_norm": 17.834708948051794,
      "learning_rate": 3.1873359392639134e-06,
      "loss": 1.0473,
      "step": 4418
    },
    {
      "epoch": 0.31881966740016593,
      "grad_norm": 4.286953433596088,
      "learning_rate": 3.1869598176895284e-06,
      "loss": 1.1403,
      "step": 4419
    },
    {
      "epoch": 0.31889181486959345,
      "grad_norm": 3.96808047917491,
      "learning_rate": 3.1865836312996605e-06,
      "loss": 1.013,
      "step": 4420
    },
    {
      "epoch": 0.31896396233902097,
      "grad_norm": 4.076307805367092,
      "learning_rate": 3.1862073801148536e-06,
      "loss": 1.0123,
      "step": 4421
    },
    {
      "epoch": 0.3190361098084485,
      "grad_norm": 8.930766950695597,
      "learning_rate": 3.1858310641556526e-06,
      "loss": 1.0813,
      "step": 4422
    },
    {
      "epoch": 0.319108257277876,
      "grad_norm": 17.555145570069424,
      "learning_rate": 3.185454683442607e-06,
      "loss": 0.9513,
      "step": 4423
    },
    {
      "epoch": 0.31918040474730347,
      "grad_norm": 4.288636233300412,
      "learning_rate": 3.1850782379962695e-06,
      "loss": 0.9886,
      "step": 4424
    },
    {
      "epoch": 0.319252552216731,
      "grad_norm": 3.92296411961997,
      "learning_rate": 3.184701727837197e-06,
      "loss": 1.0082,
      "step": 4425
    },
    {
      "epoch": 0.3193246996861585,
      "grad_norm": 5.861794096980923,
      "learning_rate": 3.184325152985948e-06,
      "loss": 1.0121,
      "step": 4426
    },
    {
      "epoch": 0.31939684715558603,
      "grad_norm": 4.649041247228574,
      "learning_rate": 3.183948513463087e-06,
      "loss": 0.9533,
      "step": 4427
    },
    {
      "epoch": 0.31946899462501355,
      "grad_norm": 5.312680768696087,
      "learning_rate": 3.1835718092891805e-06,
      "loss": 0.9097,
      "step": 4428
    },
    {
      "epoch": 0.319541142094441,
      "grad_norm": 6.056266175239622,
      "learning_rate": 3.183195040484799e-06,
      "loss": 1.0279,
      "step": 4429
    },
    {
      "epoch": 0.31961328956386853,
      "grad_norm": 15.597355916843075,
      "learning_rate": 3.1828182070705163e-06,
      "loss": 0.9934,
      "step": 4430
    },
    {
      "epoch": 0.31968543703329605,
      "grad_norm": 7.219237736330086,
      "learning_rate": 3.1824413090669097e-06,
      "loss": 1.0033,
      "step": 4431
    },
    {
      "epoch": 0.31975758450272357,
      "grad_norm": 2.965589105451159,
      "learning_rate": 3.1820643464945606e-06,
      "loss": 0.9437,
      "step": 4432
    },
    {
      "epoch": 0.3198297319721511,
      "grad_norm": 5.538025407858908,
      "learning_rate": 3.1816873193740534e-06,
      "loss": 1.0824,
      "step": 4433
    },
    {
      "epoch": 0.3199018794415786,
      "grad_norm": 5.767977598974825,
      "learning_rate": 3.1813102277259766e-06,
      "loss": 0.849,
      "step": 4434
    },
    {
      "epoch": 0.31997402691100607,
      "grad_norm": 0.7535193443700934,
      "learning_rate": 3.180933071570921e-06,
      "loss": 0.8271,
      "step": 4435
    },
    {
      "epoch": 0.3200461743804336,
      "grad_norm": 6.153458170397681,
      "learning_rate": 3.1805558509294813e-06,
      "loss": 0.966,
      "step": 4436
    },
    {
      "epoch": 0.3201183218498611,
      "grad_norm": 3.5247938163464125,
      "learning_rate": 3.1801785658222578e-06,
      "loss": 0.9808,
      "step": 4437
    },
    {
      "epoch": 0.32019046931928863,
      "grad_norm": 3.498457159861994,
      "learning_rate": 3.1798012162698518e-06,
      "loss": 1.0151,
      "step": 4438
    },
    {
      "epoch": 0.32026261678871615,
      "grad_norm": 4.119160767420524,
      "learning_rate": 3.179423802292868e-06,
      "loss": 0.9509,
      "step": 4439
    },
    {
      "epoch": 0.32033476425814367,
      "grad_norm": 2.9174412016839693,
      "learning_rate": 3.1790463239119173e-06,
      "loss": 1.1083,
      "step": 4440
    },
    {
      "epoch": 0.32040691172757113,
      "grad_norm": 7.106502950522923,
      "learning_rate": 3.178668781147611e-06,
      "loss": 0.8577,
      "step": 4441
    },
    {
      "epoch": 0.32047905919699865,
      "grad_norm": 9.791375035481442,
      "learning_rate": 3.1782911740205656e-06,
      "loss": 1.0344,
      "step": 4442
    },
    {
      "epoch": 0.32055120666642617,
      "grad_norm": 6.755822672638089,
      "learning_rate": 3.1779135025514012e-06,
      "loss": 1.0563,
      "step": 4443
    },
    {
      "epoch": 0.3206233541358537,
      "grad_norm": 6.3055687043906685,
      "learning_rate": 3.1775357667607415e-06,
      "loss": 0.8883,
      "step": 4444
    },
    {
      "epoch": 0.3206955016052812,
      "grad_norm": 4.830331242735925,
      "learning_rate": 3.1771579666692124e-06,
      "loss": 1.0149,
      "step": 4445
    },
    {
      "epoch": 0.3207676490747087,
      "grad_norm": 6.69838392210903,
      "learning_rate": 3.176780102297444e-06,
      "loss": 0.9677,
      "step": 4446
    },
    {
      "epoch": 0.3208397965441362,
      "grad_norm": 4.469933413152547,
      "learning_rate": 3.176402173666071e-06,
      "loss": 0.9692,
      "step": 4447
    },
    {
      "epoch": 0.3209119440135637,
      "grad_norm": 5.094901112803188,
      "learning_rate": 3.17602418079573e-06,
      "loss": 0.9807,
      "step": 4448
    },
    {
      "epoch": 0.32098409148299123,
      "grad_norm": 4.994910701835572,
      "learning_rate": 3.1756461237070627e-06,
      "loss": 1.0134,
      "step": 4449
    },
    {
      "epoch": 0.32105623895241875,
      "grad_norm": 3.4461743580815565,
      "learning_rate": 3.1752680024207116e-06,
      "loss": 0.8607,
      "step": 4450
    },
    {
      "epoch": 0.32112838642184627,
      "grad_norm": 0.7411022885671356,
      "learning_rate": 3.1748898169573257e-06,
      "loss": 0.8704,
      "step": 4451
    },
    {
      "epoch": 0.3212005338912738,
      "grad_norm": 9.932331012447195,
      "learning_rate": 3.1745115673375566e-06,
      "loss": 1.0124,
      "step": 4452
    },
    {
      "epoch": 0.32127268136070125,
      "grad_norm": 4.49067399775374,
      "learning_rate": 3.1741332535820587e-06,
      "loss": 0.9955,
      "step": 4453
    },
    {
      "epoch": 0.32134482883012877,
      "grad_norm": 2.8347082915756316,
      "learning_rate": 3.1737548757114906e-06,
      "loss": 0.9618,
      "step": 4454
    },
    {
      "epoch": 0.3214169762995563,
      "grad_norm": 5.254659129374423,
      "learning_rate": 3.173376433746513e-06,
      "loss": 0.9717,
      "step": 4455
    },
    {
      "epoch": 0.3214891237689838,
      "grad_norm": 4.115449889794572,
      "learning_rate": 3.1729979277077927e-06,
      "loss": 0.9768,
      "step": 4456
    },
    {
      "epoch": 0.32156127123841133,
      "grad_norm": 4.5627192614400265,
      "learning_rate": 3.1726193576159972e-06,
      "loss": 0.901,
      "step": 4457
    },
    {
      "epoch": 0.32163341870783885,
      "grad_norm": 7.470593085382397,
      "learning_rate": 3.1722407234918e-06,
      "loss": 1.0555,
      "step": 4458
    },
    {
      "epoch": 0.3217055661772663,
      "grad_norm": 5.282206612350244,
      "learning_rate": 3.1718620253558765e-06,
      "loss": 0.9597,
      "step": 4459
    },
    {
      "epoch": 0.32177771364669383,
      "grad_norm": 5.541722271097719,
      "learning_rate": 3.1714832632289056e-06,
      "loss": 0.9625,
      "step": 4460
    },
    {
      "epoch": 0.32184986111612135,
      "grad_norm": 7.653526351080883,
      "learning_rate": 3.1711044371315704e-06,
      "loss": 1.0052,
      "step": 4461
    },
    {
      "epoch": 0.32192200858554887,
      "grad_norm": 4.98679056499871,
      "learning_rate": 3.170725547084557e-06,
      "loss": 1.0134,
      "step": 4462
    },
    {
      "epoch": 0.3219941560549764,
      "grad_norm": 5.859930637717261,
      "learning_rate": 3.170346593108555e-06,
      "loss": 0.9462,
      "step": 4463
    },
    {
      "epoch": 0.3220663035244039,
      "grad_norm": 8.138930365840464,
      "learning_rate": 3.169967575224259e-06,
      "loss": 0.9879,
      "step": 4464
    },
    {
      "epoch": 0.32213845099383137,
      "grad_norm": 0.886408377620807,
      "learning_rate": 3.169588493452364e-06,
      "loss": 0.8879,
      "step": 4465
    },
    {
      "epoch": 0.3222105984632589,
      "grad_norm": 3.7468371722655887,
      "learning_rate": 3.169209347813572e-06,
      "loss": 0.9559,
      "step": 4466
    },
    {
      "epoch": 0.3222827459326864,
      "grad_norm": 6.959384526630351,
      "learning_rate": 3.168830138328585e-06,
      "loss": 0.947,
      "step": 4467
    },
    {
      "epoch": 0.32235489340211393,
      "grad_norm": 6.261624321847265,
      "learning_rate": 3.168450865018111e-06,
      "loss": 1.0182,
      "step": 4468
    },
    {
      "epoch": 0.32242704087154145,
      "grad_norm": 7.655500351931789,
      "learning_rate": 3.1680715279028613e-06,
      "loss": 0.9118,
      "step": 4469
    },
    {
      "epoch": 0.32249918834096897,
      "grad_norm": 3.903651723757578,
      "learning_rate": 3.1676921270035487e-06,
      "loss": 0.8661,
      "step": 4470
    },
    {
      "epoch": 0.32257133581039643,
      "grad_norm": 2.371798515845742,
      "learning_rate": 3.1673126623408924e-06,
      "loss": 0.9158,
      "step": 4471
    },
    {
      "epoch": 0.32264348327982395,
      "grad_norm": 4.088973425484251,
      "learning_rate": 3.1669331339356126e-06,
      "loss": 1.0732,
      "step": 4472
    },
    {
      "epoch": 0.32271563074925147,
      "grad_norm": 9.643614764448944,
      "learning_rate": 3.166553541808434e-06,
      "loss": 0.9383,
      "step": 4473
    },
    {
      "epoch": 0.322787778218679,
      "grad_norm": 13.824466544649766,
      "learning_rate": 3.166173885980085e-06,
      "loss": 0.9497,
      "step": 4474
    },
    {
      "epoch": 0.3228599256881065,
      "grad_norm": 3.764913784625627,
      "learning_rate": 3.1657941664712972e-06,
      "loss": 0.9909,
      "step": 4475
    },
    {
      "epoch": 0.32293207315753397,
      "grad_norm": 4.251974208290632,
      "learning_rate": 3.1654143833028054e-06,
      "loss": 0.9775,
      "step": 4476
    },
    {
      "epoch": 0.3230042206269615,
      "grad_norm": 5.658091882751827,
      "learning_rate": 3.1650345364953488e-06,
      "loss": 1.0091,
      "step": 4477
    },
    {
      "epoch": 0.323076368096389,
      "grad_norm": 7.311464415686879,
      "learning_rate": 3.1646546260696684e-06,
      "loss": 0.9853,
      "step": 4478
    },
    {
      "epoch": 0.32314851556581653,
      "grad_norm": 18.208700192361686,
      "learning_rate": 3.164274652046511e-06,
      "loss": 1.1041,
      "step": 4479
    },
    {
      "epoch": 0.32322066303524405,
      "grad_norm": 9.303735897366671,
      "learning_rate": 3.163894614446624e-06,
      "loss": 1.0819,
      "step": 4480
    },
    {
      "epoch": 0.32329281050467157,
      "grad_norm": 3.9372327199168318,
      "learning_rate": 3.163514513290761e-06,
      "loss": 1.0062,
      "step": 4481
    },
    {
      "epoch": 0.32336495797409903,
      "grad_norm": 5.930268712766331,
      "learning_rate": 3.163134348599678e-06,
      "loss": 0.9979,
      "step": 4482
    },
    {
      "epoch": 0.32343710544352655,
      "grad_norm": 5.380922292905747,
      "learning_rate": 3.162754120394134e-06,
      "loss": 0.9756,
      "step": 4483
    },
    {
      "epoch": 0.32350925291295407,
      "grad_norm": 6.013121244887126,
      "learning_rate": 3.1623738286948914e-06,
      "loss": 0.9242,
      "step": 4484
    },
    {
      "epoch": 0.3235814003823816,
      "grad_norm": 9.054088398266984,
      "learning_rate": 3.1619934735227166e-06,
      "loss": 1.0601,
      "step": 4485
    },
    {
      "epoch": 0.3236535478518091,
      "grad_norm": 4.398166248366085,
      "learning_rate": 3.161613054898381e-06,
      "loss": 1.0119,
      "step": 4486
    },
    {
      "epoch": 0.32372569532123663,
      "grad_norm": 6.928618923917745,
      "learning_rate": 3.161232572842656e-06,
      "loss": 0.9976,
      "step": 4487
    },
    {
      "epoch": 0.3237978427906641,
      "grad_norm": 3.3700576197738377,
      "learning_rate": 3.160852027376319e-06,
      "loss": 1.0431,
      "step": 4488
    },
    {
      "epoch": 0.3238699902600916,
      "grad_norm": 13.071283189261381,
      "learning_rate": 3.1604714185201497e-06,
      "loss": 0.9191,
      "step": 4489
    },
    {
      "epoch": 0.32394213772951913,
      "grad_norm": 6.58585322315962,
      "learning_rate": 3.1600907462949323e-06,
      "loss": 0.8302,
      "step": 4490
    },
    {
      "epoch": 0.32401428519894665,
      "grad_norm": 0.78096334920648,
      "learning_rate": 3.1597100107214536e-06,
      "loss": 0.8831,
      "step": 4491
    },
    {
      "epoch": 0.32408643266837417,
      "grad_norm": 5.010585832284527,
      "learning_rate": 3.1593292118205045e-06,
      "loss": 1.0765,
      "step": 4492
    },
    {
      "epoch": 0.3241585801378017,
      "grad_norm": 5.84472550190571,
      "learning_rate": 3.1589483496128795e-06,
      "loss": 1.0236,
      "step": 4493
    },
    {
      "epoch": 0.32423072760722915,
      "grad_norm": 5.053700841626389,
      "learning_rate": 3.1585674241193746e-06,
      "loss": 0.9856,
      "step": 4494
    },
    {
      "epoch": 0.32430287507665667,
      "grad_norm": 4.51213746873288,
      "learning_rate": 3.1581864353607917e-06,
      "loss": 0.9034,
      "step": 4495
    },
    {
      "epoch": 0.3243750225460842,
      "grad_norm": 3.4696173056169357,
      "learning_rate": 3.157805383357935e-06,
      "loss": 0.9046,
      "step": 4496
    },
    {
      "epoch": 0.3244471700155117,
      "grad_norm": 5.821016696598634,
      "learning_rate": 3.1574242681316124e-06,
      "loss": 0.8635,
      "step": 4497
    },
    {
      "epoch": 0.32451931748493923,
      "grad_norm": 3.7552981778144963,
      "learning_rate": 3.1570430897026354e-06,
      "loss": 0.9967,
      "step": 4498
    },
    {
      "epoch": 0.32459146495436675,
      "grad_norm": 3.928115075098092,
      "learning_rate": 3.1566618480918187e-06,
      "loss": 0.9389,
      "step": 4499
    },
    {
      "epoch": 0.3246636124237942,
      "grad_norm": 3.688829861083126,
      "learning_rate": 3.15628054331998e-06,
      "loss": 0.9197,
      "step": 4500
    },
    {
      "epoch": 0.32473575989322173,
      "grad_norm": 3.7616383830167623,
      "learning_rate": 3.1558991754079406e-06,
      "loss": 0.9652,
      "step": 4501
    },
    {
      "epoch": 0.32480790736264925,
      "grad_norm": 5.412314833365549,
      "learning_rate": 3.155517744376527e-06,
      "loss": 0.9098,
      "step": 4502
    },
    {
      "epoch": 0.32488005483207677,
      "grad_norm": 37.41926204343159,
      "learning_rate": 3.1551362502465663e-06,
      "loss": 1.0549,
      "step": 4503
    },
    {
      "epoch": 0.3249522023015043,
      "grad_norm": 5.8831720533589555,
      "learning_rate": 3.154754693038892e-06,
      "loss": 1.0293,
      "step": 4504
    },
    {
      "epoch": 0.3250243497709318,
      "grad_norm": 7.971995452006756,
      "learning_rate": 3.154373072774338e-06,
      "loss": 0.9138,
      "step": 4505
    },
    {
      "epoch": 0.3250964972403593,
      "grad_norm": 2.787325106894759,
      "learning_rate": 3.153991389473744e-06,
      "loss": 0.9334,
      "step": 4506
    },
    {
      "epoch": 0.3251686447097868,
      "grad_norm": 4.828159764235684,
      "learning_rate": 3.1536096431579523e-06,
      "loss": 0.9319,
      "step": 4507
    },
    {
      "epoch": 0.3252407921792143,
      "grad_norm": 4.732835374526886,
      "learning_rate": 3.1532278338478086e-06,
      "loss": 0.8876,
      "step": 4508
    },
    {
      "epoch": 0.32531293964864183,
      "grad_norm": 4.782537947572471,
      "learning_rate": 3.1528459615641615e-06,
      "loss": 0.8723,
      "step": 4509
    },
    {
      "epoch": 0.32538508711806935,
      "grad_norm": 3.902606210674567,
      "learning_rate": 3.152464026327865e-06,
      "loss": 0.9215,
      "step": 4510
    },
    {
      "epoch": 0.32545723458749687,
      "grad_norm": 5.677294283339439,
      "learning_rate": 3.152082028159773e-06,
      "loss": 1.0058,
      "step": 4511
    },
    {
      "epoch": 0.32552938205692433,
      "grad_norm": 6.457973478402796,
      "learning_rate": 3.1516999670807475e-06,
      "loss": 1.0082,
      "step": 4512
    },
    {
      "epoch": 0.32560152952635185,
      "grad_norm": 3.982589261670459,
      "learning_rate": 3.1513178431116498e-06,
      "loss": 1.0674,
      "step": 4513
    },
    {
      "epoch": 0.32567367699577937,
      "grad_norm": 3.9800460935444306,
      "learning_rate": 3.1509356562733464e-06,
      "loss": 1.1283,
      "step": 4514
    },
    {
      "epoch": 0.3257458244652069,
      "grad_norm": 6.257369922317728,
      "learning_rate": 3.150553406586708e-06,
      "loss": 0.9,
      "step": 4515
    },
    {
      "epoch": 0.3258179719346344,
      "grad_norm": 6.9482706676550885,
      "learning_rate": 3.150171094072607e-06,
      "loss": 0.949,
      "step": 4516
    },
    {
      "epoch": 0.32589011940406193,
      "grad_norm": 5.104132619731259,
      "learning_rate": 3.1497887187519203e-06,
      "loss": 1.0378,
      "step": 4517
    },
    {
      "epoch": 0.3259622668734894,
      "grad_norm": 3.639006553734153,
      "learning_rate": 3.149406280645528e-06,
      "loss": 0.9939,
      "step": 4518
    },
    {
      "epoch": 0.3260344143429169,
      "grad_norm": 5.767932295784166,
      "learning_rate": 3.149023779774314e-06,
      "loss": 1.0015,
      "step": 4519
    },
    {
      "epoch": 0.32610656181234443,
      "grad_norm": 6.096150243434568,
      "learning_rate": 3.1486412161591645e-06,
      "loss": 0.91,
      "step": 4520
    },
    {
      "epoch": 0.32617870928177195,
      "grad_norm": 3.885769184610287,
      "learning_rate": 3.1482585898209706e-06,
      "loss": 1.0903,
      "step": 4521
    },
    {
      "epoch": 0.32625085675119947,
      "grad_norm": 7.392007387974998,
      "learning_rate": 3.1478759007806258e-06,
      "loss": 1.0319,
      "step": 4522
    },
    {
      "epoch": 0.326323004220627,
      "grad_norm": 6.061648279511271,
      "learning_rate": 3.147493149059027e-06,
      "loss": 1.0581,
      "step": 4523
    },
    {
      "epoch": 0.32639515169005445,
      "grad_norm": 4.451882921967212,
      "learning_rate": 3.1471103346770757e-06,
      "loss": 1.0079,
      "step": 4524
    },
    {
      "epoch": 0.32646729915948197,
      "grad_norm": 5.420354888389585,
      "learning_rate": 3.1467274576556753e-06,
      "loss": 0.9252,
      "step": 4525
    },
    {
      "epoch": 0.3265394466289095,
      "grad_norm": 5.4147591190882185,
      "learning_rate": 3.1463445180157334e-06,
      "loss": 0.981,
      "step": 4526
    },
    {
      "epoch": 0.326611594098337,
      "grad_norm": 5.041035110808426,
      "learning_rate": 3.145961515778161e-06,
      "loss": 1.0774,
      "step": 4527
    },
    {
      "epoch": 0.32668374156776453,
      "grad_norm": 4.29873849861135,
      "learning_rate": 3.1455784509638726e-06,
      "loss": 0.9556,
      "step": 4528
    },
    {
      "epoch": 0.326755889037192,
      "grad_norm": 26.391221685084037,
      "learning_rate": 3.1451953235937854e-06,
      "loss": 0.9965,
      "step": 4529
    },
    {
      "epoch": 0.3268280365066195,
      "grad_norm": 4.02401628497793,
      "learning_rate": 3.1448121336888214e-06,
      "loss": 0.8527,
      "step": 4530
    },
    {
      "epoch": 0.32690018397604703,
      "grad_norm": 3.7154286402478416,
      "learning_rate": 3.1444288812699045e-06,
      "loss": 0.9059,
      "step": 4531
    },
    {
      "epoch": 0.32697233144547455,
      "grad_norm": 3.5238859648250282,
      "learning_rate": 3.1440455663579627e-06,
      "loss": 1.0206,
      "step": 4532
    },
    {
      "epoch": 0.32704447891490207,
      "grad_norm": 4.052980508339087,
      "learning_rate": 3.143662188973928e-06,
      "loss": 0.8446,
      "step": 4533
    },
    {
      "epoch": 0.3271166263843296,
      "grad_norm": 4.9829216156546945,
      "learning_rate": 3.1432787491387346e-06,
      "loss": 0.9193,
      "step": 4534
    },
    {
      "epoch": 0.32718877385375705,
      "grad_norm": 3.7608404510744005,
      "learning_rate": 3.142895246873321e-06,
      "loss": 1.0696,
      "step": 4535
    },
    {
      "epoch": 0.3272609213231846,
      "grad_norm": 0.7285010615061591,
      "learning_rate": 3.142511682198629e-06,
      "loss": 0.8278,
      "step": 4536
    },
    {
      "epoch": 0.3273330687926121,
      "grad_norm": 5.467726780671406,
      "learning_rate": 3.1421280551356034e-06,
      "loss": 0.9438,
      "step": 4537
    },
    {
      "epoch": 0.3274052162620396,
      "grad_norm": 14.0220452172711,
      "learning_rate": 3.141744365705193e-06,
      "loss": 0.984,
      "step": 4538
    },
    {
      "epoch": 0.32747736373146713,
      "grad_norm": 4.1525968400938496,
      "learning_rate": 3.1413606139283492e-06,
      "loss": 1.0162,
      "step": 4539
    },
    {
      "epoch": 0.32754951120089465,
      "grad_norm": 12.634090361312621,
      "learning_rate": 3.140976799826027e-06,
      "loss": 0.921,
      "step": 4540
    },
    {
      "epoch": 0.3276216586703221,
      "grad_norm": 5.002398487835961,
      "learning_rate": 3.1405929234191855e-06,
      "loss": 1.0521,
      "step": 4541
    },
    {
      "epoch": 0.32769380613974963,
      "grad_norm": 3.6801147212929983,
      "learning_rate": 3.1402089847287876e-06,
      "loss": 1.0965,
      "step": 4542
    },
    {
      "epoch": 0.32776595360917715,
      "grad_norm": 5.290486762397716,
      "learning_rate": 3.1398249837757976e-06,
      "loss": 0.9278,
      "step": 4543
    },
    {
      "epoch": 0.32783810107860467,
      "grad_norm": 5.711022998676003,
      "learning_rate": 3.1394409205811846e-06,
      "loss": 0.9452,
      "step": 4544
    },
    {
      "epoch": 0.3279102485480322,
      "grad_norm": 4.383019427142213,
      "learning_rate": 3.1390567951659204e-06,
      "loss": 0.9323,
      "step": 4545
    },
    {
      "epoch": 0.3279823960174597,
      "grad_norm": 9.426751287243286,
      "learning_rate": 3.138672607550982e-06,
      "loss": 0.9371,
      "step": 4546
    },
    {
      "epoch": 0.3280545434868872,
      "grad_norm": 5.252173019464572,
      "learning_rate": 3.138288357757348e-06,
      "loss": 1.0448,
      "step": 4547
    },
    {
      "epoch": 0.3281266909563147,
      "grad_norm": 3.3374781747168663,
      "learning_rate": 3.1379040458060004e-06,
      "loss": 0.8374,
      "step": 4548
    },
    {
      "epoch": 0.3281988384257422,
      "grad_norm": 3.804420414818529,
      "learning_rate": 3.1375196717179247e-06,
      "loss": 1.0416,
      "step": 4549
    },
    {
      "epoch": 0.32827098589516973,
      "grad_norm": 3.8822942192138163,
      "learning_rate": 3.137135235514111e-06,
      "loss": 1.0347,
      "step": 4550
    },
    {
      "epoch": 0.32834313336459725,
      "grad_norm": 4.187646891736237,
      "learning_rate": 3.1367507372155518e-06,
      "loss": 1.0445,
      "step": 4551
    },
    {
      "epoch": 0.32841528083402477,
      "grad_norm": 0.6900366024315757,
      "learning_rate": 3.1363661768432423e-06,
      "loss": 0.799,
      "step": 4552
    },
    {
      "epoch": 0.32848742830345223,
      "grad_norm": 4.358814271803641,
      "learning_rate": 3.1359815544181835e-06,
      "loss": 0.9511,
      "step": 4553
    },
    {
      "epoch": 0.32855957577287975,
      "grad_norm": 6.342597443309867,
      "learning_rate": 3.1355968699613765e-06,
      "loss": 0.9923,
      "step": 4554
    },
    {
      "epoch": 0.32863172324230727,
      "grad_norm": 4.6297259542546385,
      "learning_rate": 3.1352121234938284e-06,
      "loss": 0.9859,
      "step": 4555
    },
    {
      "epoch": 0.3287038707117348,
      "grad_norm": 4.109357362426503,
      "learning_rate": 3.134827315036549e-06,
      "loss": 1.006,
      "step": 4556
    },
    {
      "epoch": 0.3287760181811623,
      "grad_norm": 16.372997365314923,
      "learning_rate": 3.134442444610551e-06,
      "loss": 1.0688,
      "step": 4557
    },
    {
      "epoch": 0.32884816565058983,
      "grad_norm": 5.0266237966056835,
      "learning_rate": 3.13405751223685e-06,
      "loss": 0.9051,
      "step": 4558
    },
    {
      "epoch": 0.3289203131200173,
      "grad_norm": 8.107009925016524,
      "learning_rate": 3.133672517936467e-06,
      "loss": 1.0244,
      "step": 4559
    },
    {
      "epoch": 0.3289924605894448,
      "grad_norm": 6.697081929622058,
      "learning_rate": 3.1332874617304243e-06,
      "loss": 1.0576,
      "step": 4560
    },
    {
      "epoch": 0.32906460805887233,
      "grad_norm": 4.28542619826846,
      "learning_rate": 3.132902343639748e-06,
      "loss": 0.8514,
      "step": 4561
    },
    {
      "epoch": 0.32913675552829985,
      "grad_norm": 4.488919709879334,
      "learning_rate": 3.1325171636854696e-06,
      "loss": 1.0257,
      "step": 4562
    },
    {
      "epoch": 0.32920890299772737,
      "grad_norm": 3.190347633672801,
      "learning_rate": 3.1321319218886207e-06,
      "loss": 1.0323,
      "step": 4563
    },
    {
      "epoch": 0.3292810504671549,
      "grad_norm": 0.739824851095798,
      "learning_rate": 3.131746618270239e-06,
      "loss": 0.8705,
      "step": 4564
    },
    {
      "epoch": 0.32935319793658235,
      "grad_norm": 0.7238184887594075,
      "learning_rate": 3.1313612528513637e-06,
      "loss": 0.8405,
      "step": 4565
    },
    {
      "epoch": 0.3294253454060099,
      "grad_norm": 7.508468170592376,
      "learning_rate": 3.1309758256530387e-06,
      "loss": 0.8004,
      "step": 4566
    },
    {
      "epoch": 0.3294974928754374,
      "grad_norm": 6.18242830437762,
      "learning_rate": 3.1305903366963105e-06,
      "loss": 0.9468,
      "step": 4567
    },
    {
      "epoch": 0.3295696403448649,
      "grad_norm": 3.5984730078154086,
      "learning_rate": 3.1302047860022293e-06,
      "loss": 0.9396,
      "step": 4568
    },
    {
      "epoch": 0.32964178781429243,
      "grad_norm": 5.071547535627358,
      "learning_rate": 3.1298191735918487e-06,
      "loss": 1.0065,
      "step": 4569
    },
    {
      "epoch": 0.32971393528371995,
      "grad_norm": 2.976555291114548,
      "learning_rate": 3.1294334994862254e-06,
      "loss": 0.9594,
      "step": 4570
    },
    {
      "epoch": 0.3297860827531474,
      "grad_norm": 6.703909732438497,
      "learning_rate": 3.1290477637064197e-06,
      "loss": 0.9585,
      "step": 4571
    },
    {
      "epoch": 0.32985823022257493,
      "grad_norm": 4.734211833899327,
      "learning_rate": 3.1286619662734956e-06,
      "loss": 0.9408,
      "step": 4572
    },
    {
      "epoch": 0.32993037769200245,
      "grad_norm": 6.951112880386773,
      "learning_rate": 3.1282761072085188e-06,
      "loss": 0.9731,
      "step": 4573
    },
    {
      "epoch": 0.33000252516142997,
      "grad_norm": 0.8984652639328506,
      "learning_rate": 3.1278901865325613e-06,
      "loss": 0.8152,
      "step": 4574
    },
    {
      "epoch": 0.3300746726308575,
      "grad_norm": 5.272924960624877,
      "learning_rate": 3.127504204266696e-06,
      "loss": 0.9067,
      "step": 4575
    },
    {
      "epoch": 0.330146820100285,
      "grad_norm": 4.05133047031613,
      "learning_rate": 3.1271181604319994e-06,
      "loss": 1.0273,
      "step": 4576
    },
    {
      "epoch": 0.3302189675697125,
      "grad_norm": 4.848180307591899,
      "learning_rate": 3.1267320550495527e-06,
      "loss": 0.9997,
      "step": 4577
    },
    {
      "epoch": 0.33029111503914,
      "grad_norm": 5.364375592502498,
      "learning_rate": 3.12634588814044e-06,
      "loss": 1.0981,
      "step": 4578
    },
    {
      "epoch": 0.3303632625085675,
      "grad_norm": 4.849906896897026,
      "learning_rate": 3.1259596597257477e-06,
      "loss": 0.937,
      "step": 4579
    },
    {
      "epoch": 0.33043540997799503,
      "grad_norm": 0.7050335307297644,
      "learning_rate": 3.125573369826566e-06,
      "loss": 0.791,
      "step": 4580
    },
    {
      "epoch": 0.33050755744742255,
      "grad_norm": 6.382132934103944,
      "learning_rate": 3.12518701846399e-06,
      "loss": 0.8562,
      "step": 4581
    },
    {
      "epoch": 0.33057970491685,
      "grad_norm": 4.6708368787431604,
      "learning_rate": 3.1248006056591163e-06,
      "loss": 1.0464,
      "step": 4582
    },
    {
      "epoch": 0.33065185238627753,
      "grad_norm": 6.0845771806899975,
      "learning_rate": 3.1244141314330445e-06,
      "loss": 1.0407,
      "step": 4583
    },
    {
      "epoch": 0.33072399985570505,
      "grad_norm": 5.467675152464343,
      "learning_rate": 3.1240275958068805e-06,
      "loss": 1.0596,
      "step": 4584
    },
    {
      "epoch": 0.3307961473251326,
      "grad_norm": 41.090877394844284,
      "learning_rate": 3.12364099880173e-06,
      "loss": 1.007,
      "step": 4585
    },
    {
      "epoch": 0.3308682947945601,
      "grad_norm": 0.7780739503223698,
      "learning_rate": 3.123254340438704e-06,
      "loss": 0.8617,
      "step": 4586
    },
    {
      "epoch": 0.3309404422639876,
      "grad_norm": 0.760464031015948,
      "learning_rate": 3.122867620738917e-06,
      "loss": 0.8927,
      "step": 4587
    },
    {
      "epoch": 0.3310125897334151,
      "grad_norm": 9.584829152892436,
      "learning_rate": 3.122480839723486e-06,
      "loss": 0.9896,
      "step": 4588
    },
    {
      "epoch": 0.3310847372028426,
      "grad_norm": 7.395586924858281,
      "learning_rate": 3.1220939974135307e-06,
      "loss": 0.9786,
      "step": 4589
    },
    {
      "epoch": 0.3311568846722701,
      "grad_norm": 0.8081266371613784,
      "learning_rate": 3.1217070938301765e-06,
      "loss": 0.8641,
      "step": 4590
    },
    {
      "epoch": 0.33122903214169763,
      "grad_norm": 7.454450253557513,
      "learning_rate": 3.1213201289945508e-06,
      "loss": 0.9931,
      "step": 4591
    },
    {
      "epoch": 0.33130117961112515,
      "grad_norm": 5.68491749243135,
      "learning_rate": 3.1209331029277836e-06,
      "loss": 1.013,
      "step": 4592
    },
    {
      "epoch": 0.33137332708055267,
      "grad_norm": 6.576271067710474,
      "learning_rate": 3.1205460156510094e-06,
      "loss": 0.9755,
      "step": 4593
    },
    {
      "epoch": 0.33144547454998013,
      "grad_norm": 11.374738417751225,
      "learning_rate": 3.120158867185365e-06,
      "loss": 1.0713,
      "step": 4594
    },
    {
      "epoch": 0.33151762201940765,
      "grad_norm": 22.31497956504288,
      "learning_rate": 3.1197716575519916e-06,
      "loss": 0.8838,
      "step": 4595
    },
    {
      "epoch": 0.3315897694888352,
      "grad_norm": 2.7560944217015186,
      "learning_rate": 3.119384386772033e-06,
      "loss": 1.0072,
      "step": 4596
    },
    {
      "epoch": 0.3316619169582627,
      "grad_norm": 6.050677855798344,
      "learning_rate": 3.118997054866637e-06,
      "loss": 0.9532,
      "step": 4597
    },
    {
      "epoch": 0.3317340644276902,
      "grad_norm": 5.5279406976679475,
      "learning_rate": 3.1186096618569545e-06,
      "loss": 1.0392,
      "step": 4598
    },
    {
      "epoch": 0.33180621189711773,
      "grad_norm": 6.133160196154933,
      "learning_rate": 3.118222207764139e-06,
      "loss": 0.9728,
      "step": 4599
    },
    {
      "epoch": 0.3318783593665452,
      "grad_norm": 4.405582891515454,
      "learning_rate": 3.117834692609349e-06,
      "loss": 0.9793,
      "step": 4600
    },
    {
      "epoch": 0.3319505068359727,
      "grad_norm": 3.573737907375866,
      "learning_rate": 3.1174471164137436e-06,
      "loss": 1.0575,
      "step": 4601
    },
    {
      "epoch": 0.33202265430540023,
      "grad_norm": 9.377124596497701,
      "learning_rate": 3.117059479198488e-06,
      "loss": 1.0551,
      "step": 4602
    },
    {
      "epoch": 0.33209480177482775,
      "grad_norm": 4.648162782389071,
      "learning_rate": 3.1166717809847498e-06,
      "loss": 0.8993,
      "step": 4603
    },
    {
      "epoch": 0.33216694924425527,
      "grad_norm": 4.309036190664457,
      "learning_rate": 3.1162840217936996e-06,
      "loss": 1.0393,
      "step": 4604
    },
    {
      "epoch": 0.3322390967136828,
      "grad_norm": 4.511780683084116,
      "learning_rate": 3.1158962016465112e-06,
      "loss": 0.8851,
      "step": 4605
    },
    {
      "epoch": 0.33231124418311025,
      "grad_norm": 6.878622591841068,
      "learning_rate": 3.1155083205643617e-06,
      "loss": 1.0302,
      "step": 4606
    },
    {
      "epoch": 0.3323833916525378,
      "grad_norm": 10.363389101987787,
      "learning_rate": 3.115120378568432e-06,
      "loss": 1.0046,
      "step": 4607
    },
    {
      "epoch": 0.3324555391219653,
      "grad_norm": 4.273987785211227,
      "learning_rate": 3.1147323756799072e-06,
      "loss": 1.0156,
      "step": 4608
    },
    {
      "epoch": 0.3325276865913928,
      "grad_norm": 7.602612006413872,
      "learning_rate": 3.114344311919974e-06,
      "loss": 1.096,
      "step": 4609
    },
    {
      "epoch": 0.33259983406082033,
      "grad_norm": 4.758099526953848,
      "learning_rate": 3.1139561873098224e-06,
      "loss": 1.0369,
      "step": 4610
    },
    {
      "epoch": 0.33267198153024785,
      "grad_norm": 6.014670873501952,
      "learning_rate": 3.1135680018706474e-06,
      "loss": 0.9913,
      "step": 4611
    },
    {
      "epoch": 0.3327441289996753,
      "grad_norm": 8.677894606260397,
      "learning_rate": 3.1131797556236467e-06,
      "loss": 0.9438,
      "step": 4612
    },
    {
      "epoch": 0.33281627646910283,
      "grad_norm": 4.020069082624968,
      "learning_rate": 3.11279144859002e-06,
      "loss": 0.9212,
      "step": 4613
    },
    {
      "epoch": 0.33288842393853035,
      "grad_norm": 12.524141116969796,
      "learning_rate": 3.1124030807909714e-06,
      "loss": 0.9809,
      "step": 4614
    },
    {
      "epoch": 0.3329605714079579,
      "grad_norm": 3.776196973391105,
      "learning_rate": 3.1120146522477087e-06,
      "loss": 0.9596,
      "step": 4615
    },
    {
      "epoch": 0.3330327188773854,
      "grad_norm": 5.199794310389723,
      "learning_rate": 3.1116261629814425e-06,
      "loss": 0.9867,
      "step": 4616
    },
    {
      "epoch": 0.3331048663468129,
      "grad_norm": 5.345933969828437,
      "learning_rate": 3.1112376130133864e-06,
      "loss": 0.9559,
      "step": 4617
    },
    {
      "epoch": 0.3331770138162404,
      "grad_norm": 5.142826061306233,
      "learning_rate": 3.110849002364758e-06,
      "loss": 0.8787,
      "step": 4618
    },
    {
      "epoch": 0.3332491612856679,
      "grad_norm": 6.208968093817125,
      "learning_rate": 3.110460331056778e-06,
      "loss": 0.9845,
      "step": 4619
    },
    {
      "epoch": 0.3333213087550954,
      "grad_norm": 6.340614784349059,
      "learning_rate": 3.110071599110671e-06,
      "loss": 1.098,
      "step": 4620
    },
    {
      "epoch": 0.33339345622452293,
      "grad_norm": 5.340582371648187,
      "learning_rate": 3.1096828065476616e-06,
      "loss": 0.9945,
      "step": 4621
    },
    {
      "epoch": 0.33346560369395045,
      "grad_norm": 6.697693943178812,
      "learning_rate": 3.109293953388983e-06,
      "loss": 1.0454,
      "step": 4622
    },
    {
      "epoch": 0.33353775116337797,
      "grad_norm": 5.001470158924309,
      "learning_rate": 3.108905039655868e-06,
      "loss": 0.9414,
      "step": 4623
    },
    {
      "epoch": 0.33360989863280543,
      "grad_norm": 5.290288470358623,
      "learning_rate": 3.108516065369553e-06,
      "loss": 0.9435,
      "step": 4624
    },
    {
      "epoch": 0.33368204610223295,
      "grad_norm": 0.718836737665007,
      "learning_rate": 3.10812703055128e-06,
      "loss": 0.8541,
      "step": 4625
    },
    {
      "epoch": 0.3337541935716605,
      "grad_norm": 7.89081136653366,
      "learning_rate": 3.1077379352222923e-06,
      "loss": 0.9234,
      "step": 4626
    },
    {
      "epoch": 0.333826341041088,
      "grad_norm": 6.410926860862791,
      "learning_rate": 3.1073487794038364e-06,
      "loss": 1.0056,
      "step": 4627
    },
    {
      "epoch": 0.3338984885105155,
      "grad_norm": 7.146217476739252,
      "learning_rate": 3.1069595631171625e-06,
      "loss": 0.9527,
      "step": 4628
    },
    {
      "epoch": 0.333970635979943,
      "grad_norm": 8.418051678100806,
      "learning_rate": 3.106570286383525e-06,
      "loss": 1.0882,
      "step": 4629
    },
    {
      "epoch": 0.3340427834493705,
      "grad_norm": 8.739105417660125,
      "learning_rate": 3.1061809492241802e-06,
      "loss": 0.9024,
      "step": 4630
    },
    {
      "epoch": 0.334114930918798,
      "grad_norm": 5.114156276590179,
      "learning_rate": 3.1057915516603896e-06,
      "loss": 0.9844,
      "step": 4631
    },
    {
      "epoch": 0.33418707838822553,
      "grad_norm": 5.235850385295797,
      "learning_rate": 3.1054020937134154e-06,
      "loss": 0.9418,
      "step": 4632
    },
    {
      "epoch": 0.33425922585765305,
      "grad_norm": 10.22177891069461,
      "learning_rate": 3.105012575404524e-06,
      "loss": 0.9439,
      "step": 4633
    },
    {
      "epoch": 0.33433137332708057,
      "grad_norm": 5.148324504995799,
      "learning_rate": 3.104622996754988e-06,
      "loss": 0.9804,
      "step": 4634
    },
    {
      "epoch": 0.33440352079650804,
      "grad_norm": 2.715260952515004,
      "learning_rate": 3.104233357786079e-06,
      "loss": 0.9641,
      "step": 4635
    },
    {
      "epoch": 0.33447566826593556,
      "grad_norm": 5.313619697604589,
      "learning_rate": 3.1038436585190732e-06,
      "loss": 0.9509,
      "step": 4636
    },
    {
      "epoch": 0.3345478157353631,
      "grad_norm": 6.75972471139381,
      "learning_rate": 3.1034538989752524e-06,
      "loss": 0.955,
      "step": 4637
    },
    {
      "epoch": 0.3346199632047906,
      "grad_norm": 5.661590583985549,
      "learning_rate": 3.1030640791758985e-06,
      "loss": 0.95,
      "step": 4638
    },
    {
      "epoch": 0.3346921106742181,
      "grad_norm": 5.738669468658028,
      "learning_rate": 3.1026741991422986e-06,
      "loss": 0.8778,
      "step": 4639
    },
    {
      "epoch": 0.33476425814364563,
      "grad_norm": 3.899241882390121,
      "learning_rate": 3.102284258895742e-06,
      "loss": 0.9909,
      "step": 4640
    },
    {
      "epoch": 0.3348364056130731,
      "grad_norm": 3.7614114701211023,
      "learning_rate": 3.101894258457524e-06,
      "loss": 0.8688,
      "step": 4641
    },
    {
      "epoch": 0.3349085530825006,
      "grad_norm": 7.949673186488401,
      "learning_rate": 3.101504197848939e-06,
      "loss": 1.0422,
      "step": 4642
    },
    {
      "epoch": 0.33498070055192813,
      "grad_norm": 6.8095411121962846,
      "learning_rate": 3.1011140770912873e-06,
      "loss": 0.8926,
      "step": 4643
    },
    {
      "epoch": 0.33505284802135565,
      "grad_norm": 9.717667399761263,
      "learning_rate": 3.100723896205872e-06,
      "loss": 0.9349,
      "step": 4644
    },
    {
      "epoch": 0.3351249954907832,
      "grad_norm": 22.808062228885618,
      "learning_rate": 3.1003336552139995e-06,
      "loss": 1.0173,
      "step": 4645
    },
    {
      "epoch": 0.3351971429602107,
      "grad_norm": 4.900810190288972,
      "learning_rate": 3.0999433541369793e-06,
      "loss": 0.8664,
      "step": 4646
    },
    {
      "epoch": 0.33526929042963816,
      "grad_norm": 6.831113253077216,
      "learning_rate": 3.0995529929961244e-06,
      "loss": 0.8627,
      "step": 4647
    },
    {
      "epoch": 0.3353414378990657,
      "grad_norm": 4.602532949915048,
      "learning_rate": 3.0991625718127516e-06,
      "loss": 1.0307,
      "step": 4648
    },
    {
      "epoch": 0.3354135853684932,
      "grad_norm": 5.702235656792492,
      "learning_rate": 3.098772090608179e-06,
      "loss": 0.9327,
      "step": 4649
    },
    {
      "epoch": 0.3354857328379207,
      "grad_norm": 4.630062116696467,
      "learning_rate": 3.0983815494037304e-06,
      "loss": 0.9415,
      "step": 4650
    },
    {
      "epoch": 0.33555788030734823,
      "grad_norm": 5.695635667378802,
      "learning_rate": 3.0979909482207304e-06,
      "loss": 0.9899,
      "step": 4651
    },
    {
      "epoch": 0.33563002777677575,
      "grad_norm": 10.223250305150403,
      "learning_rate": 3.09760028708051e-06,
      "loss": 1.0141,
      "step": 4652
    },
    {
      "epoch": 0.3357021752462032,
      "grad_norm": 7.549330159971666,
      "learning_rate": 3.0972095660044015e-06,
      "loss": 0.9116,
      "step": 4653
    },
    {
      "epoch": 0.33577432271563074,
      "grad_norm": 4.15013831953696,
      "learning_rate": 3.096818785013741e-06,
      "loss": 0.9735,
      "step": 4654
    },
    {
      "epoch": 0.33584647018505825,
      "grad_norm": 10.741723001321034,
      "learning_rate": 3.096427944129866e-06,
      "loss": 0.9656,
      "step": 4655
    },
    {
      "epoch": 0.3359186176544858,
      "grad_norm": 4.319840527875919,
      "learning_rate": 3.0960370433741195e-06,
      "loss": 1.0154,
      "step": 4656
    },
    {
      "epoch": 0.3359907651239133,
      "grad_norm": 0.7649201632583692,
      "learning_rate": 3.095646082767848e-06,
      "loss": 0.9111,
      "step": 4657
    },
    {
      "epoch": 0.3360629125933408,
      "grad_norm": 8.812114734209217,
      "learning_rate": 3.0952550623324e-06,
      "loss": 0.9534,
      "step": 4658
    },
    {
      "epoch": 0.3361350600627683,
      "grad_norm": 10.047304040460073,
      "learning_rate": 3.0948639820891267e-06,
      "loss": 0.9056,
      "step": 4659
    },
    {
      "epoch": 0.3362072075321958,
      "grad_norm": 9.7029068042345,
      "learning_rate": 3.094472842059385e-06,
      "loss": 1.0484,
      "step": 4660
    },
    {
      "epoch": 0.3362793550016233,
      "grad_norm": 4.8332567318513435,
      "learning_rate": 3.094081642264533e-06,
      "loss": 1.0468,
      "step": 4661
    },
    {
      "epoch": 0.33635150247105083,
      "grad_norm": 0.7515125123211402,
      "learning_rate": 3.093690382725933e-06,
      "loss": 0.8443,
      "step": 4662
    },
    {
      "epoch": 0.33642364994047835,
      "grad_norm": 3.0213009877887833,
      "learning_rate": 3.093299063464949e-06,
      "loss": 0.8419,
      "step": 4663
    },
    {
      "epoch": 0.3364957974099059,
      "grad_norm": 6.062507550736277,
      "learning_rate": 3.0929076845029512e-06,
      "loss": 0.9907,
      "step": 4664
    },
    {
      "epoch": 0.33656794487933334,
      "grad_norm": 4.885310103413964,
      "learning_rate": 3.09251624586131e-06,
      "loss": 0.979,
      "step": 4665
    },
    {
      "epoch": 0.33664009234876086,
      "grad_norm": 5.227303445296064,
      "learning_rate": 3.0921247475614013e-06,
      "loss": 0.9323,
      "step": 4666
    },
    {
      "epoch": 0.3367122398181884,
      "grad_norm": 5.320631020487773,
      "learning_rate": 3.091733189624603e-06,
      "loss": 1.0551,
      "step": 4667
    },
    {
      "epoch": 0.3367843872876159,
      "grad_norm": 3.3035270511194588,
      "learning_rate": 3.0913415720722968e-06,
      "loss": 1.0353,
      "step": 4668
    },
    {
      "epoch": 0.3368565347570434,
      "grad_norm": 6.177364486634444,
      "learning_rate": 3.0909498949258666e-06,
      "loss": 1.0462,
      "step": 4669
    },
    {
      "epoch": 0.33692868222647093,
      "grad_norm": 7.195900958225171,
      "learning_rate": 3.0905581582067025e-06,
      "loss": 0.9018,
      "step": 4670
    },
    {
      "epoch": 0.3370008296958984,
      "grad_norm": 4.352550565632937,
      "learning_rate": 3.0901663619361937e-06,
      "loss": 0.9516,
      "step": 4671
    },
    {
      "epoch": 0.3370729771653259,
      "grad_norm": 3.2594309201294176,
      "learning_rate": 3.089774506135735e-06,
      "loss": 0.9388,
      "step": 4672
    },
    {
      "epoch": 0.33714512463475343,
      "grad_norm": 7.557377869745119,
      "learning_rate": 3.0893825908267255e-06,
      "loss": 0.9597,
      "step": 4673
    },
    {
      "epoch": 0.33721727210418095,
      "grad_norm": 8.641830608562623,
      "learning_rate": 3.0889906160305654e-06,
      "loss": 0.939,
      "step": 4674
    },
    {
      "epoch": 0.3372894195736085,
      "grad_norm": 4.623409306993608,
      "learning_rate": 3.088598581768659e-06,
      "loss": 1.0259,
      "step": 4675
    },
    {
      "epoch": 0.337361567043036,
      "grad_norm": 5.005042013474076,
      "learning_rate": 3.088206488062414e-06,
      "loss": 0.9377,
      "step": 4676
    },
    {
      "epoch": 0.33743371451246346,
      "grad_norm": 5.065144542804475,
      "learning_rate": 3.087814334933241e-06,
      "loss": 0.9914,
      "step": 4677
    },
    {
      "epoch": 0.337505861981891,
      "grad_norm": 6.349347731363398,
      "learning_rate": 3.087422122402554e-06,
      "loss": 0.8968,
      "step": 4678
    },
    {
      "epoch": 0.3375780094513185,
      "grad_norm": 11.386379041888025,
      "learning_rate": 3.0870298504917703e-06,
      "loss": 1.0192,
      "step": 4679
    },
    {
      "epoch": 0.337650156920746,
      "grad_norm": 4.974239364061259,
      "learning_rate": 3.086637519222311e-06,
      "loss": 0.8391,
      "step": 4680
    },
    {
      "epoch": 0.33772230439017353,
      "grad_norm": 10.583413271464428,
      "learning_rate": 3.086245128615599e-06,
      "loss": 0.961,
      "step": 4681
    },
    {
      "epoch": 0.337794451859601,
      "grad_norm": 10.34761348613285,
      "learning_rate": 3.0858526786930615e-06,
      "loss": 1.0736,
      "step": 4682
    },
    {
      "epoch": 0.3378665993290285,
      "grad_norm": 5.862896077455157,
      "learning_rate": 3.0854601694761293e-06,
      "loss": 0.9535,
      "step": 4683
    },
    {
      "epoch": 0.33793874679845604,
      "grad_norm": 6.558676632237617,
      "learning_rate": 3.0850676009862353e-06,
      "loss": 0.864,
      "step": 4684
    },
    {
      "epoch": 0.33801089426788355,
      "grad_norm": 4.3734394014878415,
      "learning_rate": 3.0846749732448163e-06,
      "loss": 1.0156,
      "step": 4685
    },
    {
      "epoch": 0.3380830417373111,
      "grad_norm": 7.621397073986829,
      "learning_rate": 3.0842822862733123e-06,
      "loss": 0.9745,
      "step": 4686
    },
    {
      "epoch": 0.3381551892067386,
      "grad_norm": 3.917893170343734,
      "learning_rate": 3.0838895400931664e-06,
      "loss": 0.9242,
      "step": 4687
    },
    {
      "epoch": 0.33822733667616606,
      "grad_norm": 0.8693354013338923,
      "learning_rate": 3.083496734725826e-06,
      "loss": 0.8843,
      "step": 4688
    },
    {
      "epoch": 0.3382994841455936,
      "grad_norm": 4.667366020841923,
      "learning_rate": 3.0831038701927394e-06,
      "loss": 0.8999,
      "step": 4689
    },
    {
      "epoch": 0.3383716316150211,
      "grad_norm": 6.861171617468512,
      "learning_rate": 3.0827109465153597e-06,
      "loss": 0.8678,
      "step": 4690
    },
    {
      "epoch": 0.3384437790844486,
      "grad_norm": 4.070558039019152,
      "learning_rate": 3.0823179637151437e-06,
      "loss": 1.0301,
      "step": 4691
    },
    {
      "epoch": 0.33851592655387613,
      "grad_norm": 4.456008994499182,
      "learning_rate": 3.0819249218135495e-06,
      "loss": 0.9453,
      "step": 4692
    },
    {
      "epoch": 0.33858807402330365,
      "grad_norm": 8.232553704089797,
      "learning_rate": 3.081531820832041e-06,
      "loss": 1.0279,
      "step": 4693
    },
    {
      "epoch": 0.3386602214927311,
      "grad_norm": 5.8486079458949165,
      "learning_rate": 3.081138660792084e-06,
      "loss": 1.0506,
      "step": 4694
    },
    {
      "epoch": 0.33873236896215864,
      "grad_norm": 0.7394304162199533,
      "learning_rate": 3.080745441715146e-06,
      "loss": 0.8741,
      "step": 4695
    },
    {
      "epoch": 0.33880451643158616,
      "grad_norm": 8.087780964376718,
      "learning_rate": 3.0803521636227005e-06,
      "loss": 0.9785,
      "step": 4696
    },
    {
      "epoch": 0.3388766639010137,
      "grad_norm": 4.561045166501232,
      "learning_rate": 3.0799588265362228e-06,
      "loss": 0.9615,
      "step": 4697
    },
    {
      "epoch": 0.3389488113704412,
      "grad_norm": 7.823754344913991,
      "learning_rate": 3.079565430477192e-06,
      "loss": 0.97,
      "step": 4698
    },
    {
      "epoch": 0.3390209588398687,
      "grad_norm": 7.192275608607875,
      "learning_rate": 3.079171975467089e-06,
      "loss": 1.0252,
      "step": 4699
    },
    {
      "epoch": 0.3390931063092962,
      "grad_norm": 9.240454620142565,
      "learning_rate": 3.078778461527399e-06,
      "loss": 0.9399,
      "step": 4700
    },
    {
      "epoch": 0.3391652537787237,
      "grad_norm": 0.6984511056210614,
      "learning_rate": 3.078384888679611e-06,
      "loss": 0.8469,
      "step": 4701
    },
    {
      "epoch": 0.3392374012481512,
      "grad_norm": 0.7569336268228487,
      "learning_rate": 3.0779912569452164e-06,
      "loss": 0.8819,
      "step": 4702
    },
    {
      "epoch": 0.33930954871757874,
      "grad_norm": 24.357568386341697,
      "learning_rate": 3.0775975663457096e-06,
      "loss": 0.8952,
      "step": 4703
    },
    {
      "epoch": 0.33938169618700625,
      "grad_norm": 5.094611159471441,
      "learning_rate": 3.077203816902589e-06,
      "loss": 0.9078,
      "step": 4704
    },
    {
      "epoch": 0.3394538436564338,
      "grad_norm": 5.560269883830305,
      "learning_rate": 3.0768100086373557e-06,
      "loss": 1.0484,
      "step": 4705
    },
    {
      "epoch": 0.33952599112586124,
      "grad_norm": 5.9060291067324,
      "learning_rate": 3.0764161415715133e-06,
      "loss": 1.0321,
      "step": 4706
    },
    {
      "epoch": 0.33959813859528876,
      "grad_norm": 6.973749257819462,
      "learning_rate": 3.0760222157265705e-06,
      "loss": 0.9431,
      "step": 4707
    },
    {
      "epoch": 0.3396702860647163,
      "grad_norm": 6.878630078570333,
      "learning_rate": 3.075628231124039e-06,
      "loss": 1.0707,
      "step": 4708
    },
    {
      "epoch": 0.3397424335341438,
      "grad_norm": 8.798977046811322,
      "learning_rate": 3.0752341877854306e-06,
      "loss": 1.0885,
      "step": 4709
    },
    {
      "epoch": 0.3398145810035713,
      "grad_norm": 8.286269023722081,
      "learning_rate": 3.074840085732264e-06,
      "loss": 0.9095,
      "step": 4710
    },
    {
      "epoch": 0.33988672847299883,
      "grad_norm": 7.821390693109991,
      "learning_rate": 3.074445924986059e-06,
      "loss": 0.9372,
      "step": 4711
    },
    {
      "epoch": 0.3399588759424263,
      "grad_norm": 3.240640367638872,
      "learning_rate": 3.074051705568339e-06,
      "loss": 1.0514,
      "step": 4712
    },
    {
      "epoch": 0.3400310234118538,
      "grad_norm": 4.8676528340833,
      "learning_rate": 3.0736574275006318e-06,
      "loss": 1.1115,
      "step": 4713
    },
    {
      "epoch": 0.34010317088128134,
      "grad_norm": 4.6207276277459215,
      "learning_rate": 3.073263090804468e-06,
      "loss": 1.0039,
      "step": 4714
    },
    {
      "epoch": 0.34017531835070886,
      "grad_norm": 4.370800073216492,
      "learning_rate": 3.072868695501378e-06,
      "loss": 1.0006,
      "step": 4715
    },
    {
      "epoch": 0.3402474658201364,
      "grad_norm": 7.912197122908881,
      "learning_rate": 3.0724742416129015e-06,
      "loss": 0.9957,
      "step": 4716
    },
    {
      "epoch": 0.3403196132895639,
      "grad_norm": 0.8349271354748729,
      "learning_rate": 3.0720797291605757e-06,
      "loss": 0.9202,
      "step": 4717
    },
    {
      "epoch": 0.34039176075899136,
      "grad_norm": 4.800801710887541,
      "learning_rate": 3.0716851581659455e-06,
      "loss": 1.0285,
      "step": 4718
    },
    {
      "epoch": 0.3404639082284189,
      "grad_norm": 9.629965950181763,
      "learning_rate": 3.0712905286505565e-06,
      "loss": 0.8984,
      "step": 4719
    },
    {
      "epoch": 0.3405360556978464,
      "grad_norm": 5.641000328362855,
      "learning_rate": 3.0708958406359564e-06,
      "loss": 1.0856,
      "step": 4720
    },
    {
      "epoch": 0.3406082031672739,
      "grad_norm": 0.6941760773801272,
      "learning_rate": 3.070501094143699e-06,
      "loss": 0.8672,
      "step": 4721
    },
    {
      "epoch": 0.34068035063670143,
      "grad_norm": 5.134989237713702,
      "learning_rate": 3.0701062891953405e-06,
      "loss": 1.0202,
      "step": 4722
    },
    {
      "epoch": 0.34075249810612895,
      "grad_norm": 0.6861234886345989,
      "learning_rate": 3.0697114258124383e-06,
      "loss": 0.8061,
      "step": 4723
    },
    {
      "epoch": 0.3408246455755564,
      "grad_norm": 5.6822680208557435,
      "learning_rate": 3.069316504016555e-06,
      "loss": 0.973,
      "step": 4724
    },
    {
      "epoch": 0.34089679304498394,
      "grad_norm": 4.96284777289892,
      "learning_rate": 3.068921523829257e-06,
      "loss": 1.0539,
      "step": 4725
    },
    {
      "epoch": 0.34096894051441146,
      "grad_norm": 4.162979987252831,
      "learning_rate": 3.0685264852721104e-06,
      "loss": 1.023,
      "step": 4726
    },
    {
      "epoch": 0.341041087983839,
      "grad_norm": 5.509789511092322,
      "learning_rate": 3.068131388366688e-06,
      "loss": 0.9476,
      "step": 4727
    },
    {
      "epoch": 0.3411132354532665,
      "grad_norm": 4.423521133111934,
      "learning_rate": 3.067736233134565e-06,
      "loss": 0.9736,
      "step": 4728
    },
    {
      "epoch": 0.341185382922694,
      "grad_norm": 2.921393206852335,
      "learning_rate": 3.067341019597318e-06,
      "loss": 0.9276,
      "step": 4729
    },
    {
      "epoch": 0.3412575303921215,
      "grad_norm": 9.960564484437256,
      "learning_rate": 3.0669457477765304e-06,
      "loss": 0.939,
      "step": 4730
    },
    {
      "epoch": 0.341329677861549,
      "grad_norm": 5.08311935141191,
      "learning_rate": 3.066550417693785e-06,
      "loss": 0.9379,
      "step": 4731
    },
    {
      "epoch": 0.3414018253309765,
      "grad_norm": 0.8560095344975047,
      "learning_rate": 3.0661550293706686e-06,
      "loss": 0.8846,
      "step": 4732
    },
    {
      "epoch": 0.34147397280040404,
      "grad_norm": 5.841882743956834,
      "learning_rate": 3.0657595828287735e-06,
      "loss": 1.0548,
      "step": 4733
    },
    {
      "epoch": 0.34154612026983155,
      "grad_norm": 6.813609461604303,
      "learning_rate": 3.065364078089693e-06,
      "loss": 0.9966,
      "step": 4734
    },
    {
      "epoch": 0.341618267739259,
      "grad_norm": 5.6758440910793375,
      "learning_rate": 3.064968515175024e-06,
      "loss": 0.8388,
      "step": 4735
    },
    {
      "epoch": 0.34169041520868654,
      "grad_norm": 14.448163210803497,
      "learning_rate": 3.064572894106366e-06,
      "loss": 1.0293,
      "step": 4736
    },
    {
      "epoch": 0.34176256267811406,
      "grad_norm": 4.394256067773037,
      "learning_rate": 3.0641772149053244e-06,
      "loss": 0.9749,
      "step": 4737
    },
    {
      "epoch": 0.3418347101475416,
      "grad_norm": 4.2745388920043546,
      "learning_rate": 3.063781477593504e-06,
      "loss": 1.0679,
      "step": 4738
    },
    {
      "epoch": 0.3419068576169691,
      "grad_norm": 5.600995533872437,
      "learning_rate": 3.0633856821925146e-06,
      "loss": 1.0157,
      "step": 4739
    },
    {
      "epoch": 0.3419790050863966,
      "grad_norm": 4.1851278389896915,
      "learning_rate": 3.06298982872397e-06,
      "loss": 1.0089,
      "step": 4740
    },
    {
      "epoch": 0.3420511525558241,
      "grad_norm": 6.517765416285593,
      "learning_rate": 3.062593917209486e-06,
      "loss": 0.9442,
      "step": 4741
    },
    {
      "epoch": 0.3421233000252516,
      "grad_norm": 7.304341778249191,
      "learning_rate": 3.0621979476706815e-06,
      "loss": 0.9443,
      "step": 4742
    },
    {
      "epoch": 0.3421954474946791,
      "grad_norm": 4.508309850907764,
      "learning_rate": 3.0618019201291795e-06,
      "loss": 0.9532,
      "step": 4743
    },
    {
      "epoch": 0.34226759496410664,
      "grad_norm": 7.2204254980270095,
      "learning_rate": 3.0614058346066046e-06,
      "loss": 0.993,
      "step": 4744
    },
    {
      "epoch": 0.34233974243353416,
      "grad_norm": 0.7319962956277472,
      "learning_rate": 3.0610096911245872e-06,
      "loss": 0.8394,
      "step": 4745
    },
    {
      "epoch": 0.3424118899029617,
      "grad_norm": 8.713886394776685,
      "learning_rate": 3.0606134897047586e-06,
      "loss": 0.8949,
      "step": 4746
    },
    {
      "epoch": 0.34248403737238914,
      "grad_norm": 4.260911510179835,
      "learning_rate": 3.060217230368752e-06,
      "loss": 1.0435,
      "step": 4747
    },
    {
      "epoch": 0.34255618484181666,
      "grad_norm": 12.32976796621031,
      "learning_rate": 3.0598209131382085e-06,
      "loss": 1.0821,
      "step": 4748
    },
    {
      "epoch": 0.3426283323112442,
      "grad_norm": 6.237163637070596,
      "learning_rate": 3.059424538034768e-06,
      "loss": 0.9863,
      "step": 4749
    },
    {
      "epoch": 0.3427004797806717,
      "grad_norm": 4.495049402561636,
      "learning_rate": 3.059028105080075e-06,
      "loss": 1.0397,
      "step": 4750
    },
    {
      "epoch": 0.3427726272500992,
      "grad_norm": 6.719261256438493,
      "learning_rate": 3.058631614295778e-06,
      "loss": 0.9985,
      "step": 4751
    },
    {
      "epoch": 0.34284477471952673,
      "grad_norm": 4.599437281148929,
      "learning_rate": 3.0582350657035276e-06,
      "loss": 0.9892,
      "step": 4752
    },
    {
      "epoch": 0.3429169221889542,
      "grad_norm": 6.023780743705915,
      "learning_rate": 3.0578384593249774e-06,
      "loss": 1.0631,
      "step": 4753
    },
    {
      "epoch": 0.3429890696583817,
      "grad_norm": 3.578033979166058,
      "learning_rate": 3.057441795181785e-06,
      "loss": 0.8978,
      "step": 4754
    },
    {
      "epoch": 0.34306121712780924,
      "grad_norm": 4.687585855333537,
      "learning_rate": 3.05704507329561e-06,
      "loss": 0.9641,
      "step": 4755
    },
    {
      "epoch": 0.34313336459723676,
      "grad_norm": 8.972473009801762,
      "learning_rate": 3.056648293688117e-06,
      "loss": 0.9919,
      "step": 4756
    },
    {
      "epoch": 0.3432055120666643,
      "grad_norm": 4.71130086278749,
      "learning_rate": 3.056251456380973e-06,
      "loss": 0.9492,
      "step": 4757
    },
    {
      "epoch": 0.3432776595360918,
      "grad_norm": 5.53207684387545,
      "learning_rate": 3.055854561395846e-06,
      "loss": 0.9978,
      "step": 4758
    },
    {
      "epoch": 0.34334980700551926,
      "grad_norm": 3.471873071706049,
      "learning_rate": 3.055457608754411e-06,
      "loss": 0.9964,
      "step": 4759
    },
    {
      "epoch": 0.3434219544749468,
      "grad_norm": 7.027930802935499,
      "learning_rate": 3.0550605984783426e-06,
      "loss": 0.9675,
      "step": 4760
    },
    {
      "epoch": 0.3434941019443743,
      "grad_norm": 5.6207142079452375,
      "learning_rate": 3.054663530589321e-06,
      "loss": 0.9647,
      "step": 4761
    },
    {
      "epoch": 0.3435662494138018,
      "grad_norm": 7.423889529228429,
      "learning_rate": 3.054266405109028e-06,
      "loss": 0.8979,
      "step": 4762
    },
    {
      "epoch": 0.34363839688322934,
      "grad_norm": 3.9204270322937464,
      "learning_rate": 3.053869222059149e-06,
      "loss": 1.0295,
      "step": 4763
    },
    {
      "epoch": 0.34371054435265685,
      "grad_norm": 5.618542164320393,
      "learning_rate": 3.0534719814613737e-06,
      "loss": 1.0089,
      "step": 4764
    },
    {
      "epoch": 0.3437826918220843,
      "grad_norm": 9.183635703461169,
      "learning_rate": 3.0530746833373933e-06,
      "loss": 1.0314,
      "step": 4765
    },
    {
      "epoch": 0.34385483929151184,
      "grad_norm": 4.91719767833392,
      "learning_rate": 3.0526773277089022e-06,
      "loss": 0.9747,
      "step": 4766
    },
    {
      "epoch": 0.34392698676093936,
      "grad_norm": 4.973827910252882,
      "learning_rate": 3.0522799145975993e-06,
      "loss": 1.001,
      "step": 4767
    },
    {
      "epoch": 0.3439991342303669,
      "grad_norm": 5.879824990503145,
      "learning_rate": 3.051882444025186e-06,
      "loss": 1.048,
      "step": 4768
    },
    {
      "epoch": 0.3440712816997944,
      "grad_norm": 8.502007640030254,
      "learning_rate": 3.051484916013366e-06,
      "loss": 0.9224,
      "step": 4769
    },
    {
      "epoch": 0.3441434291692219,
      "grad_norm": 14.893569767190009,
      "learning_rate": 3.051087330583847e-06,
      "loss": 0.9101,
      "step": 4770
    },
    {
      "epoch": 0.3442155766386494,
      "grad_norm": 5.943505392245913,
      "learning_rate": 3.0506896877583404e-06,
      "loss": 1.026,
      "step": 4771
    },
    {
      "epoch": 0.3442877241080769,
      "grad_norm": 3.7954793492120875,
      "learning_rate": 3.0502919875585595e-06,
      "loss": 0.9452,
      "step": 4772
    },
    {
      "epoch": 0.3443598715775044,
      "grad_norm": 8.39460259465378,
      "learning_rate": 3.0498942300062213e-06,
      "loss": 1.0123,
      "step": 4773
    },
    {
      "epoch": 0.34443201904693194,
      "grad_norm": 3.996059861323802,
      "learning_rate": 3.0494964151230455e-06,
      "loss": 1.0021,
      "step": 4774
    },
    {
      "epoch": 0.34450416651635946,
      "grad_norm": 4.684680548558355,
      "learning_rate": 3.049098542930755e-06,
      "loss": 1.0307,
      "step": 4775
    },
    {
      "epoch": 0.344576313985787,
      "grad_norm": 7.384018895316203,
      "learning_rate": 3.0487006134510774e-06,
      "loss": 1.0232,
      "step": 4776
    },
    {
      "epoch": 0.34464846145521444,
      "grad_norm": 6.724233261700822,
      "learning_rate": 3.048302626705742e-06,
      "loss": 0.9036,
      "step": 4777
    },
    {
      "epoch": 0.34472060892464196,
      "grad_norm": 5.660207133462624,
      "learning_rate": 3.0479045827164797e-06,
      "loss": 0.9254,
      "step": 4778
    },
    {
      "epoch": 0.3447927563940695,
      "grad_norm": 4.652376137980252,
      "learning_rate": 3.047506481505028e-06,
      "loss": 0.9017,
      "step": 4779
    },
    {
      "epoch": 0.344864903863497,
      "grad_norm": 5.762002111676195,
      "learning_rate": 3.047108323093125e-06,
      "loss": 0.9448,
      "step": 4780
    },
    {
      "epoch": 0.3449370513329245,
      "grad_norm": 4.118527507416419,
      "learning_rate": 3.0467101075025124e-06,
      "loss": 0.9726,
      "step": 4781
    },
    {
      "epoch": 0.34500919880235204,
      "grad_norm": 4.576462167022308,
      "learning_rate": 3.0463118347549354e-06,
      "loss": 0.976,
      "step": 4782
    },
    {
      "epoch": 0.3450813462717795,
      "grad_norm": 4.343136750986383,
      "learning_rate": 3.045913504872143e-06,
      "loss": 0.9887,
      "step": 4783
    },
    {
      "epoch": 0.345153493741207,
      "grad_norm": 4.775491675183415,
      "learning_rate": 3.045515117875886e-06,
      "loss": 0.9111,
      "step": 4784
    },
    {
      "epoch": 0.34522564121063454,
      "grad_norm": 11.017912758335536,
      "learning_rate": 3.0451166737879186e-06,
      "loss": 1.0026,
      "step": 4785
    },
    {
      "epoch": 0.34529778868006206,
      "grad_norm": 7.130679627459285,
      "learning_rate": 3.0447181726299987e-06,
      "loss": 1.0333,
      "step": 4786
    },
    {
      "epoch": 0.3453699361494896,
      "grad_norm": 3.416271481493547,
      "learning_rate": 3.0443196144238866e-06,
      "loss": 1.0258,
      "step": 4787
    },
    {
      "epoch": 0.34544208361891704,
      "grad_norm": 3.6138391672307897,
      "learning_rate": 3.0439209991913463e-06,
      "loss": 0.9891,
      "step": 4788
    },
    {
      "epoch": 0.34551423108834456,
      "grad_norm": 4.266661577420378,
      "learning_rate": 3.043522326954145e-06,
      "loss": 1.0014,
      "step": 4789
    },
    {
      "epoch": 0.3455863785577721,
      "grad_norm": 5.9059086453728895,
      "learning_rate": 3.043123597734052e-06,
      "loss": 1.0678,
      "step": 4790
    },
    {
      "epoch": 0.3456585260271996,
      "grad_norm": 6.650107258096628,
      "learning_rate": 3.042724811552841e-06,
      "loss": 1.0399,
      "step": 4791
    },
    {
      "epoch": 0.3457306734966271,
      "grad_norm": 5.569825567120081,
      "learning_rate": 3.0423259684322878e-06,
      "loss": 0.9759,
      "step": 4792
    },
    {
      "epoch": 0.34580282096605464,
      "grad_norm": 6.381990078594705,
      "learning_rate": 3.0419270683941723e-06,
      "loss": 0.8989,
      "step": 4793
    },
    {
      "epoch": 0.3458749684354821,
      "grad_norm": 7.24330224405151,
      "learning_rate": 3.0415281114602765e-06,
      "loss": 1.0224,
      "step": 4794
    },
    {
      "epoch": 0.3459471159049096,
      "grad_norm": 13.43423152473398,
      "learning_rate": 3.041129097652386e-06,
      "loss": 0.9679,
      "step": 4795
    },
    {
      "epoch": 0.34601926337433714,
      "grad_norm": 3.429234224923221,
      "learning_rate": 3.0407300269922904e-06,
      "loss": 0.9647,
      "step": 4796
    },
    {
      "epoch": 0.34609141084376466,
      "grad_norm": 4.387188527063439,
      "learning_rate": 3.04033089950178e-06,
      "loss": 0.9137,
      "step": 4797
    },
    {
      "epoch": 0.3461635583131922,
      "grad_norm": 7.001818556935623,
      "learning_rate": 3.0399317152026504e-06,
      "loss": 0.9952,
      "step": 4798
    },
    {
      "epoch": 0.3462357057826197,
      "grad_norm": 18.22327964836676,
      "learning_rate": 3.0395324741166997e-06,
      "loss": 1.0733,
      "step": 4799
    },
    {
      "epoch": 0.34630785325204716,
      "grad_norm": 4.103725480637633,
      "learning_rate": 3.0391331762657285e-06,
      "loss": 0.961,
      "step": 4800
    },
    {
      "epoch": 0.3463800007214747,
      "grad_norm": 7.109121196534191,
      "learning_rate": 3.038733821671542e-06,
      "loss": 0.9566,
      "step": 4801
    },
    {
      "epoch": 0.3464521481909022,
      "grad_norm": 0.6619955994742116,
      "learning_rate": 3.0383344103559464e-06,
      "loss": 0.8632,
      "step": 4802
    },
    {
      "epoch": 0.3465242956603297,
      "grad_norm": 4.777324186770042,
      "learning_rate": 3.037934942340752e-06,
      "loss": 1.0524,
      "step": 4803
    },
    {
      "epoch": 0.34659644312975724,
      "grad_norm": 4.018018669635697,
      "learning_rate": 3.0375354176477733e-06,
      "loss": 0.9622,
      "step": 4804
    },
    {
      "epoch": 0.34666859059918476,
      "grad_norm": 0.8011529571516719,
      "learning_rate": 3.0371358362988257e-06,
      "loss": 0.9046,
      "step": 4805
    },
    {
      "epoch": 0.3467407380686122,
      "grad_norm": 5.258863505595425,
      "learning_rate": 3.0367361983157297e-06,
      "loss": 0.9901,
      "step": 4806
    },
    {
      "epoch": 0.34681288553803974,
      "grad_norm": 10.098650141488008,
      "learning_rate": 3.0363365037203084e-06,
      "loss": 0.8297,
      "step": 4807
    },
    {
      "epoch": 0.34688503300746726,
      "grad_norm": 4.7117186447495225,
      "learning_rate": 3.0359367525343862e-06,
      "loss": 0.9343,
      "step": 4808
    },
    {
      "epoch": 0.3469571804768948,
      "grad_norm": 4.470536308459412,
      "learning_rate": 3.0355369447797927e-06,
      "loss": 1.0096,
      "step": 4809
    },
    {
      "epoch": 0.3470293279463223,
      "grad_norm": 3.707438728922022,
      "learning_rate": 3.0351370804783604e-06,
      "loss": 0.9906,
      "step": 4810
    },
    {
      "epoch": 0.3471014754157498,
      "grad_norm": 7.740155766617614,
      "learning_rate": 3.034737159651924e-06,
      "loss": 0.9077,
      "step": 4811
    },
    {
      "epoch": 0.3471736228851773,
      "grad_norm": 13.604132394713403,
      "learning_rate": 3.034337182322322e-06,
      "loss": 0.9986,
      "step": 4812
    },
    {
      "epoch": 0.3472457703546048,
      "grad_norm": 6.8779606773093205,
      "learning_rate": 3.0339371485113957e-06,
      "loss": 1.0395,
      "step": 4813
    },
    {
      "epoch": 0.3473179178240323,
      "grad_norm": 6.692297809839752,
      "learning_rate": 3.0335370582409887e-06,
      "loss": 0.8621,
      "step": 4814
    },
    {
      "epoch": 0.34739006529345984,
      "grad_norm": 3.359962163666724,
      "learning_rate": 3.0331369115329493e-06,
      "loss": 0.942,
      "step": 4815
    },
    {
      "epoch": 0.34746221276288736,
      "grad_norm": 3.311087541119391,
      "learning_rate": 3.0327367084091275e-06,
      "loss": 0.96,
      "step": 4816
    },
    {
      "epoch": 0.3475343602323149,
      "grad_norm": 37.04895410129684,
      "learning_rate": 3.032336448891377e-06,
      "loss": 1.02,
      "step": 4817
    },
    {
      "epoch": 0.34760650770174234,
      "grad_norm": 6.718079147558574,
      "learning_rate": 3.0319361330015554e-06,
      "loss": 1.0497,
      "step": 4818
    },
    {
      "epoch": 0.34767865517116986,
      "grad_norm": 5.055439487750461,
      "learning_rate": 3.0315357607615205e-06,
      "loss": 0.9129,
      "step": 4819
    },
    {
      "epoch": 0.3477508026405974,
      "grad_norm": 3.967447983437431,
      "learning_rate": 3.0311353321931366e-06,
      "loss": 1.054,
      "step": 4820
    },
    {
      "epoch": 0.3478229501100249,
      "grad_norm": 4.102909934922047,
      "learning_rate": 3.03073484731827e-06,
      "loss": 1.0327,
      "step": 4821
    },
    {
      "epoch": 0.3478950975794524,
      "grad_norm": 5.3788032383359585,
      "learning_rate": 3.0303343061587884e-06,
      "loss": 1.0445,
      "step": 4822
    },
    {
      "epoch": 0.34796724504887994,
      "grad_norm": 4.1762146970141085,
      "learning_rate": 3.0299337087365645e-06,
      "loss": 0.9249,
      "step": 4823
    },
    {
      "epoch": 0.3480393925183074,
      "grad_norm": 5.334974791334062,
      "learning_rate": 3.0295330550734743e-06,
      "loss": 0.9812,
      "step": 4824
    },
    {
      "epoch": 0.3481115399877349,
      "grad_norm": 4.436441429567553,
      "learning_rate": 3.029132345191394e-06,
      "loss": 0.9618,
      "step": 4825
    },
    {
      "epoch": 0.34818368745716244,
      "grad_norm": 4.600370864845161,
      "learning_rate": 3.0287315791122075e-06,
      "loss": 1.0491,
      "step": 4826
    },
    {
      "epoch": 0.34825583492658996,
      "grad_norm": 11.591114824342556,
      "learning_rate": 3.0283307568577963e-06,
      "loss": 0.9229,
      "step": 4827
    },
    {
      "epoch": 0.3483279823960175,
      "grad_norm": 5.712149724926726,
      "learning_rate": 3.02792987845005e-06,
      "loss": 0.9654,
      "step": 4828
    },
    {
      "epoch": 0.348400129865445,
      "grad_norm": 0.782969527951206,
      "learning_rate": 3.027528943910858e-06,
      "loss": 0.9349,
      "step": 4829
    },
    {
      "epoch": 0.34847227733487246,
      "grad_norm": 5.97211110014136,
      "learning_rate": 3.027127953262114e-06,
      "loss": 0.9632,
      "step": 4830
    },
    {
      "epoch": 0.3485444248043,
      "grad_norm": 0.7110965676567231,
      "learning_rate": 3.026726906525715e-06,
      "loss": 0.8647,
      "step": 4831
    },
    {
      "epoch": 0.3486165722737275,
      "grad_norm": 8.897001262769752,
      "learning_rate": 3.0263258037235604e-06,
      "loss": 1.0245,
      "step": 4832
    },
    {
      "epoch": 0.348688719743155,
      "grad_norm": 4.80304909660412,
      "learning_rate": 3.025924644877553e-06,
      "loss": 0.9891,
      "step": 4833
    },
    {
      "epoch": 0.34876086721258254,
      "grad_norm": 10.944378068941056,
      "learning_rate": 3.0255234300095983e-06,
      "loss": 1.0564,
      "step": 4834
    },
    {
      "epoch": 0.34883301468201,
      "grad_norm": 9.485925789607332,
      "learning_rate": 3.025122159141606e-06,
      "loss": 1.0027,
      "step": 4835
    },
    {
      "epoch": 0.3489051621514375,
      "grad_norm": 7.3218188776591635,
      "learning_rate": 3.024720832295487e-06,
      "loss": 0.9273,
      "step": 4836
    },
    {
      "epoch": 0.34897730962086504,
      "grad_norm": 3.5689634616699095,
      "learning_rate": 3.0243194494931567e-06,
      "loss": 1.1078,
      "step": 4837
    },
    {
      "epoch": 0.34904945709029256,
      "grad_norm": 0.7852878744907141,
      "learning_rate": 3.0239180107565337e-06,
      "loss": 0.8612,
      "step": 4838
    },
    {
      "epoch": 0.3491216045597201,
      "grad_norm": 4.570815151645568,
      "learning_rate": 3.023516516107538e-06,
      "loss": 0.9663,
      "step": 4839
    },
    {
      "epoch": 0.3491937520291476,
      "grad_norm": 12.832716163228133,
      "learning_rate": 3.0231149655680947e-06,
      "loss": 0.9762,
      "step": 4840
    },
    {
      "epoch": 0.34926589949857506,
      "grad_norm": 5.270132608790162,
      "learning_rate": 3.0227133591601305e-06,
      "loss": 0.9819,
      "step": 4841
    },
    {
      "epoch": 0.3493380469680026,
      "grad_norm": 4.446198869507192,
      "learning_rate": 3.022311696905576e-06,
      "loss": 0.9056,
      "step": 4842
    },
    {
      "epoch": 0.3494101944374301,
      "grad_norm": 4.667526211371276,
      "learning_rate": 3.0219099788263634e-06,
      "loss": 0.9175,
      "step": 4843
    },
    {
      "epoch": 0.3494823419068576,
      "grad_norm": 10.91113766340904,
      "learning_rate": 3.02150820494443e-06,
      "loss": 1.0271,
      "step": 4844
    },
    {
      "epoch": 0.34955448937628514,
      "grad_norm": 3.2274656348020385,
      "learning_rate": 3.0211063752817152e-06,
      "loss": 0.9643,
      "step": 4845
    },
    {
      "epoch": 0.34962663684571266,
      "grad_norm": 6.648246717573334,
      "learning_rate": 3.0207044898601613e-06,
      "loss": 1.0182,
      "step": 4846
    },
    {
      "epoch": 0.3496987843151401,
      "grad_norm": 5.044654759370095,
      "learning_rate": 3.0203025487017143e-06,
      "loss": 1.0118,
      "step": 4847
    },
    {
      "epoch": 0.34977093178456764,
      "grad_norm": 7.673985457075753,
      "learning_rate": 3.0199005518283212e-06,
      "loss": 0.9775,
      "step": 4848
    },
    {
      "epoch": 0.34984307925399516,
      "grad_norm": 9.313902064520567,
      "learning_rate": 3.0194984992619346e-06,
      "loss": 0.924,
      "step": 4849
    },
    {
      "epoch": 0.3499152267234227,
      "grad_norm": 5.206274963538871,
      "learning_rate": 3.01909639102451e-06,
      "loss": 0.9728,
      "step": 4850
    },
    {
      "epoch": 0.3499873741928502,
      "grad_norm": 0.6551638197421754,
      "learning_rate": 3.0186942271380028e-06,
      "loss": 0.8139,
      "step": 4851
    },
    {
      "epoch": 0.3500595216622777,
      "grad_norm": 3.3782636615762343,
      "learning_rate": 3.0182920076243754e-06,
      "loss": 0.9645,
      "step": 4852
    },
    {
      "epoch": 0.3501316691317052,
      "grad_norm": 5.293270538848031,
      "learning_rate": 3.0178897325055914e-06,
      "loss": 0.9413,
      "step": 4853
    },
    {
      "epoch": 0.3502038166011327,
      "grad_norm": 3.5516788703593036,
      "learning_rate": 3.017487401803617e-06,
      "loss": 0.9993,
      "step": 4854
    },
    {
      "epoch": 0.3502759640705602,
      "grad_norm": 3.5531006341667433,
      "learning_rate": 3.017085015540422e-06,
      "loss": 0.9807,
      "step": 4855
    },
    {
      "epoch": 0.35034811153998774,
      "grad_norm": 0.691364761228446,
      "learning_rate": 3.0166825737379796e-06,
      "loss": 0.8605,
      "step": 4856
    },
    {
      "epoch": 0.35042025900941526,
      "grad_norm": 4.029032725830177,
      "learning_rate": 3.0162800764182653e-06,
      "loss": 0.9548,
      "step": 4857
    },
    {
      "epoch": 0.3504924064788428,
      "grad_norm": 8.711763378730634,
      "learning_rate": 3.0158775236032587e-06,
      "loss": 0.9411,
      "step": 4858
    },
    {
      "epoch": 0.35056455394827024,
      "grad_norm": 4.219808036090054,
      "learning_rate": 3.0154749153149403e-06,
      "loss": 0.8959,
      "step": 4859
    },
    {
      "epoch": 0.35063670141769776,
      "grad_norm": 4.628336965586732,
      "learning_rate": 3.0150722515752963e-06,
      "loss": 1.0151,
      "step": 4860
    },
    {
      "epoch": 0.3507088488871253,
      "grad_norm": 5.045960901581391,
      "learning_rate": 3.014669532406315e-06,
      "loss": 0.9809,
      "step": 4861
    },
    {
      "epoch": 0.3507809963565528,
      "grad_norm": 0.676527262027709,
      "learning_rate": 3.014266757829986e-06,
      "loss": 0.8026,
      "step": 4862
    },
    {
      "epoch": 0.3508531438259803,
      "grad_norm": 8.37856296645269,
      "learning_rate": 3.0138639278683047e-06,
      "loss": 0.9722,
      "step": 4863
    },
    {
      "epoch": 0.35092529129540784,
      "grad_norm": 8.166010719996732,
      "learning_rate": 3.013461042543266e-06,
      "loss": 0.9972,
      "step": 4864
    },
    {
      "epoch": 0.3509974387648353,
      "grad_norm": 5.299035995911135,
      "learning_rate": 3.013058101876873e-06,
      "loss": 0.9275,
      "step": 4865
    },
    {
      "epoch": 0.3510695862342628,
      "grad_norm": 4.7460274397373885,
      "learning_rate": 3.0126551058911266e-06,
      "loss": 1.0691,
      "step": 4866
    },
    {
      "epoch": 0.35114173370369034,
      "grad_norm": 0.7781961645958786,
      "learning_rate": 3.0122520546080336e-06,
      "loss": 0.8457,
      "step": 4867
    },
    {
      "epoch": 0.35121388117311786,
      "grad_norm": 4.762926081623437,
      "learning_rate": 3.0118489480496037e-06,
      "loss": 0.9575,
      "step": 4868
    },
    {
      "epoch": 0.3512860286425454,
      "grad_norm": 5.32659585885803,
      "learning_rate": 3.011445786237848e-06,
      "loss": 0.968,
      "step": 4869
    },
    {
      "epoch": 0.3513581761119729,
      "grad_norm": 9.992134052339464,
      "learning_rate": 3.011042569194782e-06,
      "loss": 0.9885,
      "step": 4870
    },
    {
      "epoch": 0.35143032358140036,
      "grad_norm": 5.12623358045998,
      "learning_rate": 3.010639296942424e-06,
      "loss": 0.9197,
      "step": 4871
    },
    {
      "epoch": 0.3515024710508279,
      "grad_norm": 8.390172168297376,
      "learning_rate": 3.0102359695027953e-06,
      "loss": 1.0298,
      "step": 4872
    },
    {
      "epoch": 0.3515746185202554,
      "grad_norm": 15.368203250301518,
      "learning_rate": 3.009832586897921e-06,
      "loss": 0.9939,
      "step": 4873
    },
    {
      "epoch": 0.3516467659896829,
      "grad_norm": 4.167104214900534,
      "learning_rate": 3.0094291491498267e-06,
      "loss": 0.9739,
      "step": 4874
    },
    {
      "epoch": 0.35171891345911044,
      "grad_norm": 7.054255441676732,
      "learning_rate": 3.009025656280543e-06,
      "loss": 0.9787,
      "step": 4875
    },
    {
      "epoch": 0.35179106092853796,
      "grad_norm": 6.50677489289709,
      "learning_rate": 3.0086221083121033e-06,
      "loss": 1.0542,
      "step": 4876
    },
    {
      "epoch": 0.3518632083979654,
      "grad_norm": 4.982781325847135,
      "learning_rate": 3.008218505266545e-06,
      "loss": 0.9894,
      "step": 4877
    },
    {
      "epoch": 0.35193535586739294,
      "grad_norm": 61.62631131338099,
      "learning_rate": 3.007814847165906e-06,
      "loss": 0.9792,
      "step": 4878
    },
    {
      "epoch": 0.35200750333682046,
      "grad_norm": 0.7711384444342446,
      "learning_rate": 3.007411134032229e-06,
      "loss": 0.8406,
      "step": 4879
    },
    {
      "epoch": 0.352079650806248,
      "grad_norm": 14.72160062297384,
      "learning_rate": 3.0070073658875593e-06,
      "loss": 0.8801,
      "step": 4880
    },
    {
      "epoch": 0.3521517982756755,
      "grad_norm": 3.6317848638937056,
      "learning_rate": 3.006603542753945e-06,
      "loss": 0.9915,
      "step": 4881
    },
    {
      "epoch": 0.352223945745103,
      "grad_norm": 4.846546178174488,
      "learning_rate": 3.0061996646534374e-06,
      "loss": 0.9703,
      "step": 4882
    },
    {
      "epoch": 0.3522960932145305,
      "grad_norm": 3.7379191032869192,
      "learning_rate": 3.0057957316080913e-06,
      "loss": 0.9207,
      "step": 4883
    },
    {
      "epoch": 0.352368240683958,
      "grad_norm": 5.332500015165791,
      "learning_rate": 3.005391743639964e-06,
      "loss": 0.9731,
      "step": 4884
    },
    {
      "epoch": 0.3524403881533855,
      "grad_norm": 8.847710575164346,
      "learning_rate": 3.0049877007711145e-06,
      "loss": 1.0407,
      "step": 4885
    },
    {
      "epoch": 0.35251253562281304,
      "grad_norm": 3.615696379303994,
      "learning_rate": 3.0045836030236076e-06,
      "loss": 0.8983,
      "step": 4886
    },
    {
      "epoch": 0.35258468309224056,
      "grad_norm": 2.9223000936281744,
      "learning_rate": 3.0041794504195085e-06,
      "loss": 0.9197,
      "step": 4887
    },
    {
      "epoch": 0.352656830561668,
      "grad_norm": 17.29661916872684,
      "learning_rate": 3.003775242980887e-06,
      "loss": 1.1069,
      "step": 4888
    },
    {
      "epoch": 0.35272897803109554,
      "grad_norm": 4.589743183738691,
      "learning_rate": 3.0033709807298164e-06,
      "loss": 0.9455,
      "step": 4889
    },
    {
      "epoch": 0.35280112550052306,
      "grad_norm": 3.4166314348094566,
      "learning_rate": 3.00296666368837e-06,
      "loss": 1.0463,
      "step": 4890
    },
    {
      "epoch": 0.3528732729699506,
      "grad_norm": 3.809906749835729,
      "learning_rate": 3.0025622918786272e-06,
      "loss": 0.8827,
      "step": 4891
    },
    {
      "epoch": 0.3529454204393781,
      "grad_norm": 5.207600636125622,
      "learning_rate": 3.002157865322669e-06,
      "loss": 1.0106,
      "step": 4892
    },
    {
      "epoch": 0.3530175679088056,
      "grad_norm": 5.2933388218182635,
      "learning_rate": 3.0017533840425787e-06,
      "loss": 0.9484,
      "step": 4893
    },
    {
      "epoch": 0.3530897153782331,
      "grad_norm": 5.736838495022215,
      "learning_rate": 3.0013488480604457e-06,
      "loss": 0.9022,
      "step": 4894
    },
    {
      "epoch": 0.3531618628476606,
      "grad_norm": 5.256215548770959,
      "learning_rate": 3.000944257398359e-06,
      "loss": 1.0177,
      "step": 4895
    },
    {
      "epoch": 0.3532340103170881,
      "grad_norm": 0.9363328662254152,
      "learning_rate": 3.0005396120784107e-06,
      "loss": 0.9203,
      "step": 4896
    },
    {
      "epoch": 0.35330615778651564,
      "grad_norm": 6.351427965979875,
      "learning_rate": 3.000134912122699e-06,
      "loss": 0.992,
      "step": 4897
    },
    {
      "epoch": 0.35337830525594316,
      "grad_norm": 7.668783586672579,
      "learning_rate": 2.999730157553321e-06,
      "loss": 0.8902,
      "step": 4898
    },
    {
      "epoch": 0.3534504527253707,
      "grad_norm": 8.964202897845965,
      "learning_rate": 2.999325348392381e-06,
      "loss": 0.9438,
      "step": 4899
    },
    {
      "epoch": 0.35352260019479814,
      "grad_norm": 0.7936338820339709,
      "learning_rate": 2.998920484661983e-06,
      "loss": 0.8844,
      "step": 4900
    },
    {
      "epoch": 0.35359474766422566,
      "grad_norm": 4.505544214094578,
      "learning_rate": 2.998515566384235e-06,
      "loss": 0.946,
      "step": 4901
    },
    {
      "epoch": 0.3536668951336532,
      "grad_norm": 13.506909333057852,
      "learning_rate": 2.9981105935812483e-06,
      "loss": 0.9653,
      "step": 4902
    },
    {
      "epoch": 0.3537390426030807,
      "grad_norm": 7.740107467654428,
      "learning_rate": 2.9977055662751372e-06,
      "loss": 1.0184,
      "step": 4903
    },
    {
      "epoch": 0.3538111900725082,
      "grad_norm": 6.246961541695128,
      "learning_rate": 2.9973004844880186e-06,
      "loss": 0.927,
      "step": 4904
    },
    {
      "epoch": 0.35388333754193574,
      "grad_norm": 5.040350413780888,
      "learning_rate": 2.996895348242012e-06,
      "loss": 0.8909,
      "step": 4905
    },
    {
      "epoch": 0.3539554850113632,
      "grad_norm": 9.4481210122236,
      "learning_rate": 2.996490157559241e-06,
      "loss": 1.0318,
      "step": 4906
    },
    {
      "epoch": 0.3540276324807907,
      "grad_norm": 8.666440178283672,
      "learning_rate": 2.9960849124618315e-06,
      "loss": 1.027,
      "step": 4907
    },
    {
      "epoch": 0.35409977995021824,
      "grad_norm": 4.721543855498383,
      "learning_rate": 2.9956796129719116e-06,
      "loss": 1.0496,
      "step": 4908
    },
    {
      "epoch": 0.35417192741964576,
      "grad_norm": 5.63258596274265,
      "learning_rate": 2.995274259111615e-06,
      "loss": 0.9812,
      "step": 4909
    },
    {
      "epoch": 0.3542440748890733,
      "grad_norm": 11.198016290739009,
      "learning_rate": 2.9948688509030745e-06,
      "loss": 0.9532,
      "step": 4910
    },
    {
      "epoch": 0.3543162223585008,
      "grad_norm": 6.769073382795456,
      "learning_rate": 2.9944633883684293e-06,
      "loss": 0.8095,
      "step": 4911
    },
    {
      "epoch": 0.35438836982792826,
      "grad_norm": 3.541003505477743,
      "learning_rate": 2.9940578715298204e-06,
      "loss": 1.0417,
      "step": 4912
    },
    {
      "epoch": 0.3544605172973558,
      "grad_norm": 7.561643189372841,
      "learning_rate": 2.99365230040939e-06,
      "loss": 1.0082,
      "step": 4913
    },
    {
      "epoch": 0.3545326647667833,
      "grad_norm": 6.350754652461526,
      "learning_rate": 2.9932466750292866e-06,
      "loss": 1.0098,
      "step": 4914
    },
    {
      "epoch": 0.3546048122362108,
      "grad_norm": 4.346945705428396,
      "learning_rate": 2.992840995411659e-06,
      "loss": 0.9581,
      "step": 4915
    },
    {
      "epoch": 0.35467695970563834,
      "grad_norm": 0.8302536004782983,
      "learning_rate": 2.9924352615786597e-06,
      "loss": 0.8663,
      "step": 4916
    },
    {
      "epoch": 0.35474910717506586,
      "grad_norm": 5.748413198699144,
      "learning_rate": 2.992029473552445e-06,
      "loss": 0.931,
      "step": 4917
    },
    {
      "epoch": 0.3548212546444933,
      "grad_norm": 3.5230119875406993,
      "learning_rate": 2.991623631355173e-06,
      "loss": 0.9841,
      "step": 4918
    },
    {
      "epoch": 0.35489340211392084,
      "grad_norm": 11.99729253107433,
      "learning_rate": 2.991217735009005e-06,
      "loss": 0.9569,
      "step": 4919
    },
    {
      "epoch": 0.35496554958334836,
      "grad_norm": 10.329748167530022,
      "learning_rate": 2.9908117845361064e-06,
      "loss": 1.0065,
      "step": 4920
    },
    {
      "epoch": 0.3550376970527759,
      "grad_norm": 4.60513058521788,
      "learning_rate": 2.990405779958644e-06,
      "loss": 1.0157,
      "step": 4921
    },
    {
      "epoch": 0.3551098445222034,
      "grad_norm": 10.730457373947374,
      "learning_rate": 2.9899997212987882e-06,
      "loss": 0.959,
      "step": 4922
    },
    {
      "epoch": 0.3551819919916309,
      "grad_norm": 7.107712771424043,
      "learning_rate": 2.9895936085787126e-06,
      "loss": 0.8618,
      "step": 4923
    },
    {
      "epoch": 0.3552541394610584,
      "grad_norm": 14.67929822112244,
      "learning_rate": 2.9891874418205934e-06,
      "loss": 1.0464,
      "step": 4924
    },
    {
      "epoch": 0.3553262869304859,
      "grad_norm": 5.0257640808087345,
      "learning_rate": 2.9887812210466097e-06,
      "loss": 0.9934,
      "step": 4925
    },
    {
      "epoch": 0.3553984343999134,
      "grad_norm": 5.104115990627208,
      "learning_rate": 2.988374946278943e-06,
      "loss": 1.0536,
      "step": 4926
    },
    {
      "epoch": 0.35547058186934094,
      "grad_norm": 3.845853276612608,
      "learning_rate": 2.987968617539781e-06,
      "loss": 0.8757,
      "step": 4927
    },
    {
      "epoch": 0.35554272933876846,
      "grad_norm": 7.426898998596356,
      "learning_rate": 2.9875622348513096e-06,
      "loss": 0.9463,
      "step": 4928
    },
    {
      "epoch": 0.355614876808196,
      "grad_norm": 6.237423564924182,
      "learning_rate": 2.9871557982357194e-06,
      "loss": 0.9358,
      "step": 4929
    },
    {
      "epoch": 0.35568702427762344,
      "grad_norm": 4.727652129373741,
      "learning_rate": 2.9867493077152066e-06,
      "loss": 0.9618,
      "step": 4930
    },
    {
      "epoch": 0.35575917174705096,
      "grad_norm": 3.7617214119684896,
      "learning_rate": 2.9863427633119664e-06,
      "loss": 1.0539,
      "step": 4931
    },
    {
      "epoch": 0.3558313192164785,
      "grad_norm": 5.690735954866279,
      "learning_rate": 2.9859361650481994e-06,
      "loss": 0.9293,
      "step": 4932
    },
    {
      "epoch": 0.355903466685906,
      "grad_norm": 0.7632448607470722,
      "learning_rate": 2.985529512946108e-06,
      "loss": 0.8159,
      "step": 4933
    },
    {
      "epoch": 0.3559756141553335,
      "grad_norm": 0.7818708241913035,
      "learning_rate": 2.985122807027898e-06,
      "loss": 0.8324,
      "step": 4934
    },
    {
      "epoch": 0.35604776162476104,
      "grad_norm": 3.195099541392527,
      "learning_rate": 2.984716047315779e-06,
      "loss": 0.959,
      "step": 4935
    },
    {
      "epoch": 0.3561199090941885,
      "grad_norm": 4.287894334814943,
      "learning_rate": 2.984309233831961e-06,
      "loss": 1.0255,
      "step": 4936
    },
    {
      "epoch": 0.356192056563616,
      "grad_norm": 4.3117661888414505,
      "learning_rate": 2.9839023665986603e-06,
      "loss": 0.9723,
      "step": 4937
    },
    {
      "epoch": 0.35626420403304354,
      "grad_norm": 4.181441293853749,
      "learning_rate": 2.9834954456380934e-06,
      "loss": 0.9566,
      "step": 4938
    },
    {
      "epoch": 0.35633635150247106,
      "grad_norm": 12.674127757139615,
      "learning_rate": 2.9830884709724812e-06,
      "loss": 1.0118,
      "step": 4939
    },
    {
      "epoch": 0.3564084989718986,
      "grad_norm": 8.336244557811192,
      "learning_rate": 2.9826814426240465e-06,
      "loss": 1.0009,
      "step": 4940
    },
    {
      "epoch": 0.35648064644132604,
      "grad_norm": 11.980689724282623,
      "learning_rate": 2.982274360615016e-06,
      "loss": 0.8714,
      "step": 4941
    },
    {
      "epoch": 0.35655279391075356,
      "grad_norm": 10.040036355366938,
      "learning_rate": 2.9818672249676194e-06,
      "loss": 0.9536,
      "step": 4942
    },
    {
      "epoch": 0.3566249413801811,
      "grad_norm": 3.178531220918095,
      "learning_rate": 2.9814600357040882e-06,
      "loss": 0.8818,
      "step": 4943
    },
    {
      "epoch": 0.3566970888496086,
      "grad_norm": 5.22725674019883,
      "learning_rate": 2.9810527928466577e-06,
      "loss": 1.0548,
      "step": 4944
    },
    {
      "epoch": 0.3567692363190361,
      "grad_norm": 3.7358955894681687,
      "learning_rate": 2.9806454964175654e-06,
      "loss": 1.0374,
      "step": 4945
    },
    {
      "epoch": 0.35684138378846364,
      "grad_norm": 3.7098045145442384,
      "learning_rate": 2.9802381464390536e-06,
      "loss": 0.9979,
      "step": 4946
    },
    {
      "epoch": 0.3569135312578911,
      "grad_norm": 0.7194902505697892,
      "learning_rate": 2.979830742933365e-06,
      "loss": 0.8365,
      "step": 4947
    },
    {
      "epoch": 0.3569856787273186,
      "grad_norm": 6.1476842474581685,
      "learning_rate": 2.9794232859227466e-06,
      "loss": 1.0548,
      "step": 4948
    },
    {
      "epoch": 0.35705782619674614,
      "grad_norm": 8.04299151587356,
      "learning_rate": 2.9790157754294477e-06,
      "loss": 1.0266,
      "step": 4949
    },
    {
      "epoch": 0.35712997366617366,
      "grad_norm": 5.8159179482529275,
      "learning_rate": 2.9786082114757225e-06,
      "loss": 0.9744,
      "step": 4950
    },
    {
      "epoch": 0.3572021211356012,
      "grad_norm": 3.9117048095308444,
      "learning_rate": 2.978200594083825e-06,
      "loss": 0.9429,
      "step": 4951
    },
    {
      "epoch": 0.3572742686050287,
      "grad_norm": 4.463573952746905,
      "learning_rate": 2.977792923276014e-06,
      "loss": 1.0623,
      "step": 4952
    },
    {
      "epoch": 0.35734641607445616,
      "grad_norm": 4.923240641705187,
      "learning_rate": 2.977385199074552e-06,
      "loss": 1.0492,
      "step": 4953
    },
    {
      "epoch": 0.3574185635438837,
      "grad_norm": 4.830801903793138,
      "learning_rate": 2.9769774215017016e-06,
      "loss": 1.044,
      "step": 4954
    },
    {
      "epoch": 0.3574907110133112,
      "grad_norm": 5.964719360093795,
      "learning_rate": 2.9765695905797314e-06,
      "loss": 0.8919,
      "step": 4955
    },
    {
      "epoch": 0.3575628584827387,
      "grad_norm": 3.102354779652548,
      "learning_rate": 2.9761617063309106e-06,
      "loss": 1.0349,
      "step": 4956
    },
    {
      "epoch": 0.35763500595216624,
      "grad_norm": 4.0781425388002885,
      "learning_rate": 2.975753768777513e-06,
      "loss": 1.064,
      "step": 4957
    },
    {
      "epoch": 0.35770715342159376,
      "grad_norm": 10.267061130162238,
      "learning_rate": 2.975345777941814e-06,
      "loss": 0.9447,
      "step": 4958
    },
    {
      "epoch": 0.3577793008910212,
      "grad_norm": 6.4094367312132645,
      "learning_rate": 2.974937733846093e-06,
      "loss": 1.0154,
      "step": 4959
    },
    {
      "epoch": 0.35785144836044874,
      "grad_norm": 3.9420904618581063,
      "learning_rate": 2.9745296365126303e-06,
      "loss": 1.0628,
      "step": 4960
    },
    {
      "epoch": 0.35792359582987626,
      "grad_norm": 3.5107135288729445,
      "learning_rate": 2.9741214859637134e-06,
      "loss": 0.9639,
      "step": 4961
    },
    {
      "epoch": 0.3579957432993038,
      "grad_norm": 8.093516224905743,
      "learning_rate": 2.973713282221627e-06,
      "loss": 1.0303,
      "step": 4962
    },
    {
      "epoch": 0.3580678907687313,
      "grad_norm": 5.607058586063537,
      "learning_rate": 2.9733050253086634e-06,
      "loss": 1.0053,
      "step": 4963
    },
    {
      "epoch": 0.3581400382381588,
      "grad_norm": 6.275951371130354,
      "learning_rate": 2.972896715247115e-06,
      "loss": 0.9738,
      "step": 4964
    },
    {
      "epoch": 0.3582121857075863,
      "grad_norm": 4.469903970322656,
      "learning_rate": 2.9724883520592788e-06,
      "loss": 1.0579,
      "step": 4965
    },
    {
      "epoch": 0.3582843331770138,
      "grad_norm": 3.641200728182933,
      "learning_rate": 2.9720799357674534e-06,
      "loss": 0.9465,
      "step": 4966
    },
    {
      "epoch": 0.3583564806464413,
      "grad_norm": 6.9691058012449,
      "learning_rate": 2.971671466393942e-06,
      "loss": 0.9629,
      "step": 4967
    },
    {
      "epoch": 0.35842862811586884,
      "grad_norm": 3.7892031928615273,
      "learning_rate": 2.971262943961048e-06,
      "loss": 1.0917,
      "step": 4968
    },
    {
      "epoch": 0.35850077558529636,
      "grad_norm": 19.828153473433062,
      "learning_rate": 2.9708543684910808e-06,
      "loss": 0.9967,
      "step": 4969
    },
    {
      "epoch": 0.3585729230547239,
      "grad_norm": 6.296592384224675,
      "learning_rate": 2.97044574000635e-06,
      "loss": 1.0943,
      "step": 4970
    },
    {
      "epoch": 0.35864507052415134,
      "grad_norm": 4.69093136126855,
      "learning_rate": 2.97003705852917e-06,
      "loss": 1.0523,
      "step": 4971
    },
    {
      "epoch": 0.35871721799357886,
      "grad_norm": 6.520655750460656,
      "learning_rate": 2.9696283240818574e-06,
      "loss": 1.0395,
      "step": 4972
    },
    {
      "epoch": 0.3587893654630064,
      "grad_norm": 4.896947898882041,
      "learning_rate": 2.969219536686731e-06,
      "loss": 0.975,
      "step": 4973
    },
    {
      "epoch": 0.3588615129324339,
      "grad_norm": 3.3153137636213748,
      "learning_rate": 2.968810696366114e-06,
      "loss": 1.0173,
      "step": 4974
    },
    {
      "epoch": 0.3589336604018614,
      "grad_norm": 5.181086976876849,
      "learning_rate": 2.9684018031423313e-06,
      "loss": 1.0184,
      "step": 4975
    },
    {
      "epoch": 0.35900580787128894,
      "grad_norm": 7.708045115538299,
      "learning_rate": 2.9679928570377108e-06,
      "loss": 1.0302,
      "step": 4976
    },
    {
      "epoch": 0.3590779553407164,
      "grad_norm": 4.212117725085621,
      "learning_rate": 2.9675838580745837e-06,
      "loss": 1.0655,
      "step": 4977
    },
    {
      "epoch": 0.3591501028101439,
      "grad_norm": 3.37271832233822,
      "learning_rate": 2.967174806275284e-06,
      "loss": 0.9802,
      "step": 4978
    },
    {
      "epoch": 0.35922225027957144,
      "grad_norm": 5.022598315160114,
      "learning_rate": 2.9667657016621486e-06,
      "loss": 0.8816,
      "step": 4979
    },
    {
      "epoch": 0.35929439774899896,
      "grad_norm": 7.127522072384903,
      "learning_rate": 2.9663565442575175e-06,
      "loss": 0.9276,
      "step": 4980
    },
    {
      "epoch": 0.3593665452184265,
      "grad_norm": 3.8365947836188545,
      "learning_rate": 2.9659473340837324e-06,
      "loss": 0.9458,
      "step": 4981
    },
    {
      "epoch": 0.359438692687854,
      "grad_norm": 3.6502262933683784,
      "learning_rate": 2.9655380711631393e-06,
      "loss": 0.9747,
      "step": 4982
    },
    {
      "epoch": 0.35951084015728146,
      "grad_norm": 7.808781097286682,
      "learning_rate": 2.965128755518086e-06,
      "loss": 0.8773,
      "step": 4983
    },
    {
      "epoch": 0.359582987626709,
      "grad_norm": 5.040588810146937,
      "learning_rate": 2.964719387170925e-06,
      "loss": 0.9738,
      "step": 4984
    },
    {
      "epoch": 0.3596551350961365,
      "grad_norm": 6.896867687945687,
      "learning_rate": 2.964309966144009e-06,
      "loss": 1.0186,
      "step": 4985
    },
    {
      "epoch": 0.359727282565564,
      "grad_norm": 6.209992496406088,
      "learning_rate": 2.963900492459695e-06,
      "loss": 1.0195,
      "step": 4986
    },
    {
      "epoch": 0.35979943003499154,
      "grad_norm": 5.909755090586486,
      "learning_rate": 2.963490966140344e-06,
      "loss": 0.8973,
      "step": 4987
    },
    {
      "epoch": 0.35987157750441906,
      "grad_norm": 11.254322154216082,
      "learning_rate": 2.963081387208318e-06,
      "loss": 1.0468,
      "step": 4988
    },
    {
      "epoch": 0.3599437249738465,
      "grad_norm": 5.04165596868888,
      "learning_rate": 2.962671755685983e-06,
      "loss": 0.9781,
      "step": 4989
    },
    {
      "epoch": 0.36001587244327404,
      "grad_norm": 4.255405354539719,
      "learning_rate": 2.9622620715957066e-06,
      "loss": 1.008,
      "step": 4990
    },
    {
      "epoch": 0.36008801991270156,
      "grad_norm": 6.2273855395309345,
      "learning_rate": 2.9618523349598603e-06,
      "loss": 1.0362,
      "step": 4991
    },
    {
      "epoch": 0.3601601673821291,
      "grad_norm": 0.7832008418860893,
      "learning_rate": 2.9614425458008195e-06,
      "loss": 0.8394,
      "step": 4992
    },
    {
      "epoch": 0.3602323148515566,
      "grad_norm": 0.789387777531642,
      "learning_rate": 2.96103270414096e-06,
      "loss": 0.92,
      "step": 4993
    },
    {
      "epoch": 0.36030446232098406,
      "grad_norm": 7.0503492228270765,
      "learning_rate": 2.960622810002662e-06,
      "loss": 1.044,
      "step": 4994
    },
    {
      "epoch": 0.3603766097904116,
      "grad_norm": 3.904153123721501,
      "learning_rate": 2.9602128634083087e-06,
      "loss": 0.9934,
      "step": 4995
    },
    {
      "epoch": 0.3604487572598391,
      "grad_norm": 5.284831080051446,
      "learning_rate": 2.959802864380286e-06,
      "loss": 0.9705,
      "step": 4996
    },
    {
      "epoch": 0.3605209047292666,
      "grad_norm": 24.22400522620524,
      "learning_rate": 2.9593928129409814e-06,
      "loss": 1.019,
      "step": 4997
    },
    {
      "epoch": 0.36059305219869414,
      "grad_norm": 3.5622851658341683,
      "learning_rate": 2.9589827091127865e-06,
      "loss": 1.0738,
      "step": 4998
    },
    {
      "epoch": 0.36066519966812166,
      "grad_norm": 8.564998512780127,
      "learning_rate": 2.958572552918096e-06,
      "loss": 1.1107,
      "step": 4999
    },
    {
      "epoch": 0.3607373471375491,
      "grad_norm": 11.705029728593841,
      "learning_rate": 2.9581623443793074e-06,
      "loss": 0.9699,
      "step": 5000
    },
    {
      "epoch": 0.36080949460697664,
      "grad_norm": 3.5208565675469203,
      "learning_rate": 2.9577520835188198e-06,
      "loss": 0.9382,
      "step": 5001
    },
    {
      "epoch": 0.36088164207640416,
      "grad_norm": 4.951206738351122,
      "learning_rate": 2.9573417703590362e-06,
      "loss": 1.0014,
      "step": 5002
    },
    {
      "epoch": 0.3609537895458317,
      "grad_norm": 0.7171549305553583,
      "learning_rate": 2.956931404922363e-06,
      "loss": 0.9044,
      "step": 5003
    },
    {
      "epoch": 0.3610259370152592,
      "grad_norm": 3.863005378523876,
      "learning_rate": 2.9565209872312074e-06,
      "loss": 1.0277,
      "step": 5004
    },
    {
      "epoch": 0.3610980844846867,
      "grad_norm": 5.084691517642918,
      "learning_rate": 2.9561105173079828e-06,
      "loss": 1.0232,
      "step": 5005
    },
    {
      "epoch": 0.3611702319541142,
      "grad_norm": 6.618871193155373,
      "learning_rate": 2.955699995175101e-06,
      "loss": 0.8653,
      "step": 5006
    },
    {
      "epoch": 0.3612423794235417,
      "grad_norm": 4.495836238960314,
      "learning_rate": 2.9552894208549816e-06,
      "loss": 0.9101,
      "step": 5007
    },
    {
      "epoch": 0.3613145268929692,
      "grad_norm": 3.0651285218616504,
      "learning_rate": 2.9548787943700422e-06,
      "loss": 1.0082,
      "step": 5008
    },
    {
      "epoch": 0.36138667436239674,
      "grad_norm": 3.127736533754875,
      "learning_rate": 2.9544681157427076e-06,
      "loss": 0.9223,
      "step": 5009
    },
    {
      "epoch": 0.36145882183182426,
      "grad_norm": 12.819841170495005,
      "learning_rate": 2.954057384995402e-06,
      "loss": 1.0281,
      "step": 5010
    },
    {
      "epoch": 0.3615309693012518,
      "grad_norm": 5.700837980544608,
      "learning_rate": 2.9536466021505544e-06,
      "loss": 0.9656,
      "step": 5011
    },
    {
      "epoch": 0.36160311677067924,
      "grad_norm": 5.990741580153143,
      "learning_rate": 2.953235767230596e-06,
      "loss": 0.9684,
      "step": 5012
    },
    {
      "epoch": 0.36167526424010676,
      "grad_norm": 5.325360696960758,
      "learning_rate": 2.9528248802579613e-06,
      "loss": 0.9881,
      "step": 5013
    },
    {
      "epoch": 0.3617474117095343,
      "grad_norm": 3.4704673571088107,
      "learning_rate": 2.952413941255087e-06,
      "loss": 1.0904,
      "step": 5014
    },
    {
      "epoch": 0.3618195591789618,
      "grad_norm": 4.878961567378057,
      "learning_rate": 2.9520029502444132e-06,
      "loss": 0.909,
      "step": 5015
    },
    {
      "epoch": 0.3618917066483893,
      "grad_norm": 5.102862859792014,
      "learning_rate": 2.951591907248383e-06,
      "loss": 0.9363,
      "step": 5016
    },
    {
      "epoch": 0.36196385411781684,
      "grad_norm": 7.782692894191748,
      "learning_rate": 2.951180812289441e-06,
      "loss": 0.942,
      "step": 5017
    },
    {
      "epoch": 0.3620360015872443,
      "grad_norm": 9.129738609357625,
      "learning_rate": 2.9507696653900355e-06,
      "loss": 0.9396,
      "step": 5018
    },
    {
      "epoch": 0.3621081490566718,
      "grad_norm": 6.3648663929763245,
      "learning_rate": 2.950358466572618e-06,
      "loss": 0.9182,
      "step": 5019
    },
    {
      "epoch": 0.36218029652609934,
      "grad_norm": 5.64521834999464,
      "learning_rate": 2.949947215859644e-06,
      "loss": 0.8918,
      "step": 5020
    },
    {
      "epoch": 0.36225244399552686,
      "grad_norm": 4.239200559140277,
      "learning_rate": 2.949535913273569e-06,
      "loss": 0.761,
      "step": 5021
    },
    {
      "epoch": 0.3623245914649544,
      "grad_norm": 8.329117993099194,
      "learning_rate": 2.949124558836852e-06,
      "loss": 0.9133,
      "step": 5022
    },
    {
      "epoch": 0.3623967389343819,
      "grad_norm": 3.4513947307521025,
      "learning_rate": 2.948713152571957e-06,
      "loss": 1.0073,
      "step": 5023
    },
    {
      "epoch": 0.36246888640380936,
      "grad_norm": 3.3714907733961272,
      "learning_rate": 2.948301694501348e-06,
      "loss": 1.0086,
      "step": 5024
    },
    {
      "epoch": 0.3625410338732369,
      "grad_norm": 4.362849337956002,
      "learning_rate": 2.9478901846474946e-06,
      "loss": 1.0388,
      "step": 5025
    },
    {
      "epoch": 0.3626131813426644,
      "grad_norm": 3.9254719930273345,
      "learning_rate": 2.947478623032867e-06,
      "loss": 1.0195,
      "step": 5026
    },
    {
      "epoch": 0.3626853288120919,
      "grad_norm": 4.26004546521779,
      "learning_rate": 2.947067009679939e-06,
      "loss": 0.9942,
      "step": 5027
    },
    {
      "epoch": 0.36275747628151944,
      "grad_norm": 6.159268486383226,
      "learning_rate": 2.9466553446111875e-06,
      "loss": 1.0075,
      "step": 5028
    },
    {
      "epoch": 0.36282962375094696,
      "grad_norm": 4.355851694558178,
      "learning_rate": 2.9462436278490923e-06,
      "loss": 1.0305,
      "step": 5029
    },
    {
      "epoch": 0.3629017712203744,
      "grad_norm": 0.8284169528158556,
      "learning_rate": 2.9458318594161357e-06,
      "loss": 0.8894,
      "step": 5030
    },
    {
      "epoch": 0.36297391868980194,
      "grad_norm": 6.01466357982852,
      "learning_rate": 2.9454200393348022e-06,
      "loss": 0.9915,
      "step": 5031
    },
    {
      "epoch": 0.36304606615922946,
      "grad_norm": 6.866353927206354,
      "learning_rate": 2.9450081676275805e-06,
      "loss": 0.9915,
      "step": 5032
    },
    {
      "epoch": 0.363118213628657,
      "grad_norm": 5.540219722756131,
      "learning_rate": 2.944596244316961e-06,
      "loss": 1.0055,
      "step": 5033
    },
    {
      "epoch": 0.3631903610980845,
      "grad_norm": 3.5825423912846532,
      "learning_rate": 2.944184269425437e-06,
      "loss": 1.0145,
      "step": 5034
    },
    {
      "epoch": 0.363262508567512,
      "grad_norm": 9.785015124980326,
      "learning_rate": 2.943772242975506e-06,
      "loss": 0.9125,
      "step": 5035
    },
    {
      "epoch": 0.3633346560369395,
      "grad_norm": 5.700061382833207,
      "learning_rate": 2.943360164989666e-06,
      "loss": 0.9957,
      "step": 5036
    },
    {
      "epoch": 0.363406803506367,
      "grad_norm": 4.449763799678001,
      "learning_rate": 2.94294803549042e-06,
      "loss": 0.9508,
      "step": 5037
    },
    {
      "epoch": 0.3634789509757945,
      "grad_norm": 6.137491373028879,
      "learning_rate": 2.9425358545002717e-06,
      "loss": 0.9578,
      "step": 5038
    },
    {
      "epoch": 0.36355109844522204,
      "grad_norm": 7.508107190496442,
      "learning_rate": 2.94212362204173e-06,
      "loss": 0.9508,
      "step": 5039
    },
    {
      "epoch": 0.36362324591464956,
      "grad_norm": 6.090823992027308,
      "learning_rate": 2.941711338137305e-06,
      "loss": 0.9549,
      "step": 5040
    },
    {
      "epoch": 0.363695393384077,
      "grad_norm": 9.268732845936166,
      "learning_rate": 2.94129900280951e-06,
      "loss": 0.9466,
      "step": 5041
    },
    {
      "epoch": 0.36376754085350455,
      "grad_norm": 8.705456515572934,
      "learning_rate": 2.94088661608086e-06,
      "loss": 0.9727,
      "step": 5042
    },
    {
      "epoch": 0.36383968832293206,
      "grad_norm": 3.9730918142874954,
      "learning_rate": 2.940474177973876e-06,
      "loss": 0.931,
      "step": 5043
    },
    {
      "epoch": 0.3639118357923596,
      "grad_norm": 4.562525709942009,
      "learning_rate": 2.940061688511079e-06,
      "loss": 0.8408,
      "step": 5044
    },
    {
      "epoch": 0.3639839832617871,
      "grad_norm": 4.106132127065796,
      "learning_rate": 2.9396491477149922e-06,
      "loss": 1.0066,
      "step": 5045
    },
    {
      "epoch": 0.3640561307312146,
      "grad_norm": 5.160398536786833,
      "learning_rate": 2.9392365556081446e-06,
      "loss": 0.9856,
      "step": 5046
    },
    {
      "epoch": 0.3641282782006421,
      "grad_norm": 4.088837216460793,
      "learning_rate": 2.9388239122130653e-06,
      "loss": 1.0132,
      "step": 5047
    },
    {
      "epoch": 0.3642004256700696,
      "grad_norm": 4.926473159052552,
      "learning_rate": 2.9384112175522875e-06,
      "loss": 0.9502,
      "step": 5048
    },
    {
      "epoch": 0.3642725731394971,
      "grad_norm": 4.536878155429877,
      "learning_rate": 2.937998471648347e-06,
      "loss": 1.0147,
      "step": 5049
    },
    {
      "epoch": 0.36434472060892464,
      "grad_norm": 7.2888774950563855,
      "learning_rate": 2.9375856745237827e-06,
      "loss": 0.9502,
      "step": 5050
    },
    {
      "epoch": 0.36441686807835216,
      "grad_norm": 10.785111330153649,
      "learning_rate": 2.937172826201135e-06,
      "loss": 1.0034,
      "step": 5051
    },
    {
      "epoch": 0.3644890155477797,
      "grad_norm": 5.908176975665273,
      "learning_rate": 2.9367599267029494e-06,
      "loss": 1.0271,
      "step": 5052
    },
    {
      "epoch": 0.36456116301720715,
      "grad_norm": 4.241318306298385,
      "learning_rate": 2.936346976051771e-06,
      "loss": 0.9474,
      "step": 5053
    },
    {
      "epoch": 0.36463331048663467,
      "grad_norm": 4.4930563443062175,
      "learning_rate": 2.9359339742701505e-06,
      "loss": 0.9588,
      "step": 5054
    },
    {
      "epoch": 0.3647054579560622,
      "grad_norm": 4.34191746100259,
      "learning_rate": 2.9355209213806414e-06,
      "loss": 1.0509,
      "step": 5055
    },
    {
      "epoch": 0.3647776054254897,
      "grad_norm": 4.7193164611750476,
      "learning_rate": 2.935107817405797e-06,
      "loss": 0.9865,
      "step": 5056
    },
    {
      "epoch": 0.3648497528949172,
      "grad_norm": 4.439378367716368,
      "learning_rate": 2.9346946623681764e-06,
      "loss": 0.9442,
      "step": 5057
    },
    {
      "epoch": 0.36492190036434474,
      "grad_norm": 3.4648887519537372,
      "learning_rate": 2.934281456290341e-06,
      "loss": 0.9467,
      "step": 5058
    },
    {
      "epoch": 0.3649940478337722,
      "grad_norm": 3.4638375903989975,
      "learning_rate": 2.933868199194854e-06,
      "loss": 1.0196,
      "step": 5059
    },
    {
      "epoch": 0.3650661953031997,
      "grad_norm": 4.477274839702648,
      "learning_rate": 2.9334548911042807e-06,
      "loss": 0.9901,
      "step": 5060
    },
    {
      "epoch": 0.36513834277262724,
      "grad_norm": 5.989304865016748,
      "learning_rate": 2.933041532041192e-06,
      "loss": 1.0395,
      "step": 5061
    },
    {
      "epoch": 0.36521049024205476,
      "grad_norm": 13.821183042996276,
      "learning_rate": 2.9326281220281594e-06,
      "loss": 1.0178,
      "step": 5062
    },
    {
      "epoch": 0.3652826377114823,
      "grad_norm": 6.801924309811618,
      "learning_rate": 2.932214661087757e-06,
      "loss": 0.9395,
      "step": 5063
    },
    {
      "epoch": 0.3653547851809098,
      "grad_norm": 3.8037426537108394,
      "learning_rate": 2.9318011492425626e-06,
      "loss": 0.947,
      "step": 5064
    },
    {
      "epoch": 0.36542693265033727,
      "grad_norm": 3.5949625374615453,
      "learning_rate": 2.9313875865151575e-06,
      "loss": 1.0037,
      "step": 5065
    },
    {
      "epoch": 0.3654990801197648,
      "grad_norm": 4.6452283023372445,
      "learning_rate": 2.9309739729281245e-06,
      "loss": 1.0136,
      "step": 5066
    },
    {
      "epoch": 0.3655712275891923,
      "grad_norm": 4.3151064749761225,
      "learning_rate": 2.9305603085040485e-06,
      "loss": 1.0351,
      "step": 5067
    },
    {
      "epoch": 0.3656433750586198,
      "grad_norm": 5.741845028584052,
      "learning_rate": 2.930146593265519e-06,
      "loss": 1.0814,
      "step": 5068
    },
    {
      "epoch": 0.36571552252804734,
      "grad_norm": 0.8406342800152848,
      "learning_rate": 2.9297328272351273e-06,
      "loss": 0.8781,
      "step": 5069
    },
    {
      "epoch": 0.36578766999747486,
      "grad_norm": 6.317175917953302,
      "learning_rate": 2.929319010435468e-06,
      "loss": 0.8903,
      "step": 5070
    },
    {
      "epoch": 0.3658598174669023,
      "grad_norm": 0.6758211818650008,
      "learning_rate": 2.9289051428891375e-06,
      "loss": 0.8645,
      "step": 5071
    },
    {
      "epoch": 0.36593196493632985,
      "grad_norm": 3.803436261544274,
      "learning_rate": 2.9284912246187355e-06,
      "loss": 1.0496,
      "step": 5072
    },
    {
      "epoch": 0.36600411240575736,
      "grad_norm": 5.721425582261765,
      "learning_rate": 2.9280772556468657e-06,
      "loss": 0.9349,
      "step": 5073
    },
    {
      "epoch": 0.3660762598751849,
      "grad_norm": 8.903614176161792,
      "learning_rate": 2.927663235996132e-06,
      "loss": 0.8982,
      "step": 5074
    },
    {
      "epoch": 0.3661484073446124,
      "grad_norm": 4.73278761838592,
      "learning_rate": 2.9272491656891435e-06,
      "loss": 0.9126,
      "step": 5075
    },
    {
      "epoch": 0.3662205548140399,
      "grad_norm": 4.267093168240566,
      "learning_rate": 2.92683504474851e-06,
      "loss": 1.098,
      "step": 5076
    },
    {
      "epoch": 0.3662927022834674,
      "grad_norm": 12.978195096916705,
      "learning_rate": 2.9264208731968465e-06,
      "loss": 0.9707,
      "step": 5077
    },
    {
      "epoch": 0.3663648497528949,
      "grad_norm": 7.78863297696729,
      "learning_rate": 2.926006651056768e-06,
      "loss": 1.0749,
      "step": 5078
    },
    {
      "epoch": 0.3664369972223224,
      "grad_norm": 4.168362513982395,
      "learning_rate": 2.9255923783508945e-06,
      "loss": 1.0579,
      "step": 5079
    },
    {
      "epoch": 0.36650914469174994,
      "grad_norm": 5.915334668014916,
      "learning_rate": 2.9251780551018473e-06,
      "loss": 0.9936,
      "step": 5080
    },
    {
      "epoch": 0.36658129216117746,
      "grad_norm": 4.302635742386873,
      "learning_rate": 2.9247636813322514e-06,
      "loss": 0.9455,
      "step": 5081
    },
    {
      "epoch": 0.366653439630605,
      "grad_norm": 9.480309408426088,
      "learning_rate": 2.924349257064735e-06,
      "loss": 0.9648,
      "step": 5082
    },
    {
      "epoch": 0.36672558710003245,
      "grad_norm": 5.960744232204382,
      "learning_rate": 2.923934782321927e-06,
      "loss": 1.064,
      "step": 5083
    },
    {
      "epoch": 0.36679773456945997,
      "grad_norm": 6.737302351902951,
      "learning_rate": 2.9235202571264605e-06,
      "loss": 1.0483,
      "step": 5084
    },
    {
      "epoch": 0.3668698820388875,
      "grad_norm": 5.827592513480488,
      "learning_rate": 2.9231056815009723e-06,
      "loss": 0.9984,
      "step": 5085
    },
    {
      "epoch": 0.366942029508315,
      "grad_norm": 4.539606636247819,
      "learning_rate": 2.9226910554681e-06,
      "loss": 1.0204,
      "step": 5086
    },
    {
      "epoch": 0.3670141769777425,
      "grad_norm": 4.702726936603781,
      "learning_rate": 2.922276379050484e-06,
      "loss": 1.0051,
      "step": 5087
    },
    {
      "epoch": 0.36708632444717004,
      "grad_norm": 3.6678135263182035,
      "learning_rate": 2.9218616522707695e-06,
      "loss": 0.8923,
      "step": 5088
    },
    {
      "epoch": 0.3671584719165975,
      "grad_norm": 4.716577092770338,
      "learning_rate": 2.9214468751516032e-06,
      "loss": 0.992,
      "step": 5089
    },
    {
      "epoch": 0.367230619386025,
      "grad_norm": 7.716839387854296,
      "learning_rate": 2.9210320477156335e-06,
      "loss": 0.9844,
      "step": 5090
    },
    {
      "epoch": 0.36730276685545254,
      "grad_norm": 3.362982937191303,
      "learning_rate": 2.920617169985513e-06,
      "loss": 1.0067,
      "step": 5091
    },
    {
      "epoch": 0.36737491432488006,
      "grad_norm": 3.755596435356988,
      "learning_rate": 2.9202022419838974e-06,
      "loss": 0.9392,
      "step": 5092
    },
    {
      "epoch": 0.3674470617943076,
      "grad_norm": 4.816484065483011,
      "learning_rate": 2.919787263733444e-06,
      "loss": 0.9683,
      "step": 5093
    },
    {
      "epoch": 0.36751920926373505,
      "grad_norm": 4.30841357355024,
      "learning_rate": 2.9193722352568125e-06,
      "loss": 0.9688,
      "step": 5094
    },
    {
      "epoch": 0.36759135673316257,
      "grad_norm": 6.527555598091957,
      "learning_rate": 2.9189571565766667e-06,
      "loss": 0.95,
      "step": 5095
    },
    {
      "epoch": 0.3676635042025901,
      "grad_norm": 6.154745005106292,
      "learning_rate": 2.9185420277156727e-06,
      "loss": 1.02,
      "step": 5096
    },
    {
      "epoch": 0.3677356516720176,
      "grad_norm": 0.9402753756830688,
      "learning_rate": 2.9181268486964996e-06,
      "loss": 0.8419,
      "step": 5097
    },
    {
      "epoch": 0.3678077991414451,
      "grad_norm": 3.5900955859026884,
      "learning_rate": 2.917711619541817e-06,
      "loss": 1.0473,
      "step": 5098
    },
    {
      "epoch": 0.36787994661087264,
      "grad_norm": 5.34426217106498,
      "learning_rate": 2.9172963402743005e-06,
      "loss": 0.9018,
      "step": 5099
    },
    {
      "epoch": 0.3679520940803001,
      "grad_norm": 9.247687385766415,
      "learning_rate": 2.9168810109166268e-06,
      "loss": 1.0394,
      "step": 5100
    },
    {
      "epoch": 0.3680242415497276,
      "grad_norm": 5.463546806262726,
      "learning_rate": 2.9164656314914747e-06,
      "loss": 0.9567,
      "step": 5101
    },
    {
      "epoch": 0.36809638901915515,
      "grad_norm": 4.28180128571682,
      "learning_rate": 2.9160502020215277e-06,
      "loss": 0.938,
      "step": 5102
    },
    {
      "epoch": 0.36816853648858267,
      "grad_norm": 5.127542562842813,
      "learning_rate": 2.9156347225294698e-06,
      "loss": 0.9069,
      "step": 5103
    },
    {
      "epoch": 0.3682406839580102,
      "grad_norm": 5.087426127429349,
      "learning_rate": 2.9152191930379895e-06,
      "loss": 0.8255,
      "step": 5104
    },
    {
      "epoch": 0.3683128314274377,
      "grad_norm": 0.677160014187528,
      "learning_rate": 2.914803613569777e-06,
      "loss": 0.8562,
      "step": 5105
    },
    {
      "epoch": 0.36838497889686517,
      "grad_norm": 7.023096129339493,
      "learning_rate": 2.9143879841475256e-06,
      "loss": 0.9509,
      "step": 5106
    },
    {
      "epoch": 0.3684571263662927,
      "grad_norm": 8.814703916011764,
      "learning_rate": 2.9139723047939315e-06,
      "loss": 1.0811,
      "step": 5107
    },
    {
      "epoch": 0.3685292738357202,
      "grad_norm": 5.185069054609998,
      "learning_rate": 2.9135565755316932e-06,
      "loss": 0.928,
      "step": 5108
    },
    {
      "epoch": 0.3686014213051477,
      "grad_norm": 3.310890958566011,
      "learning_rate": 2.9131407963835122e-06,
      "loss": 0.982,
      "step": 5109
    },
    {
      "epoch": 0.36867356877457524,
      "grad_norm": 3.5573272628827706,
      "learning_rate": 2.9127249673720924e-06,
      "loss": 0.9358,
      "step": 5110
    },
    {
      "epoch": 0.36874571624400276,
      "grad_norm": 4.586906473706997,
      "learning_rate": 2.912309088520141e-06,
      "loss": 0.9249,
      "step": 5111
    },
    {
      "epoch": 0.3688178637134302,
      "grad_norm": 3.4731266488384884,
      "learning_rate": 2.911893159850368e-06,
      "loss": 0.9519,
      "step": 5112
    },
    {
      "epoch": 0.36889001118285775,
      "grad_norm": 5.956319437492844,
      "learning_rate": 2.911477181385485e-06,
      "loss": 1.0174,
      "step": 5113
    },
    {
      "epoch": 0.36896215865228527,
      "grad_norm": 0.7075254283419701,
      "learning_rate": 2.911061153148207e-06,
      "loss": 0.8086,
      "step": 5114
    },
    {
      "epoch": 0.3690343061217128,
      "grad_norm": 3.7031622132310704,
      "learning_rate": 2.9106450751612525e-06,
      "loss": 1.0584,
      "step": 5115
    },
    {
      "epoch": 0.3691064535911403,
      "grad_norm": 4.246565384627672,
      "learning_rate": 2.9102289474473407e-06,
      "loss": 0.9176,
      "step": 5116
    },
    {
      "epoch": 0.3691786010605678,
      "grad_norm": 6.285517670912828,
      "learning_rate": 2.9098127700291964e-06,
      "loss": 0.9836,
      "step": 5117
    },
    {
      "epoch": 0.3692507485299953,
      "grad_norm": 4.42816448099846,
      "learning_rate": 2.909396542929545e-06,
      "loss": 0.9437,
      "step": 5118
    },
    {
      "epoch": 0.3693228959994228,
      "grad_norm": 6.497850796372892,
      "learning_rate": 2.908980266171114e-06,
      "loss": 0.8757,
      "step": 5119
    },
    {
      "epoch": 0.3693950434688503,
      "grad_norm": 3.0843053436916716,
      "learning_rate": 2.9085639397766358e-06,
      "loss": 1.0158,
      "step": 5120
    },
    {
      "epoch": 0.36946719093827785,
      "grad_norm": 3.2711941627340275,
      "learning_rate": 2.908147563768845e-06,
      "loss": 0.9989,
      "step": 5121
    },
    {
      "epoch": 0.36953933840770536,
      "grad_norm": 4.946002062410325,
      "learning_rate": 2.9077311381704766e-06,
      "loss": 0.9384,
      "step": 5122
    },
    {
      "epoch": 0.3696114858771329,
      "grad_norm": 5.976409311762269,
      "learning_rate": 2.9073146630042716e-06,
      "loss": 0.9089,
      "step": 5123
    },
    {
      "epoch": 0.36968363334656035,
      "grad_norm": 3.825951531429745,
      "learning_rate": 2.906898138292972e-06,
      "loss": 1.0471,
      "step": 5124
    },
    {
      "epoch": 0.36975578081598787,
      "grad_norm": 2.412521808901232,
      "learning_rate": 2.9064815640593216e-06,
      "loss": 0.9779,
      "step": 5125
    },
    {
      "epoch": 0.3698279282854154,
      "grad_norm": 5.19958393926663,
      "learning_rate": 2.9060649403260686e-06,
      "loss": 0.9262,
      "step": 5126
    },
    {
      "epoch": 0.3699000757548429,
      "grad_norm": 3.7030765836101787,
      "learning_rate": 2.905648267115964e-06,
      "loss": 1.0176,
      "step": 5127
    },
    {
      "epoch": 0.3699722232242704,
      "grad_norm": 3.744726160119352,
      "learning_rate": 2.9052315444517597e-06,
      "loss": 0.905,
      "step": 5128
    },
    {
      "epoch": 0.37004437069369794,
      "grad_norm": 3.968401435503735,
      "learning_rate": 2.9048147723562116e-06,
      "loss": 0.986,
      "step": 5129
    },
    {
      "epoch": 0.3701165181631254,
      "grad_norm": 2.939338271219604,
      "learning_rate": 2.904397950852079e-06,
      "loss": 0.9403,
      "step": 5130
    },
    {
      "epoch": 0.3701886656325529,
      "grad_norm": 7.115059862240213,
      "learning_rate": 2.903981079962121e-06,
      "loss": 0.8948,
      "step": 5131
    },
    {
      "epoch": 0.37026081310198045,
      "grad_norm": 6.030619692108731,
      "learning_rate": 2.9035641597091044e-06,
      "loss": 0.8414,
      "step": 5132
    },
    {
      "epoch": 0.37033296057140797,
      "grad_norm": 4.0648055356891,
      "learning_rate": 2.903147190115793e-06,
      "loss": 0.8645,
      "step": 5133
    },
    {
      "epoch": 0.3704051080408355,
      "grad_norm": 4.413841687646207,
      "learning_rate": 2.902730171204956e-06,
      "loss": 0.8653,
      "step": 5134
    },
    {
      "epoch": 0.370477255510263,
      "grad_norm": 7.442391974575288,
      "learning_rate": 2.9023131029993676e-06,
      "loss": 1.0058,
      "step": 5135
    },
    {
      "epoch": 0.37054940297969047,
      "grad_norm": 5.441167887815854,
      "learning_rate": 2.9018959855218e-06,
      "loss": 1.0128,
      "step": 5136
    },
    {
      "epoch": 0.370621550449118,
      "grad_norm": 4.026548023960203,
      "learning_rate": 2.9014788187950316e-06,
      "loss": 0.8848,
      "step": 5137
    },
    {
      "epoch": 0.3706936979185455,
      "grad_norm": 5.378608734169386,
      "learning_rate": 2.9010616028418422e-06,
      "loss": 1.0238,
      "step": 5138
    },
    {
      "epoch": 0.370765845387973,
      "grad_norm": 3.4241247576329576,
      "learning_rate": 2.9006443376850146e-06,
      "loss": 1.0014,
      "step": 5139
    },
    {
      "epoch": 0.37083799285740054,
      "grad_norm": 5.452120363223663,
      "learning_rate": 2.9002270233473336e-06,
      "loss": 0.9021,
      "step": 5140
    },
    {
      "epoch": 0.37091014032682806,
      "grad_norm": 2.5896594412052214,
      "learning_rate": 2.8998096598515868e-06,
      "loss": 0.9106,
      "step": 5141
    },
    {
      "epoch": 0.3709822877962555,
      "grad_norm": 3.9237425715536265,
      "learning_rate": 2.8993922472205653e-06,
      "loss": 0.97,
      "step": 5142
    },
    {
      "epoch": 0.37105443526568305,
      "grad_norm": 4.738226504217948,
      "learning_rate": 2.898974785477064e-06,
      "loss": 0.9883,
      "step": 5143
    },
    {
      "epoch": 0.37112658273511057,
      "grad_norm": 4.229209310883498,
      "learning_rate": 2.8985572746438762e-06,
      "loss": 0.9158,
      "step": 5144
    },
    {
      "epoch": 0.3711987302045381,
      "grad_norm": 6.342871092998231,
      "learning_rate": 2.8981397147438026e-06,
      "loss": 1.0331,
      "step": 5145
    },
    {
      "epoch": 0.3712708776739656,
      "grad_norm": 4.20953742197487,
      "learning_rate": 2.8977221057996434e-06,
      "loss": 1.034,
      "step": 5146
    },
    {
      "epoch": 0.37134302514339307,
      "grad_norm": 3.0834819826602096,
      "learning_rate": 2.8973044478342037e-06,
      "loss": 1.1235,
      "step": 5147
    },
    {
      "epoch": 0.3714151726128206,
      "grad_norm": 4.966331421122017,
      "learning_rate": 2.8968867408702903e-06,
      "loss": 0.946,
      "step": 5148
    },
    {
      "epoch": 0.3714873200822481,
      "grad_norm": 5.634094684925092,
      "learning_rate": 2.8964689849307116e-06,
      "loss": 1.0133,
      "step": 5149
    },
    {
      "epoch": 0.3715594675516756,
      "grad_norm": 9.964674161262712,
      "learning_rate": 2.89605118003828e-06,
      "loss": 1.0052,
      "step": 5150
    },
    {
      "epoch": 0.37163161502110315,
      "grad_norm": 4.1692071163622915,
      "learning_rate": 2.895633326215811e-06,
      "loss": 1.0607,
      "step": 5151
    },
    {
      "epoch": 0.37170376249053066,
      "grad_norm": 3.642739005537369,
      "learning_rate": 2.8952154234861215e-06,
      "loss": 1.168,
      "step": 5152
    },
    {
      "epoch": 0.37177590995995813,
      "grad_norm": 3.5251955776202877,
      "learning_rate": 2.8947974718720315e-06,
      "loss": 0.9935,
      "step": 5153
    },
    {
      "epoch": 0.37184805742938565,
      "grad_norm": 4.085473466420289,
      "learning_rate": 2.8943794713963644e-06,
      "loss": 0.8504,
      "step": 5154
    },
    {
      "epoch": 0.37192020489881317,
      "grad_norm": 6.451760801342851,
      "learning_rate": 2.8939614220819447e-06,
      "loss": 0.907,
      "step": 5155
    },
    {
      "epoch": 0.3719923523682407,
      "grad_norm": 0.7531254732755701,
      "learning_rate": 2.8935433239516018e-06,
      "loss": 0.8415,
      "step": 5156
    },
    {
      "epoch": 0.3720644998376682,
      "grad_norm": 3.664771818627318,
      "learning_rate": 2.8931251770281647e-06,
      "loss": 1.0451,
      "step": 5157
    },
    {
      "epoch": 0.3721366473070957,
      "grad_norm": 4.171048085987516,
      "learning_rate": 2.892706981334469e-06,
      "loss": 0.84,
      "step": 5158
    },
    {
      "epoch": 0.3722087947765232,
      "grad_norm": 3.5247495792257983,
      "learning_rate": 2.892288736893349e-06,
      "loss": 0.9485,
      "step": 5159
    },
    {
      "epoch": 0.3722809422459507,
      "grad_norm": 5.514718562092884,
      "learning_rate": 2.8918704437276448e-06,
      "loss": 0.9407,
      "step": 5160
    },
    {
      "epoch": 0.3723530897153782,
      "grad_norm": 2.658859048677621,
      "learning_rate": 2.8914521018601962e-06,
      "loss": 0.9231,
      "step": 5161
    },
    {
      "epoch": 0.37242523718480575,
      "grad_norm": 8.190194887630165,
      "learning_rate": 2.8910337113138494e-06,
      "loss": 0.9847,
      "step": 5162
    },
    {
      "epoch": 0.37249738465423327,
      "grad_norm": 3.659850605484843,
      "learning_rate": 2.89061527211145e-06,
      "loss": 1.0185,
      "step": 5163
    },
    {
      "epoch": 0.3725695321236608,
      "grad_norm": 4.00782725311003,
      "learning_rate": 2.8901967842758474e-06,
      "loss": 0.9876,
      "step": 5164
    },
    {
      "epoch": 0.37264167959308825,
      "grad_norm": 2.793412009324638,
      "learning_rate": 2.8897782478298935e-06,
      "loss": 0.9936,
      "step": 5165
    },
    {
      "epoch": 0.37271382706251577,
      "grad_norm": 3.0632970901464978,
      "learning_rate": 2.8893596627964437e-06,
      "loss": 0.9739,
      "step": 5166
    },
    {
      "epoch": 0.3727859745319433,
      "grad_norm": 3.012595122169122,
      "learning_rate": 2.888941029198355e-06,
      "loss": 1.018,
      "step": 5167
    },
    {
      "epoch": 0.3728581220013708,
      "grad_norm": 3.3906822375551173,
      "learning_rate": 2.8885223470584866e-06,
      "loss": 0.9268,
      "step": 5168
    },
    {
      "epoch": 0.3729302694707983,
      "grad_norm": 11.772449231705608,
      "learning_rate": 2.888103616399703e-06,
      "loss": 0.9759,
      "step": 5169
    },
    {
      "epoch": 0.37300241694022584,
      "grad_norm": 3.4222052427864234,
      "learning_rate": 2.8876848372448686e-06,
      "loss": 0.945,
      "step": 5170
    },
    {
      "epoch": 0.3730745644096533,
      "grad_norm": 2.4845156839712095,
      "learning_rate": 2.887266009616851e-06,
      "loss": 0.908,
      "step": 5171
    },
    {
      "epoch": 0.37314671187908083,
      "grad_norm": 4.216490642105225,
      "learning_rate": 2.886847133538521e-06,
      "loss": 0.9878,
      "step": 5172
    },
    {
      "epoch": 0.37321885934850835,
      "grad_norm": 3.327152737009256,
      "learning_rate": 2.8864282090327524e-06,
      "loss": 0.939,
      "step": 5173
    },
    {
      "epoch": 0.37329100681793587,
      "grad_norm": 4.520005157602794,
      "learning_rate": 2.88600923612242e-06,
      "loss": 0.9969,
      "step": 5174
    },
    {
      "epoch": 0.3733631542873634,
      "grad_norm": 4.538907440741134,
      "learning_rate": 2.8855902148304042e-06,
      "loss": 1.1033,
      "step": 5175
    },
    {
      "epoch": 0.3734353017567909,
      "grad_norm": 4.149783503389062,
      "learning_rate": 2.8851711451795843e-06,
      "loss": 0.9449,
      "step": 5176
    },
    {
      "epoch": 0.37350744922621837,
      "grad_norm": 0.7722450079937169,
      "learning_rate": 2.8847520271928457e-06,
      "loss": 0.8825,
      "step": 5177
    },
    {
      "epoch": 0.3735795966956459,
      "grad_norm": 2.944159082064094,
      "learning_rate": 2.8843328608930737e-06,
      "loss": 1.0408,
      "step": 5178
    },
    {
      "epoch": 0.3736517441650734,
      "grad_norm": 3.8279228312767057,
      "learning_rate": 2.8839136463031575e-06,
      "loss": 1.0167,
      "step": 5179
    },
    {
      "epoch": 0.3737238916345009,
      "grad_norm": 4.178500169046273,
      "learning_rate": 2.8834943834459896e-06,
      "loss": 0.986,
      "step": 5180
    },
    {
      "epoch": 0.37379603910392845,
      "grad_norm": 3.1096954947303583,
      "learning_rate": 2.8830750723444643e-06,
      "loss": 0.9763,
      "step": 5181
    },
    {
      "epoch": 0.37386818657335597,
      "grad_norm": 2.6145379618523967,
      "learning_rate": 2.882655713021478e-06,
      "loss": 1.0407,
      "step": 5182
    },
    {
      "epoch": 0.37394033404278343,
      "grad_norm": 3.902836032551824,
      "learning_rate": 2.8822363054999303e-06,
      "loss": 0.9137,
      "step": 5183
    },
    {
      "epoch": 0.37401248151221095,
      "grad_norm": 3.1783705474909945,
      "learning_rate": 2.8818168498027243e-06,
      "loss": 0.9182,
      "step": 5184
    },
    {
      "epoch": 0.37408462898163847,
      "grad_norm": 3.9844718173362668,
      "learning_rate": 2.8813973459527643e-06,
      "loss": 0.9423,
      "step": 5185
    },
    {
      "epoch": 0.374156776451066,
      "grad_norm": 4.128491397973906,
      "learning_rate": 2.880977793972958e-06,
      "loss": 0.9091,
      "step": 5186
    },
    {
      "epoch": 0.3742289239204935,
      "grad_norm": 2.242257359709003,
      "learning_rate": 2.8805581938862165e-06,
      "loss": 1.067,
      "step": 5187
    },
    {
      "epoch": 0.374301071389921,
      "grad_norm": 2.698641283284681,
      "learning_rate": 2.880138545715451e-06,
      "loss": 1.0317,
      "step": 5188
    },
    {
      "epoch": 0.3743732188593485,
      "grad_norm": 3.306083222132722,
      "learning_rate": 2.879718849483578e-06,
      "loss": 0.9938,
      "step": 5189
    },
    {
      "epoch": 0.374445366328776,
      "grad_norm": 6.048890243768054,
      "learning_rate": 2.8792991052135147e-06,
      "loss": 0.9464,
      "step": 5190
    },
    {
      "epoch": 0.3745175137982035,
      "grad_norm": 6.717983184348228,
      "learning_rate": 2.878879312928183e-06,
      "loss": 0.893,
      "step": 5191
    },
    {
      "epoch": 0.37458966126763105,
      "grad_norm": 2.934530177294177,
      "learning_rate": 2.878459472650505e-06,
      "loss": 0.9529,
      "step": 5192
    },
    {
      "epoch": 0.37466180873705857,
      "grad_norm": 11.17977621601767,
      "learning_rate": 2.8780395844034077e-06,
      "loss": 0.9701,
      "step": 5193
    },
    {
      "epoch": 0.3747339562064861,
      "grad_norm": 3.5390095485742763,
      "learning_rate": 2.8776196482098184e-06,
      "loss": 0.8976,
      "step": 5194
    },
    {
      "epoch": 0.37480610367591355,
      "grad_norm": 6.145070105106995,
      "learning_rate": 2.8771996640926686e-06,
      "loss": 0.9265,
      "step": 5195
    },
    {
      "epoch": 0.37487825114534107,
      "grad_norm": 5.597097918715942,
      "learning_rate": 2.8767796320748938e-06,
      "loss": 1.0369,
      "step": 5196
    },
    {
      "epoch": 0.3749503986147686,
      "grad_norm": 4.127073287117195,
      "learning_rate": 2.8763595521794278e-06,
      "loss": 1.0701,
      "step": 5197
    },
    {
      "epoch": 0.3750225460841961,
      "grad_norm": 2.375364978252232,
      "learning_rate": 2.8759394244292116e-06,
      "loss": 0.9981,
      "step": 5198
    },
    {
      "epoch": 0.3750946935536236,
      "grad_norm": 2.8852097383153663,
      "learning_rate": 2.8755192488471856e-06,
      "loss": 1.0499,
      "step": 5199
    },
    {
      "epoch": 0.3751668410230511,
      "grad_norm": 3.1607289779604866,
      "learning_rate": 2.8750990254562945e-06,
      "loss": 0.9605,
      "step": 5200
    },
    {
      "epoch": 0.3752389884924786,
      "grad_norm": 5.157360165466707,
      "learning_rate": 2.8746787542794857e-06,
      "loss": 0.9521,
      "step": 5201
    },
    {
      "epoch": 0.37531113596190613,
      "grad_norm": 3.871562294266834,
      "learning_rate": 2.8742584353397075e-06,
      "loss": 1.0727,
      "step": 5202
    },
    {
      "epoch": 0.37538328343133365,
      "grad_norm": 5.631038730396687,
      "learning_rate": 2.8738380686599123e-06,
      "loss": 1.0272,
      "step": 5203
    },
    {
      "epoch": 0.37545543090076117,
      "grad_norm": 0.7608876707458079,
      "learning_rate": 2.873417654263056e-06,
      "loss": 0.8292,
      "step": 5204
    },
    {
      "epoch": 0.3755275783701887,
      "grad_norm": 5.073361836143227,
      "learning_rate": 2.8729971921720945e-06,
      "loss": 1.0134,
      "step": 5205
    },
    {
      "epoch": 0.37559972583961615,
      "grad_norm": 3.5513160898677842,
      "learning_rate": 2.8725766824099874e-06,
      "loss": 0.9679,
      "step": 5206
    },
    {
      "epoch": 0.37567187330904367,
      "grad_norm": 3.15217811120939,
      "learning_rate": 2.8721561249996982e-06,
      "loss": 1.0057,
      "step": 5207
    },
    {
      "epoch": 0.3757440207784712,
      "grad_norm": 4.142339617721478,
      "learning_rate": 2.8717355199641917e-06,
      "loss": 0.8883,
      "step": 5208
    },
    {
      "epoch": 0.3758161682478987,
      "grad_norm": 3.421974912256059,
      "learning_rate": 2.8713148673264357e-06,
      "loss": 1.0681,
      "step": 5209
    },
    {
      "epoch": 0.3758883157173262,
      "grad_norm": 3.081763581037517,
      "learning_rate": 2.8708941671094003e-06,
      "loss": 0.9659,
      "step": 5210
    },
    {
      "epoch": 0.37596046318675375,
      "grad_norm": 6.211135266112413,
      "learning_rate": 2.870473419336058e-06,
      "loss": 1.0003,
      "step": 5211
    },
    {
      "epoch": 0.3760326106561812,
      "grad_norm": 5.18026633502157,
      "learning_rate": 2.8700526240293848e-06,
      "loss": 0.9968,
      "step": 5212
    },
    {
      "epoch": 0.37610475812560873,
      "grad_norm": 3.2476281534447375,
      "learning_rate": 2.8696317812123588e-06,
      "loss": 1.0004,
      "step": 5213
    },
    {
      "epoch": 0.37617690559503625,
      "grad_norm": 3.2970231009389366,
      "learning_rate": 2.8692108909079603e-06,
      "loss": 1.0566,
      "step": 5214
    },
    {
      "epoch": 0.37624905306446377,
      "grad_norm": 4.944957268417454,
      "learning_rate": 2.8687899531391724e-06,
      "loss": 0.9058,
      "step": 5215
    },
    {
      "epoch": 0.3763212005338913,
      "grad_norm": 3.163760926745117,
      "learning_rate": 2.8683689679289822e-06,
      "loss": 1.0137,
      "step": 5216
    },
    {
      "epoch": 0.3763933480033188,
      "grad_norm": 4.328935567924882,
      "learning_rate": 2.8679479353003765e-06,
      "loss": 0.9955,
      "step": 5217
    },
    {
      "epoch": 0.37646549547274627,
      "grad_norm": 4.280233889098321,
      "learning_rate": 2.8675268552763477e-06,
      "loss": 0.9755,
      "step": 5218
    },
    {
      "epoch": 0.3765376429421738,
      "grad_norm": 4.917565387549037,
      "learning_rate": 2.8671057278798883e-06,
      "loss": 0.9147,
      "step": 5219
    },
    {
      "epoch": 0.3766097904116013,
      "grad_norm": 5.651214550086847,
      "learning_rate": 2.8666845531339957e-06,
      "loss": 0.943,
      "step": 5220
    },
    {
      "epoch": 0.3766819378810288,
      "grad_norm": 5.292491978642399,
      "learning_rate": 2.866263331061667e-06,
      "loss": 1.0078,
      "step": 5221
    },
    {
      "epoch": 0.37675408535045635,
      "grad_norm": 5.375691391590673,
      "learning_rate": 2.865842061685905e-06,
      "loss": 0.8309,
      "step": 5222
    },
    {
      "epoch": 0.37682623281988387,
      "grad_norm": 3.5695480763737386,
      "learning_rate": 2.8654207450297134e-06,
      "loss": 1.048,
      "step": 5223
    },
    {
      "epoch": 0.37689838028931133,
      "grad_norm": 5.921754860980394,
      "learning_rate": 2.864999381116098e-06,
      "loss": 0.9926,
      "step": 5224
    },
    {
      "epoch": 0.37697052775873885,
      "grad_norm": 3.705733012637271,
      "learning_rate": 2.86457796996807e-06,
      "loss": 1.0517,
      "step": 5225
    },
    {
      "epoch": 0.37704267522816637,
      "grad_norm": 22.798581745140282,
      "learning_rate": 2.8641565116086384e-06,
      "loss": 0.9766,
      "step": 5226
    },
    {
      "epoch": 0.3771148226975939,
      "grad_norm": 8.424794216642532,
      "learning_rate": 2.8637350060608186e-06,
      "loss": 1.0172,
      "step": 5227
    },
    {
      "epoch": 0.3771869701670214,
      "grad_norm": 5.119770452300858,
      "learning_rate": 2.8633134533476284e-06,
      "loss": 0.9421,
      "step": 5228
    },
    {
      "epoch": 0.3772591176364489,
      "grad_norm": 5.170376747910352,
      "learning_rate": 2.8628918534920856e-06,
      "loss": 1.0002,
      "step": 5229
    },
    {
      "epoch": 0.3773312651058764,
      "grad_norm": 4.132052317403516,
      "learning_rate": 2.862470206517213e-06,
      "loss": 1.0552,
      "step": 5230
    },
    {
      "epoch": 0.3774034125753039,
      "grad_norm": 3.0786940924787376,
      "learning_rate": 2.862048512446036e-06,
      "loss": 1.0027,
      "step": 5231
    },
    {
      "epoch": 0.37747556004473143,
      "grad_norm": 5.645725300823915,
      "learning_rate": 2.8616267713015806e-06,
      "loss": 1.0872,
      "step": 5232
    },
    {
      "epoch": 0.37754770751415895,
      "grad_norm": 3.4954178652274566,
      "learning_rate": 2.8612049831068767e-06,
      "loss": 1.1213,
      "step": 5233
    },
    {
      "epoch": 0.37761985498358647,
      "grad_norm": 2.4175372254780827,
      "learning_rate": 2.860783147884956e-06,
      "loss": 0.9487,
      "step": 5234
    },
    {
      "epoch": 0.377692002453014,
      "grad_norm": 5.5795770032245375,
      "learning_rate": 2.860361265658855e-06,
      "loss": 0.9694,
      "step": 5235
    },
    {
      "epoch": 0.37776414992244145,
      "grad_norm": 5.799972652502371,
      "learning_rate": 2.8599393364516103e-06,
      "loss": 0.9182,
      "step": 5236
    },
    {
      "epoch": 0.37783629739186897,
      "grad_norm": 3.561553026122978,
      "learning_rate": 2.8595173602862613e-06,
      "loss": 1.0306,
      "step": 5237
    },
    {
      "epoch": 0.3779084448612965,
      "grad_norm": 5.772346527716016,
      "learning_rate": 2.8590953371858515e-06,
      "loss": 0.9846,
      "step": 5238
    },
    {
      "epoch": 0.377980592330724,
      "grad_norm": 3.2681412197402864,
      "learning_rate": 2.858673267173425e-06,
      "loss": 1.051,
      "step": 5239
    },
    {
      "epoch": 0.3780527398001515,
      "grad_norm": 4.462726080801468,
      "learning_rate": 2.8582511502720313e-06,
      "loss": 0.9748,
      "step": 5240
    },
    {
      "epoch": 0.37812488726957905,
      "grad_norm": 4.675543834041444,
      "learning_rate": 2.8578289865047184e-06,
      "loss": 0.9971,
      "step": 5241
    },
    {
      "epoch": 0.3781970347390065,
      "grad_norm": 3.755986458011545,
      "learning_rate": 2.85740677589454e-06,
      "loss": 0.9101,
      "step": 5242
    },
    {
      "epoch": 0.37826918220843403,
      "grad_norm": 4.953657151766369,
      "learning_rate": 2.8569845184645526e-06,
      "loss": 1.015,
      "step": 5243
    },
    {
      "epoch": 0.37834132967786155,
      "grad_norm": 3.9924757045324317,
      "learning_rate": 2.856562214237812e-06,
      "loss": 0.8957,
      "step": 5244
    },
    {
      "epoch": 0.37841347714728907,
      "grad_norm": 4.117559021721885,
      "learning_rate": 2.8561398632373805e-06,
      "loss": 1.0543,
      "step": 5245
    },
    {
      "epoch": 0.3784856246167166,
      "grad_norm": 5.51956753405006,
      "learning_rate": 2.85571746548632e-06,
      "loss": 0.9709,
      "step": 5246
    },
    {
      "epoch": 0.37855777208614405,
      "grad_norm": 4.539092964949937,
      "learning_rate": 2.8552950210076966e-06,
      "loss": 0.9347,
      "step": 5247
    },
    {
      "epoch": 0.37862991955557157,
      "grad_norm": 3.5861085007396456,
      "learning_rate": 2.8548725298245783e-06,
      "loss": 1.0656,
      "step": 5248
    },
    {
      "epoch": 0.3787020670249991,
      "grad_norm": 3.3812263417429893,
      "learning_rate": 2.8544499919600355e-06,
      "loss": 0.947,
      "step": 5249
    },
    {
      "epoch": 0.3787742144944266,
      "grad_norm": 5.612351501811234,
      "learning_rate": 2.854027407437142e-06,
      "loss": 0.9619,
      "step": 5250
    },
    {
      "epoch": 0.37884636196385413,
      "grad_norm": 10.13706872589486,
      "learning_rate": 2.8536047762789733e-06,
      "loss": 1.0415,
      "step": 5251
    },
    {
      "epoch": 0.37891850943328165,
      "grad_norm": 5.19661656125333,
      "learning_rate": 2.853182098508608e-06,
      "loss": 1.062,
      "step": 5252
    },
    {
      "epoch": 0.3789906569027091,
      "grad_norm": 3.546249401696215,
      "learning_rate": 2.8527593741491255e-06,
      "loss": 0.9861,
      "step": 5253
    },
    {
      "epoch": 0.37906280437213663,
      "grad_norm": 3.131933379077372,
      "learning_rate": 2.8523366032236117e-06,
      "loss": 1.0137,
      "step": 5254
    },
    {
      "epoch": 0.37913495184156415,
      "grad_norm": 5.386591986588595,
      "learning_rate": 2.8519137857551507e-06,
      "loss": 0.934,
      "step": 5255
    },
    {
      "epoch": 0.37920709931099167,
      "grad_norm": 3.0438243755050087,
      "learning_rate": 2.8514909217668318e-06,
      "loss": 1.0315,
      "step": 5256
    },
    {
      "epoch": 0.3792792467804192,
      "grad_norm": 6.25126757165628,
      "learning_rate": 2.851068011281745e-06,
      "loss": 1.0295,
      "step": 5257
    },
    {
      "epoch": 0.3793513942498467,
      "grad_norm": 4.3812054902562565,
      "learning_rate": 2.8506450543229856e-06,
      "loss": 0.9911,
      "step": 5258
    },
    {
      "epoch": 0.37942354171927417,
      "grad_norm": 4.006783455549846,
      "learning_rate": 2.8502220509136483e-06,
      "loss": 0.9233,
      "step": 5259
    },
    {
      "epoch": 0.3794956891887017,
      "grad_norm": 5.056695506835718,
      "learning_rate": 2.849799001076832e-06,
      "loss": 0.9802,
      "step": 5260
    },
    {
      "epoch": 0.3795678366581292,
      "grad_norm": 11.99109382736862,
      "learning_rate": 2.8493759048356387e-06,
      "loss": 1.0245,
      "step": 5261
    },
    {
      "epoch": 0.37963998412755673,
      "grad_norm": 6.058689453628385,
      "learning_rate": 2.8489527622131708e-06,
      "loss": 0.8906,
      "step": 5262
    },
    {
      "epoch": 0.37971213159698425,
      "grad_norm": 0.7739454585801916,
      "learning_rate": 2.8485295732325355e-06,
      "loss": 0.8287,
      "step": 5263
    },
    {
      "epoch": 0.37978427906641177,
      "grad_norm": 3.8009814751667164,
      "learning_rate": 2.8481063379168415e-06,
      "loss": 0.9462,
      "step": 5264
    },
    {
      "epoch": 0.37985642653583923,
      "grad_norm": 4.52425481234818,
      "learning_rate": 2.8476830562892e-06,
      "loss": 1.0158,
      "step": 5265
    },
    {
      "epoch": 0.37992857400526675,
      "grad_norm": 0.7487680330477177,
      "learning_rate": 2.847259728372725e-06,
      "loss": 0.8567,
      "step": 5266
    },
    {
      "epoch": 0.38000072147469427,
      "grad_norm": 7.450729784968267,
      "learning_rate": 2.8468363541905324e-06,
      "loss": 1.024,
      "step": 5267
    },
    {
      "epoch": 0.3800728689441218,
      "grad_norm": 4.916333181849665,
      "learning_rate": 2.8464129337657413e-06,
      "loss": 1.0769,
      "step": 5268
    },
    {
      "epoch": 0.3801450164135493,
      "grad_norm": 6.391137734435657,
      "learning_rate": 2.8459894671214742e-06,
      "loss": 1.0655,
      "step": 5269
    },
    {
      "epoch": 0.3802171638829768,
      "grad_norm": 6.437121926473745,
      "learning_rate": 2.845565954280853e-06,
      "loss": 0.932,
      "step": 5270
    },
    {
      "epoch": 0.3802893113524043,
      "grad_norm": 6.230217104048158,
      "learning_rate": 2.845142395267005e-06,
      "loss": 0.8225,
      "step": 5271
    },
    {
      "epoch": 0.3803614588218318,
      "grad_norm": 7.6759715903563,
      "learning_rate": 2.8447187901030596e-06,
      "loss": 0.8495,
      "step": 5272
    },
    {
      "epoch": 0.38043360629125933,
      "grad_norm": 3.3007647813013716,
      "learning_rate": 2.8442951388121487e-06,
      "loss": 0.9813,
      "step": 5273
    },
    {
      "epoch": 0.38050575376068685,
      "grad_norm": 4.121530171648698,
      "learning_rate": 2.8438714414174048e-06,
      "loss": 0.9255,
      "step": 5274
    },
    {
      "epoch": 0.38057790123011437,
      "grad_norm": 4.569192028449525,
      "learning_rate": 2.8434476979419666e-06,
      "loss": 1.0796,
      "step": 5275
    },
    {
      "epoch": 0.3806500486995419,
      "grad_norm": 3.0743169320507358,
      "learning_rate": 2.8430239084089713e-06,
      "loss": 0.9194,
      "step": 5276
    },
    {
      "epoch": 0.38072219616896935,
      "grad_norm": 3.599531879718916,
      "learning_rate": 2.8426000728415606e-06,
      "loss": 0.9321,
      "step": 5277
    },
    {
      "epoch": 0.38079434363839687,
      "grad_norm": 4.87269733528454,
      "learning_rate": 2.8421761912628794e-06,
      "loss": 0.8722,
      "step": 5278
    },
    {
      "epoch": 0.3808664911078244,
      "grad_norm": 3.0738103224461435,
      "learning_rate": 2.841752263696075e-06,
      "loss": 0.9063,
      "step": 5279
    },
    {
      "epoch": 0.3809386385772519,
      "grad_norm": 0.751080886479899,
      "learning_rate": 2.8413282901642945e-06,
      "loss": 0.7895,
      "step": 5280
    },
    {
      "epoch": 0.38101078604667943,
      "grad_norm": 6.605799652600203,
      "learning_rate": 2.8409042706906912e-06,
      "loss": 0.9677,
      "step": 5281
    },
    {
      "epoch": 0.38108293351610695,
      "grad_norm": 6.182168532454601,
      "learning_rate": 2.8404802052984184e-06,
      "loss": 0.9348,
      "step": 5282
    },
    {
      "epoch": 0.3811550809855344,
      "grad_norm": 10.42060077890242,
      "learning_rate": 2.8400560940106327e-06,
      "loss": 0.9872,
      "step": 5283
    },
    {
      "epoch": 0.38122722845496193,
      "grad_norm": 3.583348754731343,
      "learning_rate": 2.839631936850493e-06,
      "loss": 1.0401,
      "step": 5284
    },
    {
      "epoch": 0.38129937592438945,
      "grad_norm": 2.918040487896195,
      "learning_rate": 2.8392077338411624e-06,
      "loss": 1.025,
      "step": 5285
    },
    {
      "epoch": 0.38137152339381697,
      "grad_norm": 7.105727783984403,
      "learning_rate": 2.8387834850058042e-06,
      "loss": 1.0168,
      "step": 5286
    },
    {
      "epoch": 0.3814436708632445,
      "grad_norm": 3.806246021187242,
      "learning_rate": 2.838359190367584e-06,
      "loss": 0.8817,
      "step": 5287
    },
    {
      "epoch": 0.381515818332672,
      "grad_norm": 5.504789867634949,
      "learning_rate": 2.8379348499496727e-06,
      "loss": 1.0208,
      "step": 5288
    },
    {
      "epoch": 0.38158796580209947,
      "grad_norm": 6.292426495007181,
      "learning_rate": 2.837510463775241e-06,
      "loss": 1.0282,
      "step": 5289
    },
    {
      "epoch": 0.381660113271527,
      "grad_norm": 3.2422294154387696,
      "learning_rate": 2.8370860318674627e-06,
      "loss": 0.9842,
      "step": 5290
    },
    {
      "epoch": 0.3817322607409545,
      "grad_norm": 2.628046538534585,
      "learning_rate": 2.836661554249516e-06,
      "loss": 0.9336,
      "step": 5291
    },
    {
      "epoch": 0.38180440821038203,
      "grad_norm": 4.099151542146071,
      "learning_rate": 2.836237030944578e-06,
      "loss": 0.8665,
      "step": 5292
    },
    {
      "epoch": 0.38187655567980955,
      "grad_norm": 3.451105830667581,
      "learning_rate": 2.835812461975832e-06,
      "loss": 1.0084,
      "step": 5293
    },
    {
      "epoch": 0.38194870314923707,
      "grad_norm": 5.121299337978565,
      "learning_rate": 2.8353878473664612e-06,
      "loss": 1.0243,
      "step": 5294
    },
    {
      "epoch": 0.38202085061866453,
      "grad_norm": 5.284057593217309,
      "learning_rate": 2.834963187139653e-06,
      "loss": 0.9737,
      "step": 5295
    },
    {
      "epoch": 0.38209299808809205,
      "grad_norm": 3.585402770886415,
      "learning_rate": 2.8345384813185956e-06,
      "loss": 1.0194,
      "step": 5296
    },
    {
      "epoch": 0.38216514555751957,
      "grad_norm": 2.3817539552888194,
      "learning_rate": 2.834113729926481e-06,
      "loss": 0.9688,
      "step": 5297
    },
    {
      "epoch": 0.3822372930269471,
      "grad_norm": 4.9521918636246705,
      "learning_rate": 2.833688932986504e-06,
      "loss": 0.9427,
      "step": 5298
    },
    {
      "epoch": 0.3823094404963746,
      "grad_norm": 3.37700395783025,
      "learning_rate": 2.83326409052186e-06,
      "loss": 0.9948,
      "step": 5299
    },
    {
      "epoch": 0.38238158796580207,
      "grad_norm": 3.628577998727143,
      "learning_rate": 2.832839202555749e-06,
      "loss": 1.0385,
      "step": 5300
    },
    {
      "epoch": 0.3824537354352296,
      "grad_norm": 8.237286568685615,
      "learning_rate": 2.8324142691113723e-06,
      "loss": 1.0267,
      "step": 5301
    },
    {
      "epoch": 0.3825258829046571,
      "grad_norm": 3.5638525553557794,
      "learning_rate": 2.831989290211933e-06,
      "loss": 0.9796,
      "step": 5302
    },
    {
      "epoch": 0.38259803037408463,
      "grad_norm": 3.8956630213972225,
      "learning_rate": 2.83156426588064e-06,
      "loss": 1.0521,
      "step": 5303
    },
    {
      "epoch": 0.38267017784351215,
      "grad_norm": 3.287192059733237,
      "learning_rate": 2.8311391961406996e-06,
      "loss": 0.9903,
      "step": 5304
    },
    {
      "epoch": 0.38274232531293967,
      "grad_norm": 2.787175926905579,
      "learning_rate": 2.830714081015325e-06,
      "loss": 0.9357,
      "step": 5305
    },
    {
      "epoch": 0.38281447278236713,
      "grad_norm": 3.4146521297007864,
      "learning_rate": 2.8302889205277293e-06,
      "loss": 0.9967,
      "step": 5306
    },
    {
      "epoch": 0.38288662025179465,
      "grad_norm": 5.860929806733987,
      "learning_rate": 2.829863714701129e-06,
      "loss": 0.9688,
      "step": 5307
    },
    {
      "epoch": 0.38295876772122217,
      "grad_norm": 8.859275965330509,
      "learning_rate": 2.8294384635587443e-06,
      "loss": 0.906,
      "step": 5308
    },
    {
      "epoch": 0.3830309151906497,
      "grad_norm": 3.89738349197203,
      "learning_rate": 2.829013167123795e-06,
      "loss": 0.995,
      "step": 5309
    },
    {
      "epoch": 0.3831030626600772,
      "grad_norm": 4.120998406776244,
      "learning_rate": 2.828587825419506e-06,
      "loss": 0.9789,
      "step": 5310
    },
    {
      "epoch": 0.38317521012950473,
      "grad_norm": 4.6366257822925,
      "learning_rate": 2.828162438469102e-06,
      "loss": 1.0413,
      "step": 5311
    },
    {
      "epoch": 0.3832473575989322,
      "grad_norm": 3.408272798804306,
      "learning_rate": 2.8277370062958134e-06,
      "loss": 1.0244,
      "step": 5312
    },
    {
      "epoch": 0.3833195050683597,
      "grad_norm": 0.8362169957169813,
      "learning_rate": 2.8273115289228715e-06,
      "loss": 0.877,
      "step": 5313
    },
    {
      "epoch": 0.38339165253778723,
      "grad_norm": 9.969507266373666,
      "learning_rate": 2.8268860063735096e-06,
      "loss": 0.9715,
      "step": 5314
    },
    {
      "epoch": 0.38346380000721475,
      "grad_norm": 3.359352998883261,
      "learning_rate": 2.826460438670964e-06,
      "loss": 0.9374,
      "step": 5315
    },
    {
      "epoch": 0.38353594747664227,
      "grad_norm": 4.211206544338811,
      "learning_rate": 2.826034825838473e-06,
      "loss": 0.9879,
      "step": 5316
    },
    {
      "epoch": 0.3836080949460698,
      "grad_norm": 9.101765773615245,
      "learning_rate": 2.825609167899278e-06,
      "loss": 0.8906,
      "step": 5317
    },
    {
      "epoch": 0.38368024241549725,
      "grad_norm": 3.9288491782554487,
      "learning_rate": 2.825183464876623e-06,
      "loss": 1.1016,
      "step": 5318
    },
    {
      "epoch": 0.38375238988492477,
      "grad_norm": 3.8555450586499163,
      "learning_rate": 2.8247577167937533e-06,
      "loss": 0.9184,
      "step": 5319
    },
    {
      "epoch": 0.3838245373543523,
      "grad_norm": 3.461037662339761,
      "learning_rate": 2.8243319236739186e-06,
      "loss": 0.9885,
      "step": 5320
    },
    {
      "epoch": 0.3838966848237798,
      "grad_norm": 3.24617630432404,
      "learning_rate": 2.823906085540368e-06,
      "loss": 0.9802,
      "step": 5321
    },
    {
      "epoch": 0.38396883229320733,
      "grad_norm": 0.7820564685150869,
      "learning_rate": 2.8234802024163566e-06,
      "loss": 0.8816,
      "step": 5322
    },
    {
      "epoch": 0.38404097976263485,
      "grad_norm": 3.3892013443979176,
      "learning_rate": 2.8230542743251396e-06,
      "loss": 0.8749,
      "step": 5323
    },
    {
      "epoch": 0.3841131272320623,
      "grad_norm": 2.837479101431764,
      "learning_rate": 2.8226283012899757e-06,
      "loss": 1.0443,
      "step": 5324
    },
    {
      "epoch": 0.38418527470148983,
      "grad_norm": 3.992923317898758,
      "learning_rate": 2.822202283334126e-06,
      "loss": 0.989,
      "step": 5325
    },
    {
      "epoch": 0.38425742217091735,
      "grad_norm": 3.75667880081433,
      "learning_rate": 2.8217762204808525e-06,
      "loss": 1.0683,
      "step": 5326
    },
    {
      "epoch": 0.38432956964034487,
      "grad_norm": 3.8839173615893623,
      "learning_rate": 2.821350112753422e-06,
      "loss": 0.9776,
      "step": 5327
    },
    {
      "epoch": 0.3844017171097724,
      "grad_norm": 4.81207610406095,
      "learning_rate": 2.8209239601751025e-06,
      "loss": 0.9563,
      "step": 5328
    },
    {
      "epoch": 0.3844738645791999,
      "grad_norm": 3.536971596657446,
      "learning_rate": 2.820497762769164e-06,
      "loss": 0.9355,
      "step": 5329
    },
    {
      "epoch": 0.3845460120486274,
      "grad_norm": 3.9788889496773776,
      "learning_rate": 2.8200715205588807e-06,
      "loss": 0.9668,
      "step": 5330
    },
    {
      "epoch": 0.3846181595180549,
      "grad_norm": 4.108438943636484,
      "learning_rate": 2.8196452335675275e-06,
      "loss": 0.9553,
      "step": 5331
    },
    {
      "epoch": 0.3846903069874824,
      "grad_norm": 4.129491730234272,
      "learning_rate": 2.819218901818382e-06,
      "loss": 1.034,
      "step": 5332
    },
    {
      "epoch": 0.38476245445690993,
      "grad_norm": 3.6958625496989317,
      "learning_rate": 2.818792525334725e-06,
      "loss": 0.8546,
      "step": 5333
    },
    {
      "epoch": 0.38483460192633745,
      "grad_norm": 3.9921262254883594,
      "learning_rate": 2.818366104139839e-06,
      "loss": 1.0161,
      "step": 5334
    },
    {
      "epoch": 0.38490674939576497,
      "grad_norm": 4.2859736091630785,
      "learning_rate": 2.81793963825701e-06,
      "loss": 1.0539,
      "step": 5335
    },
    {
      "epoch": 0.38497889686519243,
      "grad_norm": 4.648763693255705,
      "learning_rate": 2.817513127709525e-06,
      "loss": 1.0821,
      "step": 5336
    },
    {
      "epoch": 0.38505104433461995,
      "grad_norm": 3.961054030891845,
      "learning_rate": 2.817086572520674e-06,
      "loss": 0.9361,
      "step": 5337
    },
    {
      "epoch": 0.38512319180404747,
      "grad_norm": 3.683505335340368,
      "learning_rate": 2.8166599727137508e-06,
      "loss": 1.064,
      "step": 5338
    },
    {
      "epoch": 0.385195339273475,
      "grad_norm": 4.13647598548465,
      "learning_rate": 2.816233328312049e-06,
      "loss": 1.0634,
      "step": 5339
    },
    {
      "epoch": 0.3852674867429025,
      "grad_norm": 3.1229491561478815,
      "learning_rate": 2.815806639338867e-06,
      "loss": 1.0389,
      "step": 5340
    },
    {
      "epoch": 0.38533963421233003,
      "grad_norm": 3.5861961254861203,
      "learning_rate": 2.8153799058175046e-06,
      "loss": 0.9941,
      "step": 5341
    },
    {
      "epoch": 0.3854117816817575,
      "grad_norm": 4.4594955117957635,
      "learning_rate": 2.814953127771264e-06,
      "loss": 1.0079,
      "step": 5342
    },
    {
      "epoch": 0.385483929151185,
      "grad_norm": 3.22059519201158,
      "learning_rate": 2.8145263052234498e-06,
      "loss": 0.9455,
      "step": 5343
    },
    {
      "epoch": 0.38555607662061253,
      "grad_norm": 6.150981061886115,
      "learning_rate": 2.8140994381973694e-06,
      "loss": 1.1138,
      "step": 5344
    },
    {
      "epoch": 0.38562822409004005,
      "grad_norm": 2.6448979764418388,
      "learning_rate": 2.8136725267163323e-06,
      "loss": 0.9803,
      "step": 5345
    },
    {
      "epoch": 0.38570037155946757,
      "grad_norm": 2.922919591538336,
      "learning_rate": 2.813245570803651e-06,
      "loss": 1.1024,
      "step": 5346
    },
    {
      "epoch": 0.3857725190288951,
      "grad_norm": 6.329393499887332,
      "learning_rate": 2.8128185704826387e-06,
      "loss": 1.0087,
      "step": 5347
    },
    {
      "epoch": 0.38584466649832255,
      "grad_norm": 2.6178331176630785,
      "learning_rate": 2.812391525776614e-06,
      "loss": 0.9574,
      "step": 5348
    },
    {
      "epoch": 0.38591681396775007,
      "grad_norm": 5.906431165066263,
      "learning_rate": 2.811964436708895e-06,
      "loss": 1.0741,
      "step": 5349
    },
    {
      "epoch": 0.3859889614371776,
      "grad_norm": 3.3976984031520128,
      "learning_rate": 2.8115373033028045e-06,
      "loss": 0.9704,
      "step": 5350
    },
    {
      "epoch": 0.3860611089066051,
      "grad_norm": 4.61725607367648,
      "learning_rate": 2.811110125581665e-06,
      "loss": 0.9945,
      "step": 5351
    },
    {
      "epoch": 0.38613325637603263,
      "grad_norm": 5.125748184299785,
      "learning_rate": 2.8106829035688055e-06,
      "loss": 0.9338,
      "step": 5352
    },
    {
      "epoch": 0.3862054038454601,
      "grad_norm": 7.066717728244393,
      "learning_rate": 2.810255637287553e-06,
      "loss": 1.0518,
      "step": 5353
    },
    {
      "epoch": 0.3862775513148876,
      "grad_norm": 5.465188921483846,
      "learning_rate": 2.8098283267612398e-06,
      "loss": 0.9677,
      "step": 5354
    },
    {
      "epoch": 0.38634969878431513,
      "grad_norm": 3.1238784303248313,
      "learning_rate": 2.8094009720131988e-06,
      "loss": 0.9029,
      "step": 5355
    },
    {
      "epoch": 0.38642184625374265,
      "grad_norm": 2.6917412581211044,
      "learning_rate": 2.808973573066768e-06,
      "loss": 1.0231,
      "step": 5356
    },
    {
      "epoch": 0.38649399372317017,
      "grad_norm": 3.9462485836438552,
      "learning_rate": 2.8085461299452848e-06,
      "loss": 0.9988,
      "step": 5357
    },
    {
      "epoch": 0.3865661411925977,
      "grad_norm": 3.3119540934306726,
      "learning_rate": 2.8081186426720905e-06,
      "loss": 1.0952,
      "step": 5358
    },
    {
      "epoch": 0.38663828866202515,
      "grad_norm": 5.77864205588214,
      "learning_rate": 2.8076911112705286e-06,
      "loss": 0.9373,
      "step": 5359
    },
    {
      "epoch": 0.3867104361314527,
      "grad_norm": 3.8508822631589132,
      "learning_rate": 2.8072635357639455e-06,
      "loss": 1.0082,
      "step": 5360
    },
    {
      "epoch": 0.3867825836008802,
      "grad_norm": 2.7207334452388943,
      "learning_rate": 2.8068359161756884e-06,
      "loss": 1.0513,
      "step": 5361
    },
    {
      "epoch": 0.3868547310703077,
      "grad_norm": 5.244216458076008,
      "learning_rate": 2.806408252529109e-06,
      "loss": 1.0347,
      "step": 5362
    },
    {
      "epoch": 0.38692687853973523,
      "grad_norm": 0.7891360333615667,
      "learning_rate": 2.8059805448475607e-06,
      "loss": 0.8767,
      "step": 5363
    },
    {
      "epoch": 0.38699902600916275,
      "grad_norm": 3.240659496156876,
      "learning_rate": 2.805552793154398e-06,
      "loss": 1.014,
      "step": 5364
    },
    {
      "epoch": 0.3870711734785902,
      "grad_norm": 2.473629920978867,
      "learning_rate": 2.805124997472979e-06,
      "loss": 0.9329,
      "step": 5365
    },
    {
      "epoch": 0.38714332094801773,
      "grad_norm": 4.059630334119971,
      "learning_rate": 2.804697157826665e-06,
      "loss": 0.8576,
      "step": 5366
    },
    {
      "epoch": 0.38721546841744525,
      "grad_norm": 4.022937336175026,
      "learning_rate": 2.8042692742388173e-06,
      "loss": 0.8951,
      "step": 5367
    },
    {
      "epoch": 0.38728761588687277,
      "grad_norm": 2.765399471074503,
      "learning_rate": 2.8038413467328025e-06,
      "loss": 0.9603,
      "step": 5368
    },
    {
      "epoch": 0.3873597633563003,
      "grad_norm": 4.608192007554721,
      "learning_rate": 2.8034133753319874e-06,
      "loss": 1.0839,
      "step": 5369
    },
    {
      "epoch": 0.3874319108257278,
      "grad_norm": 3.4594115602513504,
      "learning_rate": 2.802985360059742e-06,
      "loss": 0.927,
      "step": 5370
    },
    {
      "epoch": 0.3875040582951553,
      "grad_norm": 4.186519678860279,
      "learning_rate": 2.802557300939439e-06,
      "loss": 0.951,
      "step": 5371
    },
    {
      "epoch": 0.3875762057645828,
      "grad_norm": 3.6794011030896034,
      "learning_rate": 2.802129197994452e-06,
      "loss": 1.0641,
      "step": 5372
    },
    {
      "epoch": 0.3876483532340103,
      "grad_norm": 2.7092375688555426,
      "learning_rate": 2.801701051248159e-06,
      "loss": 1.0237,
      "step": 5373
    },
    {
      "epoch": 0.38772050070343783,
      "grad_norm": 3.2347064009718536,
      "learning_rate": 2.80127286072394e-06,
      "loss": 0.934,
      "step": 5374
    },
    {
      "epoch": 0.38779264817286535,
      "grad_norm": 3.6789159917631555,
      "learning_rate": 2.8008446264451758e-06,
      "loss": 0.904,
      "step": 5375
    },
    {
      "epoch": 0.38786479564229287,
      "grad_norm": 2.6119448861445824,
      "learning_rate": 2.8004163484352516e-06,
      "loss": 0.9376,
      "step": 5376
    },
    {
      "epoch": 0.38793694311172033,
      "grad_norm": 3.247072221568988,
      "learning_rate": 2.799988026717553e-06,
      "loss": 0.9275,
      "step": 5377
    },
    {
      "epoch": 0.38800909058114785,
      "grad_norm": 6.178487668465865,
      "learning_rate": 2.7995596613154697e-06,
      "loss": 0.9505,
      "step": 5378
    },
    {
      "epoch": 0.3880812380505754,
      "grad_norm": 4.231893452750739,
      "learning_rate": 2.7991312522523937e-06,
      "loss": 0.9387,
      "step": 5379
    },
    {
      "epoch": 0.3881533855200029,
      "grad_norm": 2.709908237804025,
      "learning_rate": 2.798702799551718e-06,
      "loss": 0.9396,
      "step": 5380
    },
    {
      "epoch": 0.3882255329894304,
      "grad_norm": 3.807441983297547,
      "learning_rate": 2.7982743032368393e-06,
      "loss": 0.9261,
      "step": 5381
    },
    {
      "epoch": 0.38829768045885793,
      "grad_norm": 3.5424559368227935,
      "learning_rate": 2.797845763331156e-06,
      "loss": 0.922,
      "step": 5382
    },
    {
      "epoch": 0.3883698279282854,
      "grad_norm": 3.4452603221583944,
      "learning_rate": 2.797417179858069e-06,
      "loss": 0.873,
      "step": 5383
    },
    {
      "epoch": 0.3884419753977129,
      "grad_norm": 2.6780989584441204,
      "learning_rate": 2.7969885528409814e-06,
      "loss": 1.0722,
      "step": 5384
    },
    {
      "epoch": 0.38851412286714043,
      "grad_norm": 2.5875141604123586,
      "learning_rate": 2.7965598823033e-06,
      "loss": 1.1062,
      "step": 5385
    },
    {
      "epoch": 0.38858627033656795,
      "grad_norm": 4.5203003224888905,
      "learning_rate": 2.7961311682684317e-06,
      "loss": 1.0404,
      "step": 5386
    },
    {
      "epoch": 0.38865841780599547,
      "grad_norm": 4.247599653183043,
      "learning_rate": 2.795702410759788e-06,
      "loss": 0.8948,
      "step": 5387
    },
    {
      "epoch": 0.388730565275423,
      "grad_norm": 3.7917971466306435,
      "learning_rate": 2.7952736098007803e-06,
      "loss": 1.0023,
      "step": 5388
    },
    {
      "epoch": 0.38880271274485045,
      "grad_norm": 2.992794443430327,
      "learning_rate": 2.794844765414825e-06,
      "loss": 0.93,
      "step": 5389
    },
    {
      "epoch": 0.388874860214278,
      "grad_norm": 4.794473152374293,
      "learning_rate": 2.79441587762534e-06,
      "loss": 0.8359,
      "step": 5390
    },
    {
      "epoch": 0.3889470076837055,
      "grad_norm": 3.8477883563139685,
      "learning_rate": 2.7939869464557445e-06,
      "loss": 1.0152,
      "step": 5391
    },
    {
      "epoch": 0.389019155153133,
      "grad_norm": 5.425482789104859,
      "learning_rate": 2.793557971929461e-06,
      "loss": 0.9274,
      "step": 5392
    },
    {
      "epoch": 0.38909130262256053,
      "grad_norm": 4.135671510058715,
      "learning_rate": 2.7931289540699147e-06,
      "loss": 0.9872,
      "step": 5393
    },
    {
      "epoch": 0.38916345009198805,
      "grad_norm": 3.342566450053381,
      "learning_rate": 2.792699892900532e-06,
      "loss": 0.9712,
      "step": 5394
    },
    {
      "epoch": 0.3892355975614155,
      "grad_norm": 3.1857245493350237,
      "learning_rate": 2.7922707884447425e-06,
      "loss": 1.02,
      "step": 5395
    },
    {
      "epoch": 0.38930774503084303,
      "grad_norm": 2.9734585014078685,
      "learning_rate": 2.7918416407259783e-06,
      "loss": 1.0473,
      "step": 5396
    },
    {
      "epoch": 0.38937989250027055,
      "grad_norm": 4.636315396511827,
      "learning_rate": 2.791412449767673e-06,
      "loss": 1.006,
      "step": 5397
    },
    {
      "epoch": 0.38945203996969807,
      "grad_norm": 6.525443530446261,
      "learning_rate": 2.7909832155932647e-06,
      "loss": 0.9505,
      "step": 5398
    },
    {
      "epoch": 0.3895241874391256,
      "grad_norm": 2.8981170065816686,
      "learning_rate": 2.79055393822619e-06,
      "loss": 0.9754,
      "step": 5399
    },
    {
      "epoch": 0.3895963349085531,
      "grad_norm": 2.8144607490896907,
      "learning_rate": 2.7901246176898914e-06,
      "loss": 1.0595,
      "step": 5400
    },
    {
      "epoch": 0.3896684823779806,
      "grad_norm": 5.297113913940377,
      "learning_rate": 2.7896952540078123e-06,
      "loss": 0.9956,
      "step": 5401
    },
    {
      "epoch": 0.3897406298474081,
      "grad_norm": 4.034564407831098,
      "learning_rate": 2.7892658472033993e-06,
      "loss": 0.9976,
      "step": 5402
    },
    {
      "epoch": 0.3898127773168356,
      "grad_norm": 3.261205721957033,
      "learning_rate": 2.7888363973000996e-06,
      "loss": 0.9189,
      "step": 5403
    },
    {
      "epoch": 0.38988492478626313,
      "grad_norm": 2.7387304005689814,
      "learning_rate": 2.7884069043213646e-06,
      "loss": 0.9847,
      "step": 5404
    },
    {
      "epoch": 0.38995707225569065,
      "grad_norm": 0.7156004106053453,
      "learning_rate": 2.7879773682906474e-06,
      "loss": 0.8415,
      "step": 5405
    },
    {
      "epoch": 0.3900292197251181,
      "grad_norm": 4.322021587806885,
      "learning_rate": 2.7875477892314022e-06,
      "loss": 0.897,
      "step": 5406
    },
    {
      "epoch": 0.39010136719454563,
      "grad_norm": 0.8308627739654302,
      "learning_rate": 2.787118167167089e-06,
      "loss": 0.8664,
      "step": 5407
    },
    {
      "epoch": 0.39017351466397315,
      "grad_norm": 4.656675792913612,
      "learning_rate": 2.786688502121166e-06,
      "loss": 0.9735,
      "step": 5408
    },
    {
      "epoch": 0.3902456621334007,
      "grad_norm": 0.8855459717223041,
      "learning_rate": 2.7862587941170957e-06,
      "loss": 0.9344,
      "step": 5409
    },
    {
      "epoch": 0.3903178096028282,
      "grad_norm": 3.620957191962223,
      "learning_rate": 2.7858290431783437e-06,
      "loss": 1.0226,
      "step": 5410
    },
    {
      "epoch": 0.3903899570722557,
      "grad_norm": 0.6910817090442459,
      "learning_rate": 2.7853992493283766e-06,
      "loss": 0.8384,
      "step": 5411
    },
    {
      "epoch": 0.3904621045416832,
      "grad_norm": 0.7291684332326652,
      "learning_rate": 2.784969412590663e-06,
      "loss": 0.8123,
      "step": 5412
    },
    {
      "epoch": 0.3905342520111107,
      "grad_norm": 3.832975343214388,
      "learning_rate": 2.784539532988677e-06,
      "loss": 1.0244,
      "step": 5413
    },
    {
      "epoch": 0.3906063994805382,
      "grad_norm": 6.5096578008247015,
      "learning_rate": 2.784109610545891e-06,
      "loss": 0.8319,
      "step": 5414
    },
    {
      "epoch": 0.39067854694996573,
      "grad_norm": 7.712296850165852,
      "learning_rate": 2.783679645285781e-06,
      "loss": 1.0031,
      "step": 5415
    },
    {
      "epoch": 0.39075069441939325,
      "grad_norm": 3.8042469435400856,
      "learning_rate": 2.7832496372318277e-06,
      "loss": 1.0381,
      "step": 5416
    },
    {
      "epoch": 0.39082284188882077,
      "grad_norm": 3.757441321934484,
      "learning_rate": 2.782819586407511e-06,
      "loss": 0.8829,
      "step": 5417
    },
    {
      "epoch": 0.39089498935824823,
      "grad_norm": 4.841146446201996,
      "learning_rate": 2.7823894928363144e-06,
      "loss": 1.0619,
      "step": 5418
    },
    {
      "epoch": 0.39096713682767575,
      "grad_norm": 3.5135483775718934,
      "learning_rate": 2.7819593565417235e-06,
      "loss": 0.9082,
      "step": 5419
    },
    {
      "epoch": 0.3910392842971033,
      "grad_norm": 7.505106395600088,
      "learning_rate": 2.7815291775472276e-06,
      "loss": 0.9614,
      "step": 5420
    },
    {
      "epoch": 0.3911114317665308,
      "grad_norm": 3.0249100285541,
      "learning_rate": 2.7810989558763158e-06,
      "loss": 0.9066,
      "step": 5421
    },
    {
      "epoch": 0.3911835792359583,
      "grad_norm": 3.405008072136317,
      "learning_rate": 2.7806686915524823e-06,
      "loss": 0.9522,
      "step": 5422
    },
    {
      "epoch": 0.39125572670538583,
      "grad_norm": 3.4894182460318146,
      "learning_rate": 2.780238384599221e-06,
      "loss": 0.9907,
      "step": 5423
    },
    {
      "epoch": 0.3913278741748133,
      "grad_norm": 3.568929926249806,
      "learning_rate": 2.7798080350400304e-06,
      "loss": 1.0113,
      "step": 5424
    },
    {
      "epoch": 0.3914000216442408,
      "grad_norm": 4.974962297613008,
      "learning_rate": 2.779377642898409e-06,
      "loss": 0.8766,
      "step": 5425
    },
    {
      "epoch": 0.39147216911366833,
      "grad_norm": 3.2066516349530048,
      "learning_rate": 2.7789472081978596e-06,
      "loss": 0.9339,
      "step": 5426
    },
    {
      "epoch": 0.39154431658309585,
      "grad_norm": 5.914565917828449,
      "learning_rate": 2.7785167309618873e-06,
      "loss": 0.9628,
      "step": 5427
    },
    {
      "epoch": 0.39161646405252337,
      "grad_norm": 4.124303816799461,
      "learning_rate": 2.778086211213998e-06,
      "loss": 0.8989,
      "step": 5428
    },
    {
      "epoch": 0.3916886115219509,
      "grad_norm": 3.73295802602768,
      "learning_rate": 2.7776556489777015e-06,
      "loss": 1.0917,
      "step": 5429
    },
    {
      "epoch": 0.39176075899137836,
      "grad_norm": 4.608012576721831,
      "learning_rate": 2.777225044276508e-06,
      "loss": 0.9226,
      "step": 5430
    },
    {
      "epoch": 0.3918329064608059,
      "grad_norm": 4.06208094856447,
      "learning_rate": 2.7767943971339326e-06,
      "loss": 1.081,
      "step": 5431
    },
    {
      "epoch": 0.3919050539302334,
      "grad_norm": 3.3786700039340034,
      "learning_rate": 2.7763637075734904e-06,
      "loss": 0.9354,
      "step": 5432
    },
    {
      "epoch": 0.3919772013996609,
      "grad_norm": 19.999948882991315,
      "learning_rate": 2.7759329756187002e-06,
      "loss": 0.9331,
      "step": 5433
    },
    {
      "epoch": 0.39204934886908843,
      "grad_norm": 6.064961985494884,
      "learning_rate": 2.7755022012930833e-06,
      "loss": 1.0136,
      "step": 5434
    },
    {
      "epoch": 0.39212149633851595,
      "grad_norm": 3.926572137373768,
      "learning_rate": 2.775071384620161e-06,
      "loss": 0.9044,
      "step": 5435
    },
    {
      "epoch": 0.3921936438079434,
      "grad_norm": 3.324822448349583,
      "learning_rate": 2.7746405256234607e-06,
      "loss": 1.0091,
      "step": 5436
    },
    {
      "epoch": 0.39226579127737093,
      "grad_norm": 6.038142717795659,
      "learning_rate": 2.774209624326508e-06,
      "loss": 0.9247,
      "step": 5437
    },
    {
      "epoch": 0.39233793874679845,
      "grad_norm": 5.871136307958695,
      "learning_rate": 2.773778680752834e-06,
      "loss": 1.0803,
      "step": 5438
    },
    {
      "epoch": 0.392410086216226,
      "grad_norm": 3.9009167694076354,
      "learning_rate": 2.7733476949259707e-06,
      "loss": 0.9772,
      "step": 5439
    },
    {
      "epoch": 0.3924822336856535,
      "grad_norm": 0.7538996879015336,
      "learning_rate": 2.772916666869453e-06,
      "loss": 0.8762,
      "step": 5440
    },
    {
      "epoch": 0.392554381155081,
      "grad_norm": 2.769267124768789,
      "learning_rate": 2.772485596606817e-06,
      "loss": 1.0644,
      "step": 5441
    },
    {
      "epoch": 0.3926265286245085,
      "grad_norm": 11.101781726151762,
      "learning_rate": 2.772054484161602e-06,
      "loss": 0.9601,
      "step": 5442
    },
    {
      "epoch": 0.392698676093936,
      "grad_norm": 5.625304658904543,
      "learning_rate": 2.7716233295573495e-06,
      "loss": 0.9461,
      "step": 5443
    },
    {
      "epoch": 0.3927708235633635,
      "grad_norm": 0.7345039984851961,
      "learning_rate": 2.771192132817604e-06,
      "loss": 0.8143,
      "step": 5444
    },
    {
      "epoch": 0.39284297103279103,
      "grad_norm": 3.6655083184887407,
      "learning_rate": 2.7707608939659103e-06,
      "loss": 0.9926,
      "step": 5445
    },
    {
      "epoch": 0.39291511850221855,
      "grad_norm": 3.22783393840187,
      "learning_rate": 2.770329613025818e-06,
      "loss": 0.8988,
      "step": 5446
    },
    {
      "epoch": 0.39298726597164607,
      "grad_norm": 5.401697612525896,
      "learning_rate": 2.7698982900208775e-06,
      "loss": 0.982,
      "step": 5447
    },
    {
      "epoch": 0.39305941344107354,
      "grad_norm": 7.125880672169855,
      "learning_rate": 2.769466924974641e-06,
      "loss": 1.0387,
      "step": 5448
    },
    {
      "epoch": 0.39313156091050105,
      "grad_norm": 4.899464574093068,
      "learning_rate": 2.7690355179106642e-06,
      "loss": 0.9663,
      "step": 5449
    },
    {
      "epoch": 0.3932037083799286,
      "grad_norm": 3.6421859080058727,
      "learning_rate": 2.7686040688525048e-06,
      "loss": 0.8282,
      "step": 5450
    },
    {
      "epoch": 0.3932758558493561,
      "grad_norm": 5.1069445881195,
      "learning_rate": 2.768172577823723e-06,
      "loss": 0.9183,
      "step": 5451
    },
    {
      "epoch": 0.3933480033187836,
      "grad_norm": 4.619643744071999,
      "learning_rate": 2.767741044847879e-06,
      "loss": 0.9026,
      "step": 5452
    },
    {
      "epoch": 0.3934201507882111,
      "grad_norm": 2.7894153532049173,
      "learning_rate": 2.7673094699485394e-06,
      "loss": 0.9888,
      "step": 5453
    },
    {
      "epoch": 0.3934922982576386,
      "grad_norm": 3.8456961814310096,
      "learning_rate": 2.7668778531492702e-06,
      "loss": 0.9501,
      "step": 5454
    },
    {
      "epoch": 0.3935644457270661,
      "grad_norm": 2.7494034553611013,
      "learning_rate": 2.76644619447364e-06,
      "loss": 0.9797,
      "step": 5455
    },
    {
      "epoch": 0.39363659319649363,
      "grad_norm": 4.286876239389901,
      "learning_rate": 2.766014493945221e-06,
      "loss": 0.9484,
      "step": 5456
    },
    {
      "epoch": 0.39370874066592115,
      "grad_norm": 3.6931813326748615,
      "learning_rate": 2.7655827515875855e-06,
      "loss": 1.019,
      "step": 5457
    },
    {
      "epoch": 0.3937808881353487,
      "grad_norm": 5.225954485140654,
      "learning_rate": 2.7651509674243105e-06,
      "loss": 1.0074,
      "step": 5458
    },
    {
      "epoch": 0.39385303560477614,
      "grad_norm": 5.747948985638608,
      "learning_rate": 2.7647191414789733e-06,
      "loss": 0.9842,
      "step": 5459
    },
    {
      "epoch": 0.39392518307420366,
      "grad_norm": 3.0193943323279653,
      "learning_rate": 2.7642872737751553e-06,
      "loss": 0.976,
      "step": 5460
    },
    {
      "epoch": 0.3939973305436312,
      "grad_norm": 3.0021507183262592,
      "learning_rate": 2.763855364336438e-06,
      "loss": 0.9279,
      "step": 5461
    },
    {
      "epoch": 0.3940694780130587,
      "grad_norm": 3.064754668182669,
      "learning_rate": 2.763423413186407e-06,
      "loss": 1.0032,
      "step": 5462
    },
    {
      "epoch": 0.3941416254824862,
      "grad_norm": 3.865548574544825,
      "learning_rate": 2.76299142034865e-06,
      "loss": 1.0031,
      "step": 5463
    },
    {
      "epoch": 0.39421377295191373,
      "grad_norm": 0.7374389122054598,
      "learning_rate": 2.762559385846755e-06,
      "loss": 0.8114,
      "step": 5464
    },
    {
      "epoch": 0.3942859204213412,
      "grad_norm": 2.184839210791989,
      "learning_rate": 2.762127309704316e-06,
      "loss": 0.9505,
      "step": 5465
    },
    {
      "epoch": 0.3943580678907687,
      "grad_norm": 2.8881941188127116,
      "learning_rate": 2.761695191944925e-06,
      "loss": 1.0598,
      "step": 5466
    },
    {
      "epoch": 0.39443021536019623,
      "grad_norm": 10.304501662700703,
      "learning_rate": 2.7612630325921797e-06,
      "loss": 0.9255,
      "step": 5467
    },
    {
      "epoch": 0.39450236282962375,
      "grad_norm": 2.776890542010237,
      "learning_rate": 2.7608308316696787e-06,
      "loss": 0.9918,
      "step": 5468
    },
    {
      "epoch": 0.3945745102990513,
      "grad_norm": 6.332264090510035,
      "learning_rate": 2.7603985892010214e-06,
      "loss": 0.9254,
      "step": 5469
    },
    {
      "epoch": 0.3946466577684788,
      "grad_norm": 2.925522592924455,
      "learning_rate": 2.759966305209813e-06,
      "loss": 0.9646,
      "step": 5470
    },
    {
      "epoch": 0.39471880523790626,
      "grad_norm": 6.899294289145262,
      "learning_rate": 2.7595339797196575e-06,
      "loss": 0.97,
      "step": 5471
    },
    {
      "epoch": 0.3947909527073338,
      "grad_norm": 3.2750196616478027,
      "learning_rate": 2.7591016127541637e-06,
      "loss": 1.0569,
      "step": 5472
    },
    {
      "epoch": 0.3948631001767613,
      "grad_norm": 3.4033949899895166,
      "learning_rate": 2.75866920433694e-06,
      "loss": 0.9962,
      "step": 5473
    },
    {
      "epoch": 0.3949352476461888,
      "grad_norm": 6.306075953642286,
      "learning_rate": 2.7582367544916e-06,
      "loss": 0.9005,
      "step": 5474
    },
    {
      "epoch": 0.39500739511561633,
      "grad_norm": 4.252297846111664,
      "learning_rate": 2.757804263241758e-06,
      "loss": 0.9587,
      "step": 5475
    },
    {
      "epoch": 0.39507954258504385,
      "grad_norm": 3.4644104902805637,
      "learning_rate": 2.7573717306110303e-06,
      "loss": 1.0113,
      "step": 5476
    },
    {
      "epoch": 0.3951516900544713,
      "grad_norm": 4.691915441572265,
      "learning_rate": 2.756939156623036e-06,
      "loss": 0.9935,
      "step": 5477
    },
    {
      "epoch": 0.39522383752389884,
      "grad_norm": 4.025870348734462,
      "learning_rate": 2.756506541301396e-06,
      "loss": 0.9883,
      "step": 5478
    },
    {
      "epoch": 0.39529598499332635,
      "grad_norm": 8.619908253475778,
      "learning_rate": 2.756073884669735e-06,
      "loss": 0.9417,
      "step": 5479
    },
    {
      "epoch": 0.3953681324627539,
      "grad_norm": 3.4672213098459976,
      "learning_rate": 2.755641186751678e-06,
      "loss": 0.9647,
      "step": 5480
    },
    {
      "epoch": 0.3954402799321814,
      "grad_norm": 3.81554103949522,
      "learning_rate": 2.7552084475708525e-06,
      "loss": 0.8884,
      "step": 5481
    },
    {
      "epoch": 0.3955124274016089,
      "grad_norm": 5.3649945839092865,
      "learning_rate": 2.7547756671508894e-06,
      "loss": 0.8767,
      "step": 5482
    },
    {
      "epoch": 0.3955845748710364,
      "grad_norm": 4.432395040320417,
      "learning_rate": 2.7543428455154216e-06,
      "loss": 1.061,
      "step": 5483
    },
    {
      "epoch": 0.3956567223404639,
      "grad_norm": 0.969232439076379,
      "learning_rate": 2.753909982688084e-06,
      "loss": 0.8787,
      "step": 5484
    },
    {
      "epoch": 0.3957288698098914,
      "grad_norm": 4.175265759444218,
      "learning_rate": 2.753477078692512e-06,
      "loss": 1.0479,
      "step": 5485
    },
    {
      "epoch": 0.39580101727931893,
      "grad_norm": 8.003120290687066,
      "learning_rate": 2.7530441335523464e-06,
      "loss": 1.0129,
      "step": 5486
    },
    {
      "epoch": 0.39587316474874645,
      "grad_norm": 4.517576647278725,
      "learning_rate": 2.752611147291229e-06,
      "loss": 0.9905,
      "step": 5487
    },
    {
      "epoch": 0.395945312218174,
      "grad_norm": 4.015544964775733,
      "learning_rate": 2.752178119932802e-06,
      "loss": 1.0597,
      "step": 5488
    },
    {
      "epoch": 0.39601745968760144,
      "grad_norm": 3.406927758779167,
      "learning_rate": 2.7517450515007124e-06,
      "loss": 0.9735,
      "step": 5489
    },
    {
      "epoch": 0.39608960715702896,
      "grad_norm": 3.214601973460839,
      "learning_rate": 2.751311942018609e-06,
      "loss": 0.938,
      "step": 5490
    },
    {
      "epoch": 0.3961617546264565,
      "grad_norm": 4.538598987333235,
      "learning_rate": 2.7508787915101413e-06,
      "loss": 0.9463,
      "step": 5491
    },
    {
      "epoch": 0.396233902095884,
      "grad_norm": 4.457718898941747,
      "learning_rate": 2.750445599998963e-06,
      "loss": 0.8874,
      "step": 5492
    },
    {
      "epoch": 0.3963060495653115,
      "grad_norm": 3.119479835138143,
      "learning_rate": 2.7500123675087276e-06,
      "loss": 0.9745,
      "step": 5493
    },
    {
      "epoch": 0.39637819703473903,
      "grad_norm": 0.6800542477749575,
      "learning_rate": 2.749579094063094e-06,
      "loss": 0.8576,
      "step": 5494
    },
    {
      "epoch": 0.3964503445041665,
      "grad_norm": 3.8591742212370823,
      "learning_rate": 2.749145779685721e-06,
      "loss": 0.9461,
      "step": 5495
    },
    {
      "epoch": 0.396522491973594,
      "grad_norm": 2.5450627719055285,
      "learning_rate": 2.74871242440027e-06,
      "loss": 1.0404,
      "step": 5496
    },
    {
      "epoch": 0.39659463944302153,
      "grad_norm": 2.5663876699765025,
      "learning_rate": 2.7482790282304046e-06,
      "loss": 0.9818,
      "step": 5497
    },
    {
      "epoch": 0.39666678691244905,
      "grad_norm": 2.0702683929927197,
      "learning_rate": 2.747845591199792e-06,
      "loss": 0.9614,
      "step": 5498
    },
    {
      "epoch": 0.3967389343818766,
      "grad_norm": 3.145841905864294,
      "learning_rate": 2.7474121133321004e-06,
      "loss": 1.0309,
      "step": 5499
    },
    {
      "epoch": 0.3968110818513041,
      "grad_norm": 2.737836815306218,
      "learning_rate": 2.7469785946509992e-06,
      "loss": 0.9682,
      "step": 5500
    },
    {
      "epoch": 0.39688322932073156,
      "grad_norm": 2.5417489731599803,
      "learning_rate": 2.746545035180163e-06,
      "loss": 0.9522,
      "step": 5501
    },
    {
      "epoch": 0.3969553767901591,
      "grad_norm": 2.7455932682043747,
      "learning_rate": 2.746111434943266e-06,
      "loss": 1.0041,
      "step": 5502
    },
    {
      "epoch": 0.3970275242595866,
      "grad_norm": 13.452947493566175,
      "learning_rate": 2.7456777939639854e-06,
      "loss": 0.968,
      "step": 5503
    },
    {
      "epoch": 0.3970996717290141,
      "grad_norm": 4.8479114007313315,
      "learning_rate": 2.7452441122660005e-06,
      "loss": 1.0341,
      "step": 5504
    },
    {
      "epoch": 0.39717181919844163,
      "grad_norm": 4.505461663269851,
      "learning_rate": 2.7448103898729935e-06,
      "loss": 0.9951,
      "step": 5505
    },
    {
      "epoch": 0.3972439666678691,
      "grad_norm": 3.0714120230751725,
      "learning_rate": 2.7443766268086493e-06,
      "loss": 1.0198,
      "step": 5506
    },
    {
      "epoch": 0.3973161141372966,
      "grad_norm": 2.7097561155704737,
      "learning_rate": 2.7439428230966517e-06,
      "loss": 0.99,
      "step": 5507
    },
    {
      "epoch": 0.39738826160672414,
      "grad_norm": 3.7170901400823024,
      "learning_rate": 2.7435089787606913e-06,
      "loss": 1.016,
      "step": 5508
    },
    {
      "epoch": 0.39746040907615166,
      "grad_norm": 0.6261394604189532,
      "learning_rate": 2.7430750938244576e-06,
      "loss": 0.8159,
      "step": 5509
    },
    {
      "epoch": 0.3975325565455792,
      "grad_norm": 5.319132893964236,
      "learning_rate": 2.742641168311644e-06,
      "loss": 0.9742,
      "step": 5510
    },
    {
      "epoch": 0.3976047040150067,
      "grad_norm": 4.3949693792012035,
      "learning_rate": 2.7422072022459456e-06,
      "loss": 0.9173,
      "step": 5511
    },
    {
      "epoch": 0.39767685148443416,
      "grad_norm": 4.781272937993895,
      "learning_rate": 2.741773195651059e-06,
      "loss": 0.9411,
      "step": 5512
    },
    {
      "epoch": 0.3977489989538617,
      "grad_norm": 2.4484644067353325,
      "learning_rate": 2.7413391485506845e-06,
      "loss": 1.05,
      "step": 5513
    },
    {
      "epoch": 0.3978211464232892,
      "grad_norm": 3.0728174915581596,
      "learning_rate": 2.7409050609685233e-06,
      "loss": 1.0748,
      "step": 5514
    },
    {
      "epoch": 0.3978932938927167,
      "grad_norm": 3.561359423706658,
      "learning_rate": 2.7404709329282798e-06,
      "loss": 1.0426,
      "step": 5515
    },
    {
      "epoch": 0.39796544136214423,
      "grad_norm": 2.8011259880345056,
      "learning_rate": 2.740036764453659e-06,
      "loss": 1.0652,
      "step": 5516
    },
    {
      "epoch": 0.39803758883157175,
      "grad_norm": 4.371383698283882,
      "learning_rate": 2.7396025555683706e-06,
      "loss": 1.0574,
      "step": 5517
    },
    {
      "epoch": 0.3981097363009992,
      "grad_norm": 0.974048246561476,
      "learning_rate": 2.739168306296124e-06,
      "loss": 0.9277,
      "step": 5518
    },
    {
      "epoch": 0.39818188377042674,
      "grad_norm": 3.228171623905122,
      "learning_rate": 2.738734016660633e-06,
      "loss": 1.0217,
      "step": 5519
    },
    {
      "epoch": 0.39825403123985426,
      "grad_norm": 4.211303242103119,
      "learning_rate": 2.738299686685612e-06,
      "loss": 1.0689,
      "step": 5520
    },
    {
      "epoch": 0.3983261787092818,
      "grad_norm": 3.014475549028945,
      "learning_rate": 2.737865316394778e-06,
      "loss": 0.8601,
      "step": 5521
    },
    {
      "epoch": 0.3983983261787093,
      "grad_norm": 2.8880572487289853,
      "learning_rate": 2.7374309058118506e-06,
      "loss": 0.8978,
      "step": 5522
    },
    {
      "epoch": 0.3984704736481368,
      "grad_norm": 3.0985524828068503,
      "learning_rate": 2.7369964549605517e-06,
      "loss": 0.9739,
      "step": 5523
    },
    {
      "epoch": 0.3985426211175643,
      "grad_norm": 3.0720720757632107,
      "learning_rate": 2.736561963864604e-06,
      "loss": 0.9459,
      "step": 5524
    },
    {
      "epoch": 0.3986147685869918,
      "grad_norm": 3.817142973806838,
      "learning_rate": 2.7361274325477346e-06,
      "loss": 0.9693,
      "step": 5525
    },
    {
      "epoch": 0.3986869160564193,
      "grad_norm": 2.915233387053045,
      "learning_rate": 2.735692861033671e-06,
      "loss": 1.0658,
      "step": 5526
    },
    {
      "epoch": 0.39875906352584684,
      "grad_norm": 3.558268258946269,
      "learning_rate": 2.735258249346144e-06,
      "loss": 0.9637,
      "step": 5527
    },
    {
      "epoch": 0.39883121099527435,
      "grad_norm": 3.910678398032356,
      "learning_rate": 2.734823597508886e-06,
      "loss": 0.9598,
      "step": 5528
    },
    {
      "epoch": 0.3989033584647019,
      "grad_norm": 0.6394740442059549,
      "learning_rate": 2.734388905545631e-06,
      "loss": 0.7801,
      "step": 5529
    },
    {
      "epoch": 0.39897550593412934,
      "grad_norm": 4.4303645402050655,
      "learning_rate": 2.733954173480116e-06,
      "loss": 0.9629,
      "step": 5530
    },
    {
      "epoch": 0.39904765340355686,
      "grad_norm": 3.2773022836601275,
      "learning_rate": 2.7335194013360812e-06,
      "loss": 0.9964,
      "step": 5531
    },
    {
      "epoch": 0.3991198008729844,
      "grad_norm": 2.0707432712591602,
      "learning_rate": 2.7330845891372677e-06,
      "loss": 0.9214,
      "step": 5532
    },
    {
      "epoch": 0.3991919483424119,
      "grad_norm": 2.958948922023584,
      "learning_rate": 2.732649736907418e-06,
      "loss": 0.9814,
      "step": 5533
    },
    {
      "epoch": 0.3992640958118394,
      "grad_norm": 6.748884320774458,
      "learning_rate": 2.7322148446702783e-06,
      "loss": 0.8764,
      "step": 5534
    },
    {
      "epoch": 0.39933624328126693,
      "grad_norm": 1.9485897050646865,
      "learning_rate": 2.731779912449597e-06,
      "loss": 1.0372,
      "step": 5535
    },
    {
      "epoch": 0.3994083907506944,
      "grad_norm": 2.3568377462481145,
      "learning_rate": 2.7313449402691227e-06,
      "loss": 0.9338,
      "step": 5536
    },
    {
      "epoch": 0.3994805382201219,
      "grad_norm": 0.6977711402467865,
      "learning_rate": 2.7309099281526095e-06,
      "loss": 0.7829,
      "step": 5537
    },
    {
      "epoch": 0.39955268568954944,
      "grad_norm": 2.106913055284455,
      "learning_rate": 2.7304748761238107e-06,
      "loss": 0.9532,
      "step": 5538
    },
    {
      "epoch": 0.39962483315897696,
      "grad_norm": 3.38856000468259,
      "learning_rate": 2.730039784206483e-06,
      "loss": 1.0338,
      "step": 5539
    },
    {
      "epoch": 0.3996969806284045,
      "grad_norm": 7.569218701215217,
      "learning_rate": 2.7296046524243856e-06,
      "loss": 0.9565,
      "step": 5540
    },
    {
      "epoch": 0.399769128097832,
      "grad_norm": 3.3870591781217274,
      "learning_rate": 2.7291694808012782e-06,
      "loss": 1.0238,
      "step": 5541
    },
    {
      "epoch": 0.39984127556725946,
      "grad_norm": 3.798131915235729,
      "learning_rate": 2.728734269360925e-06,
      "loss": 1.0752,
      "step": 5542
    },
    {
      "epoch": 0.399913423036687,
      "grad_norm": 3.15804383319104,
      "learning_rate": 2.728299018127091e-06,
      "loss": 0.9412,
      "step": 5543
    },
    {
      "epoch": 0.3999855705061145,
      "grad_norm": 7.006520368002674,
      "learning_rate": 2.7278637271235436e-06,
      "loss": 0.9957,
      "step": 5544
    },
    {
      "epoch": 0.400057717975542,
      "grad_norm": 4.8392929736672645,
      "learning_rate": 2.727428396374053e-06,
      "loss": 0.9723,
      "step": 5545
    },
    {
      "epoch": 0.40012986544496953,
      "grad_norm": 0.7344617183699127,
      "learning_rate": 2.7269930259023907e-06,
      "loss": 0.871,
      "step": 5546
    },
    {
      "epoch": 0.40020201291439705,
      "grad_norm": 3.024446067850109,
      "learning_rate": 2.7265576157323294e-06,
      "loss": 0.9471,
      "step": 5547
    },
    {
      "epoch": 0.4002741603838245,
      "grad_norm": 2.7323248452661324,
      "learning_rate": 2.7261221658876468e-06,
      "loss": 1.0478,
      "step": 5548
    },
    {
      "epoch": 0.40034630785325204,
      "grad_norm": 3.9121642239820753,
      "learning_rate": 2.7256866763921207e-06,
      "loss": 1.0154,
      "step": 5549
    },
    {
      "epoch": 0.40041845532267956,
      "grad_norm": 0.8867936291776808,
      "learning_rate": 2.7252511472695326e-06,
      "loss": 0.8414,
      "step": 5550
    },
    {
      "epoch": 0.4004906027921071,
      "grad_norm": 2.470472579874915,
      "learning_rate": 2.724815578543663e-06,
      "loss": 0.8938,
      "step": 5551
    },
    {
      "epoch": 0.4005627502615346,
      "grad_norm": 2.4758610739919837,
      "learning_rate": 2.7243799702382984e-06,
      "loss": 0.9768,
      "step": 5552
    },
    {
      "epoch": 0.4006348977309621,
      "grad_norm": 2.546419348649398,
      "learning_rate": 2.723944322377225e-06,
      "loss": 0.9188,
      "step": 5553
    },
    {
      "epoch": 0.4007070452003896,
      "grad_norm": 4.159452423590313,
      "learning_rate": 2.723508634984232e-06,
      "loss": 1.1112,
      "step": 5554
    },
    {
      "epoch": 0.4007791926698171,
      "grad_norm": 4.925719682034933,
      "learning_rate": 2.723072908083111e-06,
      "loss": 0.9804,
      "step": 5555
    },
    {
      "epoch": 0.4008513401392446,
      "grad_norm": 3.689246846900229,
      "learning_rate": 2.722637141697655e-06,
      "loss": 1.0484,
      "step": 5556
    },
    {
      "epoch": 0.40092348760867214,
      "grad_norm": 3.0536245852937056,
      "learning_rate": 2.7222013358516597e-06,
      "loss": 1.032,
      "step": 5557
    },
    {
      "epoch": 0.40099563507809965,
      "grad_norm": 0.7902210530521951,
      "learning_rate": 2.721765490568923e-06,
      "loss": 0.8873,
      "step": 5558
    },
    {
      "epoch": 0.4010677825475271,
      "grad_norm": 4.982332295353087,
      "learning_rate": 2.7213296058732446e-06,
      "loss": 1.0171,
      "step": 5559
    },
    {
      "epoch": 0.40113993001695464,
      "grad_norm": 3.7596030301692767,
      "learning_rate": 2.720893681788427e-06,
      "loss": 1.0477,
      "step": 5560
    },
    {
      "epoch": 0.40121207748638216,
      "grad_norm": 2.6659841458764255,
      "learning_rate": 2.7204577183382747e-06,
      "loss": 0.927,
      "step": 5561
    },
    {
      "epoch": 0.4012842249558097,
      "grad_norm": 3.237747719025252,
      "learning_rate": 2.7200217155465924e-06,
      "loss": 0.9764,
      "step": 5562
    },
    {
      "epoch": 0.4013563724252372,
      "grad_norm": 2.9972920912159893,
      "learning_rate": 2.71958567343719e-06,
      "loss": 0.9228,
      "step": 5563
    },
    {
      "epoch": 0.4014285198946647,
      "grad_norm": 3.388608974785462,
      "learning_rate": 2.719149592033878e-06,
      "loss": 1.081,
      "step": 5564
    },
    {
      "epoch": 0.4015006673640922,
      "grad_norm": 5.2597603668628565,
      "learning_rate": 2.718713471360469e-06,
      "loss": 0.9787,
      "step": 5565
    },
    {
      "epoch": 0.4015728148335197,
      "grad_norm": 3.9738642863506306,
      "learning_rate": 2.7182773114407776e-06,
      "loss": 1.0043,
      "step": 5566
    },
    {
      "epoch": 0.4016449623029472,
      "grad_norm": 2.7319650547646694,
      "learning_rate": 2.717841112298621e-06,
      "loss": 0.9112,
      "step": 5567
    },
    {
      "epoch": 0.40171710977237474,
      "grad_norm": 4.059060621532088,
      "learning_rate": 2.717404873957819e-06,
      "loss": 0.9541,
      "step": 5568
    },
    {
      "epoch": 0.40178925724180226,
      "grad_norm": 2.9632918124585186,
      "learning_rate": 2.7169685964421927e-06,
      "loss": 0.951,
      "step": 5569
    },
    {
      "epoch": 0.4018614047112298,
      "grad_norm": 5.121296917153648,
      "learning_rate": 2.716532279775565e-06,
      "loss": 0.9761,
      "step": 5570
    },
    {
      "epoch": 0.40193355218065724,
      "grad_norm": 3.4717134755739756,
      "learning_rate": 2.7160959239817623e-06,
      "loss": 1.0375,
      "step": 5571
    },
    {
      "epoch": 0.40200569965008476,
      "grad_norm": 3.1408411966942174,
      "learning_rate": 2.7156595290846124e-06,
      "loss": 0.9011,
      "step": 5572
    },
    {
      "epoch": 0.4020778471195123,
      "grad_norm": 3.0394130004540516,
      "learning_rate": 2.7152230951079445e-06,
      "loss": 1.034,
      "step": 5573
    },
    {
      "epoch": 0.4021499945889398,
      "grad_norm": 3.009666446187944,
      "learning_rate": 2.714786622075591e-06,
      "loss": 0.9387,
      "step": 5574
    },
    {
      "epoch": 0.4022221420583673,
      "grad_norm": 2.36676545362302,
      "learning_rate": 2.714350110011386e-06,
      "loss": 1.0216,
      "step": 5575
    },
    {
      "epoch": 0.40229428952779483,
      "grad_norm": 0.6226058642822867,
      "learning_rate": 2.7139135589391666e-06,
      "loss": 0.7721,
      "step": 5576
    },
    {
      "epoch": 0.4023664369972223,
      "grad_norm": 3.5519701957397305,
      "learning_rate": 2.71347696888277e-06,
      "loss": 0.9353,
      "step": 5577
    },
    {
      "epoch": 0.4024385844666498,
      "grad_norm": 3.234243196759658,
      "learning_rate": 2.7130403398660374e-06,
      "loss": 1.0136,
      "step": 5578
    },
    {
      "epoch": 0.40251073193607734,
      "grad_norm": 10.273566346031382,
      "learning_rate": 2.7126036719128118e-06,
      "loss": 0.972,
      "step": 5579
    },
    {
      "epoch": 0.40258287940550486,
      "grad_norm": 2.7958929799326757,
      "learning_rate": 2.712166965046937e-06,
      "loss": 0.8609,
      "step": 5580
    },
    {
      "epoch": 0.4026550268749324,
      "grad_norm": 9.881301993631986,
      "learning_rate": 2.711730219292261e-06,
      "loss": 0.9421,
      "step": 5581
    },
    {
      "epoch": 0.4027271743443599,
      "grad_norm": 5.296567474710405,
      "learning_rate": 2.7112934346726326e-06,
      "loss": 0.9091,
      "step": 5582
    },
    {
      "epoch": 0.40279932181378736,
      "grad_norm": 0.8087064507859361,
      "learning_rate": 2.7108566112119027e-06,
      "loss": 0.8509,
      "step": 5583
    },
    {
      "epoch": 0.4028714692832149,
      "grad_norm": 3.4869599611277176,
      "learning_rate": 2.7104197489339245e-06,
      "loss": 1.0105,
      "step": 5584
    },
    {
      "epoch": 0.4029436167526424,
      "grad_norm": 6.2821320630862765,
      "learning_rate": 2.7099828478625536e-06,
      "loss": 0.8894,
      "step": 5585
    },
    {
      "epoch": 0.4030157642220699,
      "grad_norm": 2.6172621075828113,
      "learning_rate": 2.7095459080216478e-06,
      "loss": 0.9084,
      "step": 5586
    },
    {
      "epoch": 0.40308791169149744,
      "grad_norm": 4.0694159693728915,
      "learning_rate": 2.7091089294350667e-06,
      "loss": 0.9259,
      "step": 5587
    },
    {
      "epoch": 0.40316005916092496,
      "grad_norm": 3.77883670183733,
      "learning_rate": 2.7086719121266727e-06,
      "loss": 0.9767,
      "step": 5588
    },
    {
      "epoch": 0.4032322066303524,
      "grad_norm": 2.483463048000697,
      "learning_rate": 2.708234856120328e-06,
      "loss": 1.0262,
      "step": 5589
    },
    {
      "epoch": 0.40330435409977994,
      "grad_norm": 3.09560639143029,
      "learning_rate": 2.7077977614398998e-06,
      "loss": 0.92,
      "step": 5590
    },
    {
      "epoch": 0.40337650156920746,
      "grad_norm": 4.2481847139993585,
      "learning_rate": 2.7073606281092567e-06,
      "loss": 0.971,
      "step": 5591
    },
    {
      "epoch": 0.403448649038635,
      "grad_norm": 4.339788373015278,
      "learning_rate": 2.7069234561522676e-06,
      "loss": 0.9067,
      "step": 5592
    },
    {
      "epoch": 0.4035207965080625,
      "grad_norm": 3.2615880517414664,
      "learning_rate": 2.7064862455928056e-06,
      "loss": 1.0001,
      "step": 5593
    },
    {
      "epoch": 0.40359294397749,
      "grad_norm": 3.5481090667458246,
      "learning_rate": 2.706048996454745e-06,
      "loss": 0.9236,
      "step": 5594
    },
    {
      "epoch": 0.4036650914469175,
      "grad_norm": 3.614526284653519,
      "learning_rate": 2.7056117087619626e-06,
      "loss": 1.0028,
      "step": 5595
    },
    {
      "epoch": 0.403737238916345,
      "grad_norm": 4.1919798160824175,
      "learning_rate": 2.7051743825383364e-06,
      "loss": 0.9433,
      "step": 5596
    },
    {
      "epoch": 0.4038093863857725,
      "grad_norm": 5.834588197430368,
      "learning_rate": 2.7047370178077476e-06,
      "loss": 0.9381,
      "step": 5597
    },
    {
      "epoch": 0.40388153385520004,
      "grad_norm": 1.1777578576120782,
      "learning_rate": 2.70429961459408e-06,
      "loss": 0.8593,
      "step": 5598
    },
    {
      "epoch": 0.40395368132462756,
      "grad_norm": 3.046425028121541,
      "learning_rate": 2.7038621729212166e-06,
      "loss": 1.0352,
      "step": 5599
    },
    {
      "epoch": 0.4040258287940551,
      "grad_norm": 3.2143444540318074,
      "learning_rate": 2.7034246928130466e-06,
      "loss": 1.0001,
      "step": 5600
    },
    {
      "epoch": 0.40409797626348254,
      "grad_norm": 3.2253162849935553,
      "learning_rate": 2.7029871742934576e-06,
      "loss": 1.0275,
      "step": 5601
    },
    {
      "epoch": 0.40417012373291006,
      "grad_norm": 3.2940524344240276,
      "learning_rate": 2.7025496173863413e-06,
      "loss": 1.0053,
      "step": 5602
    },
    {
      "epoch": 0.4042422712023376,
      "grad_norm": 3.9123561893638485,
      "learning_rate": 2.7021120221155913e-06,
      "loss": 0.9966,
      "step": 5603
    },
    {
      "epoch": 0.4043144186717651,
      "grad_norm": 4.525383459883193,
      "learning_rate": 2.701674388505103e-06,
      "loss": 1.0109,
      "step": 5604
    },
    {
      "epoch": 0.4043865661411926,
      "grad_norm": 4.070528050314656,
      "learning_rate": 2.7012367165787732e-06,
      "loss": 0.9845,
      "step": 5605
    },
    {
      "epoch": 0.4044587136106201,
      "grad_norm": 3.5323924689317057,
      "learning_rate": 2.700799006360503e-06,
      "loss": 0.9981,
      "step": 5606
    },
    {
      "epoch": 0.4045308610800476,
      "grad_norm": 2.5265683808078396,
      "learning_rate": 2.700361257874193e-06,
      "loss": 0.8852,
      "step": 5607
    },
    {
      "epoch": 0.4046030085494751,
      "grad_norm": 4.5442247565118645,
      "learning_rate": 2.6999234711437477e-06,
      "loss": 0.9838,
      "step": 5608
    },
    {
      "epoch": 0.40467515601890264,
      "grad_norm": 3.2029264062864313,
      "learning_rate": 2.699485646193072e-06,
      "loss": 1.0018,
      "step": 5609
    },
    {
      "epoch": 0.40474730348833016,
      "grad_norm": 3.1389412233234,
      "learning_rate": 2.699047783046075e-06,
      "loss": 0.9953,
      "step": 5610
    },
    {
      "epoch": 0.4048194509577577,
      "grad_norm": 3.1029214260173585,
      "learning_rate": 2.6986098817266666e-06,
      "loss": 1.0354,
      "step": 5611
    },
    {
      "epoch": 0.40489159842718514,
      "grad_norm": 2.2465760351779633,
      "learning_rate": 2.6981719422587583e-06,
      "loss": 0.9366,
      "step": 5612
    },
    {
      "epoch": 0.40496374589661266,
      "grad_norm": 2.845797461920806,
      "learning_rate": 2.6977339646662647e-06,
      "loss": 0.9315,
      "step": 5613
    },
    {
      "epoch": 0.4050358933660402,
      "grad_norm": 4.561352311103129,
      "learning_rate": 2.697295948973102e-06,
      "loss": 0.9157,
      "step": 5614
    },
    {
      "epoch": 0.4051080408354677,
      "grad_norm": 4.995810088336095,
      "learning_rate": 2.69685789520319e-06,
      "loss": 0.8952,
      "step": 5615
    },
    {
      "epoch": 0.4051801883048952,
      "grad_norm": 3.9037735069659156,
      "learning_rate": 2.6964198033804466e-06,
      "loss": 1.0224,
      "step": 5616
    },
    {
      "epoch": 0.40525233577432274,
      "grad_norm": 7.211782023693963,
      "learning_rate": 2.6959816735287963e-06,
      "loss": 0.9151,
      "step": 5617
    },
    {
      "epoch": 0.4053244832437502,
      "grad_norm": 5.106488173496973,
      "learning_rate": 2.695543505672164e-06,
      "loss": 1.0168,
      "step": 5618
    },
    {
      "epoch": 0.4053966307131777,
      "grad_norm": 0.8439865134029458,
      "learning_rate": 2.695105299834475e-06,
      "loss": 0.8566,
      "step": 5619
    },
    {
      "epoch": 0.40546877818260524,
      "grad_norm": 6.649117386836074,
      "learning_rate": 2.6946670560396587e-06,
      "loss": 0.9278,
      "step": 5620
    },
    {
      "epoch": 0.40554092565203276,
      "grad_norm": 4.598169808562849,
      "learning_rate": 2.694228774311646e-06,
      "loss": 0.9774,
      "step": 5621
    },
    {
      "epoch": 0.4056130731214603,
      "grad_norm": 3.875370254055484,
      "learning_rate": 2.69379045467437e-06,
      "loss": 0.9295,
      "step": 5622
    },
    {
      "epoch": 0.4056852205908878,
      "grad_norm": 7.05972910850333,
      "learning_rate": 2.6933520971517655e-06,
      "loss": 0.9689,
      "step": 5623
    },
    {
      "epoch": 0.40575736806031526,
      "grad_norm": 2.906484614920998,
      "learning_rate": 2.69291370176777e-06,
      "loss": 0.8757,
      "step": 5624
    },
    {
      "epoch": 0.4058295155297428,
      "grad_norm": 2.7520143760649427,
      "learning_rate": 2.692475268546322e-06,
      "loss": 1.0703,
      "step": 5625
    },
    {
      "epoch": 0.4059016629991703,
      "grad_norm": 0.72089993237196,
      "learning_rate": 2.692036797511363e-06,
      "loss": 0.8677,
      "step": 5626
    },
    {
      "epoch": 0.4059738104685978,
      "grad_norm": 3.531204763477191,
      "learning_rate": 2.691598288686837e-06,
      "loss": 0.9517,
      "step": 5627
    },
    {
      "epoch": 0.40604595793802534,
      "grad_norm": 3.9659537729944367,
      "learning_rate": 2.691159742096688e-06,
      "loss": 0.9804,
      "step": 5628
    },
    {
      "epoch": 0.40611810540745286,
      "grad_norm": 5.156965726249204,
      "learning_rate": 2.690721157764864e-06,
      "loss": 0.8261,
      "step": 5629
    },
    {
      "epoch": 0.4061902528768803,
      "grad_norm": 3.5388409879973524,
      "learning_rate": 2.690282535715315e-06,
      "loss": 1.0,
      "step": 5630
    },
    {
      "epoch": 0.40626240034630784,
      "grad_norm": 14.96370425624482,
      "learning_rate": 2.689843875971992e-06,
      "loss": 0.9884,
      "step": 5631
    },
    {
      "epoch": 0.40633454781573536,
      "grad_norm": 2.45836570416706,
      "learning_rate": 2.6894051785588486e-06,
      "loss": 1.0816,
      "step": 5632
    },
    {
      "epoch": 0.4064066952851629,
      "grad_norm": 2.83469803051301,
      "learning_rate": 2.68896644349984e-06,
      "loss": 1.052,
      "step": 5633
    },
    {
      "epoch": 0.4064788427545904,
      "grad_norm": 2.7435646185671776,
      "learning_rate": 2.688527670818925e-06,
      "loss": 0.9489,
      "step": 5634
    },
    {
      "epoch": 0.4065509902240179,
      "grad_norm": 3.1714493790119453,
      "learning_rate": 2.688088860540062e-06,
      "loss": 1.0391,
      "step": 5635
    },
    {
      "epoch": 0.4066231376934454,
      "grad_norm": 3.419083802491819,
      "learning_rate": 2.6876500126872144e-06,
      "loss": 0.9004,
      "step": 5636
    },
    {
      "epoch": 0.4066952851628729,
      "grad_norm": 3.5717098915473864,
      "learning_rate": 2.6872111272843453e-06,
      "loss": 1.0537,
      "step": 5637
    },
    {
      "epoch": 0.4067674326323004,
      "grad_norm": 3.150464335535316,
      "learning_rate": 2.686772204355419e-06,
      "loss": 0.9713,
      "step": 5638
    },
    {
      "epoch": 0.40683958010172794,
      "grad_norm": 8.429331802910282,
      "learning_rate": 2.6863332439244064e-06,
      "loss": 0.9329,
      "step": 5639
    },
    {
      "epoch": 0.40691172757115546,
      "grad_norm": 3.336941784527809,
      "learning_rate": 2.685894246015276e-06,
      "loss": 1.0453,
      "step": 5640
    },
    {
      "epoch": 0.406983875040583,
      "grad_norm": 2.0034026764321613,
      "learning_rate": 2.685455210651999e-06,
      "loss": 0.9823,
      "step": 5641
    },
    {
      "epoch": 0.40705602251001044,
      "grad_norm": 3.337289290616733,
      "learning_rate": 2.685016137858551e-06,
      "loss": 1.0527,
      "step": 5642
    },
    {
      "epoch": 0.40712816997943796,
      "grad_norm": 5.580965574392455,
      "learning_rate": 2.684577027658908e-06,
      "loss": 1.0411,
      "step": 5643
    },
    {
      "epoch": 0.4072003174488655,
      "grad_norm": 3.7745062852575475,
      "learning_rate": 2.6841378800770474e-06,
      "loss": 0.9583,
      "step": 5644
    },
    {
      "epoch": 0.407272464918293,
      "grad_norm": 13.267524502366612,
      "learning_rate": 2.6836986951369494e-06,
      "loss": 0.8701,
      "step": 5645
    },
    {
      "epoch": 0.4073446123877205,
      "grad_norm": 2.9547151996962278,
      "learning_rate": 2.683259472862597e-06,
      "loss": 1.0017,
      "step": 5646
    },
    {
      "epoch": 0.40741675985714804,
      "grad_norm": 2.86348325009024,
      "learning_rate": 2.682820213277974e-06,
      "loss": 1.0216,
      "step": 5647
    },
    {
      "epoch": 0.4074889073265755,
      "grad_norm": 3.4523241314025195,
      "learning_rate": 2.682380916407067e-06,
      "loss": 0.9509,
      "step": 5648
    },
    {
      "epoch": 0.407561054796003,
      "grad_norm": 8.079647783968573,
      "learning_rate": 2.6819415822738643e-06,
      "loss": 0.8622,
      "step": 5649
    },
    {
      "epoch": 0.40763320226543054,
      "grad_norm": 3.1344315680650636,
      "learning_rate": 2.6815022109023555e-06,
      "loss": 0.9163,
      "step": 5650
    },
    {
      "epoch": 0.40770534973485806,
      "grad_norm": 3.1249270621370924,
      "learning_rate": 2.6810628023165345e-06,
      "loss": 0.8757,
      "step": 5651
    },
    {
      "epoch": 0.4077774972042856,
      "grad_norm": 4.821596144736157,
      "learning_rate": 2.680623356540395e-06,
      "loss": 0.9504,
      "step": 5652
    },
    {
      "epoch": 0.4078496446737131,
      "grad_norm": 3.102639883178507,
      "learning_rate": 2.6801838735979333e-06,
      "loss": 0.9911,
      "step": 5653
    },
    {
      "epoch": 0.40792179214314056,
      "grad_norm": 2.998780797535472,
      "learning_rate": 2.679744353513149e-06,
      "loss": 0.958,
      "step": 5654
    },
    {
      "epoch": 0.4079939396125681,
      "grad_norm": 2.5976934215209337,
      "learning_rate": 2.679304796310041e-06,
      "loss": 1.0275,
      "step": 5655
    },
    {
      "epoch": 0.4080660870819956,
      "grad_norm": 3.1477970958110504,
      "learning_rate": 2.6788652020126132e-06,
      "loss": 0.975,
      "step": 5656
    },
    {
      "epoch": 0.4081382345514231,
      "grad_norm": 2.9873182552711666,
      "learning_rate": 2.6784255706448693e-06,
      "loss": 1.0653,
      "step": 5657
    },
    {
      "epoch": 0.40821038202085064,
      "grad_norm": 6.147477614279405,
      "learning_rate": 2.6779859022308172e-06,
      "loss": 0.871,
      "step": 5658
    },
    {
      "epoch": 0.4082825294902781,
      "grad_norm": 3.3607016139691375,
      "learning_rate": 2.6775461967944645e-06,
      "loss": 1.0508,
      "step": 5659
    },
    {
      "epoch": 0.4083546769597056,
      "grad_norm": 3.5308082776909644,
      "learning_rate": 2.677106454359822e-06,
      "loss": 0.9579,
      "step": 5660
    },
    {
      "epoch": 0.40842682442913314,
      "grad_norm": 3.5562181467999676,
      "learning_rate": 2.676666674950903e-06,
      "loss": 1.0545,
      "step": 5661
    },
    {
      "epoch": 0.40849897189856066,
      "grad_norm": 2.7108505191096,
      "learning_rate": 2.676226858591721e-06,
      "loss": 1.1121,
      "step": 5662
    },
    {
      "epoch": 0.4085711193679882,
      "grad_norm": 12.40483304250807,
      "learning_rate": 2.675787005306294e-06,
      "loss": 0.9877,
      "step": 5663
    },
    {
      "epoch": 0.4086432668374157,
      "grad_norm": 2.838367776473212,
      "learning_rate": 2.675347115118641e-06,
      "loss": 0.9644,
      "step": 5664
    },
    {
      "epoch": 0.40871541430684316,
      "grad_norm": 3.5972287796988467,
      "learning_rate": 2.674907188052781e-06,
      "loss": 0.9174,
      "step": 5665
    },
    {
      "epoch": 0.4087875617762707,
      "grad_norm": 0.7287565349462519,
      "learning_rate": 2.674467224132738e-06,
      "loss": 0.8818,
      "step": 5666
    },
    {
      "epoch": 0.4088597092456982,
      "grad_norm": 2.4346184938072297,
      "learning_rate": 2.6740272233825373e-06,
      "loss": 0.9587,
      "step": 5667
    },
    {
      "epoch": 0.4089318567151257,
      "grad_norm": 2.8686358460296884,
      "learning_rate": 2.6735871858262054e-06,
      "loss": 0.9414,
      "step": 5668
    },
    {
      "epoch": 0.40900400418455324,
      "grad_norm": 3.9470885254524997,
      "learning_rate": 2.67314711148777e-06,
      "loss": 0.9907,
      "step": 5669
    },
    {
      "epoch": 0.40907615165398076,
      "grad_norm": 2.9822688634926453,
      "learning_rate": 2.672707000391263e-06,
      "loss": 0.8429,
      "step": 5670
    },
    {
      "epoch": 0.4091482991234082,
      "grad_norm": 2.713146881458173,
      "learning_rate": 2.672266852560717e-06,
      "loss": 0.9427,
      "step": 5671
    },
    {
      "epoch": 0.40922044659283574,
      "grad_norm": 4.302210820535872,
      "learning_rate": 2.6718266680201663e-06,
      "loss": 1.0224,
      "step": 5672
    },
    {
      "epoch": 0.40929259406226326,
      "grad_norm": 4.227048948150304,
      "learning_rate": 2.6713864467936492e-06,
      "loss": 0.9275,
      "step": 5673
    },
    {
      "epoch": 0.4093647415316908,
      "grad_norm": 3.4204060785240684,
      "learning_rate": 2.6709461889052037e-06,
      "loss": 0.9463,
      "step": 5674
    },
    {
      "epoch": 0.4094368890011183,
      "grad_norm": 5.090041606860431,
      "learning_rate": 2.6705058943788695e-06,
      "loss": 1.0375,
      "step": 5675
    },
    {
      "epoch": 0.4095090364705458,
      "grad_norm": 3.632046001336265,
      "learning_rate": 2.670065563238691e-06,
      "loss": 1.0515,
      "step": 5676
    },
    {
      "epoch": 0.4095811839399733,
      "grad_norm": 4.282027348024358,
      "learning_rate": 2.6696251955087126e-06,
      "loss": 1.0085,
      "step": 5677
    },
    {
      "epoch": 0.4096533314094008,
      "grad_norm": 2.788796206458702,
      "learning_rate": 2.6691847912129816e-06,
      "loss": 0.9637,
      "step": 5678
    },
    {
      "epoch": 0.4097254788788283,
      "grad_norm": 2.5154373383274247,
      "learning_rate": 2.668744350375546e-06,
      "loss": 0.946,
      "step": 5679
    },
    {
      "epoch": 0.40979762634825584,
      "grad_norm": 6.135850287648926,
      "learning_rate": 2.668303873020457e-06,
      "loss": 0.9244,
      "step": 5680
    },
    {
      "epoch": 0.40986977381768336,
      "grad_norm": 2.3668934856666985,
      "learning_rate": 2.6678633591717678e-06,
      "loss": 1.0041,
      "step": 5681
    },
    {
      "epoch": 0.4099419212871109,
      "grad_norm": 2.439871783405433,
      "learning_rate": 2.6674228088535325e-06,
      "loss": 0.9538,
      "step": 5682
    },
    {
      "epoch": 0.41001406875653834,
      "grad_norm": 3.180229122136817,
      "learning_rate": 2.6669822220898085e-06,
      "loss": 1.0262,
      "step": 5683
    },
    {
      "epoch": 0.41008621622596586,
      "grad_norm": 5.291932086846163,
      "learning_rate": 2.666541598904654e-06,
      "loss": 0.8526,
      "step": 5684
    },
    {
      "epoch": 0.4101583636953934,
      "grad_norm": 3.707114215722338,
      "learning_rate": 2.666100939322131e-06,
      "loss": 1.0206,
      "step": 5685
    },
    {
      "epoch": 0.4102305111648209,
      "grad_norm": 3.8468098440470544,
      "learning_rate": 2.6656602433663016e-06,
      "loss": 0.9939,
      "step": 5686
    },
    {
      "epoch": 0.4103026586342484,
      "grad_norm": 4.349140582196779,
      "learning_rate": 2.6652195110612295e-06,
      "loss": 0.966,
      "step": 5687
    },
    {
      "epoch": 0.41037480610367594,
      "grad_norm": 6.684061467689942,
      "learning_rate": 2.664778742430983e-06,
      "loss": 0.964,
      "step": 5688
    },
    {
      "epoch": 0.4104469535731034,
      "grad_norm": 3.5030376331552247,
      "learning_rate": 2.6643379374996304e-06,
      "loss": 1.0277,
      "step": 5689
    },
    {
      "epoch": 0.4105191010425309,
      "grad_norm": 4.883709878475996,
      "learning_rate": 2.663897096291242e-06,
      "loss": 0.9546,
      "step": 5690
    },
    {
      "epoch": 0.41059124851195844,
      "grad_norm": 4.030832668259961,
      "learning_rate": 2.6634562188298917e-06,
      "loss": 1.0243,
      "step": 5691
    },
    {
      "epoch": 0.41066339598138596,
      "grad_norm": 4.226146628179857,
      "learning_rate": 2.663015305139652e-06,
      "loss": 0.9161,
      "step": 5692
    },
    {
      "epoch": 0.4107355434508135,
      "grad_norm": 3.0670955199151675,
      "learning_rate": 2.6625743552446023e-06,
      "loss": 0.9382,
      "step": 5693
    },
    {
      "epoch": 0.410807690920241,
      "grad_norm": 2.7580647555820526,
      "learning_rate": 2.662133369168819e-06,
      "loss": 1.0094,
      "step": 5694
    },
    {
      "epoch": 0.41087983838966846,
      "grad_norm": 3.422958829453303,
      "learning_rate": 2.6616923469363837e-06,
      "loss": 0.9199,
      "step": 5695
    },
    {
      "epoch": 0.410951985859096,
      "grad_norm": 2.5303726559630806,
      "learning_rate": 2.6612512885713794e-06,
      "loss": 0.9983,
      "step": 5696
    },
    {
      "epoch": 0.4110241333285235,
      "grad_norm": 3.147000498385855,
      "learning_rate": 2.660810194097889e-06,
      "loss": 0.8873,
      "step": 5697
    },
    {
      "epoch": 0.411096280797951,
      "grad_norm": 2.3111116520359354,
      "learning_rate": 2.660369063540001e-06,
      "loss": 0.9661,
      "step": 5698
    },
    {
      "epoch": 0.41116842826737854,
      "grad_norm": 4.001463383970524,
      "learning_rate": 2.6599278969218024e-06,
      "loss": 0.9329,
      "step": 5699
    },
    {
      "epoch": 0.41124057573680606,
      "grad_norm": 3.279563833870243,
      "learning_rate": 2.6594866942673845e-06,
      "loss": 0.9936,
      "step": 5700
    },
    {
      "epoch": 0.4113127232062335,
      "grad_norm": 3.6662731537476345,
      "learning_rate": 2.6590454556008395e-06,
      "loss": 0.9237,
      "step": 5701
    },
    {
      "epoch": 0.41138487067566104,
      "grad_norm": 2.4539541799096307,
      "learning_rate": 2.658604180946262e-06,
      "loss": 0.9585,
      "step": 5702
    },
    {
      "epoch": 0.41145701814508856,
      "grad_norm": 3.2630095160740407,
      "learning_rate": 2.658162870327749e-06,
      "loss": 0.89,
      "step": 5703
    },
    {
      "epoch": 0.4115291656145161,
      "grad_norm": 3.3540162394340323,
      "learning_rate": 2.6577215237693967e-06,
      "loss": 0.9512,
      "step": 5704
    },
    {
      "epoch": 0.4116013130839436,
      "grad_norm": 2.9844360345321306,
      "learning_rate": 2.6572801412953077e-06,
      "loss": 0.9433,
      "step": 5705
    },
    {
      "epoch": 0.4116734605533711,
      "grad_norm": 3.811882969687969,
      "learning_rate": 2.6568387229295833e-06,
      "loss": 0.8987,
      "step": 5706
    },
    {
      "epoch": 0.4117456080227986,
      "grad_norm": 2.71837429213018,
      "learning_rate": 2.6563972686963275e-06,
      "loss": 0.9836,
      "step": 5707
    },
    {
      "epoch": 0.4118177554922261,
      "grad_norm": 5.5861106152151985,
      "learning_rate": 2.655955778619647e-06,
      "loss": 0.9372,
      "step": 5708
    },
    {
      "epoch": 0.4118899029616536,
      "grad_norm": 4.261288181446079,
      "learning_rate": 2.6555142527236493e-06,
      "loss": 1.0358,
      "step": 5709
    },
    {
      "epoch": 0.41196205043108114,
      "grad_norm": 3.138743430024035,
      "learning_rate": 2.6550726910324455e-06,
      "loss": 1.0098,
      "step": 5710
    },
    {
      "epoch": 0.41203419790050866,
      "grad_norm": 9.699131043829189,
      "learning_rate": 2.654631093570146e-06,
      "loss": 1.0097,
      "step": 5711
    },
    {
      "epoch": 0.4121063453699361,
      "grad_norm": 3.0260046181350435,
      "learning_rate": 2.6541894603608668e-06,
      "loss": 0.8963,
      "step": 5712
    },
    {
      "epoch": 0.41217849283936364,
      "grad_norm": 2.8111066969818856,
      "learning_rate": 2.653747791428723e-06,
      "loss": 0.9315,
      "step": 5713
    },
    {
      "epoch": 0.41225064030879116,
      "grad_norm": 7.722921749909461,
      "learning_rate": 2.6533060867978317e-06,
      "loss": 0.9606,
      "step": 5714
    },
    {
      "epoch": 0.4123227877782187,
      "grad_norm": 5.879675769819551,
      "learning_rate": 2.652864346492314e-06,
      "loss": 0.9086,
      "step": 5715
    },
    {
      "epoch": 0.4123949352476462,
      "grad_norm": 3.7456879301495953,
      "learning_rate": 2.65242257053629e-06,
      "loss": 1.0682,
      "step": 5716
    },
    {
      "epoch": 0.4124670827170737,
      "grad_norm": 5.3401780710997535,
      "learning_rate": 2.651980758953886e-06,
      "loss": 0.9926,
      "step": 5717
    },
    {
      "epoch": 0.4125392301865012,
      "grad_norm": 4.169704830051041,
      "learning_rate": 2.6515389117692264e-06,
      "loss": 0.9223,
      "step": 5718
    },
    {
      "epoch": 0.4126113776559287,
      "grad_norm": 4.119070473042389,
      "learning_rate": 2.651097029006438e-06,
      "loss": 0.99,
      "step": 5719
    },
    {
      "epoch": 0.4126835251253562,
      "grad_norm": 2.731908328678415,
      "learning_rate": 2.650655110689651e-06,
      "loss": 0.9471,
      "step": 5720
    },
    {
      "epoch": 0.41275567259478374,
      "grad_norm": 3.6563093392537187,
      "learning_rate": 2.6502131568429976e-06,
      "loss": 0.9383,
      "step": 5721
    },
    {
      "epoch": 0.41282782006421126,
      "grad_norm": 0.653458288474133,
      "learning_rate": 2.6497711674906105e-06,
      "loss": 0.8201,
      "step": 5722
    },
    {
      "epoch": 0.4128999675336388,
      "grad_norm": 3.2705655484415583,
      "learning_rate": 2.649329142656625e-06,
      "loss": 0.9244,
      "step": 5723
    },
    {
      "epoch": 0.41297211500306624,
      "grad_norm": 3.429765496203458,
      "learning_rate": 2.6488870823651793e-06,
      "loss": 0.9681,
      "step": 5724
    },
    {
      "epoch": 0.41304426247249376,
      "grad_norm": 0.6850595508223446,
      "learning_rate": 2.648444986640412e-06,
      "loss": 0.8258,
      "step": 5725
    },
    {
      "epoch": 0.4131164099419213,
      "grad_norm": 3.08622427464977,
      "learning_rate": 2.6480028555064636e-06,
      "loss": 0.9423,
      "step": 5726
    },
    {
      "epoch": 0.4131885574113488,
      "grad_norm": 7.646739050276748,
      "learning_rate": 2.6475606889874795e-06,
      "loss": 0.9553,
      "step": 5727
    },
    {
      "epoch": 0.4132607048807763,
      "grad_norm": 5.13288496131786,
      "learning_rate": 2.647118487107602e-06,
      "loss": 0.9809,
      "step": 5728
    },
    {
      "epoch": 0.41333285235020384,
      "grad_norm": 5.887030762198237,
      "learning_rate": 2.6466762498909805e-06,
      "loss": 1.0337,
      "step": 5729
    },
    {
      "epoch": 0.4134049998196313,
      "grad_norm": 2.6942038204684526,
      "learning_rate": 2.6462339773617628e-06,
      "loss": 0.9483,
      "step": 5730
    },
    {
      "epoch": 0.4134771472890588,
      "grad_norm": 4.003700690226349,
      "learning_rate": 2.645791669544099e-06,
      "loss": 0.9852,
      "step": 5731
    },
    {
      "epoch": 0.41354929475848634,
      "grad_norm": 4.113252732050357,
      "learning_rate": 2.6453493264621436e-06,
      "loss": 0.9805,
      "step": 5732
    },
    {
      "epoch": 0.41362144222791386,
      "grad_norm": 3.105912163705269,
      "learning_rate": 2.64490694814005e-06,
      "loss": 0.9869,
      "step": 5733
    },
    {
      "epoch": 0.4136935896973414,
      "grad_norm": 4.329431439925753,
      "learning_rate": 2.644464534601976e-06,
      "loss": 1.0153,
      "step": 5734
    },
    {
      "epoch": 0.4137657371667689,
      "grad_norm": 3.3808579645974817,
      "learning_rate": 2.644022085872079e-06,
      "loss": 0.9745,
      "step": 5735
    },
    {
      "epoch": 0.41383788463619636,
      "grad_norm": 3.238743582232943,
      "learning_rate": 2.64357960197452e-06,
      "loss": 1.0884,
      "step": 5736
    },
    {
      "epoch": 0.4139100321056239,
      "grad_norm": 2.7362995459339103,
      "learning_rate": 2.6431370829334614e-06,
      "loss": 1.1463,
      "step": 5737
    },
    {
      "epoch": 0.4139821795750514,
      "grad_norm": 3.3624090643289297,
      "learning_rate": 2.6426945287730677e-06,
      "loss": 1.0499,
      "step": 5738
    },
    {
      "epoch": 0.4140543270444789,
      "grad_norm": 2.8309868090237034,
      "learning_rate": 2.6422519395175044e-06,
      "loss": 1.0364,
      "step": 5739
    },
    {
      "epoch": 0.41412647451390644,
      "grad_norm": 3.439949550036936,
      "learning_rate": 2.641809315190941e-06,
      "loss": 0.9843,
      "step": 5740
    },
    {
      "epoch": 0.41419862198333396,
      "grad_norm": 4.100613496982585,
      "learning_rate": 2.641366655817546e-06,
      "loss": 1.0436,
      "step": 5741
    },
    {
      "epoch": 0.4142707694527614,
      "grad_norm": 2.457221237624387,
      "learning_rate": 2.6409239614214933e-06,
      "loss": 0.9819,
      "step": 5742
    },
    {
      "epoch": 0.41434291692218894,
      "grad_norm": 3.18887284577957,
      "learning_rate": 2.6404812320269548e-06,
      "loss": 1.0615,
      "step": 5743
    },
    {
      "epoch": 0.41441506439161646,
      "grad_norm": 3.1273665050676493,
      "learning_rate": 2.6400384676581073e-06,
      "loss": 1.0391,
      "step": 5744
    },
    {
      "epoch": 0.414487211861044,
      "grad_norm": 5.16256955448614,
      "learning_rate": 2.6395956683391283e-06,
      "loss": 0.9255,
      "step": 5745
    },
    {
      "epoch": 0.4145593593304715,
      "grad_norm": 4.778842494427688,
      "learning_rate": 2.6391528340941984e-06,
      "loss": 1.051,
      "step": 5746
    },
    {
      "epoch": 0.414631506799899,
      "grad_norm": 5.76002656559646,
      "learning_rate": 2.6387099649474974e-06,
      "loss": 0.9509,
      "step": 5747
    },
    {
      "epoch": 0.4147036542693265,
      "grad_norm": 9.940309523247691,
      "learning_rate": 2.6382670609232093e-06,
      "loss": 1.0135,
      "step": 5748
    },
    {
      "epoch": 0.414775801738754,
      "grad_norm": 4.505068785283066,
      "learning_rate": 2.637824122045521e-06,
      "loss": 1.0181,
      "step": 5749
    },
    {
      "epoch": 0.4148479492081815,
      "grad_norm": 2.90846184680684,
      "learning_rate": 2.6373811483386173e-06,
      "loss": 0.9922,
      "step": 5750
    },
    {
      "epoch": 0.41492009667760904,
      "grad_norm": 2.8169264822166338,
      "learning_rate": 2.6369381398266894e-06,
      "loss": 0.9535,
      "step": 5751
    },
    {
      "epoch": 0.41499224414703656,
      "grad_norm": 5.6822733915217505,
      "learning_rate": 2.6364950965339277e-06,
      "loss": 0.9059,
      "step": 5752
    },
    {
      "epoch": 0.4150643916164641,
      "grad_norm": 2.7326171456203987,
      "learning_rate": 2.6360520184845243e-06,
      "loss": 1.0736,
      "step": 5753
    },
    {
      "epoch": 0.41513653908589154,
      "grad_norm": 4.33981474309613,
      "learning_rate": 2.6356089057026755e-06,
      "loss": 0.8888,
      "step": 5754
    },
    {
      "epoch": 0.41520868655531906,
      "grad_norm": 0.6815942568160374,
      "learning_rate": 2.6351657582125768e-06,
      "loss": 0.819,
      "step": 5755
    },
    {
      "epoch": 0.4152808340247466,
      "grad_norm": 5.7763601707744785,
      "learning_rate": 2.634722576038427e-06,
      "loss": 0.9944,
      "step": 5756
    },
    {
      "epoch": 0.4153529814941741,
      "grad_norm": 2.821376103332393,
      "learning_rate": 2.634279359204428e-06,
      "loss": 0.9339,
      "step": 5757
    },
    {
      "epoch": 0.4154251289636016,
      "grad_norm": 3.6778023136648383,
      "learning_rate": 2.6338361077347806e-06,
      "loss": 0.9766,
      "step": 5758
    },
    {
      "epoch": 0.41549727643302914,
      "grad_norm": 3.7358486188895927,
      "learning_rate": 2.6333928216536894e-06,
      "loss": 1.0404,
      "step": 5759
    },
    {
      "epoch": 0.4155694239024566,
      "grad_norm": 4.918451383995133,
      "learning_rate": 2.6329495009853617e-06,
      "loss": 1.0163,
      "step": 5760
    },
    {
      "epoch": 0.4156415713718841,
      "grad_norm": 11.069279721474478,
      "learning_rate": 2.6325061457540045e-06,
      "loss": 0.9401,
      "step": 5761
    },
    {
      "epoch": 0.41571371884131164,
      "grad_norm": 3.2882644518756017,
      "learning_rate": 2.632062755983828e-06,
      "loss": 0.9659,
      "step": 5762
    },
    {
      "epoch": 0.41578586631073916,
      "grad_norm": 2.201216933323208,
      "learning_rate": 2.6316193316990443e-06,
      "loss": 0.9446,
      "step": 5763
    },
    {
      "epoch": 0.4158580137801667,
      "grad_norm": 3.06187339133263,
      "learning_rate": 2.6311758729238673e-06,
      "loss": 0.9461,
      "step": 5764
    },
    {
      "epoch": 0.41593016124959414,
      "grad_norm": 4.969423260450241,
      "learning_rate": 2.630732379682512e-06,
      "loss": 0.8969,
      "step": 5765
    },
    {
      "epoch": 0.41600230871902166,
      "grad_norm": 8.738576789397642,
      "learning_rate": 2.6302888519991964e-06,
      "loss": 1.0078,
      "step": 5766
    },
    {
      "epoch": 0.4160744561884492,
      "grad_norm": 2.7433337997337244,
      "learning_rate": 2.6298452898981404e-06,
      "loss": 0.9832,
      "step": 5767
    },
    {
      "epoch": 0.4161466036578767,
      "grad_norm": 4.680830419502259,
      "learning_rate": 2.6294016934035645e-06,
      "loss": 0.9622,
      "step": 5768
    },
    {
      "epoch": 0.4162187511273042,
      "grad_norm": 8.247693288433933,
      "learning_rate": 2.628958062539692e-06,
      "loss": 0.996,
      "step": 5769
    },
    {
      "epoch": 0.41629089859673174,
      "grad_norm": 3.0637109658888066,
      "learning_rate": 2.6285143973307477e-06,
      "loss": 0.9176,
      "step": 5770
    },
    {
      "epoch": 0.4163630460661592,
      "grad_norm": 2.5682334956081907,
      "learning_rate": 2.6280706978009592e-06,
      "loss": 0.9434,
      "step": 5771
    },
    {
      "epoch": 0.4164351935355867,
      "grad_norm": 3.483652082640536,
      "learning_rate": 2.6276269639745543e-06,
      "loss": 0.9749,
      "step": 5772
    },
    {
      "epoch": 0.41650734100501424,
      "grad_norm": 2.8511123092832533,
      "learning_rate": 2.6271831958757653e-06,
      "loss": 1.0104,
      "step": 5773
    },
    {
      "epoch": 0.41657948847444176,
      "grad_norm": 4.029848076770461,
      "learning_rate": 2.6267393935288233e-06,
      "loss": 0.8504,
      "step": 5774
    },
    {
      "epoch": 0.4166516359438693,
      "grad_norm": 4.210865027205663,
      "learning_rate": 2.626295556957963e-06,
      "loss": 1.0057,
      "step": 5775
    },
    {
      "epoch": 0.4167237834132968,
      "grad_norm": 3.1736720214014347,
      "learning_rate": 2.625851686187421e-06,
      "loss": 0.9814,
      "step": 5776
    },
    {
      "epoch": 0.41679593088272426,
      "grad_norm": 7.132537869935529,
      "learning_rate": 2.6254077812414345e-06,
      "loss": 0.9064,
      "step": 5777
    },
    {
      "epoch": 0.4168680783521518,
      "grad_norm": 4.488297398524039,
      "learning_rate": 2.6249638421442457e-06,
      "loss": 0.9905,
      "step": 5778
    },
    {
      "epoch": 0.4169402258215793,
      "grad_norm": 2.9158702489611668,
      "learning_rate": 2.6245198689200943e-06,
      "loss": 0.9473,
      "step": 5779
    },
    {
      "epoch": 0.4170123732910068,
      "grad_norm": 5.295054435125976,
      "learning_rate": 2.6240758615932238e-06,
      "loss": 0.8795,
      "step": 5780
    },
    {
      "epoch": 0.41708452076043434,
      "grad_norm": 2.4950492476378954,
      "learning_rate": 2.623631820187882e-06,
      "loss": 0.8951,
      "step": 5781
    },
    {
      "epoch": 0.41715666822986186,
      "grad_norm": 2.8314432309802138,
      "learning_rate": 2.6231877447283147e-06,
      "loss": 0.9107,
      "step": 5782
    },
    {
      "epoch": 0.4172288156992893,
      "grad_norm": 8.076564615979603,
      "learning_rate": 2.6227436352387717e-06,
      "loss": 1.0477,
      "step": 5783
    },
    {
      "epoch": 0.41730096316871684,
      "grad_norm": 2.7021405847543556,
      "learning_rate": 2.622299491743504e-06,
      "loss": 0.9697,
      "step": 5784
    },
    {
      "epoch": 0.41737311063814436,
      "grad_norm": 2.4817237382538564,
      "learning_rate": 2.6218553142667647e-06,
      "loss": 1.0254,
      "step": 5785
    },
    {
      "epoch": 0.4174452581075719,
      "grad_norm": 3.6800542110098977,
      "learning_rate": 2.6214111028328086e-06,
      "loss": 0.9874,
      "step": 5786
    },
    {
      "epoch": 0.4175174055769994,
      "grad_norm": 3.5183885351416646,
      "learning_rate": 2.6209668574658926e-06,
      "loss": 0.8583,
      "step": 5787
    },
    {
      "epoch": 0.4175895530464269,
      "grad_norm": 5.772669677883963,
      "learning_rate": 2.620522578190276e-06,
      "loss": 0.8155,
      "step": 5788
    },
    {
      "epoch": 0.4176617005158544,
      "grad_norm": 3.042237337132311,
      "learning_rate": 2.620078265030217e-06,
      "loss": 1.0051,
      "step": 5789
    },
    {
      "epoch": 0.4177338479852819,
      "grad_norm": 2.7312762331193787,
      "learning_rate": 2.619633918009981e-06,
      "loss": 0.8819,
      "step": 5790
    },
    {
      "epoch": 0.4178059954547094,
      "grad_norm": 3.321169323040041,
      "learning_rate": 2.6191895371538296e-06,
      "loss": 1.0818,
      "step": 5791
    },
    {
      "epoch": 0.41787814292413694,
      "grad_norm": 2.944247511109969,
      "learning_rate": 2.6187451224860296e-06,
      "loss": 1.0274,
      "step": 5792
    },
    {
      "epoch": 0.41795029039356446,
      "grad_norm": 3.3302329108546673,
      "learning_rate": 2.61830067403085e-06,
      "loss": 0.9738,
      "step": 5793
    },
    {
      "epoch": 0.418022437862992,
      "grad_norm": 2.861691063266115,
      "learning_rate": 2.6178561918125582e-06,
      "loss": 0.9175,
      "step": 5794
    },
    {
      "epoch": 0.41809458533241944,
      "grad_norm": 4.102384125295266,
      "learning_rate": 2.6174116758554277e-06,
      "loss": 0.8815,
      "step": 5795
    },
    {
      "epoch": 0.41816673280184696,
      "grad_norm": 3.00389355085092,
      "learning_rate": 2.6169671261837313e-06,
      "loss": 0.9573,
      "step": 5796
    },
    {
      "epoch": 0.4182388802712745,
      "grad_norm": 2.848752662352807,
      "learning_rate": 2.6165225428217436e-06,
      "loss": 0.9549,
      "step": 5797
    },
    {
      "epoch": 0.418311027740702,
      "grad_norm": 3.3443835362421868,
      "learning_rate": 2.6160779257937424e-06,
      "loss": 0.8392,
      "step": 5798
    },
    {
      "epoch": 0.4183831752101295,
      "grad_norm": 9.370987096881624,
      "learning_rate": 2.615633275124006e-06,
      "loss": 0.9607,
      "step": 5799
    },
    {
      "epoch": 0.41845532267955704,
      "grad_norm": 4.507693813090374,
      "learning_rate": 2.615188590836816e-06,
      "loss": 0.9748,
      "step": 5800
    },
    {
      "epoch": 0.4185274701489845,
      "grad_norm": 2.7693787009512127,
      "learning_rate": 2.614743872956454e-06,
      "loss": 1.0477,
      "step": 5801
    },
    {
      "epoch": 0.418599617618412,
      "grad_norm": 2.9220954689445833,
      "learning_rate": 2.614299121507205e-06,
      "loss": 1.0085,
      "step": 5802
    },
    {
      "epoch": 0.41867176508783954,
      "grad_norm": 2.590959081517338,
      "learning_rate": 2.6138543365133554e-06,
      "loss": 0.9375,
      "step": 5803
    },
    {
      "epoch": 0.41874391255726706,
      "grad_norm": 2.9109531207645825,
      "learning_rate": 2.6134095179991918e-06,
      "loss": 0.9555,
      "step": 5804
    },
    {
      "epoch": 0.4188160600266946,
      "grad_norm": 5.284583129032625,
      "learning_rate": 2.612964665989006e-06,
      "loss": 0.946,
      "step": 5805
    },
    {
      "epoch": 0.4188882074961221,
      "grad_norm": 2.3296184486067384,
      "learning_rate": 2.6125197805070887e-06,
      "loss": 0.8996,
      "step": 5806
    },
    {
      "epoch": 0.41896035496554956,
      "grad_norm": 2.834645210684925,
      "learning_rate": 2.6120748615777333e-06,
      "loss": 0.9473,
      "step": 5807
    },
    {
      "epoch": 0.4190325024349771,
      "grad_norm": 0.6977474140268296,
      "learning_rate": 2.611629909225236e-06,
      "loss": 0.8294,
      "step": 5808
    },
    {
      "epoch": 0.4191046499044046,
      "grad_norm": 2.5614829487423876,
      "learning_rate": 2.6111849234738935e-06,
      "loss": 0.98,
      "step": 5809
    },
    {
      "epoch": 0.4191767973738321,
      "grad_norm": 5.337234004448119,
      "learning_rate": 2.610739904348005e-06,
      "loss": 0.9621,
      "step": 5810
    },
    {
      "epoch": 0.41924894484325964,
      "grad_norm": 4.084987668563926,
      "learning_rate": 2.610294851871871e-06,
      "loss": 0.8964,
      "step": 5811
    },
    {
      "epoch": 0.4193210923126871,
      "grad_norm": 3.8457710721206686,
      "learning_rate": 2.6098497660697945e-06,
      "loss": 0.8776,
      "step": 5812
    },
    {
      "epoch": 0.4193932397821146,
      "grad_norm": 3.0729194425581126,
      "learning_rate": 2.6094046469660797e-06,
      "loss": 0.8696,
      "step": 5813
    },
    {
      "epoch": 0.41946538725154214,
      "grad_norm": 2.710770395718838,
      "learning_rate": 2.6089594945850327e-06,
      "loss": 1.056,
      "step": 5814
    },
    {
      "epoch": 0.41953753472096966,
      "grad_norm": 3.353135528974543,
      "learning_rate": 2.608514308950963e-06,
      "loss": 1.091,
      "step": 5815
    },
    {
      "epoch": 0.4196096821903972,
      "grad_norm": 2.3127341280800695,
      "learning_rate": 2.608069090088178e-06,
      "loss": 1.0191,
      "step": 5816
    },
    {
      "epoch": 0.4196818296598247,
      "grad_norm": 0.8012532773598046,
      "learning_rate": 2.6076238380209925e-06,
      "loss": 0.9052,
      "step": 5817
    },
    {
      "epoch": 0.41975397712925216,
      "grad_norm": 13.14331520656568,
      "learning_rate": 2.6071785527737183e-06,
      "loss": 1.0039,
      "step": 5818
    },
    {
      "epoch": 0.4198261245986797,
      "grad_norm": 4.009118177910589,
      "learning_rate": 2.6067332343706705e-06,
      "loss": 1.0274,
      "step": 5819
    },
    {
      "epoch": 0.4198982720681072,
      "grad_norm": 2.487290596893981,
      "learning_rate": 2.606287882836167e-06,
      "loss": 0.9423,
      "step": 5820
    },
    {
      "epoch": 0.4199704195375347,
      "grad_norm": 4.429018538621633,
      "learning_rate": 2.6058424981945265e-06,
      "loss": 0.7941,
      "step": 5821
    },
    {
      "epoch": 0.42004256700696224,
      "grad_norm": 3.0954782301984958,
      "learning_rate": 2.60539708047007e-06,
      "loss": 0.9245,
      "step": 5822
    },
    {
      "epoch": 0.42011471447638976,
      "grad_norm": 0.6914471328631221,
      "learning_rate": 2.60495162968712e-06,
      "loss": 0.8125,
      "step": 5823
    },
    {
      "epoch": 0.4201868619458172,
      "grad_norm": 4.391735004470022,
      "learning_rate": 2.6045061458700015e-06,
      "loss": 0.9845,
      "step": 5824
    },
    {
      "epoch": 0.42025900941524474,
      "grad_norm": 3.4379567623115808,
      "learning_rate": 2.6040606290430394e-06,
      "loss": 0.8758,
      "step": 5825
    },
    {
      "epoch": 0.42033115688467226,
      "grad_norm": 3.126036205157744,
      "learning_rate": 2.603615079230563e-06,
      "loss": 1.033,
      "step": 5826
    },
    {
      "epoch": 0.4204033043540998,
      "grad_norm": 4.167129618039262,
      "learning_rate": 2.6031694964569016e-06,
      "loss": 0.9588,
      "step": 5827
    },
    {
      "epoch": 0.4204754518235273,
      "grad_norm": 4.593602599808858,
      "learning_rate": 2.602723880746387e-06,
      "loss": 1.0291,
      "step": 5828
    },
    {
      "epoch": 0.4205475992929548,
      "grad_norm": 3.7552661794297806,
      "learning_rate": 2.6022782321233516e-06,
      "loss": 1.0899,
      "step": 5829
    },
    {
      "epoch": 0.4206197467623823,
      "grad_norm": 4.631090226808669,
      "learning_rate": 2.6018325506121324e-06,
      "loss": 0.9193,
      "step": 5830
    },
    {
      "epoch": 0.4206918942318098,
      "grad_norm": 4.952152770500717,
      "learning_rate": 2.601386836237065e-06,
      "loss": 1.0985,
      "step": 5831
    },
    {
      "epoch": 0.4207640417012373,
      "grad_norm": 4.938186211524671,
      "learning_rate": 2.6009410890224894e-06,
      "loss": 0.9483,
      "step": 5832
    },
    {
      "epoch": 0.42083618917066484,
      "grad_norm": 0.956730012779099,
      "learning_rate": 2.6004953089927445e-06,
      "loss": 0.9521,
      "step": 5833
    },
    {
      "epoch": 0.42090833664009236,
      "grad_norm": 3.081440490653117,
      "learning_rate": 2.6000494961721748e-06,
      "loss": 0.8657,
      "step": 5834
    },
    {
      "epoch": 0.4209804841095199,
      "grad_norm": 6.132247062906659,
      "learning_rate": 2.599603650585123e-06,
      "loss": 1.0831,
      "step": 5835
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 3.8419286863350015,
      "learning_rate": 2.5991577722559354e-06,
      "loss": 0.9038,
      "step": 5836
    },
    {
      "epoch": 0.42112477904837486,
      "grad_norm": 3.2227112921435834,
      "learning_rate": 2.59871186120896e-06,
      "loss": 0.9926,
      "step": 5837
    },
    {
      "epoch": 0.4211969265178024,
      "grad_norm": 6.0377532210864855,
      "learning_rate": 2.598265917468546e-06,
      "loss": 1.0723,
      "step": 5838
    },
    {
      "epoch": 0.4212690739872299,
      "grad_norm": 2.887698119661723,
      "learning_rate": 2.597819941059045e-06,
      "loss": 1.0286,
      "step": 5839
    },
    {
      "epoch": 0.4213412214566574,
      "grad_norm": 2.768428782134901,
      "learning_rate": 2.59737393200481e-06,
      "loss": 1.0527,
      "step": 5840
    },
    {
      "epoch": 0.42141336892608494,
      "grad_norm": 2.90812737319492,
      "learning_rate": 2.596927890330196e-06,
      "loss": 0.8193,
      "step": 5841
    },
    {
      "epoch": 0.4214855163955124,
      "grad_norm": 5.352661198233657,
      "learning_rate": 2.5964818160595605e-06,
      "loss": 0.9005,
      "step": 5842
    },
    {
      "epoch": 0.4215576638649399,
      "grad_norm": 2.3507981182797715,
      "learning_rate": 2.5960357092172598e-06,
      "loss": 0.9851,
      "step": 5843
    },
    {
      "epoch": 0.42162981133436744,
      "grad_norm": 4.067255611193758,
      "learning_rate": 2.5955895698276566e-06,
      "loss": 0.9328,
      "step": 5844
    },
    {
      "epoch": 0.42170195880379496,
      "grad_norm": 4.213742144907485,
      "learning_rate": 2.5951433979151117e-06,
      "loss": 0.9747,
      "step": 5845
    },
    {
      "epoch": 0.4217741062732225,
      "grad_norm": 4.185885901166907,
      "learning_rate": 2.5946971935039878e-06,
      "loss": 0.9186,
      "step": 5846
    },
    {
      "epoch": 0.42184625374265,
      "grad_norm": 0.7018366985112033,
      "learning_rate": 2.594250956618653e-06,
      "loss": 0.9238,
      "step": 5847
    },
    {
      "epoch": 0.42191840121207747,
      "grad_norm": 4.254929992960532,
      "learning_rate": 2.593804687283472e-06,
      "loss": 1.0174,
      "step": 5848
    },
    {
      "epoch": 0.421990548681505,
      "grad_norm": 6.545762871213875,
      "learning_rate": 2.593358385522816e-06,
      "loss": 0.9718,
      "step": 5849
    },
    {
      "epoch": 0.4220626961509325,
      "grad_norm": 6.089350124317327,
      "learning_rate": 2.5929120513610554e-06,
      "loss": 0.9584,
      "step": 5850
    },
    {
      "epoch": 0.42213484362036,
      "grad_norm": 9.732271166114375,
      "learning_rate": 2.592465684822562e-06,
      "loss": 0.8218,
      "step": 5851
    },
    {
      "epoch": 0.42220699108978754,
      "grad_norm": 3.781020606783102,
      "learning_rate": 2.592019285931711e-06,
      "loss": 0.9331,
      "step": 5852
    },
    {
      "epoch": 0.42227913855921506,
      "grad_norm": 3.361737081773382,
      "learning_rate": 2.5915728547128785e-06,
      "loss": 1.0682,
      "step": 5853
    },
    {
      "epoch": 0.4223512860286425,
      "grad_norm": 0.7916489691178334,
      "learning_rate": 2.591126391190442e-06,
      "loss": 0.8066,
      "step": 5854
    },
    {
      "epoch": 0.42242343349807004,
      "grad_norm": 2.566335645195076,
      "learning_rate": 2.5906798953887814e-06,
      "loss": 0.9552,
      "step": 5855
    },
    {
      "epoch": 0.42249558096749756,
      "grad_norm": 3.3698912207368963,
      "learning_rate": 2.590233367332279e-06,
      "loss": 1.0542,
      "step": 5856
    },
    {
      "epoch": 0.4225677284369251,
      "grad_norm": 3.8052995569360344,
      "learning_rate": 2.589786807045318e-06,
      "loss": 1.0017,
      "step": 5857
    },
    {
      "epoch": 0.4226398759063526,
      "grad_norm": 3.1867517827316973,
      "learning_rate": 2.5893402145522806e-06,
      "loss": 0.9133,
      "step": 5858
    },
    {
      "epoch": 0.4227120233757801,
      "grad_norm": 7.42504686539884,
      "learning_rate": 2.5888935898775573e-06,
      "loss": 0.9925,
      "step": 5859
    },
    {
      "epoch": 0.4227841708452076,
      "grad_norm": 2.9164597210809378,
      "learning_rate": 2.5884469330455348e-06,
      "loss": 0.9261,
      "step": 5860
    },
    {
      "epoch": 0.4228563183146351,
      "grad_norm": 3.4213041761329843,
      "learning_rate": 2.5880002440806037e-06,
      "loss": 0.9662,
      "step": 5861
    },
    {
      "epoch": 0.4229284657840626,
      "grad_norm": 2.180312637869289,
      "learning_rate": 2.587553523007156e-06,
      "loss": 1.0268,
      "step": 5862
    },
    {
      "epoch": 0.42300061325349014,
      "grad_norm": 5.790857199034472,
      "learning_rate": 2.587106769849585e-06,
      "loss": 1.0095,
      "step": 5863
    },
    {
      "epoch": 0.42307276072291766,
      "grad_norm": 2.5627956685012503,
      "learning_rate": 2.586659984632287e-06,
      "loss": 0.9089,
      "step": 5864
    },
    {
      "epoch": 0.4231449081923451,
      "grad_norm": 3.3117811754610513,
      "learning_rate": 2.586213167379659e-06,
      "loss": 0.8597,
      "step": 5865
    },
    {
      "epoch": 0.42321705566177265,
      "grad_norm": 2.8184805121662517,
      "learning_rate": 2.5857663181161006e-06,
      "loss": 0.9951,
      "step": 5866
    },
    {
      "epoch": 0.42328920313120016,
      "grad_norm": 0.6497157722549122,
      "learning_rate": 2.5853194368660114e-06,
      "loss": 0.8173,
      "step": 5867
    },
    {
      "epoch": 0.4233613506006277,
      "grad_norm": 3.036688262103354,
      "learning_rate": 2.584872523653794e-06,
      "loss": 0.9207,
      "step": 5868
    },
    {
      "epoch": 0.4234334980700552,
      "grad_norm": 3.2888415482123863,
      "learning_rate": 2.584425578503854e-06,
      "loss": 0.9016,
      "step": 5869
    },
    {
      "epoch": 0.4235056455394827,
      "grad_norm": 2.8131632552631816,
      "learning_rate": 2.583978601440596e-06,
      "loss": 0.9192,
      "step": 5870
    },
    {
      "epoch": 0.4235777930089102,
      "grad_norm": 4.594706117947342,
      "learning_rate": 2.583531592488429e-06,
      "loss": 0.9901,
      "step": 5871
    },
    {
      "epoch": 0.4236499404783377,
      "grad_norm": 2.8960768965422923,
      "learning_rate": 2.5830845516717618e-06,
      "loss": 0.9869,
      "step": 5872
    },
    {
      "epoch": 0.4237220879477652,
      "grad_norm": 2.300578861369806,
      "learning_rate": 2.5826374790150054e-06,
      "loss": 0.9494,
      "step": 5873
    },
    {
      "epoch": 0.42379423541719274,
      "grad_norm": 3.1446859191387135,
      "learning_rate": 2.5821903745425734e-06,
      "loss": 1.0138,
      "step": 5874
    },
    {
      "epoch": 0.42386638288662026,
      "grad_norm": 3.693069002789251,
      "learning_rate": 2.5817432382788802e-06,
      "loss": 1.0049,
      "step": 5875
    },
    {
      "epoch": 0.4239385303560478,
      "grad_norm": 2.985371847586055,
      "learning_rate": 2.581296070248342e-06,
      "loss": 0.9865,
      "step": 5876
    },
    {
      "epoch": 0.42401067782547525,
      "grad_norm": 2.3797979578443025,
      "learning_rate": 2.5808488704753773e-06,
      "loss": 1.0701,
      "step": 5877
    },
    {
      "epoch": 0.42408282529490277,
      "grad_norm": 7.459544106633367,
      "learning_rate": 2.5804016389844066e-06,
      "loss": 0.8872,
      "step": 5878
    },
    {
      "epoch": 0.4241549727643303,
      "grad_norm": 3.177663848038041,
      "learning_rate": 2.5799543757998516e-06,
      "loss": 0.9237,
      "step": 5879
    },
    {
      "epoch": 0.4242271202337578,
      "grad_norm": 3.1360516248463086,
      "learning_rate": 2.579507080946133e-06,
      "loss": 1.003,
      "step": 5880
    },
    {
      "epoch": 0.4242992677031853,
      "grad_norm": 2.59819532264827,
      "learning_rate": 2.5790597544476793e-06,
      "loss": 1.0316,
      "step": 5881
    },
    {
      "epoch": 0.42437141517261284,
      "grad_norm": 3.369944848534695,
      "learning_rate": 2.5786123963289154e-06,
      "loss": 1.0076,
      "step": 5882
    },
    {
      "epoch": 0.4244435626420403,
      "grad_norm": 2.572769150514122,
      "learning_rate": 2.578165006614271e-06,
      "loss": 0.9992,
      "step": 5883
    },
    {
      "epoch": 0.4245157101114678,
      "grad_norm": 2.6467506640707184,
      "learning_rate": 2.577717585328176e-06,
      "loss": 0.9111,
      "step": 5884
    },
    {
      "epoch": 0.42458785758089534,
      "grad_norm": 2.7109834958442645,
      "learning_rate": 2.577270132495062e-06,
      "loss": 0.9666,
      "step": 5885
    },
    {
      "epoch": 0.42466000505032286,
      "grad_norm": 2.620498885348279,
      "learning_rate": 2.5768226481393632e-06,
      "loss": 0.9603,
      "step": 5886
    },
    {
      "epoch": 0.4247321525197504,
      "grad_norm": 3.8553289911820787,
      "learning_rate": 2.5763751322855145e-06,
      "loss": 0.9596,
      "step": 5887
    },
    {
      "epoch": 0.4248042999891779,
      "grad_norm": 2.6101498852366465,
      "learning_rate": 2.575927584957954e-06,
      "loss": 0.9765,
      "step": 5888
    },
    {
      "epoch": 0.42487644745860537,
      "grad_norm": 4.940978551154493,
      "learning_rate": 2.57548000618112e-06,
      "loss": 0.9954,
      "step": 5889
    },
    {
      "epoch": 0.4249485949280329,
      "grad_norm": 2.2622506104324196,
      "learning_rate": 2.575032395979453e-06,
      "loss": 0.982,
      "step": 5890
    },
    {
      "epoch": 0.4250207423974604,
      "grad_norm": 3.440587962936086,
      "learning_rate": 2.5745847543773962e-06,
      "loss": 0.9247,
      "step": 5891
    },
    {
      "epoch": 0.4250928898668879,
      "grad_norm": 2.93297748213408,
      "learning_rate": 2.574137081399392e-06,
      "loss": 1.0578,
      "step": 5892
    },
    {
      "epoch": 0.42516503733631544,
      "grad_norm": 3.4662762318116207,
      "learning_rate": 2.573689377069888e-06,
      "loss": 0.9975,
      "step": 5893
    },
    {
      "epoch": 0.42523718480574296,
      "grad_norm": 3.0285963260521793,
      "learning_rate": 2.573241641413331e-06,
      "loss": 0.8489,
      "step": 5894
    },
    {
      "epoch": 0.4253093322751704,
      "grad_norm": 0.6770549963865705,
      "learning_rate": 2.5727938744541682e-06,
      "loss": 0.806,
      "step": 5895
    },
    {
      "epoch": 0.42538147974459795,
      "grad_norm": 4.685122891730815,
      "learning_rate": 2.5723460762168546e-06,
      "loss": 0.8323,
      "step": 5896
    },
    {
      "epoch": 0.42545362721402546,
      "grad_norm": 6.740145235845661,
      "learning_rate": 2.5718982467258387e-06,
      "loss": 1.0324,
      "step": 5897
    },
    {
      "epoch": 0.425525774683453,
      "grad_norm": 2.695727330226153,
      "learning_rate": 2.5714503860055782e-06,
      "loss": 0.847,
      "step": 5898
    },
    {
      "epoch": 0.4255979221528805,
      "grad_norm": 4.624904734687153,
      "learning_rate": 2.5710024940805265e-06,
      "loss": 0.943,
      "step": 5899
    },
    {
      "epoch": 0.425670069622308,
      "grad_norm": 3.1111989254895427,
      "learning_rate": 2.5705545709751426e-06,
      "loss": 1.0854,
      "step": 5900
    },
    {
      "epoch": 0.4257422170917355,
      "grad_norm": 2.4014983387642905,
      "learning_rate": 2.570106616713886e-06,
      "loss": 0.9526,
      "step": 5901
    },
    {
      "epoch": 0.425814364561163,
      "grad_norm": 5.2340969638144195,
      "learning_rate": 2.5696586313212175e-06,
      "loss": 1.0295,
      "step": 5902
    },
    {
      "epoch": 0.4258865120305905,
      "grad_norm": 4.106550631266855,
      "learning_rate": 2.5692106148215996e-06,
      "loss": 0.9687,
      "step": 5903
    },
    {
      "epoch": 0.42595865950001804,
      "grad_norm": 0.7695716034043996,
      "learning_rate": 2.5687625672394974e-06,
      "loss": 0.8739,
      "step": 5904
    },
    {
      "epoch": 0.42603080696944556,
      "grad_norm": 3.699241178901778,
      "learning_rate": 2.5683144885993767e-06,
      "loss": 0.9859,
      "step": 5905
    },
    {
      "epoch": 0.4261029544388731,
      "grad_norm": 2.8853799608744293,
      "learning_rate": 2.567866378925706e-06,
      "loss": 0.9743,
      "step": 5906
    },
    {
      "epoch": 0.42617510190830055,
      "grad_norm": 3.981280752209957,
      "learning_rate": 2.567418238242954e-06,
      "loss": 1.0122,
      "step": 5907
    },
    {
      "epoch": 0.42624724937772807,
      "grad_norm": 2.4089314766239873,
      "learning_rate": 2.5669700665755934e-06,
      "loss": 1.1018,
      "step": 5908
    },
    {
      "epoch": 0.4263193968471556,
      "grad_norm": 2.832675707923476,
      "learning_rate": 2.566521863948095e-06,
      "loss": 0.8529,
      "step": 5909
    },
    {
      "epoch": 0.4263915443165831,
      "grad_norm": 4.313087091703588,
      "learning_rate": 2.566073630384936e-06,
      "loss": 1.0632,
      "step": 5910
    },
    {
      "epoch": 0.4264636917860106,
      "grad_norm": 3.207201043674359,
      "learning_rate": 2.5656253659105914e-06,
      "loss": 0.9667,
      "step": 5911
    },
    {
      "epoch": 0.42653583925543814,
      "grad_norm": 3.0058181293872597,
      "learning_rate": 2.565177070549539e-06,
      "loss": 0.9404,
      "step": 5912
    },
    {
      "epoch": 0.4266079867248656,
      "grad_norm": 3.3155395669384014,
      "learning_rate": 2.5647287443262587e-06,
      "loss": 1.0965,
      "step": 5913
    },
    {
      "epoch": 0.4266801341942931,
      "grad_norm": 3.4103148118742452,
      "learning_rate": 2.564280387265233e-06,
      "loss": 0.9497,
      "step": 5914
    },
    {
      "epoch": 0.42675228166372065,
      "grad_norm": 3.213223945000961,
      "learning_rate": 2.563831999390944e-06,
      "loss": 1.0542,
      "step": 5915
    },
    {
      "epoch": 0.42682442913314816,
      "grad_norm": 2.7967498580811188,
      "learning_rate": 2.5633835807278766e-06,
      "loss": 0.9809,
      "step": 5916
    },
    {
      "epoch": 0.4268965766025757,
      "grad_norm": 4.061257744638204,
      "learning_rate": 2.5629351313005172e-06,
      "loss": 1.0596,
      "step": 5917
    },
    {
      "epoch": 0.42696872407200315,
      "grad_norm": 2.7353827663005097,
      "learning_rate": 2.5624866511333547e-06,
      "loss": 0.9724,
      "step": 5918
    },
    {
      "epoch": 0.42704087154143067,
      "grad_norm": 3.723603743452952,
      "learning_rate": 2.562038140250878e-06,
      "loss": 0.9213,
      "step": 5919
    },
    {
      "epoch": 0.4271130190108582,
      "grad_norm": 4.4232296435677325,
      "learning_rate": 2.5615895986775797e-06,
      "loss": 0.9678,
      "step": 5920
    },
    {
      "epoch": 0.4271851664802857,
      "grad_norm": 4.068410007786221,
      "learning_rate": 2.5611410264379515e-06,
      "loss": 0.9086,
      "step": 5921
    },
    {
      "epoch": 0.4272573139497132,
      "grad_norm": 2.8221688118667423,
      "learning_rate": 2.5606924235564897e-06,
      "loss": 1.0444,
      "step": 5922
    },
    {
      "epoch": 0.42732946141914074,
      "grad_norm": 2.40714336506837,
      "learning_rate": 2.5602437900576905e-06,
      "loss": 0.9242,
      "step": 5923
    },
    {
      "epoch": 0.4274016088885682,
      "grad_norm": 2.820590525948348,
      "learning_rate": 2.5597951259660513e-06,
      "loss": 0.9531,
      "step": 5924
    },
    {
      "epoch": 0.4274737563579957,
      "grad_norm": 0.7675989745125771,
      "learning_rate": 2.5593464313060726e-06,
      "loss": 0.8762,
      "step": 5925
    },
    {
      "epoch": 0.42754590382742325,
      "grad_norm": 3.542583945174194,
      "learning_rate": 2.558897706102256e-06,
      "loss": 1.0762,
      "step": 5926
    },
    {
      "epoch": 0.42761805129685077,
      "grad_norm": 3.1653561808407775,
      "learning_rate": 2.558448950379105e-06,
      "loss": 1.0141,
      "step": 5927
    },
    {
      "epoch": 0.4276901987662783,
      "grad_norm": 3.6216583634790585,
      "learning_rate": 2.5580001641611235e-06,
      "loss": 1.0784,
      "step": 5928
    },
    {
      "epoch": 0.4277623462357058,
      "grad_norm": 2.6050396384300467,
      "learning_rate": 2.557551347472819e-06,
      "loss": 1.0486,
      "step": 5929
    },
    {
      "epoch": 0.42783449370513327,
      "grad_norm": 2.432562791850933,
      "learning_rate": 2.5571025003386995e-06,
      "loss": 0.8013,
      "step": 5930
    },
    {
      "epoch": 0.4279066411745608,
      "grad_norm": 6.552911638343783,
      "learning_rate": 2.5566536227832744e-06,
      "loss": 0.9373,
      "step": 5931
    },
    {
      "epoch": 0.4279787886439883,
      "grad_norm": 2.8677733443863636,
      "learning_rate": 2.5562047148310564e-06,
      "loss": 0.9116,
      "step": 5932
    },
    {
      "epoch": 0.4280509361134158,
      "grad_norm": 2.6589730161780394,
      "learning_rate": 2.5557557765065577e-06,
      "loss": 0.8946,
      "step": 5933
    },
    {
      "epoch": 0.42812308358284334,
      "grad_norm": 5.35742952942171,
      "learning_rate": 2.5553068078342924e-06,
      "loss": 0.8839,
      "step": 5934
    },
    {
      "epoch": 0.42819523105227086,
      "grad_norm": 3.22866538474119,
      "learning_rate": 2.554857808838779e-06,
      "loss": 0.8758,
      "step": 5935
    },
    {
      "epoch": 0.4282673785216983,
      "grad_norm": 3.072765661290184,
      "learning_rate": 2.554408779544534e-06,
      "loss": 1.0137,
      "step": 5936
    },
    {
      "epoch": 0.42833952599112585,
      "grad_norm": 10.522556236879932,
      "learning_rate": 2.5539597199760785e-06,
      "loss": 1.016,
      "step": 5937
    },
    {
      "epoch": 0.42841167346055337,
      "grad_norm": 3.1598994241474743,
      "learning_rate": 2.5535106301579335e-06,
      "loss": 1.0256,
      "step": 5938
    },
    {
      "epoch": 0.4284838209299809,
      "grad_norm": 4.087885959146884,
      "learning_rate": 2.5530615101146214e-06,
      "loss": 0.9007,
      "step": 5939
    },
    {
      "epoch": 0.4285559683994084,
      "grad_norm": 2.6912544107998975,
      "learning_rate": 2.552612359870668e-06,
      "loss": 0.9971,
      "step": 5940
    },
    {
      "epoch": 0.4286281158688359,
      "grad_norm": 3.239798012312418,
      "learning_rate": 2.552163179450599e-06,
      "loss": 0.9557,
      "step": 5941
    },
    {
      "epoch": 0.4287002633382634,
      "grad_norm": 4.026935960046839,
      "learning_rate": 2.5517139688789436e-06,
      "loss": 0.9253,
      "step": 5942
    },
    {
      "epoch": 0.4287724108076909,
      "grad_norm": 3.2026147947798336,
      "learning_rate": 2.5512647281802295e-06,
      "loss": 1.0544,
      "step": 5943
    },
    {
      "epoch": 0.4288445582771184,
      "grad_norm": 2.961341850431109,
      "learning_rate": 2.5508154573789903e-06,
      "loss": 0.9716,
      "step": 5944
    },
    {
      "epoch": 0.42891670574654595,
      "grad_norm": 2.782569143802012,
      "learning_rate": 2.5503661564997577e-06,
      "loss": 1.0154,
      "step": 5945
    },
    {
      "epoch": 0.42898885321597346,
      "grad_norm": 2.0443832722901485,
      "learning_rate": 2.5499168255670664e-06,
      "loss": 1.0122,
      "step": 5946
    },
    {
      "epoch": 0.429061000685401,
      "grad_norm": 3.699606853939789,
      "learning_rate": 2.5494674646054537e-06,
      "loss": 1.0665,
      "step": 5947
    },
    {
      "epoch": 0.42913314815482845,
      "grad_norm": 2.3693237737070296,
      "learning_rate": 2.5490180736394555e-06,
      "loss": 1.0779,
      "step": 5948
    },
    {
      "epoch": 0.42920529562425597,
      "grad_norm": 2.745262834233221,
      "learning_rate": 2.548568652693614e-06,
      "loss": 1.1198,
      "step": 5949
    },
    {
      "epoch": 0.4292774430936835,
      "grad_norm": 0.7822280102739021,
      "learning_rate": 2.5481192017924687e-06,
      "loss": 0.8799,
      "step": 5950
    },
    {
      "epoch": 0.429349590563111,
      "grad_norm": 2.3317589784590234,
      "learning_rate": 2.547669720960563e-06,
      "loss": 0.8585,
      "step": 5951
    },
    {
      "epoch": 0.4294217380325385,
      "grad_norm": 2.5078654536185487,
      "learning_rate": 2.5472202102224408e-06,
      "loss": 1.034,
      "step": 5952
    },
    {
      "epoch": 0.42949388550196604,
      "grad_norm": 3.116613196527997,
      "learning_rate": 2.546770669602649e-06,
      "loss": 1.0872,
      "step": 5953
    },
    {
      "epoch": 0.4295660329713935,
      "grad_norm": 2.7053127366108067,
      "learning_rate": 2.546321099125735e-06,
      "loss": 1.0452,
      "step": 5954
    },
    {
      "epoch": 0.429638180440821,
      "grad_norm": 3.0749593034237708,
      "learning_rate": 2.5458714988162477e-06,
      "loss": 1.0811,
      "step": 5955
    },
    {
      "epoch": 0.42971032791024855,
      "grad_norm": 3.939299641438699,
      "learning_rate": 2.545421868698739e-06,
      "loss": 0.9926,
      "step": 5956
    },
    {
      "epoch": 0.42978247537967607,
      "grad_norm": 2.8816335691717536,
      "learning_rate": 2.5449722087977616e-06,
      "loss": 1.0125,
      "step": 5957
    },
    {
      "epoch": 0.4298546228491036,
      "grad_norm": 3.0361566846319468,
      "learning_rate": 2.544522519137868e-06,
      "loss": 0.963,
      "step": 5958
    },
    {
      "epoch": 0.4299267703185311,
      "grad_norm": 2.9947748775287995,
      "learning_rate": 2.544072799743617e-06,
      "loss": 1.0869,
      "step": 5959
    },
    {
      "epoch": 0.42999891778795857,
      "grad_norm": 2.9817075941995266,
      "learning_rate": 2.5436230506395637e-06,
      "loss": 0.8381,
      "step": 5960
    },
    {
      "epoch": 0.4300710652573861,
      "grad_norm": 2.9086303806995346,
      "learning_rate": 2.543173271850268e-06,
      "loss": 1.0921,
      "step": 5961
    },
    {
      "epoch": 0.4301432127268136,
      "grad_norm": 2.7529043986067974,
      "learning_rate": 2.542723463400291e-06,
      "loss": 1.0012,
      "step": 5962
    },
    {
      "epoch": 0.4302153601962411,
      "grad_norm": 2.643597166004067,
      "learning_rate": 2.542273625314194e-06,
      "loss": 0.9459,
      "step": 5963
    },
    {
      "epoch": 0.43028750766566864,
      "grad_norm": 6.88149939144891,
      "learning_rate": 2.541823757616542e-06,
      "loss": 0.9638,
      "step": 5964
    },
    {
      "epoch": 0.43035965513509616,
      "grad_norm": 3.6182468255547753,
      "learning_rate": 2.541373860331901e-06,
      "loss": 0.8889,
      "step": 5965
    },
    {
      "epoch": 0.43043180260452363,
      "grad_norm": 7.225153473276698,
      "learning_rate": 2.540923933484837e-06,
      "loss": 1.044,
      "step": 5966
    },
    {
      "epoch": 0.43050395007395115,
      "grad_norm": 2.40164328186736,
      "learning_rate": 2.5404739770999194e-06,
      "loss": 0.9248,
      "step": 5967
    },
    {
      "epoch": 0.43057609754337867,
      "grad_norm": 5.081636406164172,
      "learning_rate": 2.540023991201719e-06,
      "loss": 0.9396,
      "step": 5968
    },
    {
      "epoch": 0.4306482450128062,
      "grad_norm": 2.732874169588977,
      "learning_rate": 2.539573975814808e-06,
      "loss": 0.984,
      "step": 5969
    },
    {
      "epoch": 0.4307203924822337,
      "grad_norm": 2.4350375917889426,
      "learning_rate": 2.5391239309637585e-06,
      "loss": 0.9701,
      "step": 5970
    },
    {
      "epoch": 0.43079253995166117,
      "grad_norm": 2.6271154871119182,
      "learning_rate": 2.5386738566731476e-06,
      "loss": 0.9943,
      "step": 5971
    },
    {
      "epoch": 0.4308646874210887,
      "grad_norm": 0.5894932242190938,
      "learning_rate": 2.5382237529675516e-06,
      "loss": 0.7745,
      "step": 5972
    },
    {
      "epoch": 0.4309368348905162,
      "grad_norm": 2.329312117873937,
      "learning_rate": 2.5377736198715484e-06,
      "loss": 0.8608,
      "step": 5973
    },
    {
      "epoch": 0.4310089823599437,
      "grad_norm": 3.301570351081753,
      "learning_rate": 2.537323457409719e-06,
      "loss": 0.9776,
      "step": 5974
    },
    {
      "epoch": 0.43108112982937125,
      "grad_norm": 2.4943533069620254,
      "learning_rate": 2.536873265606644e-06,
      "loss": 1.0223,
      "step": 5975
    },
    {
      "epoch": 0.43115327729879876,
      "grad_norm": 0.8023059379188633,
      "learning_rate": 2.5364230444869077e-06,
      "loss": 0.8287,
      "step": 5976
    },
    {
      "epoch": 0.43122542476822623,
      "grad_norm": 2.077641251555719,
      "learning_rate": 2.5359727940750955e-06,
      "loss": 1.022,
      "step": 5977
    },
    {
      "epoch": 0.43129757223765375,
      "grad_norm": 2.613467906905927,
      "learning_rate": 2.5355225143957925e-06,
      "loss": 0.9881,
      "step": 5978
    },
    {
      "epoch": 0.43136971970708127,
      "grad_norm": 2.7040920402053357,
      "learning_rate": 2.535072205473587e-06,
      "loss": 1.0412,
      "step": 5979
    },
    {
      "epoch": 0.4314418671765088,
      "grad_norm": 2.4006085339582106,
      "learning_rate": 2.5346218673330695e-06,
      "loss": 0.9161,
      "step": 5980
    },
    {
      "epoch": 0.4315140146459363,
      "grad_norm": 2.667822815686457,
      "learning_rate": 2.5341714999988313e-06,
      "loss": 0.9354,
      "step": 5981
    },
    {
      "epoch": 0.4315861621153638,
      "grad_norm": 0.7060974867895544,
      "learning_rate": 2.5337211034954643e-06,
      "loss": 0.8604,
      "step": 5982
    },
    {
      "epoch": 0.4316583095847913,
      "grad_norm": 2.848626451612713,
      "learning_rate": 2.5332706778475642e-06,
      "loss": 0.9855,
      "step": 5983
    },
    {
      "epoch": 0.4317304570542188,
      "grad_norm": 3.6362994123336017,
      "learning_rate": 2.532820223079726e-06,
      "loss": 0.9528,
      "step": 5984
    },
    {
      "epoch": 0.4318026045236463,
      "grad_norm": 3.4696207414203695,
      "learning_rate": 2.532369739216548e-06,
      "loss": 0.9625,
      "step": 5985
    },
    {
      "epoch": 0.43187475199307385,
      "grad_norm": 2.096375029718746,
      "learning_rate": 2.53191922628263e-06,
      "loss": 1.0074,
      "step": 5986
    },
    {
      "epoch": 0.43194689946250137,
      "grad_norm": 2.9223012358310108,
      "learning_rate": 2.5314686843025713e-06,
      "loss": 0.982,
      "step": 5987
    },
    {
      "epoch": 0.4320190469319289,
      "grad_norm": 2.363696555224863,
      "learning_rate": 2.5310181133009753e-06,
      "loss": 0.9378,
      "step": 5988
    },
    {
      "epoch": 0.43209119440135635,
      "grad_norm": 3.1225602353518944,
      "learning_rate": 2.5305675133024464e-06,
      "loss": 1.0286,
      "step": 5989
    },
    {
      "epoch": 0.43216334187078387,
      "grad_norm": 2.118492596949845,
      "learning_rate": 2.530116884331589e-06,
      "loss": 1.0338,
      "step": 5990
    },
    {
      "epoch": 0.4322354893402114,
      "grad_norm": 7.674125139406669,
      "learning_rate": 2.529666226413011e-06,
      "loss": 1.0667,
      "step": 5991
    },
    {
      "epoch": 0.4323076368096389,
      "grad_norm": 2.7777447846890446,
      "learning_rate": 2.5292155395713216e-06,
      "loss": 1.0383,
      "step": 5992
    },
    {
      "epoch": 0.4323797842790664,
      "grad_norm": 2.6290969666090227,
      "learning_rate": 2.5287648238311306e-06,
      "loss": 1.0222,
      "step": 5993
    },
    {
      "epoch": 0.43245193174849395,
      "grad_norm": 10.67857657558647,
      "learning_rate": 2.5283140792170496e-06,
      "loss": 0.9609,
      "step": 5994
    },
    {
      "epoch": 0.4325240792179214,
      "grad_norm": 2.480001474656928,
      "learning_rate": 2.5278633057536926e-06,
      "loss": 1.0281,
      "step": 5995
    },
    {
      "epoch": 0.43259622668734893,
      "grad_norm": 2.1867101060615375,
      "learning_rate": 2.5274125034656753e-06,
      "loss": 1.0337,
      "step": 5996
    },
    {
      "epoch": 0.43266837415677645,
      "grad_norm": 4.436223450917427,
      "learning_rate": 2.526961672377612e-06,
      "loss": 1.022,
      "step": 5997
    },
    {
      "epoch": 0.43274052162620397,
      "grad_norm": 2.340369673326141,
      "learning_rate": 2.5265108125141237e-06,
      "loss": 1.0184,
      "step": 5998
    },
    {
      "epoch": 0.4328126690956315,
      "grad_norm": 2.0544485671214816,
      "learning_rate": 2.5260599238998296e-06,
      "loss": 1.0146,
      "step": 5999
    },
    {
      "epoch": 0.432884816565059,
      "grad_norm": 3.582456540618111,
      "learning_rate": 2.5256090065593487e-06,
      "loss": 0.9624,
      "step": 6000
    },
    {
      "epoch": 0.43295696403448647,
      "grad_norm": 2.422049534569121,
      "learning_rate": 2.5251580605173075e-06,
      "loss": 1.0041,
      "step": 6001
    },
    {
      "epoch": 0.433029111503914,
      "grad_norm": 3.7509947411122244,
      "learning_rate": 2.5247070857983276e-06,
      "loss": 0.8821,
      "step": 6002
    },
    {
      "epoch": 0.4331012589733415,
      "grad_norm": 3.988421969121619,
      "learning_rate": 2.524256082427036e-06,
      "loss": 1.0273,
      "step": 6003
    },
    {
      "epoch": 0.433173406442769,
      "grad_norm": 2.8360237549834744,
      "learning_rate": 2.5238050504280614e-06,
      "loss": 0.9598,
      "step": 6004
    },
    {
      "epoch": 0.43324555391219655,
      "grad_norm": 2.9632049172564545,
      "learning_rate": 2.5233539898260313e-06,
      "loss": 1.0127,
      "step": 6005
    },
    {
      "epoch": 0.43331770138162407,
      "grad_norm": 2.9089461095492504,
      "learning_rate": 2.522902900645577e-06,
      "loss": 1.0138,
      "step": 6006
    },
    {
      "epoch": 0.43338984885105153,
      "grad_norm": 2.1144133125192073,
      "learning_rate": 2.5224517829113314e-06,
      "loss": 1.024,
      "step": 6007
    },
    {
      "epoch": 0.43346199632047905,
      "grad_norm": 3.4345507367671444,
      "learning_rate": 2.522000636647929e-06,
      "loss": 0.9994,
      "step": 6008
    },
    {
      "epoch": 0.43353414378990657,
      "grad_norm": 3.024868412150789,
      "learning_rate": 2.5215494618800025e-06,
      "loss": 0.9755,
      "step": 6009
    },
    {
      "epoch": 0.4336062912593341,
      "grad_norm": 3.8110530171046073,
      "learning_rate": 2.5210982586321912e-06,
      "loss": 1.0659,
      "step": 6010
    },
    {
      "epoch": 0.4336784387287616,
      "grad_norm": 3.4842452636651524,
      "learning_rate": 2.520647026929134e-06,
      "loss": 0.9268,
      "step": 6011
    },
    {
      "epoch": 0.4337505861981891,
      "grad_norm": 3.069838004295932,
      "learning_rate": 2.5201957667954687e-06,
      "loss": 0.9632,
      "step": 6012
    },
    {
      "epoch": 0.4338227336676166,
      "grad_norm": 3.0617553428415776,
      "learning_rate": 2.5197444782558387e-06,
      "loss": 1.0123,
      "step": 6013
    },
    {
      "epoch": 0.4338948811370441,
      "grad_norm": 3.249397955664985,
      "learning_rate": 2.519293161334887e-06,
      "loss": 0.9963,
      "step": 6014
    },
    {
      "epoch": 0.4339670286064716,
      "grad_norm": 1.9672251048595575,
      "learning_rate": 2.5188418160572577e-06,
      "loss": 1.018,
      "step": 6015
    },
    {
      "epoch": 0.43403917607589915,
      "grad_norm": 2.7622532522897236,
      "learning_rate": 2.518390442447598e-06,
      "loss": 1.0174,
      "step": 6016
    },
    {
      "epoch": 0.43411132354532667,
      "grad_norm": 2.1157779294613257,
      "learning_rate": 2.517939040530555e-06,
      "loss": 1.0102,
      "step": 6017
    },
    {
      "epoch": 0.43418347101475413,
      "grad_norm": 3.485584545177757,
      "learning_rate": 2.5174876103307784e-06,
      "loss": 0.964,
      "step": 6018
    },
    {
      "epoch": 0.43425561848418165,
      "grad_norm": 3.2002450789538996,
      "learning_rate": 2.5170361518729184e-06,
      "loss": 0.9822,
      "step": 6019
    },
    {
      "epoch": 0.43432776595360917,
      "grad_norm": 2.3406939100401973,
      "learning_rate": 2.5165846651816296e-06,
      "loss": 1.0238,
      "step": 6020
    },
    {
      "epoch": 0.4343999134230367,
      "grad_norm": 3.365645690684355,
      "learning_rate": 2.5161331502815636e-06,
      "loss": 0.9722,
      "step": 6021
    },
    {
      "epoch": 0.4344720608924642,
      "grad_norm": 3.167567442888789,
      "learning_rate": 2.515681607197377e-06,
      "loss": 0.9273,
      "step": 6022
    },
    {
      "epoch": 0.4345442083618917,
      "grad_norm": 2.79566452001845,
      "learning_rate": 2.515230035953727e-06,
      "loss": 0.9209,
      "step": 6023
    },
    {
      "epoch": 0.4346163558313192,
      "grad_norm": 3.3650276355260447,
      "learning_rate": 2.5147784365752715e-06,
      "loss": 0.9515,
      "step": 6024
    },
    {
      "epoch": 0.4346885033007467,
      "grad_norm": 4.038935233201093,
      "learning_rate": 2.5143268090866718e-06,
      "loss": 0.9561,
      "step": 6025
    },
    {
      "epoch": 0.43476065077017423,
      "grad_norm": 2.4433205385727486,
      "learning_rate": 2.5138751535125898e-06,
      "loss": 0.9731,
      "step": 6026
    },
    {
      "epoch": 0.43483279823960175,
      "grad_norm": 2.5053841786255706,
      "learning_rate": 2.5134234698776865e-06,
      "loss": 0.9304,
      "step": 6027
    },
    {
      "epoch": 0.43490494570902927,
      "grad_norm": 0.7077919677046863,
      "learning_rate": 2.5129717582066288e-06,
      "loss": 0.8958,
      "step": 6028
    },
    {
      "epoch": 0.4349770931784568,
      "grad_norm": 2.366749033591562,
      "learning_rate": 2.5125200185240822e-06,
      "loss": 0.974,
      "step": 6029
    },
    {
      "epoch": 0.43504924064788425,
      "grad_norm": 2.93003773948138,
      "learning_rate": 2.5120682508547146e-06,
      "loss": 1.082,
      "step": 6030
    },
    {
      "epoch": 0.43512138811731177,
      "grad_norm": 2.669707997817293,
      "learning_rate": 2.511616455223195e-06,
      "loss": 0.8893,
      "step": 6031
    },
    {
      "epoch": 0.4351935355867393,
      "grad_norm": 2.3664613119767592,
      "learning_rate": 2.511164631654196e-06,
      "loss": 0.9559,
      "step": 6032
    },
    {
      "epoch": 0.4352656830561668,
      "grad_norm": 2.128351261697694,
      "learning_rate": 2.510712780172388e-06,
      "loss": 0.8601,
      "step": 6033
    },
    {
      "epoch": 0.4353378305255943,
      "grad_norm": 4.061833018790184,
      "learning_rate": 2.5102609008024455e-06,
      "loss": 0.8396,
      "step": 6034
    },
    {
      "epoch": 0.43540997799502185,
      "grad_norm": 2.456559613542422,
      "learning_rate": 2.5098089935690444e-06,
      "loss": 1.037,
      "step": 6035
    },
    {
      "epoch": 0.4354821254644493,
      "grad_norm": 2.656829860450072,
      "learning_rate": 2.509357058496861e-06,
      "loss": 0.9901,
      "step": 6036
    },
    {
      "epoch": 0.43555427293387683,
      "grad_norm": 3.6415064976529603,
      "learning_rate": 2.508905095610575e-06,
      "loss": 0.9865,
      "step": 6037
    },
    {
      "epoch": 0.43562642040330435,
      "grad_norm": 2.949117373309373,
      "learning_rate": 2.5084531049348655e-06,
      "loss": 0.9599,
      "step": 6038
    },
    {
      "epoch": 0.43569856787273187,
      "grad_norm": 5.118282288363355,
      "learning_rate": 2.508001086494413e-06,
      "loss": 0.9219,
      "step": 6039
    },
    {
      "epoch": 0.4357707153421594,
      "grad_norm": 2.885373515737154,
      "learning_rate": 2.507549040313903e-06,
      "loss": 0.91,
      "step": 6040
    },
    {
      "epoch": 0.4358428628115869,
      "grad_norm": 2.340171625413222,
      "learning_rate": 2.5070969664180183e-06,
      "loss": 1.0531,
      "step": 6041
    },
    {
      "epoch": 0.43591501028101437,
      "grad_norm": 5.798131102328159,
      "learning_rate": 2.506644864831445e-06,
      "loss": 0.983,
      "step": 6042
    },
    {
      "epoch": 0.4359871577504419,
      "grad_norm": 4.5981635864679635,
      "learning_rate": 2.506192735578872e-06,
      "loss": 1.0293,
      "step": 6043
    },
    {
      "epoch": 0.4360593052198694,
      "grad_norm": 2.241685871008917,
      "learning_rate": 2.505740578684987e-06,
      "loss": 1.0281,
      "step": 6044
    },
    {
      "epoch": 0.43613145268929693,
      "grad_norm": 1.9254509137637024,
      "learning_rate": 2.5052883941744816e-06,
      "loss": 0.9841,
      "step": 6045
    },
    {
      "epoch": 0.43620360015872445,
      "grad_norm": 2.423421157004603,
      "learning_rate": 2.5048361820720472e-06,
      "loss": 1.0523,
      "step": 6046
    },
    {
      "epoch": 0.43627574762815197,
      "grad_norm": 2.449560892887263,
      "learning_rate": 2.504383942402378e-06,
      "loss": 0.9122,
      "step": 6047
    },
    {
      "epoch": 0.43634789509757943,
      "grad_norm": 2.0227799101888166,
      "learning_rate": 2.5039316751901687e-06,
      "loss": 0.9765,
      "step": 6048
    },
    {
      "epoch": 0.43642004256700695,
      "grad_norm": 3.258355184678534,
      "learning_rate": 2.5034793804601163e-06,
      "loss": 1.0668,
      "step": 6049
    },
    {
      "epoch": 0.43649219003643447,
      "grad_norm": 2.07895569224664,
      "learning_rate": 2.5030270582369186e-06,
      "loss": 1.0434,
      "step": 6050
    },
    {
      "epoch": 0.436564337505862,
      "grad_norm": 1.7962618901357623,
      "learning_rate": 2.502574708545275e-06,
      "loss": 0.9917,
      "step": 6051
    },
    {
      "epoch": 0.4366364849752895,
      "grad_norm": 3.512868385409976,
      "learning_rate": 2.5021223314098877e-06,
      "loss": 1.0225,
      "step": 6052
    },
    {
      "epoch": 0.436708632444717,
      "grad_norm": 3.851724679252236,
      "learning_rate": 2.5016699268554585e-06,
      "loss": 1.015,
      "step": 6053
    },
    {
      "epoch": 0.4367807799141445,
      "grad_norm": 2.2713598871600054,
      "learning_rate": 2.5012174949066916e-06,
      "loss": 1.0097,
      "step": 6054
    },
    {
      "epoch": 0.436852927383572,
      "grad_norm": 3.530959665971806,
      "learning_rate": 2.500765035588293e-06,
      "loss": 0.9218,
      "step": 6055
    },
    {
      "epoch": 0.43692507485299953,
      "grad_norm": 2.4609823555869914,
      "learning_rate": 2.500312548924969e-06,
      "loss": 0.878,
      "step": 6056
    },
    {
      "epoch": 0.43699722232242705,
      "grad_norm": 3.386377161172388,
      "learning_rate": 2.4998600349414293e-06,
      "loss": 0.9531,
      "step": 6057
    },
    {
      "epoch": 0.43706936979185457,
      "grad_norm": 2.218293478837269,
      "learning_rate": 2.499407493662383e-06,
      "loss": 0.901,
      "step": 6058
    },
    {
      "epoch": 0.4371415172612821,
      "grad_norm": 2.325232943278809,
      "learning_rate": 2.4989549251125427e-06,
      "loss": 0.9956,
      "step": 6059
    },
    {
      "epoch": 0.43721366473070955,
      "grad_norm": 3.1286611853935673,
      "learning_rate": 2.498502329316621e-06,
      "loss": 0.9968,
      "step": 6060
    },
    {
      "epoch": 0.43728581220013707,
      "grad_norm": 3.719368634738517,
      "learning_rate": 2.4980497062993323e-06,
      "loss": 1.0274,
      "step": 6061
    },
    {
      "epoch": 0.4373579596695646,
      "grad_norm": 4.176459718546277,
      "learning_rate": 2.4975970560853936e-06,
      "loss": 0.9553,
      "step": 6062
    },
    {
      "epoch": 0.4374301071389921,
      "grad_norm": 4.565333910258723,
      "learning_rate": 2.49714437869952e-06,
      "loss": 0.8778,
      "step": 6063
    },
    {
      "epoch": 0.4375022546084196,
      "grad_norm": 2.773390734640935,
      "learning_rate": 2.4966916741664338e-06,
      "loss": 0.9367,
      "step": 6064
    },
    {
      "epoch": 0.43757440207784715,
      "grad_norm": 4.298557243495816,
      "learning_rate": 2.4962389425108536e-06,
      "loss": 0.9613,
      "step": 6065
    },
    {
      "epoch": 0.4376465495472746,
      "grad_norm": 3.4735017139918747,
      "learning_rate": 2.495786183757501e-06,
      "loss": 0.8725,
      "step": 6066
    },
    {
      "epoch": 0.43771869701670213,
      "grad_norm": 2.2774581749568923,
      "learning_rate": 2.4953333979311014e-06,
      "loss": 0.9497,
      "step": 6067
    },
    {
      "epoch": 0.43779084448612965,
      "grad_norm": 2.576552125112953,
      "learning_rate": 2.4948805850563775e-06,
      "loss": 1.0317,
      "step": 6068
    },
    {
      "epoch": 0.43786299195555717,
      "grad_norm": 2.9222203014965378,
      "learning_rate": 2.4944277451580576e-06,
      "loss": 0.9726,
      "step": 6069
    },
    {
      "epoch": 0.4379351394249847,
      "grad_norm": 1.9864218060542915,
      "learning_rate": 2.4939748782608687e-06,
      "loss": 0.9093,
      "step": 6070
    },
    {
      "epoch": 0.43800728689441215,
      "grad_norm": 3.4686445615397545,
      "learning_rate": 2.4935219843895403e-06,
      "loss": 0.9115,
      "step": 6071
    },
    {
      "epoch": 0.43807943436383967,
      "grad_norm": 2.4518241591947585,
      "learning_rate": 2.4930690635688034e-06,
      "loss": 0.9725,
      "step": 6072
    },
    {
      "epoch": 0.4381515818332672,
      "grad_norm": 3.396832527800543,
      "learning_rate": 2.49261611582339e-06,
      "loss": 1.025,
      "step": 6073
    },
    {
      "epoch": 0.4382237293026947,
      "grad_norm": 2.440486643211673,
      "learning_rate": 2.492163141178034e-06,
      "loss": 1.0103,
      "step": 6074
    },
    {
      "epoch": 0.43829587677212223,
      "grad_norm": 2.871113347615938,
      "learning_rate": 2.4917101396574706e-06,
      "loss": 0.922,
      "step": 6075
    },
    {
      "epoch": 0.43836802424154975,
      "grad_norm": 2.155463033615336,
      "learning_rate": 2.4912571112864375e-06,
      "loss": 1.0963,
      "step": 6076
    },
    {
      "epoch": 0.4384401717109772,
      "grad_norm": 5.146684130966783,
      "learning_rate": 2.4908040560896722e-06,
      "loss": 0.9988,
      "step": 6077
    },
    {
      "epoch": 0.43851231918040473,
      "grad_norm": 3.4862832630166807,
      "learning_rate": 2.490350974091913e-06,
      "loss": 1.0166,
      "step": 6078
    },
    {
      "epoch": 0.43858446664983225,
      "grad_norm": 3.999780648892863,
      "learning_rate": 2.4898978653179036e-06,
      "loss": 0.9373,
      "step": 6079
    },
    {
      "epoch": 0.43865661411925977,
      "grad_norm": 5.025876036238919,
      "learning_rate": 2.4894447297923843e-06,
      "loss": 1.0422,
      "step": 6080
    },
    {
      "epoch": 0.4387287615886873,
      "grad_norm": 3.123661212247014,
      "learning_rate": 2.488991567540101e-06,
      "loss": 0.9787,
      "step": 6081
    },
    {
      "epoch": 0.4388009090581148,
      "grad_norm": 2.869796688308522,
      "learning_rate": 2.4885383785857983e-06,
      "loss": 0.9734,
      "step": 6082
    },
    {
      "epoch": 0.43887305652754227,
      "grad_norm": 2.8305466542485114,
      "learning_rate": 2.488085162954223e-06,
      "loss": 0.9529,
      "step": 6083
    },
    {
      "epoch": 0.4389452039969698,
      "grad_norm": 3.2761717236310623,
      "learning_rate": 2.4876319206701236e-06,
      "loss": 1.0706,
      "step": 6084
    },
    {
      "epoch": 0.4390173514663973,
      "grad_norm": 4.574932560658169,
      "learning_rate": 2.48717865175825e-06,
      "loss": 1.0399,
      "step": 6085
    },
    {
      "epoch": 0.43908949893582483,
      "grad_norm": 3.0122521072900814,
      "learning_rate": 2.4867253562433546e-06,
      "loss": 0.9275,
      "step": 6086
    },
    {
      "epoch": 0.43916164640525235,
      "grad_norm": 3.0302973557910384,
      "learning_rate": 2.4862720341501893e-06,
      "loss": 0.8974,
      "step": 6087
    },
    {
      "epoch": 0.43923379387467987,
      "grad_norm": 2.5253770301258207,
      "learning_rate": 2.485818685503507e-06,
      "loss": 0.9398,
      "step": 6088
    },
    {
      "epoch": 0.43930594134410733,
      "grad_norm": 3.2338816685776095,
      "learning_rate": 2.485365310328066e-06,
      "loss": 0.9177,
      "step": 6089
    },
    {
      "epoch": 0.43937808881353485,
      "grad_norm": 2.5341212624226084,
      "learning_rate": 2.484911908648621e-06,
      "loss": 1.0046,
      "step": 6090
    },
    {
      "epoch": 0.43945023628296237,
      "grad_norm": 3.0686418875550823,
      "learning_rate": 2.4844584804899326e-06,
      "loss": 0.9229,
      "step": 6091
    },
    {
      "epoch": 0.4395223837523899,
      "grad_norm": 5.702774162475081,
      "learning_rate": 2.48400502587676e-06,
      "loss": 0.929,
      "step": 6092
    },
    {
      "epoch": 0.4395945312218174,
      "grad_norm": 0.6975955983760191,
      "learning_rate": 2.4835515448338644e-06,
      "loss": 0.8828,
      "step": 6093
    },
    {
      "epoch": 0.4396666786912449,
      "grad_norm": 2.4639385030119474,
      "learning_rate": 2.4830980373860094e-06,
      "loss": 0.9236,
      "step": 6094
    },
    {
      "epoch": 0.4397388261606724,
      "grad_norm": 3.6609100900845823,
      "learning_rate": 2.4826445035579583e-06,
      "loss": 0.862,
      "step": 6095
    },
    {
      "epoch": 0.4398109736300999,
      "grad_norm": 2.0621299411668006,
      "learning_rate": 2.482190943374478e-06,
      "loss": 1.0165,
      "step": 6096
    },
    {
      "epoch": 0.43988312109952743,
      "grad_norm": 3.10504262627722,
      "learning_rate": 2.4817373568603346e-06,
      "loss": 0.9138,
      "step": 6097
    },
    {
      "epoch": 0.43995526856895495,
      "grad_norm": 2.6032559951483134,
      "learning_rate": 2.481283744040299e-06,
      "loss": 1.0401,
      "step": 6098
    },
    {
      "epoch": 0.44002741603838247,
      "grad_norm": 3.36104127235121,
      "learning_rate": 2.480830104939139e-06,
      "loss": 1.0685,
      "step": 6099
    },
    {
      "epoch": 0.44009956350781,
      "grad_norm": 4.571101611229567,
      "learning_rate": 2.4803764395816264e-06,
      "loss": 0.9864,
      "step": 6100
    },
    {
      "epoch": 0.44017171097723745,
      "grad_norm": 2.328254440088748,
      "learning_rate": 2.479922747992535e-06,
      "loss": 1.0782,
      "step": 6101
    },
    {
      "epoch": 0.44024385844666497,
      "grad_norm": 3.266270915861203,
      "learning_rate": 2.479469030196639e-06,
      "loss": 0.8753,
      "step": 6102
    },
    {
      "epoch": 0.4403160059160925,
      "grad_norm": 3.9386057436429738,
      "learning_rate": 2.4790152862187137e-06,
      "loss": 0.9865,
      "step": 6103
    },
    {
      "epoch": 0.44038815338552,
      "grad_norm": 3.92056799763405,
      "learning_rate": 2.4785615160835382e-06,
      "loss": 1.0129,
      "step": 6104
    },
    {
      "epoch": 0.44046030085494753,
      "grad_norm": 2.01912214315365,
      "learning_rate": 2.4781077198158886e-06,
      "loss": 1.0268,
      "step": 6105
    },
    {
      "epoch": 0.44053244832437505,
      "grad_norm": 0.694567270278168,
      "learning_rate": 2.4776538974405473e-06,
      "loss": 0.7581,
      "step": 6106
    },
    {
      "epoch": 0.4406045957938025,
      "grad_norm": 2.724216183966699,
      "learning_rate": 2.477200048982294e-06,
      "loss": 0.9872,
      "step": 6107
    },
    {
      "epoch": 0.44067674326323003,
      "grad_norm": 3.0187589341183467,
      "learning_rate": 2.476746174465913e-06,
      "loss": 1.017,
      "step": 6108
    },
    {
      "epoch": 0.44074889073265755,
      "grad_norm": 0.6986865996871868,
      "learning_rate": 2.4762922739161883e-06,
      "loss": 0.8318,
      "step": 6109
    },
    {
      "epoch": 0.44082103820208507,
      "grad_norm": 3.5336270054347274,
      "learning_rate": 2.4758383473579053e-06,
      "loss": 0.9498,
      "step": 6110
    },
    {
      "epoch": 0.4408931856715126,
      "grad_norm": 2.6194417326683803,
      "learning_rate": 2.475384394815852e-06,
      "loss": 0.9945,
      "step": 6111
    },
    {
      "epoch": 0.4409653331409401,
      "grad_norm": 3.0740253237374286,
      "learning_rate": 2.4749304163148168e-06,
      "loss": 1.0262,
      "step": 6112
    },
    {
      "epoch": 0.44103748061036757,
      "grad_norm": 2.4878106503143735,
      "learning_rate": 2.4744764118795894e-06,
      "loss": 1.0088,
      "step": 6113
    },
    {
      "epoch": 0.4411096280797951,
      "grad_norm": 2.8891932885463967,
      "learning_rate": 2.4740223815349616e-06,
      "loss": 0.9811,
      "step": 6114
    },
    {
      "epoch": 0.4411817755492226,
      "grad_norm": 2.7926602279881485,
      "learning_rate": 2.4735683253057265e-06,
      "loss": 0.922,
      "step": 6115
    },
    {
      "epoch": 0.44125392301865013,
      "grad_norm": 4.267914210164289,
      "learning_rate": 2.4731142432166784e-06,
      "loss": 0.8727,
      "step": 6116
    },
    {
      "epoch": 0.44132607048807765,
      "grad_norm": 2.909499126208027,
      "learning_rate": 2.4726601352926126e-06,
      "loss": 0.9214,
      "step": 6117
    },
    {
      "epoch": 0.44139821795750517,
      "grad_norm": 4.149679626599862,
      "learning_rate": 2.4722060015583267e-06,
      "loss": 1.0256,
      "step": 6118
    },
    {
      "epoch": 0.44147036542693263,
      "grad_norm": 2.877374580626215,
      "learning_rate": 2.471751842038619e-06,
      "loss": 0.9973,
      "step": 6119
    },
    {
      "epoch": 0.44154251289636015,
      "grad_norm": 4.190084087101645,
      "learning_rate": 2.4712976567582892e-06,
      "loss": 0.9661,
      "step": 6120
    },
    {
      "epoch": 0.44161466036578767,
      "grad_norm": 2.474558311122218,
      "learning_rate": 2.4708434457421403e-06,
      "loss": 0.948,
      "step": 6121
    },
    {
      "epoch": 0.4416868078352152,
      "grad_norm": 3.378469732235057,
      "learning_rate": 2.470389209014973e-06,
      "loss": 0.989,
      "step": 6122
    },
    {
      "epoch": 0.4417589553046427,
      "grad_norm": 3.76445894667399,
      "learning_rate": 2.469934946601592e-06,
      "loss": 0.854,
      "step": 6123
    },
    {
      "epoch": 0.4418311027740702,
      "grad_norm": 2.219002722732916,
      "learning_rate": 2.469480658526803e-06,
      "loss": 0.9667,
      "step": 6124
    },
    {
      "epoch": 0.4419032502434977,
      "grad_norm": 4.299774634199707,
      "learning_rate": 2.4690263448154143e-06,
      "loss": 0.9071,
      "step": 6125
    },
    {
      "epoch": 0.4419753977129252,
      "grad_norm": 2.665401715031007,
      "learning_rate": 2.468572005492233e-06,
      "loss": 0.914,
      "step": 6126
    },
    {
      "epoch": 0.44204754518235273,
      "grad_norm": 2.731716673065323,
      "learning_rate": 2.468117640582068e-06,
      "loss": 0.9965,
      "step": 6127
    },
    {
      "epoch": 0.44211969265178025,
      "grad_norm": 3.5379061462392087,
      "learning_rate": 2.4676632501097327e-06,
      "loss": 0.8683,
      "step": 6128
    },
    {
      "epoch": 0.44219184012120777,
      "grad_norm": 3.477963009032688,
      "learning_rate": 2.467208834100037e-06,
      "loss": 0.8988,
      "step": 6129
    },
    {
      "epoch": 0.44226398759063523,
      "grad_norm": 4.072083422932334,
      "learning_rate": 2.466754392577798e-06,
      "loss": 0.9875,
      "step": 6130
    },
    {
      "epoch": 0.44233613506006275,
      "grad_norm": 0.703740888641622,
      "learning_rate": 2.4662999255678298e-06,
      "loss": 0.8296,
      "step": 6131
    },
    {
      "epoch": 0.44240828252949027,
      "grad_norm": 2.2187404363721592,
      "learning_rate": 2.465845433094947e-06,
      "loss": 0.9889,
      "step": 6132
    },
    {
      "epoch": 0.4424804299989178,
      "grad_norm": 3.7096140218081852,
      "learning_rate": 2.4653909151839712e-06,
      "loss": 1.0999,
      "step": 6133
    },
    {
      "epoch": 0.4425525774683453,
      "grad_norm": 2.9534072337437784,
      "learning_rate": 2.4649363718597196e-06,
      "loss": 0.9231,
      "step": 6134
    },
    {
      "epoch": 0.44262472493777283,
      "grad_norm": 5.667528573104253,
      "learning_rate": 2.4644818031470136e-06,
      "loss": 1.0497,
      "step": 6135
    },
    {
      "epoch": 0.4426968724072003,
      "grad_norm": 2.414500471464622,
      "learning_rate": 2.4640272090706764e-06,
      "loss": 0.9609,
      "step": 6136
    },
    {
      "epoch": 0.4427690198766278,
      "grad_norm": 2.974427746783808,
      "learning_rate": 2.463572589655531e-06,
      "loss": 1.0208,
      "step": 6137
    },
    {
      "epoch": 0.44284116734605533,
      "grad_norm": 4.658141705569928,
      "learning_rate": 2.463117944926403e-06,
      "loss": 0.962,
      "step": 6138
    },
    {
      "epoch": 0.44291331481548285,
      "grad_norm": 2.5601277419346076,
      "learning_rate": 2.4626632749081177e-06,
      "loss": 0.8802,
      "step": 6139
    },
    {
      "epoch": 0.44298546228491037,
      "grad_norm": 2.926231196887758,
      "learning_rate": 2.462208579625504e-06,
      "loss": 1.0022,
      "step": 6140
    },
    {
      "epoch": 0.4430576097543379,
      "grad_norm": 3.2516336003402686,
      "learning_rate": 2.46175385910339e-06,
      "loss": 0.9637,
      "step": 6141
    },
    {
      "epoch": 0.44312975722376535,
      "grad_norm": 2.059385317601478,
      "learning_rate": 2.461299113366608e-06,
      "loss": 0.8754,
      "step": 6142
    },
    {
      "epoch": 0.44320190469319287,
      "grad_norm": 2.883497534102432,
      "learning_rate": 2.4608443424399895e-06,
      "loss": 0.9273,
      "step": 6143
    },
    {
      "epoch": 0.4432740521626204,
      "grad_norm": 0.9162830865150158,
      "learning_rate": 2.460389546348366e-06,
      "loss": 0.8919,
      "step": 6144
    },
    {
      "epoch": 0.4433461996320479,
      "grad_norm": 3.6429524983553434,
      "learning_rate": 2.4599347251165746e-06,
      "loss": 0.979,
      "step": 6145
    },
    {
      "epoch": 0.44341834710147543,
      "grad_norm": 2.465960508081383,
      "learning_rate": 2.4594798787694502e-06,
      "loss": 0.9729,
      "step": 6146
    },
    {
      "epoch": 0.44349049457090295,
      "grad_norm": 4.323601627853444,
      "learning_rate": 2.459025007331831e-06,
      "loss": 1.0928,
      "step": 6147
    },
    {
      "epoch": 0.4435626420403304,
      "grad_norm": 2.7702338829869535,
      "learning_rate": 2.4585701108285556e-06,
      "loss": 1.0162,
      "step": 6148
    },
    {
      "epoch": 0.44363478950975793,
      "grad_norm": 2.8903105693641176,
      "learning_rate": 2.4581151892844635e-06,
      "loss": 1.1403,
      "step": 6149
    },
    {
      "epoch": 0.44370693697918545,
      "grad_norm": 0.8012025422727728,
      "learning_rate": 2.4576602427243967e-06,
      "loss": 0.8042,
      "step": 6150
    },
    {
      "epoch": 0.44377908444861297,
      "grad_norm": 2.0381651325129075,
      "learning_rate": 2.457205271173198e-06,
      "loss": 0.9824,
      "step": 6151
    },
    {
      "epoch": 0.4438512319180405,
      "grad_norm": 0.8223467354173956,
      "learning_rate": 2.4567502746557125e-06,
      "loss": 0.8708,
      "step": 6152
    },
    {
      "epoch": 0.443923379387468,
      "grad_norm": 4.255080160065037,
      "learning_rate": 2.4562952531967854e-06,
      "loss": 0.9217,
      "step": 6153
    },
    {
      "epoch": 0.4439955268568955,
      "grad_norm": 2.372506891045647,
      "learning_rate": 2.4558402068212635e-06,
      "loss": 0.9423,
      "step": 6154
    },
    {
      "epoch": 0.444067674326323,
      "grad_norm": 2.0960125441430635,
      "learning_rate": 2.4553851355539953e-06,
      "loss": 0.8584,
      "step": 6155
    },
    {
      "epoch": 0.4441398217957505,
      "grad_norm": 2.395406491500239,
      "learning_rate": 2.45493003941983e-06,
      "loss": 0.9199,
      "step": 6156
    },
    {
      "epoch": 0.44421196926517803,
      "grad_norm": 2.9347622064443355,
      "learning_rate": 2.4544749184436206e-06,
      "loss": 0.9807,
      "step": 6157
    },
    {
      "epoch": 0.44428411673460555,
      "grad_norm": 0.7720994260139836,
      "learning_rate": 2.4540197726502175e-06,
      "loss": 0.8362,
      "step": 6158
    },
    {
      "epoch": 0.44435626420403307,
      "grad_norm": 2.649852273530206,
      "learning_rate": 2.4535646020644755e-06,
      "loss": 0.985,
      "step": 6159
    },
    {
      "epoch": 0.44442841167346053,
      "grad_norm": 2.4224228762228033,
      "learning_rate": 2.45310940671125e-06,
      "loss": 0.9258,
      "step": 6160
    },
    {
      "epoch": 0.44450055914288805,
      "grad_norm": 0.8441279589096851,
      "learning_rate": 2.452654186615397e-06,
      "loss": 0.8983,
      "step": 6161
    },
    {
      "epoch": 0.44457270661231557,
      "grad_norm": 4.208758241460133,
      "learning_rate": 2.452198941801774e-06,
      "loss": 1.0049,
      "step": 6162
    },
    {
      "epoch": 0.4446448540817431,
      "grad_norm": 3.538243348433911,
      "learning_rate": 2.4517436722952407e-06,
      "loss": 0.9094,
      "step": 6163
    },
    {
      "epoch": 0.4447170015511706,
      "grad_norm": 3.104242737795917,
      "learning_rate": 2.4512883781206587e-06,
      "loss": 0.9723,
      "step": 6164
    },
    {
      "epoch": 0.44478914902059813,
      "grad_norm": 0.9424965903425391,
      "learning_rate": 2.4508330593028885e-06,
      "loss": 0.8557,
      "step": 6165
    },
    {
      "epoch": 0.4448612964900256,
      "grad_norm": 4.114247265704201,
      "learning_rate": 2.4503777158667934e-06,
      "loss": 1.0488,
      "step": 6166
    },
    {
      "epoch": 0.4449334439594531,
      "grad_norm": 7.993425767420848,
      "learning_rate": 2.4499223478372394e-06,
      "loss": 0.9598,
      "step": 6167
    },
    {
      "epoch": 0.44500559142888063,
      "grad_norm": 0.6295219156345356,
      "learning_rate": 2.4494669552390904e-06,
      "loss": 0.7752,
      "step": 6168
    },
    {
      "epoch": 0.44507773889830815,
      "grad_norm": 2.2747199252930876,
      "learning_rate": 2.4490115380972156e-06,
      "loss": 1.0164,
      "step": 6169
    },
    {
      "epoch": 0.44514988636773567,
      "grad_norm": 2.718340612698399,
      "learning_rate": 2.4485560964364834e-06,
      "loss": 0.9891,
      "step": 6170
    },
    {
      "epoch": 0.4452220338371632,
      "grad_norm": 3.9001259367734358,
      "learning_rate": 2.448100630281762e-06,
      "loss": 0.9825,
      "step": 6171
    },
    {
      "epoch": 0.44529418130659065,
      "grad_norm": 3.2241368816936067,
      "learning_rate": 2.447645139657925e-06,
      "loss": 0.9246,
      "step": 6172
    },
    {
      "epoch": 0.44536632877601817,
      "grad_norm": 3.3614352274028576,
      "learning_rate": 2.4471896245898438e-06,
      "loss": 0.8835,
      "step": 6173
    },
    {
      "epoch": 0.4454384762454457,
      "grad_norm": 3.2818727901337246,
      "learning_rate": 2.446734085102393e-06,
      "loss": 1.0453,
      "step": 6174
    },
    {
      "epoch": 0.4455106237148732,
      "grad_norm": 0.7043803981652006,
      "learning_rate": 2.4462785212204468e-06,
      "loss": 0.8248,
      "step": 6175
    },
    {
      "epoch": 0.44558277118430073,
      "grad_norm": 2.740552191816561,
      "learning_rate": 2.4458229329688826e-06,
      "loss": 0.8894,
      "step": 6176
    },
    {
      "epoch": 0.4456549186537282,
      "grad_norm": 3.861212536206193,
      "learning_rate": 2.4453673203725793e-06,
      "loss": 0.95,
      "step": 6177
    },
    {
      "epoch": 0.4457270661231557,
      "grad_norm": 2.422175431844654,
      "learning_rate": 2.4449116834564138e-06,
      "loss": 0.8584,
      "step": 6178
    },
    {
      "epoch": 0.44579921359258323,
      "grad_norm": 2.2567782557670846,
      "learning_rate": 2.44445602224527e-06,
      "loss": 0.9322,
      "step": 6179
    },
    {
      "epoch": 0.44587136106201075,
      "grad_norm": 2.993181267793128,
      "learning_rate": 2.4440003367640264e-06,
      "loss": 0.9841,
      "step": 6180
    },
    {
      "epoch": 0.44594350853143827,
      "grad_norm": 3.2780718716369956,
      "learning_rate": 2.4435446270375695e-06,
      "loss": 0.9694,
      "step": 6181
    },
    {
      "epoch": 0.4460156560008658,
      "grad_norm": 2.8655384113038362,
      "learning_rate": 2.4430888930907822e-06,
      "loss": 1.0596,
      "step": 6182
    },
    {
      "epoch": 0.44608780347029325,
      "grad_norm": 3.607381559186942,
      "learning_rate": 2.4426331349485497e-06,
      "loss": 0.9939,
      "step": 6183
    },
    {
      "epoch": 0.4461599509397208,
      "grad_norm": 3.5242839058815014,
      "learning_rate": 2.4421773526357614e-06,
      "loss": 0.978,
      "step": 6184
    },
    {
      "epoch": 0.4462320984091483,
      "grad_norm": 3.8142027257562625,
      "learning_rate": 2.4417215461773046e-06,
      "loss": 1.0494,
      "step": 6185
    },
    {
      "epoch": 0.4463042458785758,
      "grad_norm": 5.896377951693628,
      "learning_rate": 2.4412657155980693e-06,
      "loss": 0.9281,
      "step": 6186
    },
    {
      "epoch": 0.44637639334800333,
      "grad_norm": 2.678707822324337,
      "learning_rate": 2.4408098609229468e-06,
      "loss": 1.0066,
      "step": 6187
    },
    {
      "epoch": 0.44644854081743085,
      "grad_norm": 4.092067693156003,
      "learning_rate": 2.4403539821768292e-06,
      "loss": 1.0267,
      "step": 6188
    },
    {
      "epoch": 0.4465206882868583,
      "grad_norm": 3.1159117724521748,
      "learning_rate": 2.439898079384611e-06,
      "loss": 0.8708,
      "step": 6189
    },
    {
      "epoch": 0.44659283575628583,
      "grad_norm": 2.251749947638652,
      "learning_rate": 2.439442152571187e-06,
      "loss": 0.9266,
      "step": 6190
    },
    {
      "epoch": 0.44666498322571335,
      "grad_norm": 6.592762371421797,
      "learning_rate": 2.4389862017614543e-06,
      "loss": 0.9442,
      "step": 6191
    },
    {
      "epoch": 0.44673713069514087,
      "grad_norm": 2.6287478984510555,
      "learning_rate": 2.43853022698031e-06,
      "loss": 0.9221,
      "step": 6192
    },
    {
      "epoch": 0.4468092781645684,
      "grad_norm": 4.3716254162390475,
      "learning_rate": 2.438074228252653e-06,
      "loss": 0.8894,
      "step": 6193
    },
    {
      "epoch": 0.4468814256339959,
      "grad_norm": 1.979258150362187,
      "learning_rate": 2.4376182056033844e-06,
      "loss": 0.9765,
      "step": 6194
    },
    {
      "epoch": 0.4469535731034234,
      "grad_norm": 3.045054982758515,
      "learning_rate": 2.4371621590574053e-06,
      "loss": 0.9068,
      "step": 6195
    },
    {
      "epoch": 0.4470257205728509,
      "grad_norm": 3.246044392675299,
      "learning_rate": 2.436706088639619e-06,
      "loss": 0.8769,
      "step": 6196
    },
    {
      "epoch": 0.4470978680422784,
      "grad_norm": 3.7216582907004314,
      "learning_rate": 2.436249994374931e-06,
      "loss": 1.0132,
      "step": 6197
    },
    {
      "epoch": 0.44717001551170593,
      "grad_norm": 2.8268134516565304,
      "learning_rate": 2.4357938762882443e-06,
      "loss": 0.9555,
      "step": 6198
    },
    {
      "epoch": 0.44724216298113345,
      "grad_norm": 3.24017256665733,
      "learning_rate": 2.435337734404468e-06,
      "loss": 0.9537,
      "step": 6199
    },
    {
      "epoch": 0.44731431045056097,
      "grad_norm": 2.0941333704450797,
      "learning_rate": 2.4348815687485093e-06,
      "loss": 0.9864,
      "step": 6200
    },
    {
      "epoch": 0.44738645791998843,
      "grad_norm": 3.024024295295525,
      "learning_rate": 2.4344253793452784e-06,
      "loss": 0.9669,
      "step": 6201
    },
    {
      "epoch": 0.44745860538941595,
      "grad_norm": 3.3029682566139233,
      "learning_rate": 2.433969166219685e-06,
      "loss": 0.8908,
      "step": 6202
    },
    {
      "epoch": 0.4475307528588435,
      "grad_norm": 3.9728810594653714,
      "learning_rate": 2.4335129293966425e-06,
      "loss": 0.9688,
      "step": 6203
    },
    {
      "epoch": 0.447602900328271,
      "grad_norm": 6.7380004470476305,
      "learning_rate": 2.433056668901064e-06,
      "loss": 0.9655,
      "step": 6204
    },
    {
      "epoch": 0.4476750477976985,
      "grad_norm": 4.683022967787077,
      "learning_rate": 2.4326003847578627e-06,
      "loss": 1.0533,
      "step": 6205
    },
    {
      "epoch": 0.44774719526712603,
      "grad_norm": 2.6782200301234846,
      "learning_rate": 2.432144076991957e-06,
      "loss": 0.9566,
      "step": 6206
    },
    {
      "epoch": 0.4478193427365535,
      "grad_norm": 0.7533992980107176,
      "learning_rate": 2.4316877456282624e-06,
      "loss": 0.7943,
      "step": 6207
    },
    {
      "epoch": 0.447891490205981,
      "grad_norm": 3.7660899528177554,
      "learning_rate": 2.4312313906916984e-06,
      "loss": 1.0404,
      "step": 6208
    },
    {
      "epoch": 0.44796363767540853,
      "grad_norm": 2.4091255543738397,
      "learning_rate": 2.4307750122071848e-06,
      "loss": 1.0351,
      "step": 6209
    },
    {
      "epoch": 0.44803578514483605,
      "grad_norm": 4.542355523087126,
      "learning_rate": 2.430318610199642e-06,
      "loss": 0.9576,
      "step": 6210
    },
    {
      "epoch": 0.44810793261426357,
      "grad_norm": 165.70721426472355,
      "learning_rate": 2.429862184693993e-06,
      "loss": 0.9355,
      "step": 6211
    },
    {
      "epoch": 0.4481800800836911,
      "grad_norm": 5.041201588431095,
      "learning_rate": 2.429405735715161e-06,
      "loss": 0.9405,
      "step": 6212
    },
    {
      "epoch": 0.44825222755311855,
      "grad_norm": 4.903930699006581,
      "learning_rate": 2.428949263288072e-06,
      "loss": 1.0099,
      "step": 6213
    },
    {
      "epoch": 0.4483243750225461,
      "grad_norm": 4.447279349177331,
      "learning_rate": 2.4284927674376516e-06,
      "loss": 0.8218,
      "step": 6214
    },
    {
      "epoch": 0.4483965224919736,
      "grad_norm": 3.4094915801902466,
      "learning_rate": 2.4280362481888266e-06,
      "loss": 0.9598,
      "step": 6215
    },
    {
      "epoch": 0.4484686699614011,
      "grad_norm": 5.234734093034743,
      "learning_rate": 2.427579705566528e-06,
      "loss": 0.9381,
      "step": 6216
    },
    {
      "epoch": 0.44854081743082863,
      "grad_norm": 2.654869798100642,
      "learning_rate": 2.427123139595683e-06,
      "loss": 0.9492,
      "step": 6217
    },
    {
      "epoch": 0.44861296490025615,
      "grad_norm": 3.094555682456881,
      "learning_rate": 2.4266665503012252e-06,
      "loss": 0.9575,
      "step": 6218
    },
    {
      "epoch": 0.4486851123696836,
      "grad_norm": 3.5761261400082756,
      "learning_rate": 2.426209937708087e-06,
      "loss": 1.0194,
      "step": 6219
    },
    {
      "epoch": 0.44875725983911113,
      "grad_norm": 4.186673666427819,
      "learning_rate": 2.425753301841201e-06,
      "loss": 0.9814,
      "step": 6220
    },
    {
      "epoch": 0.44882940730853865,
      "grad_norm": 2.32551315551265,
      "learning_rate": 2.4252966427255042e-06,
      "loss": 1.0536,
      "step": 6221
    },
    {
      "epoch": 0.44890155477796617,
      "grad_norm": 24.90672050998485,
      "learning_rate": 2.4248399603859317e-06,
      "loss": 1.0203,
      "step": 6222
    },
    {
      "epoch": 0.4489737022473937,
      "grad_norm": 0.7757555201509544,
      "learning_rate": 2.4243832548474214e-06,
      "loss": 0.8518,
      "step": 6223
    },
    {
      "epoch": 0.44904584971682115,
      "grad_norm": 0.6581830437650363,
      "learning_rate": 2.4239265261349125e-06,
      "loss": 0.8252,
      "step": 6224
    },
    {
      "epoch": 0.4491179971862487,
      "grad_norm": 2.7032721154195487,
      "learning_rate": 2.423469774273346e-06,
      "loss": 0.9634,
      "step": 6225
    },
    {
      "epoch": 0.4491901446556762,
      "grad_norm": 3.7868735371303677,
      "learning_rate": 2.4230129992876623e-06,
      "loss": 0.8957,
      "step": 6226
    },
    {
      "epoch": 0.4492622921251037,
      "grad_norm": 2.8523085362923233,
      "learning_rate": 2.422556201202805e-06,
      "loss": 0.965,
      "step": 6227
    },
    {
      "epoch": 0.44933443959453123,
      "grad_norm": 2.813983695426941,
      "learning_rate": 2.4220993800437174e-06,
      "loss": 0.8826,
      "step": 6228
    },
    {
      "epoch": 0.44940658706395875,
      "grad_norm": 3.2372858337497745,
      "learning_rate": 2.421642535835345e-06,
      "loss": 1.0847,
      "step": 6229
    },
    {
      "epoch": 0.4494787345333862,
      "grad_norm": 3.218415918882534,
      "learning_rate": 2.4211856686026355e-06,
      "loss": 0.9919,
      "step": 6230
    },
    {
      "epoch": 0.44955088200281373,
      "grad_norm": 2.967970896431673,
      "learning_rate": 2.4207287783705354e-06,
      "loss": 1.0129,
      "step": 6231
    },
    {
      "epoch": 0.44962302947224125,
      "grad_norm": 2.4167061068067928,
      "learning_rate": 2.4202718651639935e-06,
      "loss": 0.899,
      "step": 6232
    },
    {
      "epoch": 0.4496951769416688,
      "grad_norm": 2.2841234269538337,
      "learning_rate": 2.4198149290079616e-06,
      "loss": 0.9539,
      "step": 6233
    },
    {
      "epoch": 0.4497673244110963,
      "grad_norm": 3.9160721239689846,
      "learning_rate": 2.41935796992739e-06,
      "loss": 1.0576,
      "step": 6234
    },
    {
      "epoch": 0.4498394718805238,
      "grad_norm": 3.1393367725363466,
      "learning_rate": 2.418900987947232e-06,
      "loss": 1.007,
      "step": 6235
    },
    {
      "epoch": 0.4499116193499513,
      "grad_norm": 2.8118415061649964,
      "learning_rate": 2.4184439830924426e-06,
      "loss": 1.0669,
      "step": 6236
    },
    {
      "epoch": 0.4499837668193788,
      "grad_norm": 2.6781181878411666,
      "learning_rate": 2.417986955387976e-06,
      "loss": 0.9756,
      "step": 6237
    },
    {
      "epoch": 0.4500559142888063,
      "grad_norm": 2.7253858765534367,
      "learning_rate": 2.4175299048587887e-06,
      "loss": 0.9806,
      "step": 6238
    },
    {
      "epoch": 0.45012806175823383,
      "grad_norm": 4.777321791266991,
      "learning_rate": 2.4170728315298394e-06,
      "loss": 0.8645,
      "step": 6239
    },
    {
      "epoch": 0.45020020922766135,
      "grad_norm": 2.6089065039811015,
      "learning_rate": 2.4166157354260867e-06,
      "loss": 0.9542,
      "step": 6240
    },
    {
      "epoch": 0.45027235669708887,
      "grad_norm": 2.7802696643370957,
      "learning_rate": 2.4161586165724906e-06,
      "loss": 1.0206,
      "step": 6241
    },
    {
      "epoch": 0.45034450416651634,
      "grad_norm": 2.655841840912529,
      "learning_rate": 2.415701474994013e-06,
      "loss": 1.007,
      "step": 6242
    },
    {
      "epoch": 0.45041665163594385,
      "grad_norm": 2.8369577637554846,
      "learning_rate": 2.4152443107156177e-06,
      "loss": 0.9589,
      "step": 6243
    },
    {
      "epoch": 0.4504887991053714,
      "grad_norm": 3.3469260275091477,
      "learning_rate": 2.4147871237622663e-06,
      "loss": 1.0574,
      "step": 6244
    },
    {
      "epoch": 0.4505609465747989,
      "grad_norm": 2.9941493841094156,
      "learning_rate": 2.4143299141589265e-06,
      "loss": 0.9837,
      "step": 6245
    },
    {
      "epoch": 0.4506330940442264,
      "grad_norm": 2.727378570063371,
      "learning_rate": 2.4138726819305633e-06,
      "loss": 0.9785,
      "step": 6246
    },
    {
      "epoch": 0.45070524151365393,
      "grad_norm": 4.759926713010655,
      "learning_rate": 2.413415427102145e-06,
      "loss": 0.9241,
      "step": 6247
    },
    {
      "epoch": 0.4507773889830814,
      "grad_norm": 5.141019765427441,
      "learning_rate": 2.4129581496986414e-06,
      "loss": 0.9456,
      "step": 6248
    },
    {
      "epoch": 0.4508495364525089,
      "grad_norm": 3.952277895666649,
      "learning_rate": 2.412500849745022e-06,
      "loss": 0.8778,
      "step": 6249
    },
    {
      "epoch": 0.45092168392193643,
      "grad_norm": 2.52025079432875,
      "learning_rate": 2.4120435272662576e-06,
      "loss": 1.0068,
      "step": 6250
    },
    {
      "epoch": 0.45099383139136395,
      "grad_norm": 0.843133984087337,
      "learning_rate": 2.411586182287321e-06,
      "loss": 0.9152,
      "step": 6251
    },
    {
      "epoch": 0.45106597886079147,
      "grad_norm": 2.7318754270086782,
      "learning_rate": 2.411128814833188e-06,
      "loss": 0.9235,
      "step": 6252
    },
    {
      "epoch": 0.451138126330219,
      "grad_norm": 0.757250862823831,
      "learning_rate": 2.410671424928832e-06,
      "loss": 0.842,
      "step": 6253
    },
    {
      "epoch": 0.45121027379964646,
      "grad_norm": 10.28405278274006,
      "learning_rate": 2.410214012599229e-06,
      "loss": 0.9721,
      "step": 6254
    },
    {
      "epoch": 0.451282421269074,
      "grad_norm": 3.45103425818578,
      "learning_rate": 2.4097565778693583e-06,
      "loss": 0.8397,
      "step": 6255
    },
    {
      "epoch": 0.4513545687385015,
      "grad_norm": 0.8601405461928094,
      "learning_rate": 2.4092991207641967e-06,
      "loss": 0.9811,
      "step": 6256
    },
    {
      "epoch": 0.451426716207929,
      "grad_norm": 0.9159342520222493,
      "learning_rate": 2.4088416413087256e-06,
      "loss": 0.8764,
      "step": 6257
    },
    {
      "epoch": 0.45149886367735653,
      "grad_norm": 4.5286259913069244,
      "learning_rate": 2.4083841395279267e-06,
      "loss": 0.9997,
      "step": 6258
    },
    {
      "epoch": 0.45157101114678405,
      "grad_norm": 3.534667530315365,
      "learning_rate": 2.4079266154467806e-06,
      "loss": 0.8601,
      "step": 6259
    },
    {
      "epoch": 0.4516431586162115,
      "grad_norm": 0.7152425768646035,
      "learning_rate": 2.407469069090273e-06,
      "loss": 0.8393,
      "step": 6260
    },
    {
      "epoch": 0.45171530608563903,
      "grad_norm": 2.59613561651019,
      "learning_rate": 2.407011500483388e-06,
      "loss": 0.8327,
      "step": 6261
    },
    {
      "epoch": 0.45178745355506655,
      "grad_norm": 3.5349170920052813,
      "learning_rate": 2.406553909651111e-06,
      "loss": 1.0059,
      "step": 6262
    },
    {
      "epoch": 0.4518596010244941,
      "grad_norm": 3.1193590750542866,
      "learning_rate": 2.4060962966184304e-06,
      "loss": 0.9414,
      "step": 6263
    },
    {
      "epoch": 0.4519317484939216,
      "grad_norm": 3.311782615281587,
      "learning_rate": 2.405638661410334e-06,
      "loss": 0.8774,
      "step": 6264
    },
    {
      "epoch": 0.4520038959633491,
      "grad_norm": 0.7726363068601733,
      "learning_rate": 2.405181004051812e-06,
      "loss": 0.8361,
      "step": 6265
    },
    {
      "epoch": 0.4520760434327766,
      "grad_norm": 3.0785344040376197,
      "learning_rate": 2.404723324567856e-06,
      "loss": 0.9664,
      "step": 6266
    },
    {
      "epoch": 0.4521481909022041,
      "grad_norm": 4.053910315087575,
      "learning_rate": 2.404265622983457e-06,
      "loss": 0.9041,
      "step": 6267
    },
    {
      "epoch": 0.4522203383716316,
      "grad_norm": 2.5129608832342143,
      "learning_rate": 2.4038078993236082e-06,
      "loss": 0.9256,
      "step": 6268
    },
    {
      "epoch": 0.45229248584105913,
      "grad_norm": 3.8797826477993635,
      "learning_rate": 2.4033501536133057e-06,
      "loss": 0.946,
      "step": 6269
    },
    {
      "epoch": 0.45236463331048665,
      "grad_norm": 4.11849555241163,
      "learning_rate": 2.4028923858775445e-06,
      "loss": 1.0624,
      "step": 6270
    },
    {
      "epoch": 0.45243678077991417,
      "grad_norm": 3.5592324513868676,
      "learning_rate": 2.402434596141321e-06,
      "loss": 0.9969,
      "step": 6271
    },
    {
      "epoch": 0.45250892824934164,
      "grad_norm": 5.035516292463469,
      "learning_rate": 2.4019767844296343e-06,
      "loss": 0.969,
      "step": 6272
    },
    {
      "epoch": 0.45258107571876915,
      "grad_norm": 4.71830454359543,
      "learning_rate": 2.4015189507674836e-06,
      "loss": 1.0019,
      "step": 6273
    },
    {
      "epoch": 0.4526532231881967,
      "grad_norm": 4.8355340003564775,
      "learning_rate": 2.401061095179869e-06,
      "loss": 1.0022,
      "step": 6274
    },
    {
      "epoch": 0.4527253706576242,
      "grad_norm": 2.5353634225044708,
      "learning_rate": 2.4006032176917935e-06,
      "loss": 0.964,
      "step": 6275
    },
    {
      "epoch": 0.4527975181270517,
      "grad_norm": 2.97768623186026,
      "learning_rate": 2.4001453183282587e-06,
      "loss": 1.0899,
      "step": 6276
    },
    {
      "epoch": 0.4528696655964792,
      "grad_norm": 2.6450784362239346,
      "learning_rate": 2.399687397114269e-06,
      "loss": 0.9371,
      "step": 6277
    },
    {
      "epoch": 0.4529418130659067,
      "grad_norm": 0.7051406579012877,
      "learning_rate": 2.39922945407483e-06,
      "loss": 0.7831,
      "step": 6278
    },
    {
      "epoch": 0.4530139605353342,
      "grad_norm": 3.9064991375629132,
      "learning_rate": 2.398771489234949e-06,
      "loss": 0.927,
      "step": 6279
    },
    {
      "epoch": 0.45308610800476173,
      "grad_norm": 2.651889048549707,
      "learning_rate": 2.3983135026196334e-06,
      "loss": 0.8486,
      "step": 6280
    },
    {
      "epoch": 0.45315825547418925,
      "grad_norm": 5.014247431781653,
      "learning_rate": 2.3978554942538914e-06,
      "loss": 1.1088,
      "step": 6281
    },
    {
      "epoch": 0.4532304029436168,
      "grad_norm": 2.6152551606016785,
      "learning_rate": 2.3973974641627343e-06,
      "loss": 1.0792,
      "step": 6282
    },
    {
      "epoch": 0.45330255041304424,
      "grad_norm": 3.2732813395030775,
      "learning_rate": 2.3969394123711716e-06,
      "loss": 0.9192,
      "step": 6283
    },
    {
      "epoch": 0.45337469788247176,
      "grad_norm": 2.799104963029285,
      "learning_rate": 2.396481338904218e-06,
      "loss": 0.9978,
      "step": 6284
    },
    {
      "epoch": 0.4534468453518993,
      "grad_norm": 2.878393409630704,
      "learning_rate": 2.3960232437868866e-06,
      "loss": 0.9426,
      "step": 6285
    },
    {
      "epoch": 0.4535189928213268,
      "grad_norm": 2.117646902994387,
      "learning_rate": 2.3955651270441904e-06,
      "loss": 0.9787,
      "step": 6286
    },
    {
      "epoch": 0.4535911402907543,
      "grad_norm": 3.25316421822279,
      "learning_rate": 2.395106988701148e-06,
      "loss": 1.1107,
      "step": 6287
    },
    {
      "epoch": 0.45366328776018183,
      "grad_norm": 3.852032429654679,
      "learning_rate": 2.3946488287827753e-06,
      "loss": 0.8773,
      "step": 6288
    },
    {
      "epoch": 0.4537354352296093,
      "grad_norm": 4.520268465019485,
      "learning_rate": 2.394190647314091e-06,
      "loss": 1.0316,
      "step": 6289
    },
    {
      "epoch": 0.4538075826990368,
      "grad_norm": 0.7696316648699332,
      "learning_rate": 2.393732444320115e-06,
      "loss": 0.7847,
      "step": 6290
    },
    {
      "epoch": 0.45387973016846433,
      "grad_norm": 5.869602727232662,
      "learning_rate": 2.3932742198258675e-06,
      "loss": 0.9778,
      "step": 6291
    },
    {
      "epoch": 0.45395187763789185,
      "grad_norm": 0.8867664408357472,
      "learning_rate": 2.3928159738563707e-06,
      "loss": 0.9425,
      "step": 6292
    },
    {
      "epoch": 0.4540240251073194,
      "grad_norm": 4.281531777123756,
      "learning_rate": 2.392357706436648e-06,
      "loss": 0.977,
      "step": 6293
    },
    {
      "epoch": 0.4540961725767469,
      "grad_norm": 2.754840405613434,
      "learning_rate": 2.3918994175917235e-06,
      "loss": 0.9294,
      "step": 6294
    },
    {
      "epoch": 0.45416832004617436,
      "grad_norm": 2.6786076897482793,
      "learning_rate": 2.3914411073466214e-06,
      "loss": 0.9537,
      "step": 6295
    },
    {
      "epoch": 0.4542404675156019,
      "grad_norm": 2.4245660413234806,
      "learning_rate": 2.390982775726371e-06,
      "loss": 0.9815,
      "step": 6296
    },
    {
      "epoch": 0.4543126149850294,
      "grad_norm": 0.8090423224310074,
      "learning_rate": 2.3905244227559978e-06,
      "loss": 0.8406,
      "step": 6297
    },
    {
      "epoch": 0.4543847624544569,
      "grad_norm": 3.489133177612417,
      "learning_rate": 2.3900660484605313e-06,
      "loss": 1.0027,
      "step": 6298
    },
    {
      "epoch": 0.45445690992388443,
      "grad_norm": 4.252750852834454,
      "learning_rate": 2.389607652865002e-06,
      "loss": 1.0632,
      "step": 6299
    },
    {
      "epoch": 0.45452905739331195,
      "grad_norm": 2.48574234863626,
      "learning_rate": 2.3891492359944414e-06,
      "loss": 1.0858,
      "step": 6300
    },
    {
      "epoch": 0.4546012048627394,
      "grad_norm": 3.9206673635988007,
      "learning_rate": 2.3886907978738814e-06,
      "loss": 0.9279,
      "step": 6301
    },
    {
      "epoch": 0.45467335233216694,
      "grad_norm": 3.4424089925866177,
      "learning_rate": 2.3882323385283563e-06,
      "loss": 0.9836,
      "step": 6302
    },
    {
      "epoch": 0.45474549980159445,
      "grad_norm": 5.158242690235424,
      "learning_rate": 2.3877738579828997e-06,
      "loss": 0.8399,
      "step": 6303
    },
    {
      "epoch": 0.454817647271022,
      "grad_norm": 4.296486354938061,
      "learning_rate": 2.3873153562625482e-06,
      "loss": 1.0292,
      "step": 6304
    },
    {
      "epoch": 0.4548897947404495,
      "grad_norm": 3.3288321303500394,
      "learning_rate": 2.3868568333923393e-06,
      "loss": 1.0024,
      "step": 6305
    },
    {
      "epoch": 0.454961942209877,
      "grad_norm": 3.498212766733941,
      "learning_rate": 2.386398289397311e-06,
      "loss": 0.9351,
      "step": 6306
    },
    {
      "epoch": 0.4550340896793045,
      "grad_norm": 4.339041154126915,
      "learning_rate": 2.3859397243025017e-06,
      "loss": 1.0643,
      "step": 6307
    },
    {
      "epoch": 0.455106237148732,
      "grad_norm": 4.053702821036502,
      "learning_rate": 2.3854811381329535e-06,
      "loss": 0.8314,
      "step": 6308
    },
    {
      "epoch": 0.4551783846181595,
      "grad_norm": 2.221359974420132,
      "learning_rate": 2.3850225309137074e-06,
      "loss": 0.9283,
      "step": 6309
    },
    {
      "epoch": 0.45525053208758703,
      "grad_norm": 3.096812418315943,
      "learning_rate": 2.3845639026698054e-06,
      "loss": 0.9267,
      "step": 6310
    },
    {
      "epoch": 0.45532267955701455,
      "grad_norm": 2.748840781287476,
      "learning_rate": 2.384105253426293e-06,
      "loss": 0.9716,
      "step": 6311
    },
    {
      "epoch": 0.4553948270264421,
      "grad_norm": 4.674118064881143,
      "learning_rate": 2.3836465832082147e-06,
      "loss": 0.8987,
      "step": 6312
    },
    {
      "epoch": 0.45546697449586954,
      "grad_norm": 2.7161071033343744,
      "learning_rate": 2.3831878920406157e-06,
      "loss": 0.9759,
      "step": 6313
    },
    {
      "epoch": 0.45553912196529706,
      "grad_norm": 4.677722957971453,
      "learning_rate": 2.3827291799485458e-06,
      "loss": 0.9908,
      "step": 6314
    },
    {
      "epoch": 0.4556112694347246,
      "grad_norm": 7.8822179745915815,
      "learning_rate": 2.3822704469570513e-06,
      "loss": 0.8442,
      "step": 6315
    },
    {
      "epoch": 0.4556834169041521,
      "grad_norm": 4.008113976158116,
      "learning_rate": 2.381811693091183e-06,
      "loss": 0.9867,
      "step": 6316
    },
    {
      "epoch": 0.4557555643735796,
      "grad_norm": 2.1597587675033,
      "learning_rate": 2.3813529183759914e-06,
      "loss": 0.9903,
      "step": 6317
    },
    {
      "epoch": 0.45582771184300713,
      "grad_norm": 3.0128458291512983,
      "learning_rate": 2.3808941228365293e-06,
      "loss": 0.9197,
      "step": 6318
    },
    {
      "epoch": 0.4558998593124346,
      "grad_norm": 3.2537838843006863,
      "learning_rate": 2.3804353064978488e-06,
      "loss": 0.918,
      "step": 6319
    },
    {
      "epoch": 0.4559720067818621,
      "grad_norm": 4.016572951704872,
      "learning_rate": 2.379976469385004e-06,
      "loss": 0.9927,
      "step": 6320
    },
    {
      "epoch": 0.45604415425128964,
      "grad_norm": 3.034606176776869,
      "learning_rate": 2.3795176115230516e-06,
      "loss": 0.9589,
      "step": 6321
    },
    {
      "epoch": 0.45611630172071715,
      "grad_norm": 3.3009787235215695,
      "learning_rate": 2.3790587329370467e-06,
      "loss": 0.8809,
      "step": 6322
    },
    {
      "epoch": 0.4561884491901447,
      "grad_norm": 2.271297430740937,
      "learning_rate": 2.378599833652048e-06,
      "loss": 0.9879,
      "step": 6323
    },
    {
      "epoch": 0.4562605966595722,
      "grad_norm": 2.6456253290390217,
      "learning_rate": 2.3781409136931146e-06,
      "loss": 0.9592,
      "step": 6324
    },
    {
      "epoch": 0.45633274412899966,
      "grad_norm": 3.131782800112434,
      "learning_rate": 2.377681973085304e-06,
      "loss": 1.0035,
      "step": 6325
    },
    {
      "epoch": 0.4564048915984272,
      "grad_norm": 3.372293446679149,
      "learning_rate": 2.3772230118536805e-06,
      "loss": 0.9311,
      "step": 6326
    },
    {
      "epoch": 0.4564770390678547,
      "grad_norm": 2.919265479367349,
      "learning_rate": 2.3767640300233044e-06,
      "loss": 1.1238,
      "step": 6327
    },
    {
      "epoch": 0.4565491865372822,
      "grad_norm": 3.0709982535712226,
      "learning_rate": 2.376305027619239e-06,
      "loss": 1.0479,
      "step": 6328
    },
    {
      "epoch": 0.45662133400670973,
      "grad_norm": 3.9009189696766837,
      "learning_rate": 2.3758460046665495e-06,
      "loss": 0.8694,
      "step": 6329
    },
    {
      "epoch": 0.4566934814761372,
      "grad_norm": 3.8753342022906048,
      "learning_rate": 2.375386961190301e-06,
      "loss": 0.956,
      "step": 6330
    },
    {
      "epoch": 0.4567656289455647,
      "grad_norm": 3.0511597851548022,
      "learning_rate": 2.3749278972155595e-06,
      "loss": 0.9773,
      "step": 6331
    },
    {
      "epoch": 0.45683777641499224,
      "grad_norm": 2.8144180539214085,
      "learning_rate": 2.374468812767394e-06,
      "loss": 0.937,
      "step": 6332
    },
    {
      "epoch": 0.45690992388441976,
      "grad_norm": 4.111987541758742,
      "learning_rate": 2.374009707870873e-06,
      "loss": 0.9573,
      "step": 6333
    },
    {
      "epoch": 0.4569820713538473,
      "grad_norm": 5.044986336030783,
      "learning_rate": 2.3735505825510658e-06,
      "loss": 1.0416,
      "step": 6334
    },
    {
      "epoch": 0.4570542188232748,
      "grad_norm": 3.918226026339145,
      "learning_rate": 2.3730914368330447e-06,
      "loss": 0.9136,
      "step": 6335
    },
    {
      "epoch": 0.45712636629270226,
      "grad_norm": 2.900136937821674,
      "learning_rate": 2.3726322707418813e-06,
      "loss": 0.8262,
      "step": 6336
    },
    {
      "epoch": 0.4571985137621298,
      "grad_norm": 3.308228527750356,
      "learning_rate": 2.3721730843026484e-06,
      "loss": 1.077,
      "step": 6337
    },
    {
      "epoch": 0.4572706612315573,
      "grad_norm": 2.5554304529980585,
      "learning_rate": 2.371713877540422e-06,
      "loss": 1.005,
      "step": 6338
    },
    {
      "epoch": 0.4573428087009848,
      "grad_norm": 2.5144305503297644,
      "learning_rate": 2.3712546504802767e-06,
      "loss": 0.9772,
      "step": 6339
    },
    {
      "epoch": 0.45741495617041233,
      "grad_norm": 2.9171126615040475,
      "learning_rate": 2.3707954031472886e-06,
      "loss": 0.9775,
      "step": 6340
    },
    {
      "epoch": 0.45748710363983985,
      "grad_norm": 2.7575654953953976,
      "learning_rate": 2.370336135566537e-06,
      "loss": 0.9868,
      "step": 6341
    },
    {
      "epoch": 0.4575592511092673,
      "grad_norm": 4.652684222493846,
      "learning_rate": 2.369876847763099e-06,
      "loss": 0.9478,
      "step": 6342
    },
    {
      "epoch": 0.45763139857869484,
      "grad_norm": 0.8178174539313232,
      "learning_rate": 2.369417539762056e-06,
      "loss": 0.9096,
      "step": 6343
    },
    {
      "epoch": 0.45770354604812236,
      "grad_norm": 1.7845677046262547,
      "learning_rate": 2.368958211588489e-06,
      "loss": 1.0215,
      "step": 6344
    },
    {
      "epoch": 0.4577756935175499,
      "grad_norm": 3.059206534589787,
      "learning_rate": 2.3684988632674804e-06,
      "loss": 0.9487,
      "step": 6345
    },
    {
      "epoch": 0.4578478409869774,
      "grad_norm": 0.8807737389190621,
      "learning_rate": 2.368039494824112e-06,
      "loss": 0.8745,
      "step": 6346
    },
    {
      "epoch": 0.4579199884564049,
      "grad_norm": 3.7964172087300545,
      "learning_rate": 2.3675801062834694e-06,
      "loss": 1.0471,
      "step": 6347
    },
    {
      "epoch": 0.4579921359258324,
      "grad_norm": 22.792329749504685,
      "learning_rate": 2.3671206976706385e-06,
      "loss": 1.0225,
      "step": 6348
    },
    {
      "epoch": 0.4580642833952599,
      "grad_norm": 2.5348117400319032,
      "learning_rate": 2.3666612690107045e-06,
      "loss": 1.0566,
      "step": 6349
    },
    {
      "epoch": 0.4581364308646874,
      "grad_norm": 3.461616535610468,
      "learning_rate": 2.3662018203287565e-06,
      "loss": 0.9673,
      "step": 6350
    },
    {
      "epoch": 0.45820857833411494,
      "grad_norm": 4.950914918751366,
      "learning_rate": 2.3657423516498827e-06,
      "loss": 1.0972,
      "step": 6351
    },
    {
      "epoch": 0.45828072580354245,
      "grad_norm": 2.8899406654048363,
      "learning_rate": 2.3652828629991727e-06,
      "loss": 1.042,
      "step": 6352
    },
    {
      "epoch": 0.45835287327297,
      "grad_norm": 17.32055300492696,
      "learning_rate": 2.364823354401718e-06,
      "loss": 0.9992,
      "step": 6353
    },
    {
      "epoch": 0.45842502074239744,
      "grad_norm": 2.807758573206361,
      "learning_rate": 2.36436382588261e-06,
      "loss": 1.0162,
      "step": 6354
    },
    {
      "epoch": 0.45849716821182496,
      "grad_norm": 3.8644292068936617,
      "learning_rate": 2.3639042774669427e-06,
      "loss": 0.9278,
      "step": 6355
    },
    {
      "epoch": 0.4585693156812525,
      "grad_norm": 2.8301210893801345,
      "learning_rate": 2.3634447091798102e-06,
      "loss": 0.9403,
      "step": 6356
    },
    {
      "epoch": 0.45864146315068,
      "grad_norm": 3.4576968484422603,
      "learning_rate": 2.3629851210463075e-06,
      "loss": 1.0147,
      "step": 6357
    },
    {
      "epoch": 0.4587136106201075,
      "grad_norm": 2.412103147923389,
      "learning_rate": 2.362525513091531e-06,
      "loss": 1.061,
      "step": 6358
    },
    {
      "epoch": 0.45878575808953503,
      "grad_norm": 2.052516936628413,
      "learning_rate": 2.362065885340578e-06,
      "loss": 0.9408,
      "step": 6359
    },
    {
      "epoch": 0.4588579055589625,
      "grad_norm": 3.1447283760029086,
      "learning_rate": 2.3616062378185477e-06,
      "loss": 0.8527,
      "step": 6360
    },
    {
      "epoch": 0.45893005302839,
      "grad_norm": 2.140390647812717,
      "learning_rate": 2.3611465705505383e-06,
      "loss": 0.9065,
      "step": 6361
    },
    {
      "epoch": 0.45900220049781754,
      "grad_norm": 3.3944700035829762,
      "learning_rate": 2.360686883561653e-06,
      "loss": 0.9671,
      "step": 6362
    },
    {
      "epoch": 0.45907434796724506,
      "grad_norm": 5.4858813703159575,
      "learning_rate": 2.360227176876992e-06,
      "loss": 0.8398,
      "step": 6363
    },
    {
      "epoch": 0.4591464954366726,
      "grad_norm": 3.818600635760394,
      "learning_rate": 2.3597674505216572e-06,
      "loss": 0.8773,
      "step": 6364
    },
    {
      "epoch": 0.4592186429061001,
      "grad_norm": 2.384395386294986,
      "learning_rate": 2.3593077045207546e-06,
      "loss": 0.8935,
      "step": 6365
    },
    {
      "epoch": 0.45929079037552756,
      "grad_norm": 3.964899072120198,
      "learning_rate": 2.3588479388993883e-06,
      "loss": 1.0151,
      "step": 6366
    },
    {
      "epoch": 0.4593629378449551,
      "grad_norm": 3.124095022771188,
      "learning_rate": 2.3583881536826644e-06,
      "loss": 0.8429,
      "step": 6367
    },
    {
      "epoch": 0.4594350853143826,
      "grad_norm": 2.737557178056945,
      "learning_rate": 2.3579283488956906e-06,
      "loss": 0.8607,
      "step": 6368
    },
    {
      "epoch": 0.4595072327838101,
      "grad_norm": 2.3636488447326633,
      "learning_rate": 2.3574685245635743e-06,
      "loss": 1.0002,
      "step": 6369
    },
    {
      "epoch": 0.45957938025323763,
      "grad_norm": 10.235653024329842,
      "learning_rate": 2.3570086807114253e-06,
      "loss": 0.9755,
      "step": 6370
    },
    {
      "epoch": 0.45965152772266515,
      "grad_norm": 2.372384287320685,
      "learning_rate": 2.3565488173643535e-06,
      "loss": 1.0078,
      "step": 6371
    },
    {
      "epoch": 0.4597236751920926,
      "grad_norm": 72.67685863636375,
      "learning_rate": 2.356088934547471e-06,
      "loss": 0.9222,
      "step": 6372
    },
    {
      "epoch": 0.45979582266152014,
      "grad_norm": 3.547668502709173,
      "learning_rate": 2.3556290322858904e-06,
      "loss": 0.9166,
      "step": 6373
    },
    {
      "epoch": 0.45986797013094766,
      "grad_norm": 3.6612723001652077,
      "learning_rate": 2.3551691106047234e-06,
      "loss": 0.9548,
      "step": 6374
    },
    {
      "epoch": 0.4599401176003752,
      "grad_norm": 2.6410267541202614,
      "learning_rate": 2.354709169529088e-06,
      "loss": 0.8754,
      "step": 6375
    },
    {
      "epoch": 0.4600122650698027,
      "grad_norm": 5.0650424930639,
      "learning_rate": 2.354249209084096e-06,
      "loss": 1.0009,
      "step": 6376
    },
    {
      "epoch": 0.4600844125392302,
      "grad_norm": 2.59318545528689,
      "learning_rate": 2.3537892292948674e-06,
      "loss": 1.0367,
      "step": 6377
    },
    {
      "epoch": 0.4601565600086577,
      "grad_norm": 2.6257076217920226,
      "learning_rate": 2.3533292301865183e-06,
      "loss": 0.9557,
      "step": 6378
    },
    {
      "epoch": 0.4602287074780852,
      "grad_norm": 3.4387257558008892,
      "learning_rate": 2.3528692117841676e-06,
      "loss": 0.9369,
      "step": 6379
    },
    {
      "epoch": 0.4603008549475127,
      "grad_norm": 0.6298248737509362,
      "learning_rate": 2.3524091741129363e-06,
      "loss": 0.8165,
      "step": 6380
    },
    {
      "epoch": 0.46037300241694024,
      "grad_norm": 0.7939858514212202,
      "learning_rate": 2.3519491171979436e-06,
      "loss": 0.8702,
      "step": 6381
    },
    {
      "epoch": 0.46044514988636775,
      "grad_norm": 2.7147636566422833,
      "learning_rate": 2.3514890410643137e-06,
      "loss": 0.8469,
      "step": 6382
    },
    {
      "epoch": 0.4605172973557952,
      "grad_norm": 4.73210044080947,
      "learning_rate": 2.351028945737167e-06,
      "loss": 1.0724,
      "step": 6383
    },
    {
      "epoch": 0.46058944482522274,
      "grad_norm": 2.335050586748771,
      "learning_rate": 2.3505688312416302e-06,
      "loss": 0.9434,
      "step": 6384
    },
    {
      "epoch": 0.46066159229465026,
      "grad_norm": 2.4654069308506212,
      "learning_rate": 2.3501086976028264e-06,
      "loss": 0.9778,
      "step": 6385
    },
    {
      "epoch": 0.4607337397640778,
      "grad_norm": 4.567080980612418,
      "learning_rate": 2.3496485448458834e-06,
      "loss": 1.0571,
      "step": 6386
    },
    {
      "epoch": 0.4608058872335053,
      "grad_norm": 3.2142843730863166,
      "learning_rate": 2.349188372995927e-06,
      "loss": 1.0265,
      "step": 6387
    },
    {
      "epoch": 0.4608780347029328,
      "grad_norm": 5.792462008454157,
      "learning_rate": 2.3487281820780863e-06,
      "loss": 1.0281,
      "step": 6388
    },
    {
      "epoch": 0.4609501821723603,
      "grad_norm": 4.3245839530742725,
      "learning_rate": 2.348267972117491e-06,
      "loss": 0.8982,
      "step": 6389
    },
    {
      "epoch": 0.4610223296417878,
      "grad_norm": 3.0174589757018975,
      "learning_rate": 2.3478077431392713e-06,
      "loss": 1.0155,
      "step": 6390
    },
    {
      "epoch": 0.4610944771112153,
      "grad_norm": 3.6566812635864716,
      "learning_rate": 2.347347495168558e-06,
      "loss": 1.0647,
      "step": 6391
    },
    {
      "epoch": 0.46116662458064284,
      "grad_norm": 2.32412909326812,
      "learning_rate": 2.3468872282304834e-06,
      "loss": 0.9732,
      "step": 6392
    },
    {
      "epoch": 0.46123877205007036,
      "grad_norm": 4.0294194759886395,
      "learning_rate": 2.3464269423501814e-06,
      "loss": 0.9599,
      "step": 6393
    },
    {
      "epoch": 0.4613109195194979,
      "grad_norm": 4.990222908499546,
      "learning_rate": 2.3459666375527873e-06,
      "loss": 0.9288,
      "step": 6394
    },
    {
      "epoch": 0.46138306698892534,
      "grad_norm": 4.73098905743269,
      "learning_rate": 2.345506313863435e-06,
      "loss": 0.9654,
      "step": 6395
    },
    {
      "epoch": 0.46145521445835286,
      "grad_norm": 4.924397914894647,
      "learning_rate": 2.3450459713072625e-06,
      "loss": 0.9456,
      "step": 6396
    },
    {
      "epoch": 0.4615273619277804,
      "grad_norm": 3.403606123669507,
      "learning_rate": 2.344585609909407e-06,
      "loss": 0.9697,
      "step": 6397
    },
    {
      "epoch": 0.4615995093972079,
      "grad_norm": 2.6158819329919147,
      "learning_rate": 2.3441252296950054e-06,
      "loss": 0.9931,
      "step": 6398
    },
    {
      "epoch": 0.4616716568666354,
      "grad_norm": 3.151332839669881,
      "learning_rate": 2.3436648306892004e-06,
      "loss": 0.9807,
      "step": 6399
    },
    {
      "epoch": 0.46174380433606294,
      "grad_norm": 3.733133788112053,
      "learning_rate": 2.3432044129171306e-06,
      "loss": 0.8682,
      "step": 6400
    },
    {
      "epoch": 0.4618159518054904,
      "grad_norm": 0.7634299204727655,
      "learning_rate": 2.3427439764039385e-06,
      "loss": 0.8562,
      "step": 6401
    },
    {
      "epoch": 0.4618880992749179,
      "grad_norm": 3.1244272851661195,
      "learning_rate": 2.3422835211747673e-06,
      "loss": 1.0643,
      "step": 6402
    },
    {
      "epoch": 0.46196024674434544,
      "grad_norm": 4.398703097900726,
      "learning_rate": 2.3418230472547587e-06,
      "loss": 0.9439,
      "step": 6403
    },
    {
      "epoch": 0.46203239421377296,
      "grad_norm": 0.7233071360114359,
      "learning_rate": 2.34136255466906e-06,
      "loss": 0.8365,
      "step": 6404
    },
    {
      "epoch": 0.4621045416832005,
      "grad_norm": 3.103847478717757,
      "learning_rate": 2.3409020434428155e-06,
      "loss": 0.9903,
      "step": 6405
    },
    {
      "epoch": 0.462176689152628,
      "grad_norm": 3.6833951700350718,
      "learning_rate": 2.340441513601173e-06,
      "loss": 0.9663,
      "step": 6406
    },
    {
      "epoch": 0.46224883662205546,
      "grad_norm": 0.6606638384674131,
      "learning_rate": 2.3399809651692786e-06,
      "loss": 0.8649,
      "step": 6407
    },
    {
      "epoch": 0.462320984091483,
      "grad_norm": 3.0017064327618526,
      "learning_rate": 2.339520398172283e-06,
      "loss": 0.9231,
      "step": 6408
    },
    {
      "epoch": 0.4623931315609105,
      "grad_norm": 3.6252088815424766,
      "learning_rate": 2.339059812635335e-06,
      "loss": 0.9959,
      "step": 6409
    },
    {
      "epoch": 0.462465279030338,
      "grad_norm": 5.962717892806053,
      "learning_rate": 2.3385992085835854e-06,
      "loss": 0.8526,
      "step": 6410
    },
    {
      "epoch": 0.46253742649976554,
      "grad_norm": 3.41105075566026,
      "learning_rate": 2.3381385860421874e-06,
      "loss": 0.9671,
      "step": 6411
    },
    {
      "epoch": 0.46260957396919306,
      "grad_norm": 3.6430366616133423,
      "learning_rate": 2.337677945036293e-06,
      "loss": 0.9821,
      "step": 6412
    },
    {
      "epoch": 0.4626817214386205,
      "grad_norm": 2.9758264627825124,
      "learning_rate": 2.337217285591055e-06,
      "loss": 0.902,
      "step": 6413
    },
    {
      "epoch": 0.46275386890804804,
      "grad_norm": 2.0524406186619735,
      "learning_rate": 2.3367566077316303e-06,
      "loss": 0.9891,
      "step": 6414
    },
    {
      "epoch": 0.46282601637747556,
      "grad_norm": 2.924674095842908,
      "learning_rate": 2.3362959114831727e-06,
      "loss": 1.0117,
      "step": 6415
    },
    {
      "epoch": 0.4628981638469031,
      "grad_norm": 5.347950494666924,
      "learning_rate": 2.3358351968708417e-06,
      "loss": 0.9743,
      "step": 6416
    },
    {
      "epoch": 0.4629703113163306,
      "grad_norm": 3.7537200278178995,
      "learning_rate": 2.3353744639197934e-06,
      "loss": 0.982,
      "step": 6417
    },
    {
      "epoch": 0.4630424587857581,
      "grad_norm": 0.6664872722855646,
      "learning_rate": 2.334913712655187e-06,
      "loss": 0.8084,
      "step": 6418
    },
    {
      "epoch": 0.4631146062551856,
      "grad_norm": 0.8368530251179767,
      "learning_rate": 2.334452943102182e-06,
      "loss": 0.8888,
      "step": 6419
    },
    {
      "epoch": 0.4631867537246131,
      "grad_norm": 4.853208106140596,
      "learning_rate": 2.3339921552859406e-06,
      "loss": 0.8824,
      "step": 6420
    },
    {
      "epoch": 0.4632589011940406,
      "grad_norm": 0.7771091969935573,
      "learning_rate": 2.333531349231624e-06,
      "loss": 0.9156,
      "step": 6421
    },
    {
      "epoch": 0.46333104866346814,
      "grad_norm": 3.0121504774765824,
      "learning_rate": 2.333070524964395e-06,
      "loss": 0.9481,
      "step": 6422
    },
    {
      "epoch": 0.46340319613289566,
      "grad_norm": 7.371893276888906,
      "learning_rate": 2.3326096825094176e-06,
      "loss": 1.0069,
      "step": 6423
    },
    {
      "epoch": 0.4634753436023232,
      "grad_norm": 4.56036326918949,
      "learning_rate": 2.3321488218918567e-06,
      "loss": 0.9491,
      "step": 6424
    },
    {
      "epoch": 0.46354749107175064,
      "grad_norm": 0.7536659846130529,
      "learning_rate": 2.331687943136878e-06,
      "loss": 0.8703,
      "step": 6425
    },
    {
      "epoch": 0.46361963854117816,
      "grad_norm": 4.326915365358963,
      "learning_rate": 2.3312270462696486e-06,
      "loss": 0.9926,
      "step": 6426
    },
    {
      "epoch": 0.4636917860106057,
      "grad_norm": 3.698866314801132,
      "learning_rate": 2.3307661313153362e-06,
      "loss": 0.9585,
      "step": 6427
    },
    {
      "epoch": 0.4637639334800332,
      "grad_norm": 2.475458615130661,
      "learning_rate": 2.3303051982991106e-06,
      "loss": 1.0066,
      "step": 6428
    },
    {
      "epoch": 0.4638360809494607,
      "grad_norm": 2.538526554764473,
      "learning_rate": 2.329844247246141e-06,
      "loss": 0.9967,
      "step": 6429
    },
    {
      "epoch": 0.4639082284188882,
      "grad_norm": 3.1097924033790836,
      "learning_rate": 2.329383278181597e-06,
      "loss": 0.9216,
      "step": 6430
    },
    {
      "epoch": 0.4639803758883157,
      "grad_norm": 4.244801427287794,
      "learning_rate": 2.3289222911306518e-06,
      "loss": 0.9624,
      "step": 6431
    },
    {
      "epoch": 0.4640525233577432,
      "grad_norm": 10.037014929807162,
      "learning_rate": 2.328461286118478e-06,
      "loss": 0.9883,
      "step": 6432
    },
    {
      "epoch": 0.46412467082717074,
      "grad_norm": 2.8263637896732132,
      "learning_rate": 2.3280002631702496e-06,
      "loss": 1.0364,
      "step": 6433
    },
    {
      "epoch": 0.46419681829659826,
      "grad_norm": 2.3607912171019803,
      "learning_rate": 2.3275392223111414e-06,
      "loss": 0.9769,
      "step": 6434
    },
    {
      "epoch": 0.4642689657660258,
      "grad_norm": 2.6750032727943434,
      "learning_rate": 2.3270781635663277e-06,
      "loss": 0.9546,
      "step": 6435
    },
    {
      "epoch": 0.46434111323545324,
      "grad_norm": 13.290237652883306,
      "learning_rate": 2.3266170869609875e-06,
      "loss": 0.8729,
      "step": 6436
    },
    {
      "epoch": 0.46441326070488076,
      "grad_norm": 4.133494791343455,
      "learning_rate": 2.3261559925202962e-06,
      "loss": 0.92,
      "step": 6437
    },
    {
      "epoch": 0.4644854081743083,
      "grad_norm": 3.7955181695862263,
      "learning_rate": 2.325694880269434e-06,
      "loss": 0.9797,
      "step": 6438
    },
    {
      "epoch": 0.4645575556437358,
      "grad_norm": 3.5989603501722294,
      "learning_rate": 2.3252337502335804e-06,
      "loss": 1.0555,
      "step": 6439
    },
    {
      "epoch": 0.4646297031131633,
      "grad_norm": 0.6963645177246381,
      "learning_rate": 2.324772602437915e-06,
      "loss": 0.8405,
      "step": 6440
    },
    {
      "epoch": 0.46470185058259084,
      "grad_norm": 5.524393951179806,
      "learning_rate": 2.3243114369076203e-06,
      "loss": 0.9841,
      "step": 6441
    },
    {
      "epoch": 0.4647739980520183,
      "grad_norm": 4.3142537407245145,
      "learning_rate": 2.323850253667878e-06,
      "loss": 0.9184,
      "step": 6442
    },
    {
      "epoch": 0.4648461455214458,
      "grad_norm": 3.311349921038329,
      "learning_rate": 2.3233890527438734e-06,
      "loss": 0.9422,
      "step": 6443
    },
    {
      "epoch": 0.46491829299087334,
      "grad_norm": 1.6565847238554994,
      "learning_rate": 2.322927834160789e-06,
      "loss": 1.0176,
      "step": 6444
    },
    {
      "epoch": 0.46499044046030086,
      "grad_norm": 8.429962182011062,
      "learning_rate": 2.3224665979438116e-06,
      "loss": 0.9753,
      "step": 6445
    },
    {
      "epoch": 0.4650625879297284,
      "grad_norm": 3.6851065675396057,
      "learning_rate": 2.3220053441181267e-06,
      "loss": 0.9605,
      "step": 6446
    },
    {
      "epoch": 0.4651347353991559,
      "grad_norm": 4.340871392240241,
      "learning_rate": 2.3215440727089216e-06,
      "loss": 0.9252,
      "step": 6447
    },
    {
      "epoch": 0.46520688286858336,
      "grad_norm": 3.237792931931929,
      "learning_rate": 2.3210827837413855e-06,
      "loss": 0.971,
      "step": 6448
    },
    {
      "epoch": 0.4652790303380109,
      "grad_norm": 3.779303814391239,
      "learning_rate": 2.3206214772407066e-06,
      "loss": 1.0124,
      "step": 6449
    },
    {
      "epoch": 0.4653511778074384,
      "grad_norm": 3.0582749162823823,
      "learning_rate": 2.3201601532320765e-06,
      "loss": 0.8922,
      "step": 6450
    },
    {
      "epoch": 0.4654233252768659,
      "grad_norm": 3.094778024974448,
      "learning_rate": 2.3196988117406854e-06,
      "loss": 0.9807,
      "step": 6451
    },
    {
      "epoch": 0.46549547274629344,
      "grad_norm": 5.091203673311416,
      "learning_rate": 2.319237452791725e-06,
      "loss": 1.0409,
      "step": 6452
    },
    {
      "epoch": 0.46556762021572096,
      "grad_norm": 2.4783287128165123,
      "learning_rate": 2.31877607641039e-06,
      "loss": 0.9996,
      "step": 6453
    },
    {
      "epoch": 0.4656397676851484,
      "grad_norm": 3.1191143304699445,
      "learning_rate": 2.3183146826218725e-06,
      "loss": 1.0775,
      "step": 6454
    },
    {
      "epoch": 0.46571191515457594,
      "grad_norm": 4.380849796578918,
      "learning_rate": 2.3178532714513697e-06,
      "loss": 1.0747,
      "step": 6455
    },
    {
      "epoch": 0.46578406262400346,
      "grad_norm": 3.3988795902734252,
      "learning_rate": 2.3173918429240764e-06,
      "loss": 0.9189,
      "step": 6456
    },
    {
      "epoch": 0.465856210093431,
      "grad_norm": 3.32708853042633,
      "learning_rate": 2.3169303970651888e-06,
      "loss": 0.9103,
      "step": 6457
    },
    {
      "epoch": 0.4659283575628585,
      "grad_norm": 2.7641599406161594,
      "learning_rate": 2.316468933899905e-06,
      "loss": 1.0065,
      "step": 6458
    },
    {
      "epoch": 0.466000505032286,
      "grad_norm": 2.939901790405837,
      "learning_rate": 2.3160074534534254e-06,
      "loss": 0.9479,
      "step": 6459
    },
    {
      "epoch": 0.4660726525017135,
      "grad_norm": 4.335340524078449,
      "learning_rate": 2.315545955750948e-06,
      "loss": 0.8995,
      "step": 6460
    },
    {
      "epoch": 0.466144799971141,
      "grad_norm": 3.1913582881092335,
      "learning_rate": 2.3150844408176743e-06,
      "loss": 1.0496,
      "step": 6461
    },
    {
      "epoch": 0.4662169474405685,
      "grad_norm": 2.575187720706007,
      "learning_rate": 2.3146229086788053e-06,
      "loss": 0.8823,
      "step": 6462
    },
    {
      "epoch": 0.46628909490999604,
      "grad_norm": 7.555966920811554,
      "learning_rate": 2.3141613593595447e-06,
      "loss": 0.9146,
      "step": 6463
    },
    {
      "epoch": 0.46636124237942356,
      "grad_norm": 0.7076920851117839,
      "learning_rate": 2.313699792885094e-06,
      "loss": 0.8439,
      "step": 6464
    },
    {
      "epoch": 0.4664333898488511,
      "grad_norm": 7.497122657695501,
      "learning_rate": 2.31323820928066e-06,
      "loss": 0.9612,
      "step": 6465
    },
    {
      "epoch": 0.46650553731827854,
      "grad_norm": 2.5511963113745564,
      "learning_rate": 2.312776608571446e-06,
      "loss": 1.0408,
      "step": 6466
    },
    {
      "epoch": 0.46657768478770606,
      "grad_norm": 3.8195571620762108,
      "learning_rate": 2.3123149907826604e-06,
      "loss": 1.0976,
      "step": 6467
    },
    {
      "epoch": 0.4666498322571336,
      "grad_norm": 2.7362659128701394,
      "learning_rate": 2.311853355939509e-06,
      "loss": 0.9185,
      "step": 6468
    },
    {
      "epoch": 0.4667219797265611,
      "grad_norm": 4.140499012307919,
      "learning_rate": 2.3113917040671998e-06,
      "loss": 0.8843,
      "step": 6469
    },
    {
      "epoch": 0.4667941271959886,
      "grad_norm": 3.730253746667426,
      "learning_rate": 2.3109300351909424e-06,
      "loss": 0.8423,
      "step": 6470
    },
    {
      "epoch": 0.46686627466541614,
      "grad_norm": 3.951698015513147,
      "learning_rate": 2.310468349335947e-06,
      "loss": 1.0438,
      "step": 6471
    },
    {
      "epoch": 0.4669384221348436,
      "grad_norm": 3.588470428748908,
      "learning_rate": 2.3100066465274246e-06,
      "loss": 0.976,
      "step": 6472
    },
    {
      "epoch": 0.4670105696042711,
      "grad_norm": 4.21709584902956,
      "learning_rate": 2.3095449267905865e-06,
      "loss": 0.9291,
      "step": 6473
    },
    {
      "epoch": 0.46708271707369864,
      "grad_norm": 4.733267374429824,
      "learning_rate": 2.309083190150646e-06,
      "loss": 0.8668,
      "step": 6474
    },
    {
      "epoch": 0.46715486454312616,
      "grad_norm": 2.7151091301835644,
      "learning_rate": 2.308621436632817e-06,
      "loss": 0.9889,
      "step": 6475
    },
    {
      "epoch": 0.4672270120125537,
      "grad_norm": 7.25939411714215,
      "learning_rate": 2.308159666262313e-06,
      "loss": 0.9677,
      "step": 6476
    },
    {
      "epoch": 0.4672991594819812,
      "grad_norm": 3.782302583836116,
      "learning_rate": 2.307697879064351e-06,
      "loss": 0.9451,
      "step": 6477
    },
    {
      "epoch": 0.46737130695140866,
      "grad_norm": 3.534456535768245,
      "learning_rate": 2.3072360750641473e-06,
      "loss": 0.9116,
      "step": 6478
    },
    {
      "epoch": 0.4674434544208362,
      "grad_norm": 6.776728234387457,
      "learning_rate": 2.3067742542869177e-06,
      "loss": 0.9466,
      "step": 6479
    },
    {
      "epoch": 0.4675156018902637,
      "grad_norm": 3.709194440552712,
      "learning_rate": 2.3063124167578823e-06,
      "loss": 1.0208,
      "step": 6480
    },
    {
      "epoch": 0.4675877493596912,
      "grad_norm": 4.49737366223756,
      "learning_rate": 2.3058505625022603e-06,
      "loss": 1.0186,
      "step": 6481
    },
    {
      "epoch": 0.46765989682911874,
      "grad_norm": 4.286698709687528,
      "learning_rate": 2.3053886915452704e-06,
      "loss": 0.9998,
      "step": 6482
    },
    {
      "epoch": 0.4677320442985462,
      "grad_norm": 3.8137185072497237,
      "learning_rate": 2.304926803912135e-06,
      "loss": 1.0589,
      "step": 6483
    },
    {
      "epoch": 0.4678041917679737,
      "grad_norm": 0.7406551805616209,
      "learning_rate": 2.3044648996280757e-06,
      "loss": 0.8364,
      "step": 6484
    },
    {
      "epoch": 0.46787633923740124,
      "grad_norm": 2.8396380532343315,
      "learning_rate": 2.3040029787183147e-06,
      "loss": 1.0508,
      "step": 6485
    },
    {
      "epoch": 0.46794848670682876,
      "grad_norm": 3.467117337741397,
      "learning_rate": 2.3035410412080768e-06,
      "loss": 0.913,
      "step": 6486
    },
    {
      "epoch": 0.4680206341762563,
      "grad_norm": 3.945879904859216,
      "learning_rate": 2.3030790871225873e-06,
      "loss": 0.9796,
      "step": 6487
    },
    {
      "epoch": 0.4680927816456838,
      "grad_norm": 8.589437394464085,
      "learning_rate": 2.3026171164870694e-06,
      "loss": 0.9308,
      "step": 6488
    },
    {
      "epoch": 0.46816492911511126,
      "grad_norm": 2.601729379061423,
      "learning_rate": 2.3021551293267518e-06,
      "loss": 1.0104,
      "step": 6489
    },
    {
      "epoch": 0.4682370765845388,
      "grad_norm": 2.5666215826591534,
      "learning_rate": 2.301693125666861e-06,
      "loss": 0.9554,
      "step": 6490
    },
    {
      "epoch": 0.4683092240539663,
      "grad_norm": 2.592696469991069,
      "learning_rate": 2.3012311055326246e-06,
      "loss": 0.9886,
      "step": 6491
    },
    {
      "epoch": 0.4683813715233938,
      "grad_norm": 3.617829828101385,
      "learning_rate": 2.300769068949274e-06,
      "loss": 0.9818,
      "step": 6492
    },
    {
      "epoch": 0.46845351899282134,
      "grad_norm": 0.7334641834763903,
      "learning_rate": 2.300307015942037e-06,
      "loss": 0.8561,
      "step": 6493
    },
    {
      "epoch": 0.46852566646224886,
      "grad_norm": 0.7110154664601412,
      "learning_rate": 2.299844946536146e-06,
      "loss": 0.836,
      "step": 6494
    },
    {
      "epoch": 0.4685978139316763,
      "grad_norm": 3.9033024768168945,
      "learning_rate": 2.2993828607568324e-06,
      "loss": 1.0424,
      "step": 6495
    },
    {
      "epoch": 0.46866996140110384,
      "grad_norm": 8.664810960119636,
      "learning_rate": 2.298920758629329e-06,
      "loss": 0.9927,
      "step": 6496
    },
    {
      "epoch": 0.46874210887053136,
      "grad_norm": 3.3688613731643993,
      "learning_rate": 2.2984586401788696e-06,
      "loss": 1.0688,
      "step": 6497
    },
    {
      "epoch": 0.4688142563399589,
      "grad_norm": 10.32142170006848,
      "learning_rate": 2.297996505430688e-06,
      "loss": 1.003,
      "step": 6498
    },
    {
      "epoch": 0.4688864038093864,
      "grad_norm": 3.1421119623954388,
      "learning_rate": 2.297534354410022e-06,
      "loss": 1.0279,
      "step": 6499
    },
    {
      "epoch": 0.4689585512788139,
      "grad_norm": 7.220562067529433,
      "learning_rate": 2.2970721871421055e-06,
      "loss": 0.9521,
      "step": 6500
    },
    {
      "epoch": 0.4690306987482414,
      "grad_norm": 3.00642136939891,
      "learning_rate": 2.296610003652176e-06,
      "loss": 1.0333,
      "step": 6501
    },
    {
      "epoch": 0.4691028462176689,
      "grad_norm": 21.854232608585374,
      "learning_rate": 2.2961478039654725e-06,
      "loss": 0.9796,
      "step": 6502
    },
    {
      "epoch": 0.4691749936870964,
      "grad_norm": 3.797430052825663,
      "learning_rate": 2.2956855881072337e-06,
      "loss": 1.0483,
      "step": 6503
    },
    {
      "epoch": 0.46924714115652394,
      "grad_norm": 3.014611266524564,
      "learning_rate": 2.2952233561027003e-06,
      "loss": 0.9876,
      "step": 6504
    },
    {
      "epoch": 0.46931928862595146,
      "grad_norm": 6.495041349659682,
      "learning_rate": 2.294761107977112e-06,
      "loss": 1.0332,
      "step": 6505
    },
    {
      "epoch": 0.469391436095379,
      "grad_norm": 2.607048047874412,
      "learning_rate": 2.2942988437557096e-06,
      "loss": 0.9656,
      "step": 6506
    },
    {
      "epoch": 0.46946358356480644,
      "grad_norm": 2.5582158632262737,
      "learning_rate": 2.293836563463738e-06,
      "loss": 0.9332,
      "step": 6507
    },
    {
      "epoch": 0.46953573103423396,
      "grad_norm": 3.803248828636756,
      "learning_rate": 2.293374267126439e-06,
      "loss": 0.9592,
      "step": 6508
    },
    {
      "epoch": 0.4696078785036615,
      "grad_norm": 2.6187705646881887,
      "learning_rate": 2.2929119547690577e-06,
      "loss": 1.0188,
      "step": 6509
    },
    {
      "epoch": 0.469680025973089,
      "grad_norm": 3.6150512978698477,
      "learning_rate": 2.292449626416838e-06,
      "loss": 0.8935,
      "step": 6510
    },
    {
      "epoch": 0.4697521734425165,
      "grad_norm": 3.46259400828849,
      "learning_rate": 2.291987282095028e-06,
      "loss": 0.921,
      "step": 6511
    },
    {
      "epoch": 0.46982432091194404,
      "grad_norm": 3.1349344653713023,
      "learning_rate": 2.2915249218288726e-06,
      "loss": 1.0174,
      "step": 6512
    },
    {
      "epoch": 0.4698964683813715,
      "grad_norm": 0.8304461213238057,
      "learning_rate": 2.2910625456436205e-06,
      "loss": 0.9066,
      "step": 6513
    },
    {
      "epoch": 0.469968615850799,
      "grad_norm": 2.644855969551161,
      "learning_rate": 2.2906001535645205e-06,
      "loss": 0.9737,
      "step": 6514
    },
    {
      "epoch": 0.47004076332022654,
      "grad_norm": 0.7574822797685505,
      "learning_rate": 2.290137745616821e-06,
      "loss": 0.8088,
      "step": 6515
    },
    {
      "epoch": 0.47011291078965406,
      "grad_norm": 2.344072345183558,
      "learning_rate": 2.2896753218257746e-06,
      "loss": 1.0007,
      "step": 6516
    },
    {
      "epoch": 0.4701850582590816,
      "grad_norm": 22.579746238289456,
      "learning_rate": 2.289212882216631e-06,
      "loss": 1.0194,
      "step": 6517
    },
    {
      "epoch": 0.4702572057285091,
      "grad_norm": 5.092350492534695,
      "learning_rate": 2.2887504268146415e-06,
      "loss": 0.9677,
      "step": 6518
    },
    {
      "epoch": 0.47032935319793656,
      "grad_norm": 2.709262121328505,
      "learning_rate": 2.288287955645061e-06,
      "loss": 0.8806,
      "step": 6519
    },
    {
      "epoch": 0.4704015006673641,
      "grad_norm": 3.7663817852169688,
      "learning_rate": 2.2878254687331426e-06,
      "loss": 0.9808,
      "step": 6520
    },
    {
      "epoch": 0.4704736481367916,
      "grad_norm": 3.5922782206186445,
      "learning_rate": 2.28736296610414e-06,
      "loss": 0.9323,
      "step": 6521
    },
    {
      "epoch": 0.4705457956062191,
      "grad_norm": 3.1813909293325238,
      "learning_rate": 2.2869004477833103e-06,
      "loss": 0.8978,
      "step": 6522
    },
    {
      "epoch": 0.47061794307564664,
      "grad_norm": 2.8735059090084443,
      "learning_rate": 2.2864379137959095e-06,
      "loss": 1.0019,
      "step": 6523
    },
    {
      "epoch": 0.47069009054507416,
      "grad_norm": 2.4691708363002043,
      "learning_rate": 2.285975364167194e-06,
      "loss": 0.8118,
      "step": 6524
    },
    {
      "epoch": 0.4707622380145016,
      "grad_norm": 2.16599008804829,
      "learning_rate": 2.2855127989224223e-06,
      "loss": 0.9496,
      "step": 6525
    },
    {
      "epoch": 0.47083438548392914,
      "grad_norm": 2.84064322319942,
      "learning_rate": 2.285050218086854e-06,
      "loss": 0.9677,
      "step": 6526
    },
    {
      "epoch": 0.47090653295335666,
      "grad_norm": 3.0549011936223747,
      "learning_rate": 2.284587621685748e-06,
      "loss": 0.8836,
      "step": 6527
    },
    {
      "epoch": 0.4709786804227842,
      "grad_norm": 2.942426405823351,
      "learning_rate": 2.284125009744366e-06,
      "loss": 0.9793,
      "step": 6528
    },
    {
      "epoch": 0.4710508278922117,
      "grad_norm": 3.948352088502584,
      "learning_rate": 2.283662382287969e-06,
      "loss": 0.9343,
      "step": 6529
    },
    {
      "epoch": 0.4711229753616392,
      "grad_norm": 3.153488159438102,
      "learning_rate": 2.283199739341819e-06,
      "loss": 1.0132,
      "step": 6530
    },
    {
      "epoch": 0.4711951228310667,
      "grad_norm": 3.2207860340298846,
      "learning_rate": 2.2827370809311798e-06,
      "loss": 0.894,
      "step": 6531
    },
    {
      "epoch": 0.4712672703004942,
      "grad_norm": 3.210598892958266,
      "learning_rate": 2.2822744070813153e-06,
      "loss": 0.932,
      "step": 6532
    },
    {
      "epoch": 0.4713394177699217,
      "grad_norm": 4.240384162917446,
      "learning_rate": 2.2818117178174905e-06,
      "loss": 1.0813,
      "step": 6533
    },
    {
      "epoch": 0.47141156523934924,
      "grad_norm": 2.936604464763803,
      "learning_rate": 2.2813490131649714e-06,
      "loss": 0.9977,
      "step": 6534
    },
    {
      "epoch": 0.47148371270877676,
      "grad_norm": 3.02080742835853,
      "learning_rate": 2.2808862931490234e-06,
      "loss": 1.0638,
      "step": 6535
    },
    {
      "epoch": 0.4715558601782042,
      "grad_norm": 2.304913732772062,
      "learning_rate": 2.280423557794915e-06,
      "loss": 1.0081,
      "step": 6536
    },
    {
      "epoch": 0.47162800764763174,
      "grad_norm": 2.7845304195299425,
      "learning_rate": 2.2799608071279143e-06,
      "loss": 0.9441,
      "step": 6537
    },
    {
      "epoch": 0.47170015511705926,
      "grad_norm": 5.170577056378453,
      "learning_rate": 2.2794980411732906e-06,
      "loss": 0.9418,
      "step": 6538
    },
    {
      "epoch": 0.4717723025864868,
      "grad_norm": 4.056896159425856,
      "learning_rate": 2.279035259956313e-06,
      "loss": 1.0325,
      "step": 6539
    },
    {
      "epoch": 0.4718444500559143,
      "grad_norm": 2.668650415782572,
      "learning_rate": 2.278572463502253e-06,
      "loss": 0.941,
      "step": 6540
    },
    {
      "epoch": 0.4719165975253418,
      "grad_norm": 7.1283606918817215,
      "learning_rate": 2.278109651836382e-06,
      "loss": 1.0459,
      "step": 6541
    },
    {
      "epoch": 0.4719887449947693,
      "grad_norm": 2.2267640139758833,
      "learning_rate": 2.2776468249839724e-06,
      "loss": 1.0048,
      "step": 6542
    },
    {
      "epoch": 0.4720608924641968,
      "grad_norm": 3.816361894613342,
      "learning_rate": 2.277183982970298e-06,
      "loss": 0.879,
      "step": 6543
    },
    {
      "epoch": 0.4721330399336243,
      "grad_norm": 3.203051608105466,
      "learning_rate": 2.276721125820632e-06,
      "loss": 0.9703,
      "step": 6544
    },
    {
      "epoch": 0.47220518740305184,
      "grad_norm": 2.9967406369062464,
      "learning_rate": 2.276258253560249e-06,
      "loss": 0.8159,
      "step": 6545
    },
    {
      "epoch": 0.47227733487247936,
      "grad_norm": 10.413634283558224,
      "learning_rate": 2.2757953662144266e-06,
      "loss": 0.9917,
      "step": 6546
    },
    {
      "epoch": 0.4723494823419069,
      "grad_norm": 4.5658249954191135,
      "learning_rate": 2.2753324638084398e-06,
      "loss": 0.9291,
      "step": 6547
    },
    {
      "epoch": 0.47242162981133434,
      "grad_norm": 3.33061676591316,
      "learning_rate": 2.274869546367566e-06,
      "loss": 1.0129,
      "step": 6548
    },
    {
      "epoch": 0.47249377728076186,
      "grad_norm": 2.9303745939077683,
      "learning_rate": 2.2744066139170846e-06,
      "loss": 0.9801,
      "step": 6549
    },
    {
      "epoch": 0.4725659247501894,
      "grad_norm": 2.2601811744853118,
      "learning_rate": 2.2739436664822728e-06,
      "loss": 0.9663,
      "step": 6550
    },
    {
      "epoch": 0.4726380722196169,
      "grad_norm": 0.7494740628900888,
      "learning_rate": 2.2734807040884123e-06,
      "loss": 0.7964,
      "step": 6551
    },
    {
      "epoch": 0.4727102196890444,
      "grad_norm": 2.8582921713783724,
      "learning_rate": 2.2730177267607824e-06,
      "loss": 0.864,
      "step": 6552
    },
    {
      "epoch": 0.47278236715847194,
      "grad_norm": 4.0491293736669345,
      "learning_rate": 2.2725547345246657e-06,
      "loss": 1.013,
      "step": 6553
    },
    {
      "epoch": 0.4728545146278994,
      "grad_norm": 4.126623585547496,
      "learning_rate": 2.272091727405343e-06,
      "loss": 0.932,
      "step": 6554
    },
    {
      "epoch": 0.4729266620973269,
      "grad_norm": 4.455413337482001,
      "learning_rate": 2.2716287054280996e-06,
      "loss": 0.9601,
      "step": 6555
    },
    {
      "epoch": 0.47299880956675444,
      "grad_norm": 5.428712430714627,
      "learning_rate": 2.2711656686182175e-06,
      "loss": 0.9712,
      "step": 6556
    },
    {
      "epoch": 0.47307095703618196,
      "grad_norm": 4.6531929374946195,
      "learning_rate": 2.2707026170009815e-06,
      "loss": 0.9107,
      "step": 6557
    },
    {
      "epoch": 0.4731431045056095,
      "grad_norm": 2.606125595819454,
      "learning_rate": 2.2702395506016784e-06,
      "loss": 1.0165,
      "step": 6558
    },
    {
      "epoch": 0.473215251975037,
      "grad_norm": 51.1565482772306,
      "learning_rate": 2.2697764694455935e-06,
      "loss": 0.9282,
      "step": 6559
    },
    {
      "epoch": 0.47328739944446446,
      "grad_norm": 3.413039414248375,
      "learning_rate": 2.2693133735580144e-06,
      "loss": 1.0362,
      "step": 6560
    },
    {
      "epoch": 0.473359546913892,
      "grad_norm": 5.077621481948062,
      "learning_rate": 2.2688502629642295e-06,
      "loss": 1.0766,
      "step": 6561
    },
    {
      "epoch": 0.4734316943833195,
      "grad_norm": 2.8926781863073834,
      "learning_rate": 2.268387137689526e-06,
      "loss": 0.9746,
      "step": 6562
    },
    {
      "epoch": 0.473503841852747,
      "grad_norm": 3.6214419036043695,
      "learning_rate": 2.267923997759195e-06,
      "loss": 0.9514,
      "step": 6563
    },
    {
      "epoch": 0.47357598932217454,
      "grad_norm": 2.946347163673473,
      "learning_rate": 2.2674608431985257e-06,
      "loss": 1.022,
      "step": 6564
    },
    {
      "epoch": 0.47364813679160206,
      "grad_norm": 2.3732068168836706,
      "learning_rate": 2.266997674032811e-06,
      "loss": 0.8938,
      "step": 6565
    },
    {
      "epoch": 0.4737202842610295,
      "grad_norm": 3.19187372862322,
      "learning_rate": 2.266534490287341e-06,
      "loss": 1.0028,
      "step": 6566
    },
    {
      "epoch": 0.47379243173045704,
      "grad_norm": 2.455354685596376,
      "learning_rate": 2.266071291987409e-06,
      "loss": 0.9434,
      "step": 6567
    },
    {
      "epoch": 0.47386457919988456,
      "grad_norm": 2.6732745158256552,
      "learning_rate": 2.2656080791583093e-06,
      "loss": 0.8629,
      "step": 6568
    },
    {
      "epoch": 0.4739367266693121,
      "grad_norm": 4.548619725394727,
      "learning_rate": 2.265144851825335e-06,
      "loss": 1.1179,
      "step": 6569
    },
    {
      "epoch": 0.4740088741387396,
      "grad_norm": 11.697144797391063,
      "learning_rate": 2.264681610013782e-06,
      "loss": 1.0172,
      "step": 6570
    },
    {
      "epoch": 0.4740810216081671,
      "grad_norm": 3.1868496399652404,
      "learning_rate": 2.2642183537489464e-06,
      "loss": 0.9884,
      "step": 6571
    },
    {
      "epoch": 0.4741531690775946,
      "grad_norm": 2.1326935563788627,
      "learning_rate": 2.2637550830561244e-06,
      "loss": 1.0029,
      "step": 6572
    },
    {
      "epoch": 0.4742253165470221,
      "grad_norm": 3.203961370180497,
      "learning_rate": 2.263291797960614e-06,
      "loss": 0.805,
      "step": 6573
    },
    {
      "epoch": 0.4742974640164496,
      "grad_norm": 2.591734413250969,
      "learning_rate": 2.2628284984877127e-06,
      "loss": 0.9276,
      "step": 6574
    },
    {
      "epoch": 0.47436961148587714,
      "grad_norm": 4.209782542866976,
      "learning_rate": 2.2623651846627203e-06,
      "loss": 1.0574,
      "step": 6575
    },
    {
      "epoch": 0.47444175895530466,
      "grad_norm": 3.012130689341492,
      "learning_rate": 2.261901856510936e-06,
      "loss": 1.0723,
      "step": 6576
    },
    {
      "epoch": 0.4745139064247322,
      "grad_norm": 3.8063904634562293,
      "learning_rate": 2.2614385140576616e-06,
      "loss": 0.9668,
      "step": 6577
    },
    {
      "epoch": 0.47458605389415964,
      "grad_norm": 5.706709062175196,
      "learning_rate": 2.2609751573281978e-06,
      "loss": 1.0143,
      "step": 6578
    },
    {
      "epoch": 0.47465820136358716,
      "grad_norm": 2.7326083334399747,
      "learning_rate": 2.260511786347846e-06,
      "loss": 0.9809,
      "step": 6579
    },
    {
      "epoch": 0.4747303488330147,
      "grad_norm": 3.7517139332755356,
      "learning_rate": 2.2600484011419107e-06,
      "loss": 0.8702,
      "step": 6580
    },
    {
      "epoch": 0.4748024963024422,
      "grad_norm": 3.084275505525071,
      "learning_rate": 2.2595850017356935e-06,
      "loss": 0.9063,
      "step": 6581
    },
    {
      "epoch": 0.4748746437718697,
      "grad_norm": 2.7593031358703195,
      "learning_rate": 2.259121588154502e-06,
      "loss": 0.9052,
      "step": 6582
    },
    {
      "epoch": 0.47494679124129724,
      "grad_norm": 3.8606176190570505,
      "learning_rate": 2.2586581604236392e-06,
      "loss": 0.9975,
      "step": 6583
    },
    {
      "epoch": 0.4750189387107247,
      "grad_norm": 7.606241634618192,
      "learning_rate": 2.2581947185684116e-06,
      "loss": 1.0096,
      "step": 6584
    },
    {
      "epoch": 0.4750910861801522,
      "grad_norm": 3.2430958406141333,
      "learning_rate": 2.257731262614126e-06,
      "loss": 0.95,
      "step": 6585
    },
    {
      "epoch": 0.47516323364957974,
      "grad_norm": 4.350677893423532,
      "learning_rate": 2.2572677925860905e-06,
      "loss": 0.9176,
      "step": 6586
    },
    {
      "epoch": 0.47523538111900726,
      "grad_norm": 28.206392643113723,
      "learning_rate": 2.2568043085096132e-06,
      "loss": 0.9065,
      "step": 6587
    },
    {
      "epoch": 0.4753075285884348,
      "grad_norm": 3.6710128746045956,
      "learning_rate": 2.2563408104100037e-06,
      "loss": 1.0066,
      "step": 6588
    },
    {
      "epoch": 0.47537967605786224,
      "grad_norm": 2.9918292356504854,
      "learning_rate": 2.2558772983125707e-06,
      "loss": 1.0042,
      "step": 6589
    },
    {
      "epoch": 0.47545182352728976,
      "grad_norm": 3.659590670308162,
      "learning_rate": 2.2554137722426263e-06,
      "loss": 1.0541,
      "step": 6590
    },
    {
      "epoch": 0.4755239709967173,
      "grad_norm": 4.401018649009089,
      "learning_rate": 2.254950232225481e-06,
      "loss": 0.9198,
      "step": 6591
    },
    {
      "epoch": 0.4755961184661448,
      "grad_norm": 4.585996143284543,
      "learning_rate": 2.254486678286448e-06,
      "loss": 1.1053,
      "step": 6592
    },
    {
      "epoch": 0.4756682659355723,
      "grad_norm": 4.492166907221821,
      "learning_rate": 2.2540231104508376e-06,
      "loss": 1.0647,
      "step": 6593
    },
    {
      "epoch": 0.47574041340499984,
      "grad_norm": 2.012786404150756,
      "learning_rate": 2.2535595287439677e-06,
      "loss": 1.0618,
      "step": 6594
    },
    {
      "epoch": 0.4758125608744273,
      "grad_norm": 2.97699980639738,
      "learning_rate": 2.25309593319115e-06,
      "loss": 1.0108,
      "step": 6595
    },
    {
      "epoch": 0.4758847083438548,
      "grad_norm": 0.6837668935303247,
      "learning_rate": 2.2526323238176992e-06,
      "loss": 0.839,
      "step": 6596
    },
    {
      "epoch": 0.47595685581328234,
      "grad_norm": 2.564079728147103,
      "learning_rate": 2.2521687006489337e-06,
      "loss": 1.0765,
      "step": 6597
    },
    {
      "epoch": 0.47602900328270986,
      "grad_norm": 2.6191940586633238,
      "learning_rate": 2.2517050637101683e-06,
      "loss": 1.0144,
      "step": 6598
    },
    {
      "epoch": 0.4761011507521374,
      "grad_norm": 2.3447373408618364,
      "learning_rate": 2.2512414130267213e-06,
      "loss": 0.96,
      "step": 6599
    },
    {
      "epoch": 0.4761732982215649,
      "grad_norm": 2.632179252016491,
      "learning_rate": 2.2507777486239116e-06,
      "loss": 0.9927,
      "step": 6600
    },
    {
      "epoch": 0.47624544569099236,
      "grad_norm": 2.2293934810675555,
      "learning_rate": 2.2503140705270564e-06,
      "loss": 0.9903,
      "step": 6601
    },
    {
      "epoch": 0.4763175931604199,
      "grad_norm": 3.649269809582109,
      "learning_rate": 2.2498503787614767e-06,
      "loss": 0.9868,
      "step": 6602
    },
    {
      "epoch": 0.4763897406298474,
      "grad_norm": 2.0385601257933526,
      "learning_rate": 2.249386673352492e-06,
      "loss": 1.0427,
      "step": 6603
    },
    {
      "epoch": 0.4764618880992749,
      "grad_norm": 5.2744049315151935,
      "learning_rate": 2.2489229543254256e-06,
      "loss": 0.941,
      "step": 6604
    },
    {
      "epoch": 0.47653403556870244,
      "grad_norm": 3.8622374016341916,
      "learning_rate": 2.2484592217055983e-06,
      "loss": 0.9153,
      "step": 6605
    },
    {
      "epoch": 0.47660618303812996,
      "grad_norm": 2.7944581228856378,
      "learning_rate": 2.2479954755183313e-06,
      "loss": 1.012,
      "step": 6606
    },
    {
      "epoch": 0.4766783305075574,
      "grad_norm": 3.1753037435063423,
      "learning_rate": 2.2475317157889502e-06,
      "loss": 0.9172,
      "step": 6607
    },
    {
      "epoch": 0.47675047797698494,
      "grad_norm": 2.4962041648358704,
      "learning_rate": 2.2470679425427775e-06,
      "loss": 0.9417,
      "step": 6608
    },
    {
      "epoch": 0.47682262544641246,
      "grad_norm": 2.403547974506266,
      "learning_rate": 2.24660415580514e-06,
      "loss": 0.9569,
      "step": 6609
    },
    {
      "epoch": 0.47689477291584,
      "grad_norm": 5.073404318721857,
      "learning_rate": 2.246140355601363e-06,
      "loss": 0.9698,
      "step": 6610
    },
    {
      "epoch": 0.4769669203852675,
      "grad_norm": 3.1277501021659515,
      "learning_rate": 2.245676541956771e-06,
      "loss": 0.9394,
      "step": 6611
    },
    {
      "epoch": 0.477039067854695,
      "grad_norm": 3.232770143861074,
      "learning_rate": 2.245212714896694e-06,
      "loss": 1.0233,
      "step": 6612
    },
    {
      "epoch": 0.4771112153241225,
      "grad_norm": 5.828587081631711,
      "learning_rate": 2.244748874446457e-06,
      "loss": 0.8509,
      "step": 6613
    },
    {
      "epoch": 0.47718336279355,
      "grad_norm": 5.303839771698664,
      "learning_rate": 2.2442850206313904e-06,
      "loss": 0.9605,
      "step": 6614
    },
    {
      "epoch": 0.4772555102629775,
      "grad_norm": 3.5485845134722904,
      "learning_rate": 2.243821153476823e-06,
      "loss": 0.8851,
      "step": 6615
    },
    {
      "epoch": 0.47732765773240504,
      "grad_norm": 3.31838682094852,
      "learning_rate": 2.2433572730080857e-06,
      "loss": 0.9975,
      "step": 6616
    },
    {
      "epoch": 0.47739980520183256,
      "grad_norm": 4.4022183201547,
      "learning_rate": 2.2428933792505087e-06,
      "loss": 0.9752,
      "step": 6617
    },
    {
      "epoch": 0.4774719526712601,
      "grad_norm": 2.6102299917035325,
      "learning_rate": 2.242429472229422e-06,
      "loss": 0.964,
      "step": 6618
    },
    {
      "epoch": 0.47754410014068754,
      "grad_norm": 2.6635310294533534,
      "learning_rate": 2.2419655519701606e-06,
      "loss": 0.9847,
      "step": 6619
    },
    {
      "epoch": 0.47761624761011506,
      "grad_norm": 2.5103799862114298,
      "learning_rate": 2.241501618498055e-06,
      "loss": 0.9786,
      "step": 6620
    },
    {
      "epoch": 0.4776883950795426,
      "grad_norm": 4.038835116982724,
      "learning_rate": 2.2410376718384413e-06,
      "loss": 0.9501,
      "step": 6621
    },
    {
      "epoch": 0.4777605425489701,
      "grad_norm": 5.444552003139705,
      "learning_rate": 2.2405737120166524e-06,
      "loss": 1.0175,
      "step": 6622
    },
    {
      "epoch": 0.4778326900183976,
      "grad_norm": 3.223440659743816,
      "learning_rate": 2.2401097390580224e-06,
      "loss": 0.9298,
      "step": 6623
    },
    {
      "epoch": 0.47790483748782514,
      "grad_norm": 5.161973400829293,
      "learning_rate": 2.23964575298789e-06,
      "loss": 0.9723,
      "step": 6624
    },
    {
      "epoch": 0.4779769849572526,
      "grad_norm": 4.887020062153482,
      "learning_rate": 2.2391817538315893e-06,
      "loss": 1.0912,
      "step": 6625
    },
    {
      "epoch": 0.4780491324266801,
      "grad_norm": 12.001740647275009,
      "learning_rate": 2.2387177416144586e-06,
      "loss": 0.9068,
      "step": 6626
    },
    {
      "epoch": 0.47812127989610764,
      "grad_norm": 7.779158590755619,
      "learning_rate": 2.2382537163618364e-06,
      "loss": 0.9853,
      "step": 6627
    },
    {
      "epoch": 0.47819342736553516,
      "grad_norm": 4.611925977432148,
      "learning_rate": 2.2377896780990597e-06,
      "loss": 1.0514,
      "step": 6628
    },
    {
      "epoch": 0.4782655748349627,
      "grad_norm": 3.083886191488464,
      "learning_rate": 2.2373256268514694e-06,
      "loss": 0.9666,
      "step": 6629
    },
    {
      "epoch": 0.4783377223043902,
      "grad_norm": 3.282783949018366,
      "learning_rate": 2.236861562644405e-06,
      "loss": 0.9343,
      "step": 6630
    },
    {
      "epoch": 0.47840986977381766,
      "grad_norm": 3.2565214311224584,
      "learning_rate": 2.2363974855032074e-06,
      "loss": 0.9526,
      "step": 6631
    },
    {
      "epoch": 0.4784820172432452,
      "grad_norm": 0.8784114207627507,
      "learning_rate": 2.235933395453219e-06,
      "loss": 0.9426,
      "step": 6632
    },
    {
      "epoch": 0.4785541647126727,
      "grad_norm": 3.3718040310353232,
      "learning_rate": 2.23546929251978e-06,
      "loss": 0.8919,
      "step": 6633
    },
    {
      "epoch": 0.4786263121821002,
      "grad_norm": 3.282739646281154,
      "learning_rate": 2.2350051767282356e-06,
      "loss": 0.9206,
      "step": 6634
    },
    {
      "epoch": 0.47869845965152774,
      "grad_norm": 2.778089978369578,
      "learning_rate": 2.2345410481039278e-06,
      "loss": 1.021,
      "step": 6635
    },
    {
      "epoch": 0.4787706071209552,
      "grad_norm": 2.216324568525128,
      "learning_rate": 2.2340769066722026e-06,
      "loss": 0.951,
      "step": 6636
    },
    {
      "epoch": 0.4788427545903827,
      "grad_norm": 2.5724656382824715,
      "learning_rate": 2.233612752458403e-06,
      "loss": 1.0142,
      "step": 6637
    },
    {
      "epoch": 0.47891490205981024,
      "grad_norm": 3.5411712898423744,
      "learning_rate": 2.2331485854878764e-06,
      "loss": 0.9437,
      "step": 6638
    },
    {
      "epoch": 0.47898704952923776,
      "grad_norm": 3.1513850421712983,
      "learning_rate": 2.2326844057859685e-06,
      "loss": 1.0437,
      "step": 6639
    },
    {
      "epoch": 0.4790591969986653,
      "grad_norm": 3.3404805964920006,
      "learning_rate": 2.2322202133780265e-06,
      "loss": 1.0161,
      "step": 6640
    },
    {
      "epoch": 0.4791313444680928,
      "grad_norm": 3.4926201810481685,
      "learning_rate": 2.2317560082893986e-06,
      "loss": 0.9403,
      "step": 6641
    },
    {
      "epoch": 0.47920349193752027,
      "grad_norm": 3.415583151851513,
      "learning_rate": 2.2312917905454323e-06,
      "loss": 0.971,
      "step": 6642
    },
    {
      "epoch": 0.4792756394069478,
      "grad_norm": 2.950962142510188,
      "learning_rate": 2.230827560171478e-06,
      "loss": 0.9187,
      "step": 6643
    },
    {
      "epoch": 0.4793477868763753,
      "grad_norm": 2.5116676811852416,
      "learning_rate": 2.2303633171928856e-06,
      "loss": 0.9376,
      "step": 6644
    },
    {
      "epoch": 0.4794199343458028,
      "grad_norm": 4.124300117073014,
      "learning_rate": 2.229899061635004e-06,
      "loss": 0.9753,
      "step": 6645
    },
    {
      "epoch": 0.47949208181523034,
      "grad_norm": 4.567396071347283,
      "learning_rate": 2.229434793523187e-06,
      "loss": 0.8183,
      "step": 6646
    },
    {
      "epoch": 0.47956422928465786,
      "grad_norm": 4.503799529850116,
      "learning_rate": 2.228970512882784e-06,
      "loss": 1.0541,
      "step": 6647
    },
    {
      "epoch": 0.4796363767540853,
      "grad_norm": 8.839813110999708,
      "learning_rate": 2.2285062197391503e-06,
      "loss": 1.0298,
      "step": 6648
    },
    {
      "epoch": 0.47970852422351284,
      "grad_norm": 3.608826434571671,
      "learning_rate": 2.228041914117637e-06,
      "loss": 0.9728,
      "step": 6649
    },
    {
      "epoch": 0.47978067169294036,
      "grad_norm": 3.0935040000444505,
      "learning_rate": 2.227577596043599e-06,
      "loss": 1.08,
      "step": 6650
    },
    {
      "epoch": 0.4798528191623679,
      "grad_norm": 3.326963839995734,
      "learning_rate": 2.227113265542391e-06,
      "loss": 1.0096,
      "step": 6651
    },
    {
      "epoch": 0.4799249666317954,
      "grad_norm": 3.713207185224497,
      "learning_rate": 2.2266489226393682e-06,
      "loss": 1.0396,
      "step": 6652
    },
    {
      "epoch": 0.4799971141012229,
      "grad_norm": 3.1595272763284195,
      "learning_rate": 2.2261845673598873e-06,
      "loss": 0.9733,
      "step": 6653
    },
    {
      "epoch": 0.4800692615706504,
      "grad_norm": 3.319051920632678,
      "learning_rate": 2.225720199729304e-06,
      "loss": 0.907,
      "step": 6654
    },
    {
      "epoch": 0.4801414090400779,
      "grad_norm": 9.160473991690681,
      "learning_rate": 2.2252558197729765e-06,
      "loss": 1.0293,
      "step": 6655
    },
    {
      "epoch": 0.4802135565095054,
      "grad_norm": 4.4961018102393,
      "learning_rate": 2.2247914275162626e-06,
      "loss": 0.9998,
      "step": 6656
    },
    {
      "epoch": 0.48028570397893294,
      "grad_norm": 3.2867767292600467,
      "learning_rate": 2.22432702298452e-06,
      "loss": 0.9794,
      "step": 6657
    },
    {
      "epoch": 0.48035785144836046,
      "grad_norm": 3.343569741782585,
      "learning_rate": 2.22386260620311e-06,
      "loss": 0.971,
      "step": 6658
    },
    {
      "epoch": 0.480429998917788,
      "grad_norm": 2.9665276491261268,
      "learning_rate": 2.2233981771973915e-06,
      "loss": 0.9036,
      "step": 6659
    },
    {
      "epoch": 0.48050214638721545,
      "grad_norm": 3.286231483009728,
      "learning_rate": 2.2229337359927256e-06,
      "loss": 0.9241,
      "step": 6660
    },
    {
      "epoch": 0.48057429385664296,
      "grad_norm": 2.299406252807015,
      "learning_rate": 2.222469282614474e-06,
      "loss": 0.9959,
      "step": 6661
    },
    {
      "epoch": 0.4806464413260705,
      "grad_norm": 5.180001780093561,
      "learning_rate": 2.222004817087998e-06,
      "loss": 1.0362,
      "step": 6662
    },
    {
      "epoch": 0.480718588795498,
      "grad_norm": 3.6599592646432084,
      "learning_rate": 2.2215403394386607e-06,
      "loss": 1.0339,
      "step": 6663
    },
    {
      "epoch": 0.4807907362649255,
      "grad_norm": 3.7796382779420785,
      "learning_rate": 2.221075849691826e-06,
      "loss": 0.9698,
      "step": 6664
    },
    {
      "epoch": 0.48086288373435304,
      "grad_norm": 4.697981445587662,
      "learning_rate": 2.2206113478728574e-06,
      "loss": 0.9169,
      "step": 6665
    },
    {
      "epoch": 0.4809350312037805,
      "grad_norm": 4.193599756224391,
      "learning_rate": 2.2201468340071197e-06,
      "loss": 0.9956,
      "step": 6666
    },
    {
      "epoch": 0.481007178673208,
      "grad_norm": 3.5893530442809967,
      "learning_rate": 2.219682308119978e-06,
      "loss": 0.9859,
      "step": 6667
    },
    {
      "epoch": 0.48107932614263554,
      "grad_norm": 4.491374544306143,
      "learning_rate": 2.2192177702367997e-06,
      "loss": 0.9074,
      "step": 6668
    },
    {
      "epoch": 0.48115147361206306,
      "grad_norm": 3.857624475494356,
      "learning_rate": 2.218753220382949e-06,
      "loss": 1.0423,
      "step": 6669
    },
    {
      "epoch": 0.4812236210814906,
      "grad_norm": 6.5267166393472325,
      "learning_rate": 2.218288658583796e-06,
      "loss": 1.0473,
      "step": 6670
    },
    {
      "epoch": 0.4812957685509181,
      "grad_norm": 3.406632008261161,
      "learning_rate": 2.2178240848647074e-06,
      "loss": 0.995,
      "step": 6671
    },
    {
      "epoch": 0.48136791602034557,
      "grad_norm": 22.548317126220514,
      "learning_rate": 2.2173594992510507e-06,
      "loss": 0.9431,
      "step": 6672
    },
    {
      "epoch": 0.4814400634897731,
      "grad_norm": 6.24231827733783,
      "learning_rate": 2.216894901768198e-06,
      "loss": 0.9452,
      "step": 6673
    },
    {
      "epoch": 0.4815122109592006,
      "grad_norm": 4.264678514713894,
      "learning_rate": 2.2164302924415158e-06,
      "loss": 0.9532,
      "step": 6674
    },
    {
      "epoch": 0.4815843584286281,
      "grad_norm": 3.672812484212807,
      "learning_rate": 2.215965671296378e-06,
      "loss": 0.8798,
      "step": 6675
    },
    {
      "epoch": 0.48165650589805564,
      "grad_norm": 4.6674032084191035,
      "learning_rate": 2.215501038358154e-06,
      "loss": 0.9851,
      "step": 6676
    },
    {
      "epoch": 0.48172865336748316,
      "grad_norm": 0.849472596131218,
      "learning_rate": 2.215036393652216e-06,
      "loss": 0.9156,
      "step": 6677
    },
    {
      "epoch": 0.4818008008369106,
      "grad_norm": 4.186565921284035,
      "learning_rate": 2.214571737203936e-06,
      "loss": 0.9551,
      "step": 6678
    },
    {
      "epoch": 0.48187294830633814,
      "grad_norm": 3.1134798950996845,
      "learning_rate": 2.2141070690386885e-06,
      "loss": 1.0018,
      "step": 6679
    },
    {
      "epoch": 0.48194509577576566,
      "grad_norm": 3.0753362580335155,
      "learning_rate": 2.2136423891818463e-06,
      "loss": 1.0185,
      "step": 6680
    },
    {
      "epoch": 0.4820172432451932,
      "grad_norm": 3.988979775347617,
      "learning_rate": 2.213177697658784e-06,
      "loss": 0.9629,
      "step": 6681
    },
    {
      "epoch": 0.4820893907146207,
      "grad_norm": 4.430629300440831,
      "learning_rate": 2.212712994494877e-06,
      "loss": 0.9461,
      "step": 6682
    },
    {
      "epoch": 0.4821615381840482,
      "grad_norm": 3.167811454508604,
      "learning_rate": 2.212248279715501e-06,
      "loss": 0.9841,
      "step": 6683
    },
    {
      "epoch": 0.4822336856534757,
      "grad_norm": 2.7179143213237316,
      "learning_rate": 2.2117835533460313e-06,
      "loss": 0.9319,
      "step": 6684
    },
    {
      "epoch": 0.4823058331229032,
      "grad_norm": 4.574345092592277,
      "learning_rate": 2.2113188154118466e-06,
      "loss": 1.0205,
      "step": 6685
    },
    {
      "epoch": 0.4823779805923307,
      "grad_norm": 3.15843865077246,
      "learning_rate": 2.210854065938323e-06,
      "loss": 1.0206,
      "step": 6686
    },
    {
      "epoch": 0.48245012806175824,
      "grad_norm": 2.7976360457733946,
      "learning_rate": 2.21038930495084e-06,
      "loss": 0.9434,
      "step": 6687
    },
    {
      "epoch": 0.48252227553118576,
      "grad_norm": 2.485299568157015,
      "learning_rate": 2.2099245324747762e-06,
      "loss": 0.9246,
      "step": 6688
    },
    {
      "epoch": 0.4825944230006132,
      "grad_norm": 3.1178956721226596,
      "learning_rate": 2.20945974853551e-06,
      "loss": 0.9026,
      "step": 6689
    },
    {
      "epoch": 0.48266657047004075,
      "grad_norm": 4.332093330399406,
      "learning_rate": 2.2089949531584224e-06,
      "loss": 0.9289,
      "step": 6690
    },
    {
      "epoch": 0.48273871793946826,
      "grad_norm": 4.40805903965343,
      "learning_rate": 2.2085301463688943e-06,
      "loss": 0.9554,
      "step": 6691
    },
    {
      "epoch": 0.4828108654088958,
      "grad_norm": 5.3037414156565355,
      "learning_rate": 2.208065328192307e-06,
      "loss": 0.9031,
      "step": 6692
    },
    {
      "epoch": 0.4828830128783233,
      "grad_norm": 5.070544221399896,
      "learning_rate": 2.207600498654042e-06,
      "loss": 0.9329,
      "step": 6693
    },
    {
      "epoch": 0.4829551603477508,
      "grad_norm": 6.3970178093411105,
      "learning_rate": 2.2071356577794826e-06,
      "loss": 0.9623,
      "step": 6694
    },
    {
      "epoch": 0.4830273078171783,
      "grad_norm": 0.6645789662479747,
      "learning_rate": 2.2066708055940118e-06,
      "loss": 0.8247,
      "step": 6695
    },
    {
      "epoch": 0.4830994552866058,
      "grad_norm": 0.7449343795807029,
      "learning_rate": 2.2062059421230124e-06,
      "loss": 0.8269,
      "step": 6696
    },
    {
      "epoch": 0.4831716027560333,
      "grad_norm": 4.612953581385635,
      "learning_rate": 2.2057410673918712e-06,
      "loss": 0.9444,
      "step": 6697
    },
    {
      "epoch": 0.48324375022546084,
      "grad_norm": 0.799351565007812,
      "learning_rate": 2.2052761814259718e-06,
      "loss": 0.8262,
      "step": 6698
    },
    {
      "epoch": 0.48331589769488836,
      "grad_norm": 4.358803769768114,
      "learning_rate": 2.204811284250699e-06,
      "loss": 0.9474,
      "step": 6699
    },
    {
      "epoch": 0.4833880451643159,
      "grad_norm": 6.3436336412704515,
      "learning_rate": 2.204346375891441e-06,
      "loss": 1.0263,
      "step": 6700
    },
    {
      "epoch": 0.48346019263374335,
      "grad_norm": 8.543460812589357,
      "learning_rate": 2.2038814563735837e-06,
      "loss": 1.0031,
      "step": 6701
    },
    {
      "epoch": 0.48353234010317087,
      "grad_norm": 16.389253658969555,
      "learning_rate": 2.2034165257225145e-06,
      "loss": 1.1022,
      "step": 6702
    },
    {
      "epoch": 0.4836044875725984,
      "grad_norm": 3.6110822594949106,
      "learning_rate": 2.202951583963622e-06,
      "loss": 0.9129,
      "step": 6703
    },
    {
      "epoch": 0.4836766350420259,
      "grad_norm": 5.876606822943393,
      "learning_rate": 2.202486631122295e-06,
      "loss": 0.9476,
      "step": 6704
    },
    {
      "epoch": 0.4837487825114534,
      "grad_norm": 3.4534587439909794,
      "learning_rate": 2.202021667223923e-06,
      "loss": 0.9694,
      "step": 6705
    },
    {
      "epoch": 0.48382092998088094,
      "grad_norm": 5.94614944883008,
      "learning_rate": 2.2015566922938944e-06,
      "loss": 0.9828,
      "step": 6706
    },
    {
      "epoch": 0.4838930774503084,
      "grad_norm": 5.201691117763447,
      "learning_rate": 2.2010917063576022e-06,
      "loss": 1.0069,
      "step": 6707
    },
    {
      "epoch": 0.4839652249197359,
      "grad_norm": 4.096427915199865,
      "learning_rate": 2.2006267094404355e-06,
      "loss": 0.8354,
      "step": 6708
    },
    {
      "epoch": 0.48403737238916344,
      "grad_norm": 0.7107152486645377,
      "learning_rate": 2.200161701567788e-06,
      "loss": 0.8305,
      "step": 6709
    },
    {
      "epoch": 0.48410951985859096,
      "grad_norm": 3.1918088922775087,
      "learning_rate": 2.1996966827650502e-06,
      "loss": 0.8799,
      "step": 6710
    },
    {
      "epoch": 0.4841816673280185,
      "grad_norm": 2.878296330723446,
      "learning_rate": 2.199231653057615e-06,
      "loss": 0.925,
      "step": 6711
    },
    {
      "epoch": 0.484253814797446,
      "grad_norm": 4.033854508158811,
      "learning_rate": 2.1987666124708784e-06,
      "loss": 0.8949,
      "step": 6712
    },
    {
      "epoch": 0.48432596226687347,
      "grad_norm": 5.115783778425603,
      "learning_rate": 2.1983015610302317e-06,
      "loss": 0.8808,
      "step": 6713
    },
    {
      "epoch": 0.484398109736301,
      "grad_norm": 5.306477083305319,
      "learning_rate": 2.197836498761071e-06,
      "loss": 1.0343,
      "step": 6714
    },
    {
      "epoch": 0.4844702572057285,
      "grad_norm": 5.009363562082427,
      "learning_rate": 2.1973714256887923e-06,
      "loss": 0.9287,
      "step": 6715
    },
    {
      "epoch": 0.484542404675156,
      "grad_norm": 3.304592122173795,
      "learning_rate": 2.1969063418387897e-06,
      "loss": 1.0204,
      "step": 6716
    },
    {
      "epoch": 0.48461455214458354,
      "grad_norm": 2.132696351182905,
      "learning_rate": 2.196441247236461e-06,
      "loss": 0.9298,
      "step": 6717
    },
    {
      "epoch": 0.48468669961401106,
      "grad_norm": 4.154660257893599,
      "learning_rate": 2.195976141907203e-06,
      "loss": 1.021,
      "step": 6718
    },
    {
      "epoch": 0.4847588470834385,
      "grad_norm": 3.689313151832953,
      "learning_rate": 2.1955110258764136e-06,
      "loss": 0.9965,
      "step": 6719
    },
    {
      "epoch": 0.48483099455286605,
      "grad_norm": 0.6402835866600697,
      "learning_rate": 2.1950458991694908e-06,
      "loss": 0.8126,
      "step": 6720
    },
    {
      "epoch": 0.48490314202229357,
      "grad_norm": 8.46023277665892,
      "learning_rate": 2.1945807618118333e-06,
      "loss": 1.0015,
      "step": 6721
    },
    {
      "epoch": 0.4849752894917211,
      "grad_norm": 3.405085093237392,
      "learning_rate": 2.194115613828841e-06,
      "loss": 0.9035,
      "step": 6722
    },
    {
      "epoch": 0.4850474369611486,
      "grad_norm": 3.7908713516226347,
      "learning_rate": 2.193650455245913e-06,
      "loss": 0.9691,
      "step": 6723
    },
    {
      "epoch": 0.4851195844305761,
      "grad_norm": 8.371473338483113,
      "learning_rate": 2.193185286088451e-06,
      "loss": 1.0063,
      "step": 6724
    },
    {
      "epoch": 0.4851917319000036,
      "grad_norm": 3.7976711364718003,
      "learning_rate": 2.1927201063818552e-06,
      "loss": 1.058,
      "step": 6725
    },
    {
      "epoch": 0.4852638793694311,
      "grad_norm": 6.105629946288558,
      "learning_rate": 2.192254916151528e-06,
      "loss": 0.8464,
      "step": 6726
    },
    {
      "epoch": 0.4853360268388586,
      "grad_norm": 5.204669528404502,
      "learning_rate": 2.1917897154228723e-06,
      "loss": 1.0312,
      "step": 6727
    },
    {
      "epoch": 0.48540817430828614,
      "grad_norm": 6.460728191043555,
      "learning_rate": 2.1913245042212895e-06,
      "loss": 0.9247,
      "step": 6728
    },
    {
      "epoch": 0.48548032177771366,
      "grad_norm": 5.573091517287204,
      "learning_rate": 2.190859282572184e-06,
      "loss": 1.1192,
      "step": 6729
    },
    {
      "epoch": 0.4855524692471412,
      "grad_norm": 3.8268937602651,
      "learning_rate": 2.1903940505009595e-06,
      "loss": 0.9025,
      "step": 6730
    },
    {
      "epoch": 0.48562461671656865,
      "grad_norm": 5.477520520142666,
      "learning_rate": 2.1899288080330206e-06,
      "loss": 0.9364,
      "step": 6731
    },
    {
      "epoch": 0.48569676418599617,
      "grad_norm": 2.4514248538838594,
      "learning_rate": 2.1894635551937728e-06,
      "loss": 1.1121,
      "step": 6732
    },
    {
      "epoch": 0.4857689116554237,
      "grad_norm": 4.407717629445849,
      "learning_rate": 2.1889982920086214e-06,
      "loss": 0.9497,
      "step": 6733
    },
    {
      "epoch": 0.4858410591248512,
      "grad_norm": 14.612134181290262,
      "learning_rate": 2.1885330185029733e-06,
      "loss": 1.0295,
      "step": 6734
    },
    {
      "epoch": 0.4859132065942787,
      "grad_norm": 2.947846230967318,
      "learning_rate": 2.1880677347022343e-06,
      "loss": 0.9681,
      "step": 6735
    },
    {
      "epoch": 0.48598535406370624,
      "grad_norm": 4.743076900932472,
      "learning_rate": 2.1876024406318133e-06,
      "loss": 1.0258,
      "step": 6736
    },
    {
      "epoch": 0.4860575015331337,
      "grad_norm": 2.8439180093956717,
      "learning_rate": 2.1871371363171175e-06,
      "loss": 1.0541,
      "step": 6737
    },
    {
      "epoch": 0.4861296490025612,
      "grad_norm": 3.6856114352039544,
      "learning_rate": 2.1866718217835545e-06,
      "loss": 0.9384,
      "step": 6738
    },
    {
      "epoch": 0.48620179647198875,
      "grad_norm": 3.0125056919661444,
      "learning_rate": 2.1862064970565353e-06,
      "loss": 0.9902,
      "step": 6739
    },
    {
      "epoch": 0.48627394394141626,
      "grad_norm": 6.290411346036422,
      "learning_rate": 2.185741162161468e-06,
      "loss": 0.9151,
      "step": 6740
    },
    {
      "epoch": 0.4863460914108438,
      "grad_norm": 4.042161000227207,
      "learning_rate": 2.1852758171237637e-06,
      "loss": 0.9866,
      "step": 6741
    },
    {
      "epoch": 0.48641823888027125,
      "grad_norm": 2.584470303864294,
      "learning_rate": 2.184810461968833e-06,
      "loss": 0.9063,
      "step": 6742
    },
    {
      "epoch": 0.48649038634969877,
      "grad_norm": 3.146146712367716,
      "learning_rate": 2.1843450967220872e-06,
      "loss": 1.0035,
      "step": 6743
    },
    {
      "epoch": 0.4865625338191263,
      "grad_norm": 3.6193652590246903,
      "learning_rate": 2.1838797214089373e-06,
      "loss": 1.0806,
      "step": 6744
    },
    {
      "epoch": 0.4866346812885538,
      "grad_norm": 3.617528648025764,
      "learning_rate": 2.183414336054797e-06,
      "loss": 0.92,
      "step": 6745
    },
    {
      "epoch": 0.4867068287579813,
      "grad_norm": 4.674154790676294,
      "learning_rate": 2.1829489406850798e-06,
      "loss": 0.9855,
      "step": 6746
    },
    {
      "epoch": 0.48677897622740884,
      "grad_norm": 5.67378375090221,
      "learning_rate": 2.1824835353251967e-06,
      "loss": 0.8605,
      "step": 6747
    },
    {
      "epoch": 0.4868511236968363,
      "grad_norm": 3.547930186602999,
      "learning_rate": 2.182018120000564e-06,
      "loss": 0.9759,
      "step": 6748
    },
    {
      "epoch": 0.4869232711662638,
      "grad_norm": 2.70896175625358,
      "learning_rate": 2.1815526947365957e-06,
      "loss": 0.9294,
      "step": 6749
    },
    {
      "epoch": 0.48699541863569135,
      "grad_norm": 2.8498944982442853,
      "learning_rate": 2.1810872595587057e-06,
      "loss": 0.8647,
      "step": 6750
    },
    {
      "epoch": 0.48706756610511887,
      "grad_norm": 3.8151683537244776,
      "learning_rate": 2.180621814492312e-06,
      "loss": 0.9772,
      "step": 6751
    },
    {
      "epoch": 0.4871397135745464,
      "grad_norm": 3.324283298892251,
      "learning_rate": 2.1801563595628293e-06,
      "loss": 0.9671,
      "step": 6752
    },
    {
      "epoch": 0.4872118610439739,
      "grad_norm": 5.473981872372478,
      "learning_rate": 2.1796908947956747e-06,
      "loss": 1.0292,
      "step": 6753
    },
    {
      "epoch": 0.48728400851340137,
      "grad_norm": 7.425359225333421,
      "learning_rate": 2.179225420216266e-06,
      "loss": 0.9681,
      "step": 6754
    },
    {
      "epoch": 0.4873561559828289,
      "grad_norm": 4.0630722816566776,
      "learning_rate": 2.17875993585002e-06,
      "loss": 1.0664,
      "step": 6755
    },
    {
      "epoch": 0.4874283034522564,
      "grad_norm": 2.792119250453397,
      "learning_rate": 2.1782944417223556e-06,
      "loss": 0.9931,
      "step": 6756
    },
    {
      "epoch": 0.4875004509216839,
      "grad_norm": 3.7593800532813604,
      "learning_rate": 2.177828937858692e-06,
      "loss": 1.028,
      "step": 6757
    },
    {
      "epoch": 0.48757259839111144,
      "grad_norm": 5.399000124726963,
      "learning_rate": 2.1773634242844485e-06,
      "loss": 0.9525,
      "step": 6758
    },
    {
      "epoch": 0.48764474586053896,
      "grad_norm": 3.307665049384302,
      "learning_rate": 2.176897901025045e-06,
      "loss": 1.073,
      "step": 6759
    },
    {
      "epoch": 0.4877168933299664,
      "grad_norm": 4.764125300337863,
      "learning_rate": 2.1764323681059013e-06,
      "loss": 0.9785,
      "step": 6760
    },
    {
      "epoch": 0.48778904079939395,
      "grad_norm": 3.1486259006518793,
      "learning_rate": 2.1759668255524402e-06,
      "loss": 0.9901,
      "step": 6761
    },
    {
      "epoch": 0.48786118826882147,
      "grad_norm": 4.5043487728451534,
      "learning_rate": 2.1755012733900807e-06,
      "loss": 0.947,
      "step": 6762
    },
    {
      "epoch": 0.487933335738249,
      "grad_norm": 4.514405821845927,
      "learning_rate": 2.1750357116442477e-06,
      "loss": 0.9595,
      "step": 6763
    },
    {
      "epoch": 0.4880054832076765,
      "grad_norm": 3.547629658410994,
      "learning_rate": 2.1745701403403624e-06,
      "loss": 0.9769,
      "step": 6764
    },
    {
      "epoch": 0.488077630677104,
      "grad_norm": 2.764350209376179,
      "learning_rate": 2.174104559503847e-06,
      "loss": 1.0079,
      "step": 6765
    },
    {
      "epoch": 0.4881497781465315,
      "grad_norm": 3.934774151613602,
      "learning_rate": 2.173638969160127e-06,
      "loss": 0.9429,
      "step": 6766
    },
    {
      "epoch": 0.488221925615959,
      "grad_norm": 2.4729936077314503,
      "learning_rate": 2.173173369334625e-06,
      "loss": 0.9539,
      "step": 6767
    },
    {
      "epoch": 0.4882940730853865,
      "grad_norm": 3.113245716155433,
      "learning_rate": 2.172707760052767e-06,
      "loss": 0.8414,
      "step": 6768
    },
    {
      "epoch": 0.48836622055481405,
      "grad_norm": 4.7696472203857585,
      "learning_rate": 2.172242141339977e-06,
      "loss": 0.9301,
      "step": 6769
    },
    {
      "epoch": 0.48843836802424156,
      "grad_norm": 3.548728021900987,
      "learning_rate": 2.171776513221682e-06,
      "loss": 0.9771,
      "step": 6770
    },
    {
      "epoch": 0.4885105154936691,
      "grad_norm": 3.081430741709862,
      "learning_rate": 2.171310875723307e-06,
      "loss": 0.937,
      "step": 6771
    },
    {
      "epoch": 0.48858266296309655,
      "grad_norm": 2.266674982317462,
      "learning_rate": 2.170845228870279e-06,
      "loss": 0.9862,
      "step": 6772
    },
    {
      "epoch": 0.48865481043252407,
      "grad_norm": 3.5500224260508366,
      "learning_rate": 2.1703795726880262e-06,
      "loss": 0.9695,
      "step": 6773
    },
    {
      "epoch": 0.4887269579019516,
      "grad_norm": 3.3966884979454908,
      "learning_rate": 2.1699139072019745e-06,
      "loss": 1.0617,
      "step": 6774
    },
    {
      "epoch": 0.4887991053713791,
      "grad_norm": 3.931253716447344,
      "learning_rate": 2.1694482324375543e-06,
      "loss": 1.0662,
      "step": 6775
    },
    {
      "epoch": 0.4888712528408066,
      "grad_norm": 6.6343398565366885,
      "learning_rate": 2.1689825484201937e-06,
      "loss": 0.9215,
      "step": 6776
    },
    {
      "epoch": 0.48894340031023414,
      "grad_norm": 6.167039430366354,
      "learning_rate": 2.168516855175321e-06,
      "loss": 1.0891,
      "step": 6777
    },
    {
      "epoch": 0.4890155477796616,
      "grad_norm": 2.364034132399613,
      "learning_rate": 2.168051152728367e-06,
      "loss": 0.9781,
      "step": 6778
    },
    {
      "epoch": 0.4890876952490891,
      "grad_norm": 3.768984360627772,
      "learning_rate": 2.1675854411047613e-06,
      "loss": 0.9204,
      "step": 6779
    },
    {
      "epoch": 0.48915984271851665,
      "grad_norm": 3.264443361697513,
      "learning_rate": 2.1671197203299354e-06,
      "loss": 0.9488,
      "step": 6780
    },
    {
      "epoch": 0.48923199018794417,
      "grad_norm": 4.063168984130923,
      "learning_rate": 2.1666539904293205e-06,
      "loss": 1.0427,
      "step": 6781
    },
    {
      "epoch": 0.4893041376573717,
      "grad_norm": 4.332333938577067,
      "learning_rate": 2.1661882514283475e-06,
      "loss": 1.0019,
      "step": 6782
    },
    {
      "epoch": 0.4893762851267992,
      "grad_norm": 4.762616117568704,
      "learning_rate": 2.165722503352449e-06,
      "loss": 0.9514,
      "step": 6783
    },
    {
      "epoch": 0.48944843259622667,
      "grad_norm": 3.214084095156775,
      "learning_rate": 2.1652567462270585e-06,
      "loss": 0.983,
      "step": 6784
    },
    {
      "epoch": 0.4895205800656542,
      "grad_norm": 2.6850245408227758,
      "learning_rate": 2.1647909800776093e-06,
      "loss": 0.8726,
      "step": 6785
    },
    {
      "epoch": 0.4895927275350817,
      "grad_norm": 2.81959461152591,
      "learning_rate": 2.1643252049295345e-06,
      "loss": 0.9411,
      "step": 6786
    },
    {
      "epoch": 0.4896648750045092,
      "grad_norm": 3.9240445523080565,
      "learning_rate": 2.163859420808268e-06,
      "loss": 0.8479,
      "step": 6787
    },
    {
      "epoch": 0.48973702247393675,
      "grad_norm": 3.3535685193484333,
      "learning_rate": 2.163393627739246e-06,
      "loss": 0.8849,
      "step": 6788
    },
    {
      "epoch": 0.4898091699433642,
      "grad_norm": 3.9579613092506225,
      "learning_rate": 2.162927825747902e-06,
      "loss": 0.9366,
      "step": 6789
    },
    {
      "epoch": 0.48988131741279173,
      "grad_norm": 4.512429766755639,
      "learning_rate": 2.162462014859673e-06,
      "loss": 0.9831,
      "step": 6790
    },
    {
      "epoch": 0.48995346488221925,
      "grad_norm": 2.500718871711831,
      "learning_rate": 2.1619961950999944e-06,
      "loss": 0.9939,
      "step": 6791
    },
    {
      "epoch": 0.49002561235164677,
      "grad_norm": 0.7544370768873507,
      "learning_rate": 2.1615303664943037e-06,
      "loss": 0.8962,
      "step": 6792
    },
    {
      "epoch": 0.4900977598210743,
      "grad_norm": 11.457398162595643,
      "learning_rate": 2.161064529068038e-06,
      "loss": 1.0072,
      "step": 6793
    },
    {
      "epoch": 0.4901699072905018,
      "grad_norm": 2.918466140046121,
      "learning_rate": 2.160598682846634e-06,
      "loss": 0.951,
      "step": 6794
    },
    {
      "epoch": 0.49024205475992927,
      "grad_norm": 14.534051834217689,
      "learning_rate": 2.16013282785553e-06,
      "loss": 0.9782,
      "step": 6795
    },
    {
      "epoch": 0.4903142022293568,
      "grad_norm": 5.919520410752086,
      "learning_rate": 2.159666964120166e-06,
      "loss": 0.9711,
      "step": 6796
    },
    {
      "epoch": 0.4903863496987843,
      "grad_norm": 4.770951293197766,
      "learning_rate": 2.15920109166598e-06,
      "loss": 1.0817,
      "step": 6797
    },
    {
      "epoch": 0.4904584971682118,
      "grad_norm": 3.4883731090928602,
      "learning_rate": 2.1587352105184115e-06,
      "loss": 0.9288,
      "step": 6798
    },
    {
      "epoch": 0.49053064463763935,
      "grad_norm": 9.164360958614909,
      "learning_rate": 2.158269320702901e-06,
      "loss": 0.9789,
      "step": 6799
    },
    {
      "epoch": 0.49060279210706687,
      "grad_norm": 5.374660924817741,
      "learning_rate": 2.1578034222448885e-06,
      "loss": 0.8821,
      "step": 6800
    },
    {
      "epoch": 0.49067493957649433,
      "grad_norm": 5.180276460376668,
      "learning_rate": 2.1573375151698148e-06,
      "loss": 1.0491,
      "step": 6801
    },
    {
      "epoch": 0.49074708704592185,
      "grad_norm": 19.26603684677391,
      "learning_rate": 2.156871599503123e-06,
      "loss": 0.9786,
      "step": 6802
    },
    {
      "epoch": 0.49081923451534937,
      "grad_norm": 4.259271718128306,
      "learning_rate": 2.1564056752702532e-06,
      "loss": 0.8424,
      "step": 6803
    },
    {
      "epoch": 0.4908913819847769,
      "grad_norm": 8.113718034854372,
      "learning_rate": 2.155939742496648e-06,
      "loss": 1.049,
      "step": 6804
    },
    {
      "epoch": 0.4909635294542044,
      "grad_norm": 7.360831665236957,
      "learning_rate": 2.155473801207752e-06,
      "loss": 0.9677,
      "step": 6805
    },
    {
      "epoch": 0.4910356769236319,
      "grad_norm": 4.089099834593843,
      "learning_rate": 2.155007851429006e-06,
      "loss": 0.8881,
      "step": 6806
    },
    {
      "epoch": 0.4911078243930594,
      "grad_norm": 4.5101141119686226,
      "learning_rate": 2.154541893185856e-06,
      "loss": 0.9273,
      "step": 6807
    },
    {
      "epoch": 0.4911799718624869,
      "grad_norm": 4.793005559977334,
      "learning_rate": 2.154075926503745e-06,
      "loss": 0.878,
      "step": 6808
    },
    {
      "epoch": 0.4912521193319144,
      "grad_norm": 4.9599527658243865,
      "learning_rate": 2.153609951408118e-06,
      "loss": 0.8798,
      "step": 6809
    },
    {
      "epoch": 0.49132426680134195,
      "grad_norm": 4.687121363288382,
      "learning_rate": 2.15314396792442e-06,
      "loss": 1.068,
      "step": 6810
    },
    {
      "epoch": 0.49139641427076947,
      "grad_norm": 4.253804691668201,
      "learning_rate": 2.1526779760780967e-06,
      "loss": 1.0445,
      "step": 6811
    },
    {
      "epoch": 0.491468561740197,
      "grad_norm": 4.328306779590267,
      "learning_rate": 2.152211975894595e-06,
      "loss": 0.9415,
      "step": 6812
    },
    {
      "epoch": 0.49154070920962445,
      "grad_norm": 4.496062357351309,
      "learning_rate": 2.1517459673993596e-06,
      "loss": 1.027,
      "step": 6813
    },
    {
      "epoch": 0.49161285667905197,
      "grad_norm": 5.825810113328661,
      "learning_rate": 2.15127995061784e-06,
      "loss": 0.9119,
      "step": 6814
    },
    {
      "epoch": 0.4916850041484795,
      "grad_norm": 2.887222514505351,
      "learning_rate": 2.150813925575482e-06,
      "loss": 0.995,
      "step": 6815
    },
    {
      "epoch": 0.491757151617907,
      "grad_norm": 6.780617117464482,
      "learning_rate": 2.150347892297733e-06,
      "loss": 0.9921,
      "step": 6816
    },
    {
      "epoch": 0.4918292990873345,
      "grad_norm": 5.533408568150383,
      "learning_rate": 2.149881850810043e-06,
      "loss": 1.0138,
      "step": 6817
    },
    {
      "epoch": 0.49190144655676205,
      "grad_norm": 3.936854354966569,
      "learning_rate": 2.14941580113786e-06,
      "loss": 0.8815,
      "step": 6818
    },
    {
      "epoch": 0.4919735940261895,
      "grad_norm": 2.5985052800700172,
      "learning_rate": 2.1489497433066327e-06,
      "loss": 0.9572,
      "step": 6819
    },
    {
      "epoch": 0.49204574149561703,
      "grad_norm": 2.7009253399729793,
      "learning_rate": 2.1484836773418115e-06,
      "loss": 0.9161,
      "step": 6820
    },
    {
      "epoch": 0.49211788896504455,
      "grad_norm": 3.5696346382368724,
      "learning_rate": 2.148017603268846e-06,
      "loss": 0.9349,
      "step": 6821
    },
    {
      "epoch": 0.49219003643447207,
      "grad_norm": 3.4256487378162905,
      "learning_rate": 2.1475515211131875e-06,
      "loss": 0.9603,
      "step": 6822
    },
    {
      "epoch": 0.4922621839038996,
      "grad_norm": 3.2363072028027693,
      "learning_rate": 2.147085430900287e-06,
      "loss": 0.9781,
      "step": 6823
    },
    {
      "epoch": 0.4923343313733271,
      "grad_norm": 4.164566553465927,
      "learning_rate": 2.146619332655595e-06,
      "loss": 1.0725,
      "step": 6824
    },
    {
      "epoch": 0.49240647884275457,
      "grad_norm": 6.975393370333922,
      "learning_rate": 2.1461532264045648e-06,
      "loss": 0.9633,
      "step": 6825
    },
    {
      "epoch": 0.4924786263121821,
      "grad_norm": 6.2389431616005,
      "learning_rate": 2.1456871121726472e-06,
      "loss": 0.9109,
      "step": 6826
    },
    {
      "epoch": 0.4925507737816096,
      "grad_norm": 3.1276080120654877,
      "learning_rate": 2.145220989985296e-06,
      "loss": 1.0475,
      "step": 6827
    },
    {
      "epoch": 0.4926229212510371,
      "grad_norm": 2.663846988854521,
      "learning_rate": 2.1447548598679636e-06,
      "loss": 0.9387,
      "step": 6828
    },
    {
      "epoch": 0.49269506872046465,
      "grad_norm": 3.3686511759514755,
      "learning_rate": 2.144288721846105e-06,
      "loss": 1.0875,
      "step": 6829
    },
    {
      "epoch": 0.49276721618989217,
      "grad_norm": 3.3296144403169254,
      "learning_rate": 2.1438225759451735e-06,
      "loss": 1.0172,
      "step": 6830
    },
    {
      "epoch": 0.49283936365931963,
      "grad_norm": 14.860020893231127,
      "learning_rate": 2.143356422190622e-06,
      "loss": 0.9944,
      "step": 6831
    },
    {
      "epoch": 0.49291151112874715,
      "grad_norm": 4.857413236321229,
      "learning_rate": 2.142890260607909e-06,
      "loss": 1.052,
      "step": 6832
    },
    {
      "epoch": 0.49298365859817467,
      "grad_norm": 4.431079356535608,
      "learning_rate": 2.1424240912224866e-06,
      "loss": 0.9037,
      "step": 6833
    },
    {
      "epoch": 0.4930558060676022,
      "grad_norm": 2.8946506820523195,
      "learning_rate": 2.141957914059812e-06,
      "loss": 1.0547,
      "step": 6834
    },
    {
      "epoch": 0.4931279535370297,
      "grad_norm": 0.9424393869129798,
      "learning_rate": 2.1414917291453412e-06,
      "loss": 0.8253,
      "step": 6835
    },
    {
      "epoch": 0.4932001010064572,
      "grad_norm": 3.8974488252677615,
      "learning_rate": 2.141025536504531e-06,
      "loss": 0.9141,
      "step": 6836
    },
    {
      "epoch": 0.4932722484758847,
      "grad_norm": 3.151352358967282,
      "learning_rate": 2.140559336162839e-06,
      "loss": 0.8098,
      "step": 6837
    },
    {
      "epoch": 0.4933443959453122,
      "grad_norm": 4.349151107566316,
      "learning_rate": 2.1400931281457207e-06,
      "loss": 0.9156,
      "step": 6838
    },
    {
      "epoch": 0.49341654341473973,
      "grad_norm": 3.09794763491009,
      "learning_rate": 2.1396269124786363e-06,
      "loss": 1.0147,
      "step": 6839
    },
    {
      "epoch": 0.49348869088416725,
      "grad_norm": 3.212780500107363,
      "learning_rate": 2.139160689187042e-06,
      "loss": 0.918,
      "step": 6840
    },
    {
      "epoch": 0.49356083835359477,
      "grad_norm": 3.411206480246968,
      "learning_rate": 2.138694458296399e-06,
      "loss": 1.037,
      "step": 6841
    },
    {
      "epoch": 0.49363298582302223,
      "grad_norm": 3.654295961592951,
      "learning_rate": 2.1382282198321642e-06,
      "loss": 1.0595,
      "step": 6842
    },
    {
      "epoch": 0.49370513329244975,
      "grad_norm": 5.777209710454885,
      "learning_rate": 2.137761973819798e-06,
      "loss": 1.0183,
      "step": 6843
    },
    {
      "epoch": 0.49377728076187727,
      "grad_norm": 3.673423278339721,
      "learning_rate": 2.1372957202847606e-06,
      "loss": 1.0145,
      "step": 6844
    },
    {
      "epoch": 0.4938494282313048,
      "grad_norm": 13.178650114774891,
      "learning_rate": 2.1368294592525122e-06,
      "loss": 0.8229,
      "step": 6845
    },
    {
      "epoch": 0.4939215757007323,
      "grad_norm": 4.633688523836868,
      "learning_rate": 2.136363190748513e-06,
      "loss": 0.9684,
      "step": 6846
    },
    {
      "epoch": 0.4939937231701598,
      "grad_norm": 3.527485824545664,
      "learning_rate": 2.135896914798226e-06,
      "loss": 0.967,
      "step": 6847
    },
    {
      "epoch": 0.4940658706395873,
      "grad_norm": 3.3584451497460157,
      "learning_rate": 2.1354306314271106e-06,
      "loss": 0.9133,
      "step": 6848
    },
    {
      "epoch": 0.4941380181090148,
      "grad_norm": 3.438180890706485,
      "learning_rate": 2.13496434066063e-06,
      "loss": 1.0069,
      "step": 6849
    },
    {
      "epoch": 0.49421016557844233,
      "grad_norm": 7.72372585023067,
      "learning_rate": 2.134498042524246e-06,
      "loss": 0.9603,
      "step": 6850
    },
    {
      "epoch": 0.49428231304786985,
      "grad_norm": 2.4965727201780945,
      "learning_rate": 2.1340317370434224e-06,
      "loss": 0.9929,
      "step": 6851
    },
    {
      "epoch": 0.49435446051729737,
      "grad_norm": 3.454771728040956,
      "learning_rate": 2.1335654242436216e-06,
      "loss": 0.9498,
      "step": 6852
    },
    {
      "epoch": 0.4944266079867249,
      "grad_norm": 13.05180483209123,
      "learning_rate": 2.133099104150307e-06,
      "loss": 0.873,
      "step": 6853
    },
    {
      "epoch": 0.49449875545615235,
      "grad_norm": 5.639604048120761,
      "learning_rate": 2.132632776788944e-06,
      "loss": 0.89,
      "step": 6854
    },
    {
      "epoch": 0.49457090292557987,
      "grad_norm": 4.313949562481778,
      "learning_rate": 2.1321664421849947e-06,
      "loss": 1.0764,
      "step": 6855
    },
    {
      "epoch": 0.4946430503950074,
      "grad_norm": 6.3074139037967525,
      "learning_rate": 2.1317001003639262e-06,
      "loss": 0.9795,
      "step": 6856
    },
    {
      "epoch": 0.4947151978644349,
      "grad_norm": 3.133128618762949,
      "learning_rate": 2.1312337513512025e-06,
      "loss": 1.0621,
      "step": 6857
    },
    {
      "epoch": 0.4947873453338624,
      "grad_norm": 2.5439190707619663,
      "learning_rate": 2.130767395172289e-06,
      "loss": 0.9479,
      "step": 6858
    },
    {
      "epoch": 0.49485949280328995,
      "grad_norm": 3.403781941656612,
      "learning_rate": 2.1303010318526532e-06,
      "loss": 1.0733,
      "step": 6859
    },
    {
      "epoch": 0.4949316402727174,
      "grad_norm": 2.506099508052842,
      "learning_rate": 2.12983466141776e-06,
      "loss": 1.0439,
      "step": 6860
    },
    {
      "epoch": 0.49500378774214493,
      "grad_norm": 3.9970672107855054,
      "learning_rate": 2.129368283893076e-06,
      "loss": 1.0391,
      "step": 6861
    },
    {
      "epoch": 0.49507593521157245,
      "grad_norm": 2.7735116868103837,
      "learning_rate": 2.1289018993040694e-06,
      "loss": 0.8973,
      "step": 6862
    },
    {
      "epoch": 0.49514808268099997,
      "grad_norm": 5.405043627256752,
      "learning_rate": 2.128435507676207e-06,
      "loss": 1.0216,
      "step": 6863
    },
    {
      "epoch": 0.4952202301504275,
      "grad_norm": 12.298270896390909,
      "learning_rate": 2.127969109034958e-06,
      "loss": 0.8672,
      "step": 6864
    },
    {
      "epoch": 0.495292377619855,
      "grad_norm": 3.1011229234804034,
      "learning_rate": 2.127502703405788e-06,
      "loss": 0.9377,
      "step": 6865
    },
    {
      "epoch": 0.49536452508928247,
      "grad_norm": 3.399854483014864,
      "learning_rate": 2.1270362908141688e-06,
      "loss": 0.9979,
      "step": 6866
    },
    {
      "epoch": 0.49543667255871,
      "grad_norm": 4.87751665934695,
      "learning_rate": 2.126569871285567e-06,
      "loss": 0.9621,
      "step": 6867
    },
    {
      "epoch": 0.4955088200281375,
      "grad_norm": 3.3045602327555144,
      "learning_rate": 2.1261034448454537e-06,
      "loss": 0.9688,
      "step": 6868
    },
    {
      "epoch": 0.49558096749756503,
      "grad_norm": 6.4294584843023275,
      "learning_rate": 2.1256370115192982e-06,
      "loss": 1.0179,
      "step": 6869
    },
    {
      "epoch": 0.49565311496699255,
      "grad_norm": 3.905520928055875,
      "learning_rate": 2.1251705713325698e-06,
      "loss": 0.9575,
      "step": 6870
    },
    {
      "epoch": 0.49572526243642007,
      "grad_norm": 3.7700541553604485,
      "learning_rate": 2.1247041243107407e-06,
      "loss": 1.0166,
      "step": 6871
    },
    {
      "epoch": 0.49579740990584753,
      "grad_norm": 3.7633171132101104,
      "learning_rate": 2.124237670479281e-06,
      "loss": 0.9001,
      "step": 6872
    },
    {
      "epoch": 0.49586955737527505,
      "grad_norm": 3.277608103769725,
      "learning_rate": 2.123771209863661e-06,
      "loss": 1.0764,
      "step": 6873
    },
    {
      "epoch": 0.49594170484470257,
      "grad_norm": 4.974033833142089,
      "learning_rate": 2.1233047424893543e-06,
      "loss": 0.9321,
      "step": 6874
    },
    {
      "epoch": 0.4960138523141301,
      "grad_norm": 3.894027509030832,
      "learning_rate": 2.122838268381832e-06,
      "loss": 0.952,
      "step": 6875
    },
    {
      "epoch": 0.4960859997835576,
      "grad_norm": 3.903554245593497,
      "learning_rate": 2.122371787566567e-06,
      "loss": 1.0347,
      "step": 6876
    },
    {
      "epoch": 0.4961581472529851,
      "grad_norm": 2.2837722630242996,
      "learning_rate": 2.1219053000690308e-06,
      "loss": 1.0625,
      "step": 6877
    },
    {
      "epoch": 0.4962302947224126,
      "grad_norm": 3.4178318890010035,
      "learning_rate": 2.121438805914698e-06,
      "loss": 0.9183,
      "step": 6878
    },
    {
      "epoch": 0.4963024421918401,
      "grad_norm": 6.477453596154281,
      "learning_rate": 2.1209723051290407e-06,
      "loss": 1.0353,
      "step": 6879
    },
    {
      "epoch": 0.49637458966126763,
      "grad_norm": 7.419948608960788,
      "learning_rate": 2.1205057977375345e-06,
      "loss": 0.928,
      "step": 6880
    },
    {
      "epoch": 0.49644673713069515,
      "grad_norm": 4.660654941425317,
      "learning_rate": 2.1200392837656533e-06,
      "loss": 1.0158,
      "step": 6881
    },
    {
      "epoch": 0.49651888460012267,
      "grad_norm": 4.767429898031182,
      "learning_rate": 2.1195727632388696e-06,
      "loss": 0.9274,
      "step": 6882
    },
    {
      "epoch": 0.4965910320695502,
      "grad_norm": 2.9239824008438107,
      "learning_rate": 2.1191062361826613e-06,
      "loss": 0.8827,
      "step": 6883
    },
    {
      "epoch": 0.49666317953897765,
      "grad_norm": 3.0024836114313866,
      "learning_rate": 2.1186397026225016e-06,
      "loss": 1.0122,
      "step": 6884
    },
    {
      "epoch": 0.49673532700840517,
      "grad_norm": 3.6501624136546487,
      "learning_rate": 2.118173162583867e-06,
      "loss": 0.9641,
      "step": 6885
    },
    {
      "epoch": 0.4968074744778327,
      "grad_norm": 3.1162543943812557,
      "learning_rate": 2.117706616092234e-06,
      "loss": 0.9646,
      "step": 6886
    },
    {
      "epoch": 0.4968796219472602,
      "grad_norm": 4.145161687736737,
      "learning_rate": 2.1172400631730785e-06,
      "loss": 0.9083,
      "step": 6887
    },
    {
      "epoch": 0.4969517694166877,
      "grad_norm": 0.743746234980438,
      "learning_rate": 2.1167735038518766e-06,
      "loss": 0.8355,
      "step": 6888
    },
    {
      "epoch": 0.49702391688611525,
      "grad_norm": 2.6766919890267302,
      "learning_rate": 2.1163069381541057e-06,
      "loss": 1.0423,
      "step": 6889
    },
    {
      "epoch": 0.4970960643555427,
      "grad_norm": 4.135533841232295,
      "learning_rate": 2.115840366105244e-06,
      "loss": 1.0016,
      "step": 6890
    },
    {
      "epoch": 0.49716821182497023,
      "grad_norm": 6.810314142764967,
      "learning_rate": 2.115373787730769e-06,
      "loss": 0.9825,
      "step": 6891
    },
    {
      "epoch": 0.49724035929439775,
      "grad_norm": 5.050705911379821,
      "learning_rate": 2.1149072030561577e-06,
      "loss": 0.862,
      "step": 6892
    },
    {
      "epoch": 0.49731250676382527,
      "grad_norm": 0.6744872255960778,
      "learning_rate": 2.1144406121068904e-06,
      "loss": 0.8221,
      "step": 6893
    },
    {
      "epoch": 0.4973846542332528,
      "grad_norm": 3.4484519581390103,
      "learning_rate": 2.113974014908444e-06,
      "loss": 1.0029,
      "step": 6894
    },
    {
      "epoch": 0.49745680170268025,
      "grad_norm": 9.99525911484786,
      "learning_rate": 2.113507411486299e-06,
      "loss": 0.9704,
      "step": 6895
    },
    {
      "epoch": 0.49752894917210777,
      "grad_norm": 2.8540356088514027,
      "learning_rate": 2.1130408018659345e-06,
      "loss": 0.9795,
      "step": 6896
    },
    {
      "epoch": 0.4976010966415353,
      "grad_norm": 3.7762780407114196,
      "learning_rate": 2.11257418607283e-06,
      "loss": 0.9809,
      "step": 6897
    },
    {
      "epoch": 0.4976732441109628,
      "grad_norm": 4.688661151757087,
      "learning_rate": 2.112107564132466e-06,
      "loss": 1.1078,
      "step": 6898
    },
    {
      "epoch": 0.49774539158039033,
      "grad_norm": 3.232928556163184,
      "learning_rate": 2.1116409360703233e-06,
      "loss": 1.0646,
      "step": 6899
    },
    {
      "epoch": 0.49781753904981785,
      "grad_norm": 2.8639085194453164,
      "learning_rate": 2.1111743019118823e-06,
      "loss": 0.9043,
      "step": 6900
    },
    {
      "epoch": 0.4978896865192453,
      "grad_norm": 0.7123777451951775,
      "learning_rate": 2.1107076616826236e-06,
      "loss": 0.9049,
      "step": 6901
    },
    {
      "epoch": 0.49796183398867283,
      "grad_norm": 4.833514221138917,
      "learning_rate": 2.1102410154080297e-06,
      "loss": 0.9201,
      "step": 6902
    },
    {
      "epoch": 0.49803398145810035,
      "grad_norm": 3.0839290215306296,
      "learning_rate": 2.1097743631135824e-06,
      "loss": 0.998,
      "step": 6903
    },
    {
      "epoch": 0.49810612892752787,
      "grad_norm": 4.2622794980826155,
      "learning_rate": 2.1093077048247626e-06,
      "loss": 0.9679,
      "step": 6904
    },
    {
      "epoch": 0.4981782763969554,
      "grad_norm": 0.7765567188316389,
      "learning_rate": 2.108841040567055e-06,
      "loss": 0.8222,
      "step": 6905
    },
    {
      "epoch": 0.4982504238663829,
      "grad_norm": 0.6646856637776007,
      "learning_rate": 2.1083743703659394e-06,
      "loss": 0.8933,
      "step": 6906
    },
    {
      "epoch": 0.49832257133581037,
      "grad_norm": 12.245486303714243,
      "learning_rate": 2.107907694246902e-06,
      "loss": 0.9717,
      "step": 6907
    },
    {
      "epoch": 0.4983947188052379,
      "grad_norm": 4.847658610746806,
      "learning_rate": 2.107441012235425e-06,
      "loss": 0.9409,
      "step": 6908
    },
    {
      "epoch": 0.4984668662746654,
      "grad_norm": 6.20912230255343,
      "learning_rate": 2.1069743243569904e-06,
      "loss": 0.9737,
      "step": 6909
    },
    {
      "epoch": 0.49853901374409293,
      "grad_norm": 7.603856022033071,
      "learning_rate": 2.1065076306370857e-06,
      "loss": 1.0433,
      "step": 6910
    },
    {
      "epoch": 0.49861116121352045,
      "grad_norm": 6.468616428931203,
      "learning_rate": 2.106040931101193e-06,
      "loss": 1.0415,
      "step": 6911
    },
    {
      "epoch": 0.49868330868294797,
      "grad_norm": 3.3888383368246298,
      "learning_rate": 2.1055742257747978e-06,
      "loss": 0.941,
      "step": 6912
    },
    {
      "epoch": 0.49875545615237543,
      "grad_norm": 3.0702106177434616,
      "learning_rate": 2.1051075146833844e-06,
      "loss": 1.0052,
      "step": 6913
    },
    {
      "epoch": 0.49882760362180295,
      "grad_norm": 4.083721194368558,
      "learning_rate": 2.104640797852439e-06,
      "loss": 0.9975,
      "step": 6914
    },
    {
      "epoch": 0.49889975109123047,
      "grad_norm": 6.241771622077961,
      "learning_rate": 2.1041740753074473e-06,
      "loss": 0.8972,
      "step": 6915
    },
    {
      "epoch": 0.498971898560658,
      "grad_norm": 3.6919533944706506,
      "learning_rate": 2.1037073470738942e-06,
      "loss": 0.9142,
      "step": 6916
    },
    {
      "epoch": 0.4990440460300855,
      "grad_norm": 4.776183591818764,
      "learning_rate": 2.1032406131772674e-06,
      "loss": 0.8904,
      "step": 6917
    },
    {
      "epoch": 0.49911619349951303,
      "grad_norm": 2.5017643424764984,
      "learning_rate": 2.1027738736430534e-06,
      "loss": 0.9536,
      "step": 6918
    },
    {
      "epoch": 0.4991883409689405,
      "grad_norm": 3.468362872541597,
      "learning_rate": 2.102307128496737e-06,
      "loss": 0.8911,
      "step": 6919
    },
    {
      "epoch": 0.499260488438368,
      "grad_norm": 5.082094114414702,
      "learning_rate": 2.1018403777638088e-06,
      "loss": 1.0713,
      "step": 6920
    },
    {
      "epoch": 0.49933263590779553,
      "grad_norm": 5.732921946352379,
      "learning_rate": 2.101373621469753e-06,
      "loss": 0.9172,
      "step": 6921
    },
    {
      "epoch": 0.49940478337722305,
      "grad_norm": 4.019949043292582,
      "learning_rate": 2.10090685964006e-06,
      "loss": 0.9663,
      "step": 6922
    },
    {
      "epoch": 0.49947693084665057,
      "grad_norm": 3.758689603532801,
      "learning_rate": 2.100440092300217e-06,
      "loss": 0.9375,
      "step": 6923
    },
    {
      "epoch": 0.4995490783160781,
      "grad_norm": 5.1435775857723165,
      "learning_rate": 2.0999733194757124e-06,
      "loss": 1.0264,
      "step": 6924
    },
    {
      "epoch": 0.49962122578550555,
      "grad_norm": 4.277178310870207,
      "learning_rate": 2.0995065411920345e-06,
      "loss": 0.962,
      "step": 6925
    },
    {
      "epoch": 0.49969337325493307,
      "grad_norm": 4.524816536740701,
      "learning_rate": 2.0990397574746724e-06,
      "loss": 0.9772,
      "step": 6926
    },
    {
      "epoch": 0.4997655207243606,
      "grad_norm": 3.7860402432180513,
      "learning_rate": 2.0985729683491172e-06,
      "loss": 0.9776,
      "step": 6927
    },
    {
      "epoch": 0.4998376681937881,
      "grad_norm": 3.999126935092328,
      "learning_rate": 2.0981061738408552e-06,
      "loss": 0.9144,
      "step": 6928
    },
    {
      "epoch": 0.49990981566321563,
      "grad_norm": 0.6595643228219945,
      "learning_rate": 2.0976393739753797e-06,
      "loss": 0.7971,
      "step": 6929
    },
    {
      "epoch": 0.49998196313264315,
      "grad_norm": 8.580261964866935,
      "learning_rate": 2.0971725687781794e-06,
      "loss": 0.9792,
      "step": 6930
    },
    {
      "epoch": 0.5000541106020706,
      "grad_norm": 3.2860016341759675,
      "learning_rate": 2.0967057582747433e-06,
      "loss": 0.9516,
      "step": 6931
    },
    {
      "epoch": 0.5001262580714981,
      "grad_norm": 6.237163942874435,
      "learning_rate": 2.0962389424905654e-06,
      "loss": 0.854,
      "step": 6932
    },
    {
      "epoch": 0.5001984055409257,
      "grad_norm": 5.0468239796424585,
      "learning_rate": 2.0957721214511343e-06,
      "loss": 1.0425,
      "step": 6933
    },
    {
      "epoch": 0.5002705530103532,
      "grad_norm": 2.5591035058456684,
      "learning_rate": 2.0953052951819426e-06,
      "loss": 0.9053,
      "step": 6934
    },
    {
      "epoch": 0.5003427004797807,
      "grad_norm": 5.6354684692183055,
      "learning_rate": 2.0948384637084814e-06,
      "loss": 0.9018,
      "step": 6935
    },
    {
      "epoch": 0.5004148479492082,
      "grad_norm": 9.521077114687609,
      "learning_rate": 2.0943716270562425e-06,
      "loss": 0.9793,
      "step": 6936
    },
    {
      "epoch": 0.5004869954186357,
      "grad_norm": 5.446978496152599,
      "learning_rate": 2.0939047852507187e-06,
      "loss": 0.9953,
      "step": 6937
    },
    {
      "epoch": 0.5005591428880632,
      "grad_norm": 0.6911598456159391,
      "learning_rate": 2.093437938317402e-06,
      "loss": 0.8313,
      "step": 6938
    },
    {
      "epoch": 0.5006312903574908,
      "grad_norm": 3.5072166429559837,
      "learning_rate": 2.0929710862817855e-06,
      "loss": 0.9438,
      "step": 6939
    },
    {
      "epoch": 0.5007034378269182,
      "grad_norm": 5.719864533427494,
      "learning_rate": 2.092504229169362e-06,
      "loss": 1.0058,
      "step": 6940
    },
    {
      "epoch": 0.5007755852963457,
      "grad_norm": 6.371480830611791,
      "learning_rate": 2.0920373670056253e-06,
      "loss": 1.1002,
      "step": 6941
    },
    {
      "epoch": 0.5008477327657732,
      "grad_norm": 4.45267438783722,
      "learning_rate": 2.0915704998160686e-06,
      "loss": 0.9321,
      "step": 6942
    },
    {
      "epoch": 0.5009198802352007,
      "grad_norm": 5.092662297295197,
      "learning_rate": 2.0911036276261855e-06,
      "loss": 1.0233,
      "step": 6943
    },
    {
      "epoch": 0.5009920277046283,
      "grad_norm": 2.899926638497428,
      "learning_rate": 2.0906367504614712e-06,
      "loss": 0.9716,
      "step": 6944
    },
    {
      "epoch": 0.5010641751740558,
      "grad_norm": 3.867679079196691,
      "learning_rate": 2.090169868347418e-06,
      "loss": 0.8611,
      "step": 6945
    },
    {
      "epoch": 0.5011363226434833,
      "grad_norm": 9.02791589354151,
      "learning_rate": 2.0897029813095243e-06,
      "loss": 1.0057,
      "step": 6946
    },
    {
      "epoch": 0.5012084701129108,
      "grad_norm": 3.5551412774837305,
      "learning_rate": 2.0892360893732817e-06,
      "loss": 1.0051,
      "step": 6947
    },
    {
      "epoch": 0.5012806175823383,
      "grad_norm": 2.85677164936083,
      "learning_rate": 2.088769192564187e-06,
      "loss": 0.9441,
      "step": 6948
    },
    {
      "epoch": 0.5013527650517658,
      "grad_norm": 9.244096676488072,
      "learning_rate": 2.0883022909077343e-06,
      "loss": 1.0536,
      "step": 6949
    },
    {
      "epoch": 0.5014249125211934,
      "grad_norm": 4.365839495949723,
      "learning_rate": 2.0878353844294213e-06,
      "loss": 1.0177,
      "step": 6950
    },
    {
      "epoch": 0.5014970599906208,
      "grad_norm": 3.2383185040547837,
      "learning_rate": 2.0873684731547428e-06,
      "loss": 1.0742,
      "step": 6951
    },
    {
      "epoch": 0.5015692074600483,
      "grad_norm": 3.9947345170861492,
      "learning_rate": 2.0869015571091956e-06,
      "loss": 0.9229,
      "step": 6952
    },
    {
      "epoch": 0.5016413549294758,
      "grad_norm": 5.477199109308316,
      "learning_rate": 2.086434636318276e-06,
      "loss": 1.0447,
      "step": 6953
    },
    {
      "epoch": 0.5017135023989033,
      "grad_norm": 3.366265332186714,
      "learning_rate": 2.0859677108074807e-06,
      "loss": 0.9964,
      "step": 6954
    },
    {
      "epoch": 0.5017856498683309,
      "grad_norm": 3.8913008627552794,
      "learning_rate": 2.085500780602307e-06,
      "loss": 0.925,
      "step": 6955
    },
    {
      "epoch": 0.5018577973377584,
      "grad_norm": 3.4973770940207123,
      "learning_rate": 2.0850338457282525e-06,
      "loss": 1.0885,
      "step": 6956
    },
    {
      "epoch": 0.5019299448071859,
      "grad_norm": 3.5290673186015016,
      "learning_rate": 2.0845669062108146e-06,
      "loss": 0.8889,
      "step": 6957
    },
    {
      "epoch": 0.5020020922766134,
      "grad_norm": 3.011090441918685,
      "learning_rate": 2.08409996207549e-06,
      "loss": 0.945,
      "step": 6958
    },
    {
      "epoch": 0.5020742397460409,
      "grad_norm": 6.610662528810877,
      "learning_rate": 2.0836330133477787e-06,
      "loss": 0.9419,
      "step": 6959
    },
    {
      "epoch": 0.5021463872154684,
      "grad_norm": 3.0463087289288726,
      "learning_rate": 2.083166060053178e-06,
      "loss": 1.0534,
      "step": 6960
    },
    {
      "epoch": 0.502218534684896,
      "grad_norm": 2.632674488106562,
      "learning_rate": 2.082699102217186e-06,
      "loss": 1.0053,
      "step": 6961
    },
    {
      "epoch": 0.5022906821543235,
      "grad_norm": 5.24592777537847,
      "learning_rate": 2.082232139865303e-06,
      "loss": 0.9224,
      "step": 6962
    },
    {
      "epoch": 0.5023628296237509,
      "grad_norm": 4.697864112080741,
      "learning_rate": 2.0817651730230273e-06,
      "loss": 0.8841,
      "step": 6963
    },
    {
      "epoch": 0.5024349770931784,
      "grad_norm": 4.9513797031702005,
      "learning_rate": 2.0812982017158576e-06,
      "loss": 0.9425,
      "step": 6964
    },
    {
      "epoch": 0.5025071245626059,
      "grad_norm": 5.109792149864742,
      "learning_rate": 2.0808312259692942e-06,
      "loss": 1.0084,
      "step": 6965
    },
    {
      "epoch": 0.5025792720320335,
      "grad_norm": 3.84559512628477,
      "learning_rate": 2.080364245808837e-06,
      "loss": 0.9215,
      "step": 6966
    },
    {
      "epoch": 0.502651419501461,
      "grad_norm": 3.89003519390435,
      "learning_rate": 2.079897261259985e-06,
      "loss": 1.1616,
      "step": 6967
    },
    {
      "epoch": 0.5027235669708885,
      "grad_norm": 2.99413282140821,
      "learning_rate": 2.0794302723482407e-06,
      "loss": 0.9521,
      "step": 6968
    },
    {
      "epoch": 0.502795714440316,
      "grad_norm": 3.3634502443750702,
      "learning_rate": 2.0789632790991027e-06,
      "loss": 0.9873,
      "step": 6969
    },
    {
      "epoch": 0.5028678619097435,
      "grad_norm": 2.6317358334104677,
      "learning_rate": 2.078496281538072e-06,
      "loss": 0.9051,
      "step": 6970
    },
    {
      "epoch": 0.502940009379171,
      "grad_norm": 5.324491542844193,
      "learning_rate": 2.07802927969065e-06,
      "loss": 0.8816,
      "step": 6971
    },
    {
      "epoch": 0.5030121568485986,
      "grad_norm": 2.6506064206951625,
      "learning_rate": 2.077562273582338e-06,
      "loss": 1.0191,
      "step": 6972
    },
    {
      "epoch": 0.5030843043180261,
      "grad_norm": 4.626164882467484,
      "learning_rate": 2.0770952632386374e-06,
      "loss": 1.0575,
      "step": 6973
    },
    {
      "epoch": 0.5031564517874536,
      "grad_norm": 5.43998766589169,
      "learning_rate": 2.0766282486850503e-06,
      "loss": 0.9604,
      "step": 6974
    },
    {
      "epoch": 0.503228599256881,
      "grad_norm": 2.6915075894836424,
      "learning_rate": 2.0761612299470777e-06,
      "loss": 0.9973,
      "step": 6975
    },
    {
      "epoch": 0.5033007467263085,
      "grad_norm": 4.593157671673223,
      "learning_rate": 2.075694207050222e-06,
      "loss": 0.9064,
      "step": 6976
    },
    {
      "epoch": 0.503372894195736,
      "grad_norm": 3.9070265341434496,
      "learning_rate": 2.075227180019986e-06,
      "loss": 0.9389,
      "step": 6977
    },
    {
      "epoch": 0.5034450416651636,
      "grad_norm": 3.886850387414348,
      "learning_rate": 2.0747601488818734e-06,
      "loss": 1.0013,
      "step": 6978
    },
    {
      "epoch": 0.5035171891345911,
      "grad_norm": 6.056334831344405,
      "learning_rate": 2.0742931136613846e-06,
      "loss": 0.9458,
      "step": 6979
    },
    {
      "epoch": 0.5035893366040186,
      "grad_norm": 3.039775382326793,
      "learning_rate": 2.0738260743840243e-06,
      "loss": 0.9445,
      "step": 6980
    },
    {
      "epoch": 0.5036614840734461,
      "grad_norm": 14.585972022254271,
      "learning_rate": 2.073359031075295e-06,
      "loss": 0.964,
      "step": 6981
    },
    {
      "epoch": 0.5037336315428736,
      "grad_norm": 3.208874264648848,
      "learning_rate": 2.0728919837607007e-06,
      "loss": 0.9808,
      "step": 6982
    },
    {
      "epoch": 0.5038057790123012,
      "grad_norm": 4.3854904833068336,
      "learning_rate": 2.0724249324657454e-06,
      "loss": 0.9684,
      "step": 6983
    },
    {
      "epoch": 0.5038779264817287,
      "grad_norm": 3.0337885029553497,
      "learning_rate": 2.0719578772159326e-06,
      "loss": 0.9933,
      "step": 6984
    },
    {
      "epoch": 0.5039500739511562,
      "grad_norm": 5.043276233001075,
      "learning_rate": 2.071490818036766e-06,
      "loss": 0.9265,
      "step": 6985
    },
    {
      "epoch": 0.5040222214205837,
      "grad_norm": 4.505283432997918,
      "learning_rate": 2.0710237549537508e-06,
      "loss": 0.8403,
      "step": 6986
    },
    {
      "epoch": 0.5040943688900111,
      "grad_norm": 3.6251123345168508,
      "learning_rate": 2.070556687992391e-06,
      "loss": 1.0297,
      "step": 6987
    },
    {
      "epoch": 0.5041665163594387,
      "grad_norm": 2.7577090149393717,
      "learning_rate": 2.0700896171781914e-06,
      "loss": 0.9582,
      "step": 6988
    },
    {
      "epoch": 0.5042386638288662,
      "grad_norm": 4.0387826966520946,
      "learning_rate": 2.069622542536658e-06,
      "loss": 1.0013,
      "step": 6989
    },
    {
      "epoch": 0.5043108112982937,
      "grad_norm": 4.978587650273426,
      "learning_rate": 2.069155464093295e-06,
      "loss": 0.9862,
      "step": 6990
    },
    {
      "epoch": 0.5043829587677212,
      "grad_norm": 5.546608502070805,
      "learning_rate": 2.068688381873608e-06,
      "loss": 0.9707,
      "step": 6991
    },
    {
      "epoch": 0.5044551062371487,
      "grad_norm": 9.498808936694132,
      "learning_rate": 2.068221295903103e-06,
      "loss": 0.8813,
      "step": 6992
    },
    {
      "epoch": 0.5045272537065763,
      "grad_norm": 0.8033296554396818,
      "learning_rate": 2.0677542062072855e-06,
      "loss": 0.8219,
      "step": 6993
    },
    {
      "epoch": 0.5045994011760038,
      "grad_norm": 4.540821568434222,
      "learning_rate": 2.067287112811661e-06,
      "loss": 0.9336,
      "step": 6994
    },
    {
      "epoch": 0.5046715486454313,
      "grad_norm": 2.7192206139863084,
      "learning_rate": 2.0668200157417376e-06,
      "loss": 0.9667,
      "step": 6995
    },
    {
      "epoch": 0.5047436961148588,
      "grad_norm": 4.898912908673472,
      "learning_rate": 2.06635291502302e-06,
      "loss": 1.0048,
      "step": 6996
    },
    {
      "epoch": 0.5048158435842863,
      "grad_norm": 7.85816536727582,
      "learning_rate": 2.0658858106810145e-06,
      "loss": 0.9745,
      "step": 6997
    },
    {
      "epoch": 0.5048879910537137,
      "grad_norm": 3.405358994166458,
      "learning_rate": 2.0654187027412297e-06,
      "loss": 1.0233,
      "step": 6998
    },
    {
      "epoch": 0.5049601385231413,
      "grad_norm": 2.385977022634347,
      "learning_rate": 2.0649515912291715e-06,
      "loss": 0.8739,
      "step": 6999
    },
    {
      "epoch": 0.5050322859925688,
      "grad_norm": 5.024584791954751,
      "learning_rate": 2.064484476170347e-06,
      "loss": 1.0336,
      "step": 7000
    },
    {
      "epoch": 0.5051044334619963,
      "grad_norm": 2.9469001194519433,
      "learning_rate": 2.0640173575902645e-06,
      "loss": 0.9843,
      "step": 7001
    },
    {
      "epoch": 0.5051765809314238,
      "grad_norm": 3.5345119837650723,
      "learning_rate": 2.0635502355144307e-06,
      "loss": 0.8839,
      "step": 7002
    },
    {
      "epoch": 0.5052487284008513,
      "grad_norm": 3.684025566219457,
      "learning_rate": 2.063083109968354e-06,
      "loss": 0.9213,
      "step": 7003
    },
    {
      "epoch": 0.5053208758702789,
      "grad_norm": 3.5313861111765927,
      "learning_rate": 2.062615980977542e-06,
      "loss": 1.0632,
      "step": 7004
    },
    {
      "epoch": 0.5053930233397064,
      "grad_norm": 4.335803990413558,
      "learning_rate": 2.0621488485675035e-06,
      "loss": 0.9803,
      "step": 7005
    },
    {
      "epoch": 0.5054651708091339,
      "grad_norm": 3.412291322692547,
      "learning_rate": 2.061681712763746e-06,
      "loss": 0.8494,
      "step": 7006
    },
    {
      "epoch": 0.5055373182785614,
      "grad_norm": 3.402325810802305,
      "learning_rate": 2.0612145735917784e-06,
      "loss": 1.0002,
      "step": 7007
    },
    {
      "epoch": 0.5056094657479889,
      "grad_norm": 3.3014939481132948,
      "learning_rate": 2.0607474310771103e-06,
      "loss": 0.9205,
      "step": 7008
    },
    {
      "epoch": 0.5056816132174164,
      "grad_norm": 2.722597578557666,
      "learning_rate": 2.0602802852452486e-06,
      "loss": 0.8584,
      "step": 7009
    },
    {
      "epoch": 0.5057537606868439,
      "grad_norm": 4.271366042742503,
      "learning_rate": 2.059813136121705e-06,
      "loss": 0.8818,
      "step": 7010
    },
    {
      "epoch": 0.5058259081562714,
      "grad_norm": 3.3863685717264533,
      "learning_rate": 2.0593459837319868e-06,
      "loss": 0.8425,
      "step": 7011
    },
    {
      "epoch": 0.5058980556256989,
      "grad_norm": 3.1482636283022765,
      "learning_rate": 2.058878828101604e-06,
      "loss": 0.9134,
      "step": 7012
    },
    {
      "epoch": 0.5059702030951264,
      "grad_norm": 7.611956587314342,
      "learning_rate": 2.058411669256067e-06,
      "loss": 0.9178,
      "step": 7013
    },
    {
      "epoch": 0.5060423505645539,
      "grad_norm": 9.917071379622875,
      "learning_rate": 2.057944507220884e-06,
      "loss": 0.9121,
      "step": 7014
    },
    {
      "epoch": 0.5061144980339815,
      "grad_norm": 2.5716616907110756,
      "learning_rate": 2.057477342021567e-06,
      "loss": 0.9502,
      "step": 7015
    },
    {
      "epoch": 0.506186645503409,
      "grad_norm": 3.035647790645367,
      "learning_rate": 2.057010173683624e-06,
      "loss": 0.9876,
      "step": 7016
    },
    {
      "epoch": 0.5062587929728365,
      "grad_norm": 4.729022030852271,
      "learning_rate": 2.0565430022325678e-06,
      "loss": 1.0426,
      "step": 7017
    },
    {
      "epoch": 0.506330940442264,
      "grad_norm": 3.7922558728417233,
      "learning_rate": 2.056075827693907e-06,
      "loss": 0.9542,
      "step": 7018
    },
    {
      "epoch": 0.5064030879116915,
      "grad_norm": 3.927154513387713,
      "learning_rate": 2.0556086500931525e-06,
      "loss": 1.0513,
      "step": 7019
    },
    {
      "epoch": 0.506475235381119,
      "grad_norm": 3.101924385078893,
      "learning_rate": 2.055141469455816e-06,
      "loss": 0.9814,
      "step": 7020
    },
    {
      "epoch": 0.5065473828505466,
      "grad_norm": 4.830582175541897,
      "learning_rate": 2.054674285807407e-06,
      "loss": 1.0607,
      "step": 7021
    },
    {
      "epoch": 0.506619530319974,
      "grad_norm": 2.922631313744781,
      "learning_rate": 2.0542070991734387e-06,
      "loss": 0.9252,
      "step": 7022
    },
    {
      "epoch": 0.5066916777894015,
      "grad_norm": 3.3268484613541465,
      "learning_rate": 2.053739909579421e-06,
      "loss": 1.092,
      "step": 7023
    },
    {
      "epoch": 0.506763825258829,
      "grad_norm": 6.037743428063584,
      "learning_rate": 2.0532727170508655e-06,
      "loss": 1.0869,
      "step": 7024
    },
    {
      "epoch": 0.5068359727282565,
      "grad_norm": 6.8352055969794,
      "learning_rate": 2.0528055216132845e-06,
      "loss": 0.9951,
      "step": 7025
    },
    {
      "epoch": 0.506908120197684,
      "grad_norm": 2.8268108370590355,
      "learning_rate": 2.052338323292189e-06,
      "loss": 0.9574,
      "step": 7026
    },
    {
      "epoch": 0.5069802676671116,
      "grad_norm": 8.086093182378098,
      "learning_rate": 2.051871122113092e-06,
      "loss": 0.7916,
      "step": 7027
    },
    {
      "epoch": 0.5070524151365391,
      "grad_norm": 2.967303114211025,
      "learning_rate": 2.051403918101505e-06,
      "loss": 1.0579,
      "step": 7028
    },
    {
      "epoch": 0.5071245626059666,
      "grad_norm": 4.210785758658622,
      "learning_rate": 2.05093671128294e-06,
      "loss": 0.9312,
      "step": 7029
    },
    {
      "epoch": 0.5071967100753941,
      "grad_norm": 2.7328960669834026,
      "learning_rate": 2.05046950168291e-06,
      "loss": 1.0521,
      "step": 7030
    },
    {
      "epoch": 0.5072688575448216,
      "grad_norm": 5.681353255236909,
      "learning_rate": 2.0500022893269268e-06,
      "loss": 0.9726,
      "step": 7031
    },
    {
      "epoch": 0.5073410050142492,
      "grad_norm": 4.6444682124217715,
      "learning_rate": 2.0495350742405044e-06,
      "loss": 1.0213,
      "step": 7032
    },
    {
      "epoch": 0.5074131524836767,
      "grad_norm": 3.4787693520164296,
      "learning_rate": 2.0490678564491543e-06,
      "loss": 0.9202,
      "step": 7033
    },
    {
      "epoch": 0.5074852999531041,
      "grad_norm": 3.7138349605408885,
      "learning_rate": 2.048600635978391e-06,
      "loss": 0.9941,
      "step": 7034
    },
    {
      "epoch": 0.5075574474225316,
      "grad_norm": 2.767843272794084,
      "learning_rate": 2.0481334128537266e-06,
      "loss": 0.9679,
      "step": 7035
    },
    {
      "epoch": 0.5076295948919591,
      "grad_norm": 3.011298203452655,
      "learning_rate": 2.047666187100674e-06,
      "loss": 0.904,
      "step": 7036
    },
    {
      "epoch": 0.5077017423613867,
      "grad_norm": 5.356162484699284,
      "learning_rate": 2.047198958744748e-06,
      "loss": 0.9507,
      "step": 7037
    },
    {
      "epoch": 0.5077738898308142,
      "grad_norm": 5.060347111070691,
      "learning_rate": 2.0467317278114616e-06,
      "loss": 1.0482,
      "step": 7038
    },
    {
      "epoch": 0.5078460373002417,
      "grad_norm": 3.4744975268763714,
      "learning_rate": 2.0462644943263287e-06,
      "loss": 1.0271,
      "step": 7039
    },
    {
      "epoch": 0.5079181847696692,
      "grad_norm": 3.0746615525980907,
      "learning_rate": 2.045797258314863e-06,
      "loss": 0.9333,
      "step": 7040
    },
    {
      "epoch": 0.5079903322390967,
      "grad_norm": 5.098574547616475,
      "learning_rate": 2.045330019802578e-06,
      "loss": 0.978,
      "step": 7041
    },
    {
      "epoch": 0.5080624797085243,
      "grad_norm": 3.0911441193436344,
      "learning_rate": 2.0448627788149884e-06,
      "loss": 0.9732,
      "step": 7042
    },
    {
      "epoch": 0.5081346271779518,
      "grad_norm": 3.818600136271461,
      "learning_rate": 2.044395535377609e-06,
      "loss": 0.8396,
      "step": 7043
    },
    {
      "epoch": 0.5082067746473793,
      "grad_norm": 3.9761870381210165,
      "learning_rate": 2.043928289515954e-06,
      "loss": 0.99,
      "step": 7044
    },
    {
      "epoch": 0.5082789221168068,
      "grad_norm": 4.858992092342031,
      "learning_rate": 2.043461041255537e-06,
      "loss": 1.0353,
      "step": 7045
    },
    {
      "epoch": 0.5083510695862342,
      "grad_norm": 4.381774453024452,
      "learning_rate": 2.042993790621874e-06,
      "loss": 0.87,
      "step": 7046
    },
    {
      "epoch": 0.5084232170556617,
      "grad_norm": 5.277295138504368,
      "learning_rate": 2.0425265376404797e-06,
      "loss": 1.1088,
      "step": 7047
    },
    {
      "epoch": 0.5084953645250893,
      "grad_norm": 3.98326520236392,
      "learning_rate": 2.0420592823368673e-06,
      "loss": 0.9465,
      "step": 7048
    },
    {
      "epoch": 0.5085675119945168,
      "grad_norm": 4.5351005241919795,
      "learning_rate": 2.041592024736554e-06,
      "loss": 1.0443,
      "step": 7049
    },
    {
      "epoch": 0.5086396594639443,
      "grad_norm": 11.863784794425323,
      "learning_rate": 2.0411247648650547e-06,
      "loss": 0.9097,
      "step": 7050
    },
    {
      "epoch": 0.5087118069333718,
      "grad_norm": 3.5452828888141985,
      "learning_rate": 2.0406575027478827e-06,
      "loss": 0.9851,
      "step": 7051
    },
    {
      "epoch": 0.5087839544027993,
      "grad_norm": 4.134799530077688,
      "learning_rate": 2.0401902384105564e-06,
      "loss": 0.9698,
      "step": 7052
    },
    {
      "epoch": 0.5088561018722269,
      "grad_norm": 6.413822220682112,
      "learning_rate": 2.0397229718785893e-06,
      "loss": 0.9307,
      "step": 7053
    },
    {
      "epoch": 0.5089282493416544,
      "grad_norm": 11.785011237506906,
      "learning_rate": 2.039255703177498e-06,
      "loss": 0.9816,
      "step": 7054
    },
    {
      "epoch": 0.5090003968110819,
      "grad_norm": 3.8279249489341938,
      "learning_rate": 2.038788432332798e-06,
      "loss": 1.1294,
      "step": 7055
    },
    {
      "epoch": 0.5090725442805094,
      "grad_norm": 2.9566900139553094,
      "learning_rate": 2.0383211593700054e-06,
      "loss": 0.8361,
      "step": 7056
    },
    {
      "epoch": 0.5091446917499368,
      "grad_norm": 4.502115811865268,
      "learning_rate": 2.037853884314636e-06,
      "loss": 0.964,
      "step": 7057
    },
    {
      "epoch": 0.5092168392193643,
      "grad_norm": 5.644833503207217,
      "learning_rate": 2.0373866071922053e-06,
      "loss": 0.9151,
      "step": 7058
    },
    {
      "epoch": 0.5092889866887919,
      "grad_norm": 6.719013155729436,
      "learning_rate": 2.0369193280282318e-06,
      "loss": 0.8975,
      "step": 7059
    },
    {
      "epoch": 0.5093611341582194,
      "grad_norm": 10.393440129104759,
      "learning_rate": 2.036452046848229e-06,
      "loss": 0.9338,
      "step": 7060
    },
    {
      "epoch": 0.5094332816276469,
      "grad_norm": 3.6958515830523004,
      "learning_rate": 2.0359847636777165e-06,
      "loss": 1.0187,
      "step": 7061
    },
    {
      "epoch": 0.5095054290970744,
      "grad_norm": 4.952587591915041,
      "learning_rate": 2.035517478542209e-06,
      "loss": 1.07,
      "step": 7062
    },
    {
      "epoch": 0.5095775765665019,
      "grad_norm": 5.135088968815571,
      "learning_rate": 2.035050191467222e-06,
      "loss": 0.9412,
      "step": 7063
    },
    {
      "epoch": 0.5096497240359295,
      "grad_norm": 5.372764122365246,
      "learning_rate": 2.0345829024782754e-06,
      "loss": 0.985,
      "step": 7064
    },
    {
      "epoch": 0.509721871505357,
      "grad_norm": 7.970299663581974,
      "learning_rate": 2.0341156116008836e-06,
      "loss": 0.9238,
      "step": 7065
    },
    {
      "epoch": 0.5097940189747845,
      "grad_norm": 3.0662264825888674,
      "learning_rate": 2.0336483188605648e-06,
      "loss": 1.0425,
      "step": 7066
    },
    {
      "epoch": 0.509866166444212,
      "grad_norm": 0.7202909784970402,
      "learning_rate": 2.033181024282836e-06,
      "loss": 0.8646,
      "step": 7067
    },
    {
      "epoch": 0.5099383139136395,
      "grad_norm": 7.25065083706526,
      "learning_rate": 2.0327137278932144e-06,
      "loss": 0.9503,
      "step": 7068
    },
    {
      "epoch": 0.5100104613830669,
      "grad_norm": 3.9357905234365,
      "learning_rate": 2.0322464297172166e-06,
      "loss": 0.9997,
      "step": 7069
    },
    {
      "epoch": 0.5100826088524945,
      "grad_norm": 3.2952649854234357,
      "learning_rate": 2.0317791297803615e-06,
      "loss": 0.8679,
      "step": 7070
    },
    {
      "epoch": 0.510154756321922,
      "grad_norm": 5.774375333483559,
      "learning_rate": 2.0313118281081653e-06,
      "loss": 1.0068,
      "step": 7071
    },
    {
      "epoch": 0.5102269037913495,
      "grad_norm": 4.0048069680470695,
      "learning_rate": 2.0308445247261455e-06,
      "loss": 1.0086,
      "step": 7072
    },
    {
      "epoch": 0.510299051260777,
      "grad_norm": 6.662733189150282,
      "learning_rate": 2.0303772196598214e-06,
      "loss": 0.8128,
      "step": 7073
    },
    {
      "epoch": 0.5103711987302045,
      "grad_norm": 13.349047428964443,
      "learning_rate": 2.0299099129347095e-06,
      "loss": 0.9528,
      "step": 7074
    },
    {
      "epoch": 0.510443346199632,
      "grad_norm": 3.520690388382945,
      "learning_rate": 2.0294426045763275e-06,
      "loss": 0.9374,
      "step": 7075
    },
    {
      "epoch": 0.5105154936690596,
      "grad_norm": 5.381555507374711,
      "learning_rate": 2.0289752946101944e-06,
      "loss": 0.9565,
      "step": 7076
    },
    {
      "epoch": 0.5105876411384871,
      "grad_norm": 3.4423916777368118,
      "learning_rate": 2.028507983061827e-06,
      "loss": 0.9649,
      "step": 7077
    },
    {
      "epoch": 0.5106597886079146,
      "grad_norm": 5.8950398948037765,
      "learning_rate": 2.028040669956745e-06,
      "loss": 0.9209,
      "step": 7078
    },
    {
      "epoch": 0.5107319360773421,
      "grad_norm": 4.490929681241454,
      "learning_rate": 2.027573355320466e-06,
      "loss": 1.0267,
      "step": 7079
    },
    {
      "epoch": 0.5108040835467696,
      "grad_norm": 7.366222265771649,
      "learning_rate": 2.027106039178507e-06,
      "loss": 0.9672,
      "step": 7080
    },
    {
      "epoch": 0.5108762310161971,
      "grad_norm": 5.055244426911552,
      "learning_rate": 2.0266387215563884e-06,
      "loss": 0.9907,
      "step": 7081
    },
    {
      "epoch": 0.5109483784856246,
      "grad_norm": 4.888576090564338,
      "learning_rate": 2.0261714024796273e-06,
      "loss": 1.0209,
      "step": 7082
    },
    {
      "epoch": 0.5110205259550521,
      "grad_norm": 3.826029550457194,
      "learning_rate": 2.025704081973744e-06,
      "loss": 1.0808,
      "step": 7083
    },
    {
      "epoch": 0.5110926734244796,
      "grad_norm": 3.92682715001112,
      "learning_rate": 2.025236760064255e-06,
      "loss": 0.9639,
      "step": 7084
    },
    {
      "epoch": 0.5111648208939071,
      "grad_norm": 3.5682671696317905,
      "learning_rate": 2.02476943677668e-06,
      "loss": 1.0522,
      "step": 7085
    },
    {
      "epoch": 0.5112369683633347,
      "grad_norm": 4.504921023812308,
      "learning_rate": 2.0243021121365378e-06,
      "loss": 1.002,
      "step": 7086
    },
    {
      "epoch": 0.5113091158327622,
      "grad_norm": 4.850079541684749,
      "learning_rate": 2.0238347861693466e-06,
      "loss": 0.9965,
      "step": 7087
    },
    {
      "epoch": 0.5113812633021897,
      "grad_norm": 14.821612295564847,
      "learning_rate": 2.0233674589006267e-06,
      "loss": 0.9553,
      "step": 7088
    },
    {
      "epoch": 0.5114534107716172,
      "grad_norm": 4.130909934309386,
      "learning_rate": 2.0229001303558965e-06,
      "loss": 0.9566,
      "step": 7089
    },
    {
      "epoch": 0.5115255582410447,
      "grad_norm": 5.087300717061596,
      "learning_rate": 2.0224328005606743e-06,
      "loss": 0.8781,
      "step": 7090
    },
    {
      "epoch": 0.5115977057104723,
      "grad_norm": 3.4752757230640507,
      "learning_rate": 2.021965469540481e-06,
      "loss": 1.0246,
      "step": 7091
    },
    {
      "epoch": 0.5116698531798998,
      "grad_norm": 7.825460441132052,
      "learning_rate": 2.0214981373208336e-06,
      "loss": 0.9409,
      "step": 7092
    },
    {
      "epoch": 0.5117420006493272,
      "grad_norm": 3.0474417917327026,
      "learning_rate": 2.0210308039272527e-06,
      "loss": 0.9548,
      "step": 7093
    },
    {
      "epoch": 0.5118141481187547,
      "grad_norm": 5.658685149471528,
      "learning_rate": 2.020563469385258e-06,
      "loss": 0.9725,
      "step": 7094
    },
    {
      "epoch": 0.5118862955881822,
      "grad_norm": 3.6166300992977543,
      "learning_rate": 2.0200961337203683e-06,
      "loss": 0.9371,
      "step": 7095
    },
    {
      "epoch": 0.5119584430576097,
      "grad_norm": 5.016244440063772,
      "learning_rate": 2.019628796958103e-06,
      "loss": 1.0811,
      "step": 7096
    },
    {
      "epoch": 0.5120305905270373,
      "grad_norm": 3.3286473397136285,
      "learning_rate": 2.0191614591239808e-06,
      "loss": 0.8887,
      "step": 7097
    },
    {
      "epoch": 0.5121027379964648,
      "grad_norm": 7.64868587601052,
      "learning_rate": 2.0186941202435233e-06,
      "loss": 0.9172,
      "step": 7098
    },
    {
      "epoch": 0.5121748854658923,
      "grad_norm": 7.280005090732681,
      "learning_rate": 2.0182267803422485e-06,
      "loss": 0.9044,
      "step": 7099
    },
    {
      "epoch": 0.5122470329353198,
      "grad_norm": 5.310846767116244,
      "learning_rate": 2.0177594394456773e-06,
      "loss": 0.7508,
      "step": 7100
    },
    {
      "epoch": 0.5123191804047473,
      "grad_norm": 4.421891606707442,
      "learning_rate": 2.0172920975793288e-06,
      "loss": 0.9079,
      "step": 7101
    },
    {
      "epoch": 0.5123913278741749,
      "grad_norm": 3.471003994327534,
      "learning_rate": 2.0168247547687223e-06,
      "loss": 0.8949,
      "step": 7102
    },
    {
      "epoch": 0.5124634753436024,
      "grad_norm": 9.05324655542747,
      "learning_rate": 2.016357411039379e-06,
      "loss": 1.0042,
      "step": 7103
    },
    {
      "epoch": 0.5125356228130298,
      "grad_norm": 5.613695611174246,
      "learning_rate": 2.0158900664168175e-06,
      "loss": 0.9111,
      "step": 7104
    },
    {
      "epoch": 0.5126077702824573,
      "grad_norm": 3.8119268768278105,
      "learning_rate": 2.0154227209265584e-06,
      "loss": 0.9771,
      "step": 7105
    },
    {
      "epoch": 0.5126799177518848,
      "grad_norm": 5.629497002039496,
      "learning_rate": 2.014955374594122e-06,
      "loss": 0.8913,
      "step": 7106
    },
    {
      "epoch": 0.5127520652213123,
      "grad_norm": 0.7492802663296201,
      "learning_rate": 2.0144880274450275e-06,
      "loss": 0.8792,
      "step": 7107
    },
    {
      "epoch": 0.5128242126907399,
      "grad_norm": 10.097001913035173,
      "learning_rate": 2.014020679504796e-06,
      "loss": 0.9223,
      "step": 7108
    },
    {
      "epoch": 0.5128963601601674,
      "grad_norm": 6.201064221506503,
      "learning_rate": 2.013553330798946e-06,
      "loss": 0.8788,
      "step": 7109
    },
    {
      "epoch": 0.5129685076295949,
      "grad_norm": 7.454606586721594,
      "learning_rate": 2.0130859813529994e-06,
      "loss": 0.9953,
      "step": 7110
    },
    {
      "epoch": 0.5130406550990224,
      "grad_norm": 21.222629288014346,
      "learning_rate": 2.0126186311924763e-06,
      "loss": 0.9926,
      "step": 7111
    },
    {
      "epoch": 0.5131128025684499,
      "grad_norm": 0.7327625347094819,
      "learning_rate": 2.0121512803428953e-06,
      "loss": 0.8279,
      "step": 7112
    },
    {
      "epoch": 0.5131849500378775,
      "grad_norm": 0.75898773459625,
      "learning_rate": 2.0116839288297788e-06,
      "loss": 0.8045,
      "step": 7113
    },
    {
      "epoch": 0.513257097507305,
      "grad_norm": 5.386438839661422,
      "learning_rate": 2.0112165766786457e-06,
      "loss": 0.8578,
      "step": 7114
    },
    {
      "epoch": 0.5133292449767325,
      "grad_norm": 4.641666796334412,
      "learning_rate": 2.010749223915017e-06,
      "loss": 0.8628,
      "step": 7115
    },
    {
      "epoch": 0.5134013924461599,
      "grad_norm": 3.8874538014110818,
      "learning_rate": 2.0102818705644133e-06,
      "loss": 0.8717,
      "step": 7116
    },
    {
      "epoch": 0.5134735399155874,
      "grad_norm": 4.190280048416248,
      "learning_rate": 2.0098145166523546e-06,
      "loss": 0.9624,
      "step": 7117
    },
    {
      "epoch": 0.5135456873850149,
      "grad_norm": 8.024776239335665,
      "learning_rate": 2.0093471622043613e-06,
      "loss": 0.8622,
      "step": 7118
    },
    {
      "epoch": 0.5136178348544425,
      "grad_norm": 3.659230509054601,
      "learning_rate": 2.0088798072459543e-06,
      "loss": 0.8663,
      "step": 7119
    },
    {
      "epoch": 0.51368998232387,
      "grad_norm": 3.307183948274515,
      "learning_rate": 2.008412451802654e-06,
      "loss": 0.8884,
      "step": 7120
    },
    {
      "epoch": 0.5137621297932975,
      "grad_norm": 5.35587688201409,
      "learning_rate": 2.00794509589998e-06,
      "loss": 0.8492,
      "step": 7121
    },
    {
      "epoch": 0.513834277262725,
      "grad_norm": 12.98439909086965,
      "learning_rate": 2.007477739563455e-06,
      "loss": 1.0491,
      "step": 7122
    },
    {
      "epoch": 0.5139064247321525,
      "grad_norm": 4.536399283245606,
      "learning_rate": 2.007010382818598e-06,
      "loss": 0.8272,
      "step": 7123
    },
    {
      "epoch": 0.51397857220158,
      "grad_norm": 7.3826432788972856,
      "learning_rate": 2.006543025690929e-06,
      "loss": 1.0339,
      "step": 7124
    },
    {
      "epoch": 0.5140507196710076,
      "grad_norm": 4.980510684566107,
      "learning_rate": 2.0060756682059704e-06,
      "loss": 0.9524,
      "step": 7125
    },
    {
      "epoch": 0.5141228671404351,
      "grad_norm": 4.567031908859732,
      "learning_rate": 2.0056083103892415e-06,
      "loss": 0.8882,
      "step": 7126
    },
    {
      "epoch": 0.5141950146098626,
      "grad_norm": 5.2419404427857526,
      "learning_rate": 2.005140952266264e-06,
      "loss": 0.9886,
      "step": 7127
    },
    {
      "epoch": 0.51426716207929,
      "grad_norm": 5.0609183018724515,
      "learning_rate": 2.004673593862558e-06,
      "loss": 0.9514,
      "step": 7128
    },
    {
      "epoch": 0.5143393095487175,
      "grad_norm": 9.653094223531587,
      "learning_rate": 2.004206235203644e-06,
      "loss": 0.8631,
      "step": 7129
    },
    {
      "epoch": 0.5144114570181451,
      "grad_norm": 3.1192517628036085,
      "learning_rate": 2.0037388763150437e-06,
      "loss": 0.9287,
      "step": 7130
    },
    {
      "epoch": 0.5144836044875726,
      "grad_norm": 3.605280415584896,
      "learning_rate": 2.0032715172222765e-06,
      "loss": 0.997,
      "step": 7131
    },
    {
      "epoch": 0.5145557519570001,
      "grad_norm": 3.9487945600496905,
      "learning_rate": 2.002804157950865e-06,
      "loss": 1.0162,
      "step": 7132
    },
    {
      "epoch": 0.5146278994264276,
      "grad_norm": 3.259265017825476,
      "learning_rate": 2.002336798526328e-06,
      "loss": 1.0357,
      "step": 7133
    },
    {
      "epoch": 0.5147000468958551,
      "grad_norm": 3.874271570466014,
      "learning_rate": 2.0018694389741867e-06,
      "loss": 0.9952,
      "step": 7134
    },
    {
      "epoch": 0.5147721943652827,
      "grad_norm": 4.443289190723181,
      "learning_rate": 2.0014020793199633e-06,
      "loss": 0.8941,
      "step": 7135
    },
    {
      "epoch": 0.5148443418347102,
      "grad_norm": 5.431918206061177,
      "learning_rate": 2.0009347195891767e-06,
      "loss": 0.9966,
      "step": 7136
    },
    {
      "epoch": 0.5149164893041377,
      "grad_norm": 3.4845243002889505,
      "learning_rate": 2.0004673598073486e-06,
      "loss": 0.8516,
      "step": 7137
    },
    {
      "epoch": 0.5149886367735652,
      "grad_norm": 15.537196166673048,
      "learning_rate": 1.9999999999999995e-06,
      "loss": 0.9937,
      "step": 7138
    },
    {
      "epoch": 0.5150607842429927,
      "grad_norm": 5.38306425305917,
      "learning_rate": 1.999532640192651e-06,
      "loss": 0.8332,
      "step": 7139
    },
    {
      "epoch": 0.5151329317124201,
      "grad_norm": 0.6803156044240412,
      "learning_rate": 1.9990652804108236e-06,
      "loss": 0.7373,
      "step": 7140
    },
    {
      "epoch": 0.5152050791818477,
      "grad_norm": 6.279697814765882,
      "learning_rate": 1.998597920680037e-06,
      "loss": 0.9718,
      "step": 7141
    },
    {
      "epoch": 0.5152772266512752,
      "grad_norm": 5.53501221032592,
      "learning_rate": 1.998130561025813e-06,
      "loss": 0.9336,
      "step": 7142
    },
    {
      "epoch": 0.5153493741207027,
      "grad_norm": 3.277775259789256,
      "learning_rate": 1.997663201473672e-06,
      "loss": 1.0065,
      "step": 7143
    },
    {
      "epoch": 0.5154215215901302,
      "grad_norm": 3.515359419222091,
      "learning_rate": 1.9971958420491356e-06,
      "loss": 0.9943,
      "step": 7144
    },
    {
      "epoch": 0.5154936690595577,
      "grad_norm": 4.384198573091313,
      "learning_rate": 1.9967284827777233e-06,
      "loss": 0.9868,
      "step": 7145
    },
    {
      "epoch": 0.5155658165289853,
      "grad_norm": 5.899844616929013,
      "learning_rate": 1.996261123684956e-06,
      "loss": 0.9916,
      "step": 7146
    },
    {
      "epoch": 0.5156379639984128,
      "grad_norm": 0.8133885953126653,
      "learning_rate": 1.995793764796356e-06,
      "loss": 0.8595,
      "step": 7147
    },
    {
      "epoch": 0.5157101114678403,
      "grad_norm": 3.318643307452551,
      "learning_rate": 1.9953264061374417e-06,
      "loss": 1.0429,
      "step": 7148
    },
    {
      "epoch": 0.5157822589372678,
      "grad_norm": 4.712512409386119,
      "learning_rate": 1.9948590477337363e-06,
      "loss": 0.9877,
      "step": 7149
    },
    {
      "epoch": 0.5158544064066953,
      "grad_norm": 7.279580117158469,
      "learning_rate": 1.9943916896107588e-06,
      "loss": 0.9248,
      "step": 7150
    },
    {
      "epoch": 0.5159265538761227,
      "grad_norm": 6.564769824627654,
      "learning_rate": 1.99392433179403e-06,
      "loss": 0.8908,
      "step": 7151
    },
    {
      "epoch": 0.5159987013455503,
      "grad_norm": 4.989911009581726,
      "learning_rate": 1.9934569743090714e-06,
      "loss": 0.9816,
      "step": 7152
    },
    {
      "epoch": 0.5160708488149778,
      "grad_norm": 6.3268230841501145,
      "learning_rate": 1.9929896171814024e-06,
      "loss": 1.0661,
      "step": 7153
    },
    {
      "epoch": 0.5161429962844053,
      "grad_norm": 14.044651759370618,
      "learning_rate": 1.9925222604365454e-06,
      "loss": 1.0382,
      "step": 7154
    },
    {
      "epoch": 0.5162151437538328,
      "grad_norm": 4.455553965093932,
      "learning_rate": 1.99205490410002e-06,
      "loss": 1.0132,
      "step": 7155
    },
    {
      "epoch": 0.5162872912232603,
      "grad_norm": 8.110398068142553,
      "learning_rate": 1.9915875481973462e-06,
      "loss": 0.9775,
      "step": 7156
    },
    {
      "epoch": 0.5163594386926879,
      "grad_norm": 6.246460789911543,
      "learning_rate": 1.991120192754046e-06,
      "loss": 1.0496,
      "step": 7157
    },
    {
      "epoch": 0.5164315861621154,
      "grad_norm": 3.511117987753442,
      "learning_rate": 1.9906528377956386e-06,
      "loss": 0.9043,
      "step": 7158
    },
    {
      "epoch": 0.5165037336315429,
      "grad_norm": 5.657949287216947,
      "learning_rate": 1.9901854833476456e-06,
      "loss": 1.0152,
      "step": 7159
    },
    {
      "epoch": 0.5165758811009704,
      "grad_norm": 3.7565150568871184,
      "learning_rate": 1.989718129435586e-06,
      "loss": 1.0302,
      "step": 7160
    },
    {
      "epoch": 0.5166480285703979,
      "grad_norm": 3.919292922184778,
      "learning_rate": 1.9892507760849827e-06,
      "loss": 1.0982,
      "step": 7161
    },
    {
      "epoch": 0.5167201760398255,
      "grad_norm": 6.185353841320154,
      "learning_rate": 1.9887834233213546e-06,
      "loss": 0.9135,
      "step": 7162
    },
    {
      "epoch": 0.5167923235092529,
      "grad_norm": 6.352130334300104,
      "learning_rate": 1.988316071170221e-06,
      "loss": 0.9858,
      "step": 7163
    },
    {
      "epoch": 0.5168644709786804,
      "grad_norm": 3.5930140529562657,
      "learning_rate": 1.9878487196571045e-06,
      "loss": 0.9639,
      "step": 7164
    },
    {
      "epoch": 0.5169366184481079,
      "grad_norm": 3.597463795239174,
      "learning_rate": 1.9873813688075235e-06,
      "loss": 0.9714,
      "step": 7165
    },
    {
      "epoch": 0.5170087659175354,
      "grad_norm": 4.290523537299509,
      "learning_rate": 1.9869140186470004e-06,
      "loss": 0.838,
      "step": 7166
    },
    {
      "epoch": 0.5170809133869629,
      "grad_norm": 4.488041777567627,
      "learning_rate": 1.9864466692010542e-06,
      "loss": 1.005,
      "step": 7167
    },
    {
      "epoch": 0.5171530608563905,
      "grad_norm": 3.8048586953752253,
      "learning_rate": 1.9859793204952043e-06,
      "loss": 1.0103,
      "step": 7168
    },
    {
      "epoch": 0.517225208325818,
      "grad_norm": 5.119984661615762,
      "learning_rate": 1.9855119725549727e-06,
      "loss": 1.0034,
      "step": 7169
    },
    {
      "epoch": 0.5172973557952455,
      "grad_norm": 5.014092422163074,
      "learning_rate": 1.9850446254058783e-06,
      "loss": 0.8765,
      "step": 7170
    },
    {
      "epoch": 0.517369503264673,
      "grad_norm": 3.0191883914034983,
      "learning_rate": 1.9845772790734414e-06,
      "loss": 0.9833,
      "step": 7171
    },
    {
      "epoch": 0.5174416507341005,
      "grad_norm": 3.061595861134997,
      "learning_rate": 1.9841099335831827e-06,
      "loss": 0.9345,
      "step": 7172
    },
    {
      "epoch": 0.517513798203528,
      "grad_norm": 3.8936848691633434,
      "learning_rate": 1.983642588960621e-06,
      "loss": 1.0129,
      "step": 7173
    },
    {
      "epoch": 0.5175859456729556,
      "grad_norm": 4.396057677608245,
      "learning_rate": 1.983175245231278e-06,
      "loss": 0.9615,
      "step": 7174
    },
    {
      "epoch": 0.517658093142383,
      "grad_norm": 2.8212263576343433,
      "learning_rate": 1.982707902420671e-06,
      "loss": 1.0558,
      "step": 7175
    },
    {
      "epoch": 0.5177302406118105,
      "grad_norm": 9.03982253246473,
      "learning_rate": 1.9822405605543225e-06,
      "loss": 0.9512,
      "step": 7176
    },
    {
      "epoch": 0.517802388081238,
      "grad_norm": 19.089204687354137,
      "learning_rate": 1.9817732196577518e-06,
      "loss": 0.9492,
      "step": 7177
    },
    {
      "epoch": 0.5178745355506655,
      "grad_norm": 3.1947705991531503,
      "learning_rate": 1.9813058797564765e-06,
      "loss": 0.9326,
      "step": 7178
    },
    {
      "epoch": 0.5179466830200931,
      "grad_norm": 3.7154557198350093,
      "learning_rate": 1.980838540876019e-06,
      "loss": 0.8884,
      "step": 7179
    },
    {
      "epoch": 0.5180188304895206,
      "grad_norm": 0.7907130286624815,
      "learning_rate": 1.980371203041897e-06,
      "loss": 0.9136,
      "step": 7180
    },
    {
      "epoch": 0.5180909779589481,
      "grad_norm": 3.7228688742728098,
      "learning_rate": 1.979903866279632e-06,
      "loss": 0.9674,
      "step": 7181
    },
    {
      "epoch": 0.5181631254283756,
      "grad_norm": 3.457390959393188,
      "learning_rate": 1.979436530614742e-06,
      "loss": 1.0485,
      "step": 7182
    },
    {
      "epoch": 0.5182352728978031,
      "grad_norm": 2.3257944616032846,
      "learning_rate": 1.978969196072747e-06,
      "loss": 0.9361,
      "step": 7183
    },
    {
      "epoch": 0.5183074203672307,
      "grad_norm": 4.082992046392366,
      "learning_rate": 1.9785018626791667e-06,
      "loss": 1.1087,
      "step": 7184
    },
    {
      "epoch": 0.5183795678366582,
      "grad_norm": 4.4940338851540815,
      "learning_rate": 1.9780345304595192e-06,
      "loss": 0.9378,
      "step": 7185
    },
    {
      "epoch": 0.5184517153060857,
      "grad_norm": 8.46886351111658,
      "learning_rate": 1.9775671994393255e-06,
      "loss": 0.9305,
      "step": 7186
    },
    {
      "epoch": 0.5185238627755131,
      "grad_norm": 3.6460550440631554,
      "learning_rate": 1.9770998696441033e-06,
      "loss": 0.8475,
      "step": 7187
    },
    {
      "epoch": 0.5185960102449406,
      "grad_norm": 4.23353438813568,
      "learning_rate": 1.976632541099373e-06,
      "loss": 0.9732,
      "step": 7188
    },
    {
      "epoch": 0.5186681577143681,
      "grad_norm": 2.848922217847431,
      "learning_rate": 1.9761652138306536e-06,
      "loss": 0.9699,
      "step": 7189
    },
    {
      "epoch": 0.5187403051837957,
      "grad_norm": 3.0306603556258342,
      "learning_rate": 1.975697887863462e-06,
      "loss": 0.8857,
      "step": 7190
    },
    {
      "epoch": 0.5188124526532232,
      "grad_norm": 4.395255148710439,
      "learning_rate": 1.9752305632233204e-06,
      "loss": 1.0186,
      "step": 7191
    },
    {
      "epoch": 0.5188846001226507,
      "grad_norm": 6.17435515585038,
      "learning_rate": 1.974763239935745e-06,
      "loss": 1.0194,
      "step": 7192
    },
    {
      "epoch": 0.5189567475920782,
      "grad_norm": 5.23305319423017,
      "learning_rate": 1.9742959180262564e-06,
      "loss": 0.8962,
      "step": 7193
    },
    {
      "epoch": 0.5190288950615057,
      "grad_norm": 3.6184294774445647,
      "learning_rate": 1.9738285975203725e-06,
      "loss": 0.9697,
      "step": 7194
    },
    {
      "epoch": 0.5191010425309333,
      "grad_norm": 3.0063725182832823,
      "learning_rate": 1.9733612784436114e-06,
      "loss": 1.0196,
      "step": 7195
    },
    {
      "epoch": 0.5191731900003608,
      "grad_norm": 4.824747527637517,
      "learning_rate": 1.9728939608214927e-06,
      "loss": 0.9408,
      "step": 7196
    },
    {
      "epoch": 0.5192453374697883,
      "grad_norm": 2.9360861114845958,
      "learning_rate": 1.972426644679534e-06,
      "loss": 1.0463,
      "step": 7197
    },
    {
      "epoch": 0.5193174849392158,
      "grad_norm": 4.121939708788888,
      "learning_rate": 1.971959330043255e-06,
      "loss": 0.8675,
      "step": 7198
    },
    {
      "epoch": 0.5193896324086432,
      "grad_norm": 5.514330833696599,
      "learning_rate": 1.9714920169381728e-06,
      "loss": 1.058,
      "step": 7199
    },
    {
      "epoch": 0.5194617798780707,
      "grad_norm": 3.086032218928086,
      "learning_rate": 1.971024705389806e-06,
      "loss": 1.0027,
      "step": 7200
    },
    {
      "epoch": 0.5195339273474983,
      "grad_norm": 3.181845642565842,
      "learning_rate": 1.9705573954236727e-06,
      "loss": 1.0725,
      "step": 7201
    },
    {
      "epoch": 0.5196060748169258,
      "grad_norm": 2.8084063938345727,
      "learning_rate": 1.9700900870652907e-06,
      "loss": 0.9352,
      "step": 7202
    },
    {
      "epoch": 0.5196782222863533,
      "grad_norm": 2.8119240065028728,
      "learning_rate": 1.969622780340179e-06,
      "loss": 0.916,
      "step": 7203
    },
    {
      "epoch": 0.5197503697557808,
      "grad_norm": 3.3724997407476214,
      "learning_rate": 1.969155475273854e-06,
      "loss": 0.8978,
      "step": 7204
    },
    {
      "epoch": 0.5198225172252083,
      "grad_norm": 18.441351226485377,
      "learning_rate": 1.968688171891835e-06,
      "loss": 0.9734,
      "step": 7205
    },
    {
      "epoch": 0.5198946646946359,
      "grad_norm": 3.5558080898116287,
      "learning_rate": 1.968220870219639e-06,
      "loss": 1.0312,
      "step": 7206
    },
    {
      "epoch": 0.5199668121640634,
      "grad_norm": 0.705521990688142,
      "learning_rate": 1.9677535702827836e-06,
      "loss": 0.8381,
      "step": 7207
    },
    {
      "epoch": 0.5200389596334909,
      "grad_norm": 4.367288168632335,
      "learning_rate": 1.967286272106786e-06,
      "loss": 1.0,
      "step": 7208
    },
    {
      "epoch": 0.5201111071029184,
      "grad_norm": 4.44009452513005,
      "learning_rate": 1.966818975717164e-06,
      "loss": 1.0152,
      "step": 7209
    },
    {
      "epoch": 0.5201832545723458,
      "grad_norm": 5.7264572487769705,
      "learning_rate": 1.9663516811394355e-06,
      "loss": 1.0016,
      "step": 7210
    },
    {
      "epoch": 0.5202554020417733,
      "grad_norm": 5.82494703276951,
      "learning_rate": 1.9658843883991166e-06,
      "loss": 0.9008,
      "step": 7211
    },
    {
      "epoch": 0.5203275495112009,
      "grad_norm": 4.180942924411025,
      "learning_rate": 1.965417097521725e-06,
      "loss": 0.9472,
      "step": 7212
    },
    {
      "epoch": 0.5203996969806284,
      "grad_norm": 3.4482948733496293,
      "learning_rate": 1.9649498085327777e-06,
      "loss": 0.9596,
      "step": 7213
    },
    {
      "epoch": 0.5204718444500559,
      "grad_norm": 3.3534571843062344,
      "learning_rate": 1.9644825214577913e-06,
      "loss": 0.9084,
      "step": 7214
    },
    {
      "epoch": 0.5205439919194834,
      "grad_norm": 5.4136233461300804,
      "learning_rate": 1.9640152363222837e-06,
      "loss": 1.022,
      "step": 7215
    },
    {
      "epoch": 0.5206161393889109,
      "grad_norm": 2.8651286952754362,
      "learning_rate": 1.9635479531517706e-06,
      "loss": 0.9191,
      "step": 7216
    },
    {
      "epoch": 0.5206882868583385,
      "grad_norm": 3.9300637643314933,
      "learning_rate": 1.963080671971768e-06,
      "loss": 0.9663,
      "step": 7217
    },
    {
      "epoch": 0.520760434327766,
      "grad_norm": 2.741064031705466,
      "learning_rate": 1.9626133928077945e-06,
      "loss": 0.879,
      "step": 7218
    },
    {
      "epoch": 0.5208325817971935,
      "grad_norm": 4.795958589872974,
      "learning_rate": 1.9621461156853643e-06,
      "loss": 1.0362,
      "step": 7219
    },
    {
      "epoch": 0.520904729266621,
      "grad_norm": 5.983380825240318,
      "learning_rate": 1.961678840629995e-06,
      "loss": 0.9825,
      "step": 7220
    },
    {
      "epoch": 0.5209768767360485,
      "grad_norm": 5.425678776761359,
      "learning_rate": 1.9612115676672024e-06,
      "loss": 0.977,
      "step": 7221
    },
    {
      "epoch": 0.5210490242054759,
      "grad_norm": 3.655759990493531,
      "learning_rate": 1.9607442968225023e-06,
      "loss": 0.9004,
      "step": 7222
    },
    {
      "epoch": 0.5211211716749035,
      "grad_norm": 3.565401602190124,
      "learning_rate": 1.9602770281214105e-06,
      "loss": 1.0784,
      "step": 7223
    },
    {
      "epoch": 0.521193319144331,
      "grad_norm": 3.7620699868067744,
      "learning_rate": 1.959809761589444e-06,
      "loss": 0.945,
      "step": 7224
    },
    {
      "epoch": 0.5212654666137585,
      "grad_norm": 3.6571796246153485,
      "learning_rate": 1.959342497252117e-06,
      "loss": 0.9416,
      "step": 7225
    },
    {
      "epoch": 0.521337614083186,
      "grad_norm": 3.237894106409864,
      "learning_rate": 1.9588752351349455e-06,
      "loss": 1.0042,
      "step": 7226
    },
    {
      "epoch": 0.5214097615526135,
      "grad_norm": 5.877319851884652,
      "learning_rate": 1.958407975263446e-06,
      "loss": 0.9759,
      "step": 7227
    },
    {
      "epoch": 0.5214819090220411,
      "grad_norm": 2.8758099285873064,
      "learning_rate": 1.957940717663133e-06,
      "loss": 1.0385,
      "step": 7228
    },
    {
      "epoch": 0.5215540564914686,
      "grad_norm": 4.155416993150373,
      "learning_rate": 1.9574734623595206e-06,
      "loss": 0.9347,
      "step": 7229
    },
    {
      "epoch": 0.5216262039608961,
      "grad_norm": 4.267521363323889,
      "learning_rate": 1.9570062093781257e-06,
      "loss": 1.0753,
      "step": 7230
    },
    {
      "epoch": 0.5216983514303236,
      "grad_norm": 3.2345442428169435,
      "learning_rate": 1.9565389587444627e-06,
      "loss": 1.0131,
      "step": 7231
    },
    {
      "epoch": 0.5217704988997511,
      "grad_norm": 4.284238343177506,
      "learning_rate": 1.9560717104840457e-06,
      "loss": 0.8746,
      "step": 7232
    },
    {
      "epoch": 0.5218426463691787,
      "grad_norm": 0.79648183959737,
      "learning_rate": 1.955604464622391e-06,
      "loss": 0.8863,
      "step": 7233
    },
    {
      "epoch": 0.5219147938386061,
      "grad_norm": 4.740570144486822,
      "learning_rate": 1.9551372211850114e-06,
      "loss": 1.0408,
      "step": 7234
    },
    {
      "epoch": 0.5219869413080336,
      "grad_norm": 3.4527390208225475,
      "learning_rate": 1.954669980197422e-06,
      "loss": 0.9678,
      "step": 7235
    },
    {
      "epoch": 0.5220590887774611,
      "grad_norm": 3.586778463183574,
      "learning_rate": 1.9542027416851374e-06,
      "loss": 0.9476,
      "step": 7236
    },
    {
      "epoch": 0.5221312362468886,
      "grad_norm": 8.116353760418395,
      "learning_rate": 1.9537355056736715e-06,
      "loss": 0.9362,
      "step": 7237
    },
    {
      "epoch": 0.5222033837163161,
      "grad_norm": 4.0285997772378845,
      "learning_rate": 1.953268272188538e-06,
      "loss": 0.9641,
      "step": 7238
    },
    {
      "epoch": 0.5222755311857437,
      "grad_norm": 3.1887429001989696,
      "learning_rate": 1.9528010412552517e-06,
      "loss": 0.9369,
      "step": 7239
    },
    {
      "epoch": 0.5223476786551712,
      "grad_norm": 4.103972506246996,
      "learning_rate": 1.952333812899326e-06,
      "loss": 0.9558,
      "step": 7240
    },
    {
      "epoch": 0.5224198261245987,
      "grad_norm": 5.617321962490083,
      "learning_rate": 1.951866587146273e-06,
      "loss": 0.9416,
      "step": 7241
    },
    {
      "epoch": 0.5224919735940262,
      "grad_norm": 0.7263775661924056,
      "learning_rate": 1.951399364021609e-06,
      "loss": 0.7901,
      "step": 7242
    },
    {
      "epoch": 0.5225641210634537,
      "grad_norm": 3.2265905790053946,
      "learning_rate": 1.9509321435508455e-06,
      "loss": 0.9764,
      "step": 7243
    },
    {
      "epoch": 0.5226362685328813,
      "grad_norm": 2.9531398550800314,
      "learning_rate": 1.9504649257594954e-06,
      "loss": 1.0016,
      "step": 7244
    },
    {
      "epoch": 0.5227084160023088,
      "grad_norm": 5.4224395897728845,
      "learning_rate": 1.949997710673073e-06,
      "loss": 0.9508,
      "step": 7245
    },
    {
      "epoch": 0.5227805634717362,
      "grad_norm": 4.057441731502871,
      "learning_rate": 1.94953049831709e-06,
      "loss": 0.9435,
      "step": 7246
    },
    {
      "epoch": 0.5228527109411637,
      "grad_norm": 4.188783591240202,
      "learning_rate": 1.94906328871706e-06,
      "loss": 1.0096,
      "step": 7247
    },
    {
      "epoch": 0.5229248584105912,
      "grad_norm": 7.460175128752686,
      "learning_rate": 1.948596081898495e-06,
      "loss": 0.8907,
      "step": 7248
    },
    {
      "epoch": 0.5229970058800187,
      "grad_norm": 5.556206732840009,
      "learning_rate": 1.948128877886908e-06,
      "loss": 1.0415,
      "step": 7249
    },
    {
      "epoch": 0.5230691533494463,
      "grad_norm": 5.004082729494292,
      "learning_rate": 1.9476616767078107e-06,
      "loss": 0.9543,
      "step": 7250
    },
    {
      "epoch": 0.5231413008188738,
      "grad_norm": 5.165131104865039,
      "learning_rate": 1.9471944783867153e-06,
      "loss": 0.9068,
      "step": 7251
    },
    {
      "epoch": 0.5232134482883013,
      "grad_norm": 4.2437300713904085,
      "learning_rate": 1.9467272829491347e-06,
      "loss": 0.9515,
      "step": 7252
    },
    {
      "epoch": 0.5232855957577288,
      "grad_norm": 3.497045223398552,
      "learning_rate": 1.9462600904205787e-06,
      "loss": 1.0073,
      "step": 7253
    },
    {
      "epoch": 0.5233577432271563,
      "grad_norm": 6.962411503939947,
      "learning_rate": 1.9457929008265616e-06,
      "loss": 1.0364,
      "step": 7254
    },
    {
      "epoch": 0.5234298906965839,
      "grad_norm": 12.730334505659172,
      "learning_rate": 1.945325714192593e-06,
      "loss": 0.9531,
      "step": 7255
    },
    {
      "epoch": 0.5235020381660114,
      "grad_norm": 5.971759456872169,
      "learning_rate": 1.9448585305441843e-06,
      "loss": 0.8981,
      "step": 7256
    },
    {
      "epoch": 0.5235741856354388,
      "grad_norm": 4.723177418061421,
      "learning_rate": 1.9443913499068477e-06,
      "loss": 0.9668,
      "step": 7257
    },
    {
      "epoch": 0.5236463331048663,
      "grad_norm": 9.705470196409777,
      "learning_rate": 1.9439241723060933e-06,
      "loss": 1.0312,
      "step": 7258
    },
    {
      "epoch": 0.5237184805742938,
      "grad_norm": 4.333726033732314,
      "learning_rate": 1.943456997767432e-06,
      "loss": 0.9227,
      "step": 7259
    },
    {
      "epoch": 0.5237906280437213,
      "grad_norm": 12.40646954013088,
      "learning_rate": 1.9429898263163757e-06,
      "loss": 0.997,
      "step": 7260
    },
    {
      "epoch": 0.5238627755131489,
      "grad_norm": 4.272002319135287,
      "learning_rate": 1.942522657978433e-06,
      "loss": 1.0684,
      "step": 7261
    },
    {
      "epoch": 0.5239349229825764,
      "grad_norm": 4.807194745652795,
      "learning_rate": 1.9420554927791156e-06,
      "loss": 0.9529,
      "step": 7262
    },
    {
      "epoch": 0.5240070704520039,
      "grad_norm": 5.263539249256796,
      "learning_rate": 1.941588330743933e-06,
      "loss": 1.0455,
      "step": 7263
    },
    {
      "epoch": 0.5240792179214314,
      "grad_norm": 3.721040549209302,
      "learning_rate": 1.941121171898396e-06,
      "loss": 0.9439,
      "step": 7264
    },
    {
      "epoch": 0.5241513653908589,
      "grad_norm": 5.657303348483075,
      "learning_rate": 1.940654016268013e-06,
      "loss": 0.9654,
      "step": 7265
    },
    {
      "epoch": 0.5242235128602865,
      "grad_norm": 4.2496357930026125,
      "learning_rate": 1.940186863878295e-06,
      "loss": 0.9415,
      "step": 7266
    },
    {
      "epoch": 0.524295660329714,
      "grad_norm": 3.846250263619499,
      "learning_rate": 1.9397197147547512e-06,
      "loss": 1.0007,
      "step": 7267
    },
    {
      "epoch": 0.5243678077991415,
      "grad_norm": 0.8816752577821619,
      "learning_rate": 1.93925256892289e-06,
      "loss": 0.8177,
      "step": 7268
    },
    {
      "epoch": 0.5244399552685689,
      "grad_norm": 3.32756875990996,
      "learning_rate": 1.9387854264082214e-06,
      "loss": 0.9561,
      "step": 7269
    },
    {
      "epoch": 0.5245121027379964,
      "grad_norm": 8.426707054636939,
      "learning_rate": 1.938318287236254e-06,
      "loss": 0.9576,
      "step": 7270
    },
    {
      "epoch": 0.5245842502074239,
      "grad_norm": 3.9498579072142097,
      "learning_rate": 1.9378511514324963e-06,
      "loss": 0.9956,
      "step": 7271
    },
    {
      "epoch": 0.5246563976768515,
      "grad_norm": 0.6959336967004621,
      "learning_rate": 1.937384019022458e-06,
      "loss": 0.8144,
      "step": 7272
    },
    {
      "epoch": 0.524728545146279,
      "grad_norm": 6.733863758056035,
      "learning_rate": 1.936916890031646e-06,
      "loss": 0.8423,
      "step": 7273
    },
    {
      "epoch": 0.5248006926157065,
      "grad_norm": 5.199297989629074,
      "learning_rate": 1.936449764485569e-06,
      "loss": 0.8455,
      "step": 7274
    },
    {
      "epoch": 0.524872840085134,
      "grad_norm": 14.125526620955975,
      "learning_rate": 1.9359826424097354e-06,
      "loss": 0.8689,
      "step": 7275
    },
    {
      "epoch": 0.5249449875545615,
      "grad_norm": 5.0933251701076525,
      "learning_rate": 1.935515523829653e-06,
      "loss": 0.9324,
      "step": 7276
    },
    {
      "epoch": 0.5250171350239891,
      "grad_norm": 6.467362259620839,
      "learning_rate": 1.9350484087708287e-06,
      "loss": 1.0413,
      "step": 7277
    },
    {
      "epoch": 0.5250892824934166,
      "grad_norm": 3.486578684973806,
      "learning_rate": 1.93458129725877e-06,
      "loss": 0.9776,
      "step": 7278
    },
    {
      "epoch": 0.5251614299628441,
      "grad_norm": 3.599759591870365,
      "learning_rate": 1.9341141893189858e-06,
      "loss": 0.9945,
      "step": 7279
    },
    {
      "epoch": 0.5252335774322716,
      "grad_norm": 10.130783877984067,
      "learning_rate": 1.93364708497698e-06,
      "loss": 1.004,
      "step": 7280
    },
    {
      "epoch": 0.525305724901699,
      "grad_norm": 4.471633514345859,
      "learning_rate": 1.9331799842582627e-06,
      "loss": 0.9105,
      "step": 7281
    },
    {
      "epoch": 0.5253778723711265,
      "grad_norm": 3.2802064235209656,
      "learning_rate": 1.932712887188339e-06,
      "loss": 0.9655,
      "step": 7282
    },
    {
      "epoch": 0.5254500198405541,
      "grad_norm": 3.5642373382885015,
      "learning_rate": 1.9322457937927143e-06,
      "loss": 1.0498,
      "step": 7283
    },
    {
      "epoch": 0.5255221673099816,
      "grad_norm": 5.1956506145850385,
      "learning_rate": 1.9317787040968973e-06,
      "loss": 1.0674,
      "step": 7284
    },
    {
      "epoch": 0.5255943147794091,
      "grad_norm": 6.0829039470031745,
      "learning_rate": 1.931311618126392e-06,
      "loss": 1.0118,
      "step": 7285
    },
    {
      "epoch": 0.5256664622488366,
      "grad_norm": 3.3290846609647233,
      "learning_rate": 1.930844535906705e-06,
      "loss": 0.9936,
      "step": 7286
    },
    {
      "epoch": 0.5257386097182641,
      "grad_norm": 3.5210039146377743,
      "learning_rate": 1.9303774574633423e-06,
      "loss": 1.0433,
      "step": 7287
    },
    {
      "epoch": 0.5258107571876917,
      "grad_norm": 4.226809115451266,
      "learning_rate": 1.9299103828218084e-06,
      "loss": 0.9782,
      "step": 7288
    },
    {
      "epoch": 0.5258829046571192,
      "grad_norm": 3.787882135331958,
      "learning_rate": 1.9294433120076094e-06,
      "loss": 1.0082,
      "step": 7289
    },
    {
      "epoch": 0.5259550521265467,
      "grad_norm": 3.2328114439057756,
      "learning_rate": 1.9289762450462495e-06,
      "loss": 0.9622,
      "step": 7290
    },
    {
      "epoch": 0.5260271995959742,
      "grad_norm": 4.345603492537656,
      "learning_rate": 1.9285091819632344e-06,
      "loss": 0.9041,
      "step": 7291
    },
    {
      "epoch": 0.5260993470654017,
      "grad_norm": 4.5077615136712526,
      "learning_rate": 1.9280421227840676e-06,
      "loss": 0.8966,
      "step": 7292
    },
    {
      "epoch": 0.5261714945348291,
      "grad_norm": 0.7014636439488817,
      "learning_rate": 1.927575067534255e-06,
      "loss": 0.803,
      "step": 7293
    },
    {
      "epoch": 0.5262436420042567,
      "grad_norm": 3.747311327601981,
      "learning_rate": 1.9271080162392995e-06,
      "loss": 1.0145,
      "step": 7294
    },
    {
      "epoch": 0.5263157894736842,
      "grad_norm": 4.318606931577332,
      "learning_rate": 1.926640968924705e-06,
      "loss": 1.1018,
      "step": 7295
    },
    {
      "epoch": 0.5263879369431117,
      "grad_norm": 4.054715253848896,
      "learning_rate": 1.926173925615976e-06,
      "loss": 0.9693,
      "step": 7296
    },
    {
      "epoch": 0.5264600844125392,
      "grad_norm": 3.7678236498212523,
      "learning_rate": 1.9257068863386156e-06,
      "loss": 0.8897,
      "step": 7297
    },
    {
      "epoch": 0.5265322318819667,
      "grad_norm": 3.7572641587244116,
      "learning_rate": 1.925239851118127e-06,
      "loss": 1.0002,
      "step": 7298
    },
    {
      "epoch": 0.5266043793513943,
      "grad_norm": 3.771075218128279,
      "learning_rate": 1.9247728199800137e-06,
      "loss": 1.0561,
      "step": 7299
    },
    {
      "epoch": 0.5266765268208218,
      "grad_norm": 4.059768227787702,
      "learning_rate": 1.924305792949778e-06,
      "loss": 1.0174,
      "step": 7300
    },
    {
      "epoch": 0.5267486742902493,
      "grad_norm": 6.003712776953466,
      "learning_rate": 1.9238387700529226e-06,
      "loss": 0.9883,
      "step": 7301
    },
    {
      "epoch": 0.5268208217596768,
      "grad_norm": 6.351989206242459,
      "learning_rate": 1.9233717513149495e-06,
      "loss": 1.0887,
      "step": 7302
    },
    {
      "epoch": 0.5268929692291043,
      "grad_norm": 3.616644602296622,
      "learning_rate": 1.9229047367613624e-06,
      "loss": 0.9885,
      "step": 7303
    },
    {
      "epoch": 0.5269651166985317,
      "grad_norm": 6.6695438216431535,
      "learning_rate": 1.922437726417662e-06,
      "loss": 0.994,
      "step": 7304
    },
    {
      "epoch": 0.5270372641679593,
      "grad_norm": 4.868909895380684,
      "learning_rate": 1.9219707203093496e-06,
      "loss": 0.9693,
      "step": 7305
    },
    {
      "epoch": 0.5271094116373868,
      "grad_norm": 3.4495463750298305,
      "learning_rate": 1.9215037184619284e-06,
      "loss": 0.8833,
      "step": 7306
    },
    {
      "epoch": 0.5271815591068143,
      "grad_norm": 5.681285774949243,
      "learning_rate": 1.921036720900897e-06,
      "loss": 0.9803,
      "step": 7307
    },
    {
      "epoch": 0.5272537065762418,
      "grad_norm": 3.005202233337523,
      "learning_rate": 1.9205697276517595e-06,
      "loss": 0.9563,
      "step": 7308
    },
    {
      "epoch": 0.5273258540456693,
      "grad_norm": 0.6442262563700669,
      "learning_rate": 1.920102738740015e-06,
      "loss": 0.8181,
      "step": 7309
    },
    {
      "epoch": 0.5273980015150969,
      "grad_norm": 3.9812496119428653,
      "learning_rate": 1.919635754191163e-06,
      "loss": 0.9974,
      "step": 7310
    },
    {
      "epoch": 0.5274701489845244,
      "grad_norm": 4.440424425614047,
      "learning_rate": 1.919168774030706e-06,
      "loss": 0.94,
      "step": 7311
    },
    {
      "epoch": 0.5275422964539519,
      "grad_norm": 2.680681227887892,
      "learning_rate": 1.9187017982841426e-06,
      "loss": 0.969,
      "step": 7312
    },
    {
      "epoch": 0.5276144439233794,
      "grad_norm": 3.735956854552738,
      "learning_rate": 1.918234826976973e-06,
      "loss": 0.9901,
      "step": 7313
    },
    {
      "epoch": 0.5276865913928069,
      "grad_norm": 3.3088816908168903,
      "learning_rate": 1.917767860134697e-06,
      "loss": 1.0089,
      "step": 7314
    },
    {
      "epoch": 0.5277587388622345,
      "grad_norm": 3.6632413615054147,
      "learning_rate": 1.917300897782814e-06,
      "loss": 0.9541,
      "step": 7315
    },
    {
      "epoch": 0.5278308863316619,
      "grad_norm": 6.772598302612867,
      "learning_rate": 1.9168339399468223e-06,
      "loss": 0.9019,
      "step": 7316
    },
    {
      "epoch": 0.5279030338010894,
      "grad_norm": 0.8754742903689842,
      "learning_rate": 1.9163669866522216e-06,
      "loss": 0.8744,
      "step": 7317
    },
    {
      "epoch": 0.5279751812705169,
      "grad_norm": 4.790366662631492,
      "learning_rate": 1.91590003792451e-06,
      "loss": 0.9817,
      "step": 7318
    },
    {
      "epoch": 0.5280473287399444,
      "grad_norm": 15.024989356927387,
      "learning_rate": 1.9154330937891857e-06,
      "loss": 0.9543,
      "step": 7319
    },
    {
      "epoch": 0.5281194762093719,
      "grad_norm": 0.6049774884142061,
      "learning_rate": 1.9149661542717478e-06,
      "loss": 0.758,
      "step": 7320
    },
    {
      "epoch": 0.5281916236787995,
      "grad_norm": 5.699717384158388,
      "learning_rate": 1.9144992193976933e-06,
      "loss": 0.9526,
      "step": 7321
    },
    {
      "epoch": 0.528263771148227,
      "grad_norm": 30.340496293436455,
      "learning_rate": 1.914032289192519e-06,
      "loss": 0.9624,
      "step": 7322
    },
    {
      "epoch": 0.5283359186176545,
      "grad_norm": 2.548207314841,
      "learning_rate": 1.9135653636817243e-06,
      "loss": 1.0858,
      "step": 7323
    },
    {
      "epoch": 0.528408066087082,
      "grad_norm": 3.275523829609152,
      "learning_rate": 1.9130984428908042e-06,
      "loss": 1.0154,
      "step": 7324
    },
    {
      "epoch": 0.5284802135565095,
      "grad_norm": 4.264502968129527,
      "learning_rate": 1.912631526845257e-06,
      "loss": 0.9396,
      "step": 7325
    },
    {
      "epoch": 0.5285523610259371,
      "grad_norm": 5.064017739432121,
      "learning_rate": 1.912164615570579e-06,
      "loss": 0.887,
      "step": 7326
    },
    {
      "epoch": 0.5286245084953646,
      "grad_norm": 5.9682837498890295,
      "learning_rate": 1.9116977090922655e-06,
      "loss": 0.8646,
      "step": 7327
    },
    {
      "epoch": 0.528696655964792,
      "grad_norm": 7.465673858623218,
      "learning_rate": 1.9112308074358133e-06,
      "loss": 0.9418,
      "step": 7328
    },
    {
      "epoch": 0.5287688034342195,
      "grad_norm": 4.335941239089041,
      "learning_rate": 1.9107639106267186e-06,
      "loss": 0.9842,
      "step": 7329
    },
    {
      "epoch": 0.528840950903647,
      "grad_norm": 4.58044041744073,
      "learning_rate": 1.910297018690476e-06,
      "loss": 0.996,
      "step": 7330
    },
    {
      "epoch": 0.5289130983730745,
      "grad_norm": 4.0547237210870115,
      "learning_rate": 1.9098301316525818e-06,
      "loss": 0.9383,
      "step": 7331
    },
    {
      "epoch": 0.5289852458425021,
      "grad_norm": 4.048788082441885,
      "learning_rate": 1.909363249538529e-06,
      "loss": 1.0203,
      "step": 7332
    },
    {
      "epoch": 0.5290573933119296,
      "grad_norm": 2.868706989173674,
      "learning_rate": 1.9088963723738147e-06,
      "loss": 0.8903,
      "step": 7333
    },
    {
      "epoch": 0.5291295407813571,
      "grad_norm": 2.8023891542072903,
      "learning_rate": 1.9084295001839316e-06,
      "loss": 0.9577,
      "step": 7334
    },
    {
      "epoch": 0.5292016882507846,
      "grad_norm": 3.347995807604062,
      "learning_rate": 1.907962632994375e-06,
      "loss": 1.0326,
      "step": 7335
    },
    {
      "epoch": 0.5292738357202121,
      "grad_norm": 6.241518903883819,
      "learning_rate": 1.907495770830638e-06,
      "loss": 0.9217,
      "step": 7336
    },
    {
      "epoch": 0.5293459831896397,
      "grad_norm": 3.6802111210593407,
      "learning_rate": 1.9070289137182146e-06,
      "loss": 0.9687,
      "step": 7337
    },
    {
      "epoch": 0.5294181306590672,
      "grad_norm": 0.7470826391063329,
      "learning_rate": 1.9065620616825982e-06,
      "loss": 0.8101,
      "step": 7338
    },
    {
      "epoch": 0.5294902781284947,
      "grad_norm": 3.4632621174289646,
      "learning_rate": 1.9060952147492814e-06,
      "loss": 0.9858,
      "step": 7339
    },
    {
      "epoch": 0.5295624255979221,
      "grad_norm": 3.1271446498185322,
      "learning_rate": 1.9056283729437574e-06,
      "loss": 0.9631,
      "step": 7340
    },
    {
      "epoch": 0.5296345730673496,
      "grad_norm": 3.1648377775589323,
      "learning_rate": 1.9051615362915187e-06,
      "loss": 0.9374,
      "step": 7341
    },
    {
      "epoch": 0.5297067205367771,
      "grad_norm": 2.231337823929194,
      "learning_rate": 1.9046947048180577e-06,
      "loss": 0.9149,
      "step": 7342
    },
    {
      "epoch": 0.5297788680062047,
      "grad_norm": 4.362658504896605,
      "learning_rate": 1.9042278785488662e-06,
      "loss": 0.981,
      "step": 7343
    },
    {
      "epoch": 0.5298510154756322,
      "grad_norm": 0.7241525778224845,
      "learning_rate": 1.9037610575094348e-06,
      "loss": 0.8647,
      "step": 7344
    },
    {
      "epoch": 0.5299231629450597,
      "grad_norm": 5.412871434811638,
      "learning_rate": 1.9032942417252565e-06,
      "loss": 0.9168,
      "step": 7345
    },
    {
      "epoch": 0.5299953104144872,
      "grad_norm": 4.081610699030164,
      "learning_rate": 1.9028274312218209e-06,
      "loss": 0.9888,
      "step": 7346
    },
    {
      "epoch": 0.5300674578839147,
      "grad_norm": 0.7188636855732623,
      "learning_rate": 1.9023606260246204e-06,
      "loss": 0.8591,
      "step": 7347
    },
    {
      "epoch": 0.5301396053533423,
      "grad_norm": 7.598377215150018,
      "learning_rate": 1.9018938261591448e-06,
      "loss": 0.8903,
      "step": 7348
    },
    {
      "epoch": 0.5302117528227698,
      "grad_norm": 5.23764655515887,
      "learning_rate": 1.9014270316508832e-06,
      "loss": 0.9529,
      "step": 7349
    },
    {
      "epoch": 0.5302839002921973,
      "grad_norm": 5.401648001508628,
      "learning_rate": 1.9009602425253274e-06,
      "loss": 0.8241,
      "step": 7350
    },
    {
      "epoch": 0.5303560477616248,
      "grad_norm": 6.108196330214613,
      "learning_rate": 1.9004934588079655e-06,
      "loss": 1.004,
      "step": 7351
    },
    {
      "epoch": 0.5304281952310522,
      "grad_norm": 5.253879022350452,
      "learning_rate": 1.9000266805242878e-06,
      "loss": 0.9959,
      "step": 7352
    },
    {
      "epoch": 0.5305003427004797,
      "grad_norm": 3.8134828144955155,
      "learning_rate": 1.8995599076997832e-06,
      "loss": 0.9882,
      "step": 7353
    },
    {
      "epoch": 0.5305724901699073,
      "grad_norm": 3.3769987333801983,
      "learning_rate": 1.8990931403599397e-06,
      "loss": 0.924,
      "step": 7354
    },
    {
      "epoch": 0.5306446376393348,
      "grad_norm": 9.241469290035642,
      "learning_rate": 1.898626378530247e-06,
      "loss": 0.9273,
      "step": 7355
    },
    {
      "epoch": 0.5307167851087623,
      "grad_norm": 4.218709987874491,
      "learning_rate": 1.8981596222361913e-06,
      "loss": 0.8887,
      "step": 7356
    },
    {
      "epoch": 0.5307889325781898,
      "grad_norm": 3.3676767369982876,
      "learning_rate": 1.8976928715032626e-06,
      "loss": 0.9599,
      "step": 7357
    },
    {
      "epoch": 0.5308610800476173,
      "grad_norm": 3.7992041708156417,
      "learning_rate": 1.8972261263569466e-06,
      "loss": 0.9375,
      "step": 7358
    },
    {
      "epoch": 0.5309332275170449,
      "grad_norm": 5.5320725341256285,
      "learning_rate": 1.8967593868227325e-06,
      "loss": 1.0185,
      "step": 7359
    },
    {
      "epoch": 0.5310053749864724,
      "grad_norm": 6.014485992006871,
      "learning_rate": 1.8962926529261058e-06,
      "loss": 1.0738,
      "step": 7360
    },
    {
      "epoch": 0.5310775224558999,
      "grad_norm": 3.3838576889235044,
      "learning_rate": 1.8958259246925527e-06,
      "loss": 0.8521,
      "step": 7361
    },
    {
      "epoch": 0.5311496699253274,
      "grad_norm": 4.888652757278839,
      "learning_rate": 1.895359202147561e-06,
      "loss": 0.9646,
      "step": 7362
    },
    {
      "epoch": 0.5312218173947548,
      "grad_norm": 4.9414234266623405,
      "learning_rate": 1.8948924853166154e-06,
      "loss": 0.9721,
      "step": 7363
    },
    {
      "epoch": 0.5312939648641823,
      "grad_norm": 3.3020778298958886,
      "learning_rate": 1.8944257742252023e-06,
      "loss": 0.9193,
      "step": 7364
    },
    {
      "epoch": 0.5313661123336099,
      "grad_norm": 3.770858104195911,
      "learning_rate": 1.8939590688988072e-06,
      "loss": 1.0851,
      "step": 7365
    },
    {
      "epoch": 0.5314382598030374,
      "grad_norm": 4.056822815405034,
      "learning_rate": 1.8934923693629141e-06,
      "loss": 1.0483,
      "step": 7366
    },
    {
      "epoch": 0.5315104072724649,
      "grad_norm": 4.833086051315344,
      "learning_rate": 1.8930256756430092e-06,
      "loss": 1.0002,
      "step": 7367
    },
    {
      "epoch": 0.5315825547418924,
      "grad_norm": 3.2893860718955548,
      "learning_rate": 1.8925589877645752e-06,
      "loss": 0.9128,
      "step": 7368
    },
    {
      "epoch": 0.5316547022113199,
      "grad_norm": 5.3538275387404095,
      "learning_rate": 1.892092305753098e-06,
      "loss": 1.0272,
      "step": 7369
    },
    {
      "epoch": 0.5317268496807475,
      "grad_norm": 3.439054519110245,
      "learning_rate": 1.8916256296340604e-06,
      "loss": 0.9901,
      "step": 7370
    },
    {
      "epoch": 0.531798997150175,
      "grad_norm": 3.199361832880944,
      "learning_rate": 1.8911589594329453e-06,
      "loss": 0.9295,
      "step": 7371
    },
    {
      "epoch": 0.5318711446196025,
      "grad_norm": 4.605102628046822,
      "learning_rate": 1.8906922951752374e-06,
      "loss": 0.914,
      "step": 7372
    },
    {
      "epoch": 0.53194329208903,
      "grad_norm": 3.71322387933504,
      "learning_rate": 1.8902256368864176e-06,
      "loss": 0.9975,
      "step": 7373
    },
    {
      "epoch": 0.5320154395584575,
      "grad_norm": 5.757246012880814,
      "learning_rate": 1.8897589845919701e-06,
      "loss": 1.1496,
      "step": 7374
    },
    {
      "epoch": 0.532087587027885,
      "grad_norm": 0.8311481356052044,
      "learning_rate": 1.8892923383173767e-06,
      "loss": 0.8314,
      "step": 7375
    },
    {
      "epoch": 0.5321597344973125,
      "grad_norm": 4.39628458946055,
      "learning_rate": 1.888825698088118e-06,
      "loss": 1.0208,
      "step": 7376
    },
    {
      "epoch": 0.53223188196674,
      "grad_norm": 4.779029879265672,
      "learning_rate": 1.888359063929677e-06,
      "loss": 0.9947,
      "step": 7377
    },
    {
      "epoch": 0.5323040294361675,
      "grad_norm": 3.1699545171705816,
      "learning_rate": 1.8878924358675338e-06,
      "loss": 1.012,
      "step": 7378
    },
    {
      "epoch": 0.532376176905595,
      "grad_norm": 5.451357317250612,
      "learning_rate": 1.8874258139271702e-06,
      "loss": 0.9451,
      "step": 7379
    },
    {
      "epoch": 0.5324483243750225,
      "grad_norm": 2.4585752745286613,
      "learning_rate": 1.8869591981340653e-06,
      "loss": 1.0163,
      "step": 7380
    },
    {
      "epoch": 0.5325204718444501,
      "grad_norm": 2.8039677368462894,
      "learning_rate": 1.886492588513701e-06,
      "loss": 0.7893,
      "step": 7381
    },
    {
      "epoch": 0.5325926193138776,
      "grad_norm": 6.385857704285679,
      "learning_rate": 1.8860259850915564e-06,
      "loss": 1.0147,
      "step": 7382
    },
    {
      "epoch": 0.5326647667833051,
      "grad_norm": 5.500005201857448,
      "learning_rate": 1.8855593878931098e-06,
      "loss": 0.9571,
      "step": 7383
    },
    {
      "epoch": 0.5327369142527326,
      "grad_norm": 3.454340656041462,
      "learning_rate": 1.8850927969438421e-06,
      "loss": 1.018,
      "step": 7384
    },
    {
      "epoch": 0.5328090617221601,
      "grad_norm": 7.444806525058703,
      "learning_rate": 1.8846262122692308e-06,
      "loss": 1.0111,
      "step": 7385
    },
    {
      "epoch": 0.5328812091915877,
      "grad_norm": 6.712307560327061,
      "learning_rate": 1.8841596338947557e-06,
      "loss": 0.9499,
      "step": 7386
    },
    {
      "epoch": 0.5329533566610151,
      "grad_norm": 2.762553173951573,
      "learning_rate": 1.883693061845894e-06,
      "loss": 0.8789,
      "step": 7387
    },
    {
      "epoch": 0.5330255041304426,
      "grad_norm": 13.223704606704985,
      "learning_rate": 1.8832264961481232e-06,
      "loss": 0.9065,
      "step": 7388
    },
    {
      "epoch": 0.5330976515998701,
      "grad_norm": 0.6661648177682038,
      "learning_rate": 1.8827599368269218e-06,
      "loss": 0.8373,
      "step": 7389
    },
    {
      "epoch": 0.5331697990692976,
      "grad_norm": 4.0978675693246425,
      "learning_rate": 1.8822933839077656e-06,
      "loss": 1.0065,
      "step": 7390
    },
    {
      "epoch": 0.5332419465387251,
      "grad_norm": 5.570490894430368,
      "learning_rate": 1.8818268374161329e-06,
      "loss": 0.9032,
      "step": 7391
    },
    {
      "epoch": 0.5333140940081527,
      "grad_norm": 4.711056734666966,
      "learning_rate": 1.8813602973774982e-06,
      "loss": 0.8939,
      "step": 7392
    },
    {
      "epoch": 0.5333862414775802,
      "grad_norm": 4.486285184954495,
      "learning_rate": 1.8808937638173387e-06,
      "loss": 1.0367,
      "step": 7393
    },
    {
      "epoch": 0.5334583889470077,
      "grad_norm": 3.1989820530189146,
      "learning_rate": 1.8804272367611302e-06,
      "loss": 1.0539,
      "step": 7394
    },
    {
      "epoch": 0.5335305364164352,
      "grad_norm": 2.7938156017819527,
      "learning_rate": 1.879960716234347e-06,
      "loss": 0.9313,
      "step": 7395
    },
    {
      "epoch": 0.5336026838858627,
      "grad_norm": 5.7145186649251,
      "learning_rate": 1.8794942022624653e-06,
      "loss": 0.9065,
      "step": 7396
    },
    {
      "epoch": 0.5336748313552903,
      "grad_norm": 3.9312717891992914,
      "learning_rate": 1.8790276948709593e-06,
      "loss": 0.9416,
      "step": 7397
    },
    {
      "epoch": 0.5337469788247178,
      "grad_norm": 5.184855878807895,
      "learning_rate": 1.878561194085302e-06,
      "loss": 0.9609,
      "step": 7398
    },
    {
      "epoch": 0.5338191262941452,
      "grad_norm": 0.7435719477276216,
      "learning_rate": 1.8780946999309695e-06,
      "loss": 0.8485,
      "step": 7399
    },
    {
      "epoch": 0.5338912737635727,
      "grad_norm": 3.343420279815337,
      "learning_rate": 1.877628212433433e-06,
      "loss": 1.0394,
      "step": 7400
    },
    {
      "epoch": 0.5339634212330002,
      "grad_norm": 2.8335298769230453,
      "learning_rate": 1.8771617316181678e-06,
      "loss": 0.993,
      "step": 7401
    },
    {
      "epoch": 0.5340355687024277,
      "grad_norm": 6.460439162311017,
      "learning_rate": 1.8766952575106455e-06,
      "loss": 0.9247,
      "step": 7402
    },
    {
      "epoch": 0.5341077161718553,
      "grad_norm": 3.9221362752626225,
      "learning_rate": 1.876228790136339e-06,
      "loss": 1.03,
      "step": 7403
    },
    {
      "epoch": 0.5341798636412828,
      "grad_norm": 4.04229028878538,
      "learning_rate": 1.8757623295207194e-06,
      "loss": 1.0682,
      "step": 7404
    },
    {
      "epoch": 0.5342520111107103,
      "grad_norm": 4.525528867385669,
      "learning_rate": 1.8752958756892593e-06,
      "loss": 1.0064,
      "step": 7405
    },
    {
      "epoch": 0.5343241585801378,
      "grad_norm": 2.8086793167598416,
      "learning_rate": 1.8748294286674303e-06,
      "loss": 0.8973,
      "step": 7406
    },
    {
      "epoch": 0.5343963060495653,
      "grad_norm": 3.778401586664639,
      "learning_rate": 1.8743629884807018e-06,
      "loss": 0.9436,
      "step": 7407
    },
    {
      "epoch": 0.5344684535189929,
      "grad_norm": 2.6299521645666744,
      "learning_rate": 1.8738965551545463e-06,
      "loss": 1.019,
      "step": 7408
    },
    {
      "epoch": 0.5345406009884204,
      "grad_norm": 0.8192871224703311,
      "learning_rate": 1.8734301287144333e-06,
      "loss": 0.8535,
      "step": 7409
    },
    {
      "epoch": 0.5346127484578478,
      "grad_norm": 6.546795464444531,
      "learning_rate": 1.8729637091858313e-06,
      "loss": 0.9028,
      "step": 7410
    },
    {
      "epoch": 0.5346848959272753,
      "grad_norm": 3.058008130598518,
      "learning_rate": 1.8724972965942118e-06,
      "loss": 0.9042,
      "step": 7411
    },
    {
      "epoch": 0.5347570433967028,
      "grad_norm": 0.6881114884590961,
      "learning_rate": 1.8720308909650423e-06,
      "loss": 0.8149,
      "step": 7412
    },
    {
      "epoch": 0.5348291908661303,
      "grad_norm": 0.8120223989145612,
      "learning_rate": 1.8715644923237925e-06,
      "loss": 0.856,
      "step": 7413
    },
    {
      "epoch": 0.5349013383355579,
      "grad_norm": 0.7279271671547335,
      "learning_rate": 1.8710981006959308e-06,
      "loss": 0.8354,
      "step": 7414
    },
    {
      "epoch": 0.5349734858049854,
      "grad_norm": 10.688224901745517,
      "learning_rate": 1.8706317161069241e-06,
      "loss": 1.037,
      "step": 7415
    },
    {
      "epoch": 0.5350456332744129,
      "grad_norm": 3.8600165571341916,
      "learning_rate": 1.8701653385822402e-06,
      "loss": 1.0283,
      "step": 7416
    },
    {
      "epoch": 0.5351177807438404,
      "grad_norm": 5.979977259257784,
      "learning_rate": 1.8696989681473468e-06,
      "loss": 0.9827,
      "step": 7417
    },
    {
      "epoch": 0.5351899282132679,
      "grad_norm": 5.91052866163921,
      "learning_rate": 1.8692326048277108e-06,
      "loss": 0.928,
      "step": 7418
    },
    {
      "epoch": 0.5352620756826955,
      "grad_norm": 3.070924343656628,
      "learning_rate": 1.8687662486487976e-06,
      "loss": 0.9083,
      "step": 7419
    },
    {
      "epoch": 0.535334223152123,
      "grad_norm": 4.5402070012564195,
      "learning_rate": 1.868299899636074e-06,
      "loss": 0.9526,
      "step": 7420
    },
    {
      "epoch": 0.5354063706215505,
      "grad_norm": 0.8210977338264143,
      "learning_rate": 1.8678335578150056e-06,
      "loss": 0.8794,
      "step": 7421
    },
    {
      "epoch": 0.5354785180909779,
      "grad_norm": 5.026458733471507,
      "learning_rate": 1.8673672232110564e-06,
      "loss": 0.9669,
      "step": 7422
    },
    {
      "epoch": 0.5355506655604054,
      "grad_norm": 3.6137647480566404,
      "learning_rate": 1.866900895849693e-06,
      "loss": 0.8984,
      "step": 7423
    },
    {
      "epoch": 0.535622813029833,
      "grad_norm": 3.026618486712742,
      "learning_rate": 1.8664345757563784e-06,
      "loss": 0.9735,
      "step": 7424
    },
    {
      "epoch": 0.5356949604992605,
      "grad_norm": 2.3386614683144944,
      "learning_rate": 1.8659682629565779e-06,
      "loss": 0.9393,
      "step": 7425
    },
    {
      "epoch": 0.535767107968688,
      "grad_norm": 3.114194116332773,
      "learning_rate": 1.8655019574757542e-06,
      "loss": 0.9313,
      "step": 7426
    },
    {
      "epoch": 0.5358392554381155,
      "grad_norm": 3.2167776045832936,
      "learning_rate": 1.8650356593393703e-06,
      "loss": 0.9931,
      "step": 7427
    },
    {
      "epoch": 0.535911402907543,
      "grad_norm": 12.472678882495908,
      "learning_rate": 1.8645693685728896e-06,
      "loss": 0.8813,
      "step": 7428
    },
    {
      "epoch": 0.5359835503769705,
      "grad_norm": 4.100228810322306,
      "learning_rate": 1.8641030852017742e-06,
      "loss": 0.944,
      "step": 7429
    },
    {
      "epoch": 0.5360556978463981,
      "grad_norm": 4.414513596972101,
      "learning_rate": 1.8636368092514869e-06,
      "loss": 0.9253,
      "step": 7430
    },
    {
      "epoch": 0.5361278453158256,
      "grad_norm": 4.201984917404363,
      "learning_rate": 1.863170540747488e-06,
      "loss": 0.9516,
      "step": 7431
    },
    {
      "epoch": 0.5361999927852531,
      "grad_norm": 3.399882112591053,
      "learning_rate": 1.8627042797152394e-06,
      "loss": 0.8923,
      "step": 7432
    },
    {
      "epoch": 0.5362721402546806,
      "grad_norm": 4.472354955394965,
      "learning_rate": 1.8622380261802025e-06,
      "loss": 1.0158,
      "step": 7433
    },
    {
      "epoch": 0.536344287724108,
      "grad_norm": 8.667159824523148,
      "learning_rate": 1.8617717801678356e-06,
      "loss": 0.9818,
      "step": 7434
    },
    {
      "epoch": 0.5364164351935355,
      "grad_norm": 3.309207935817208,
      "learning_rate": 1.8613055417036014e-06,
      "loss": 0.9786,
      "step": 7435
    },
    {
      "epoch": 0.5364885826629631,
      "grad_norm": 3.3998011867513718,
      "learning_rate": 1.860839310812958e-06,
      "loss": 0.964,
      "step": 7436
    },
    {
      "epoch": 0.5365607301323906,
      "grad_norm": 3.2350184588369992,
      "learning_rate": 1.860373087521364e-06,
      "loss": 0.9366,
      "step": 7437
    },
    {
      "epoch": 0.5366328776018181,
      "grad_norm": 2.918438364256116,
      "learning_rate": 1.8599068718542793e-06,
      "loss": 0.8947,
      "step": 7438
    },
    {
      "epoch": 0.5367050250712456,
      "grad_norm": 3.8309833676528044,
      "learning_rate": 1.8594406638371616e-06,
      "loss": 0.9733,
      "step": 7439
    },
    {
      "epoch": 0.5367771725406731,
      "grad_norm": 4.546482973797469,
      "learning_rate": 1.8589744634954687e-06,
      "loss": 0.9863,
      "step": 7440
    },
    {
      "epoch": 0.5368493200101007,
      "grad_norm": 4.498822164038879,
      "learning_rate": 1.8585082708546588e-06,
      "loss": 0.9777,
      "step": 7441
    },
    {
      "epoch": 0.5369214674795282,
      "grad_norm": 3.1819223708892137,
      "learning_rate": 1.858042085940188e-06,
      "loss": 1.0434,
      "step": 7442
    },
    {
      "epoch": 0.5369936149489557,
      "grad_norm": 3.355032878094959,
      "learning_rate": 1.8575759087775134e-06,
      "loss": 0.7735,
      "step": 7443
    },
    {
      "epoch": 0.5370657624183832,
      "grad_norm": 2.496194136008322,
      "learning_rate": 1.8571097393920914e-06,
      "loss": 0.9774,
      "step": 7444
    },
    {
      "epoch": 0.5371379098878107,
      "grad_norm": 3.6215625117972863,
      "learning_rate": 1.8566435778093776e-06,
      "loss": 1.0568,
      "step": 7445
    },
    {
      "epoch": 0.5372100573572381,
      "grad_norm": 3.2698862117836294,
      "learning_rate": 1.8561774240548267e-06,
      "loss": 0.9556,
      "step": 7446
    },
    {
      "epoch": 0.5372822048266657,
      "grad_norm": 3.9597046869154986,
      "learning_rate": 1.855711278153895e-06,
      "loss": 0.9103,
      "step": 7447
    },
    {
      "epoch": 0.5373543522960932,
      "grad_norm": 5.90928063521546,
      "learning_rate": 1.8552451401320367e-06,
      "loss": 1.0525,
      "step": 7448
    },
    {
      "epoch": 0.5374264997655207,
      "grad_norm": 4.47614620596532,
      "learning_rate": 1.854779010014704e-06,
      "loss": 0.9808,
      "step": 7449
    },
    {
      "epoch": 0.5374986472349482,
      "grad_norm": 3.2067989957579717,
      "learning_rate": 1.854312887827353e-06,
      "loss": 0.9499,
      "step": 7450
    },
    {
      "epoch": 0.5375707947043757,
      "grad_norm": 3.5456793698251277,
      "learning_rate": 1.8538467735954355e-06,
      "loss": 0.9933,
      "step": 7451
    },
    {
      "epoch": 0.5376429421738033,
      "grad_norm": 4.154149033570562,
      "learning_rate": 1.8533806673444048e-06,
      "loss": 0.9905,
      "step": 7452
    },
    {
      "epoch": 0.5377150896432308,
      "grad_norm": 4.136485668672598,
      "learning_rate": 1.8529145690997132e-06,
      "loss": 0.9366,
      "step": 7453
    },
    {
      "epoch": 0.5377872371126583,
      "grad_norm": 4.79797410332901,
      "learning_rate": 1.8524484788868123e-06,
      "loss": 0.9134,
      "step": 7454
    },
    {
      "epoch": 0.5378593845820858,
      "grad_norm": 4.878184330033879,
      "learning_rate": 1.8519823967311537e-06,
      "loss": 0.8382,
      "step": 7455
    },
    {
      "epoch": 0.5379315320515133,
      "grad_norm": 6.286180070240454,
      "learning_rate": 1.8515163226581885e-06,
      "loss": 0.7756,
      "step": 7456
    },
    {
      "epoch": 0.5380036795209409,
      "grad_norm": 3.7742144493626526,
      "learning_rate": 1.8510502566933677e-06,
      "loss": 0.9283,
      "step": 7457
    },
    {
      "epoch": 0.5380758269903683,
      "grad_norm": 5.012438751379225,
      "learning_rate": 1.8505841988621404e-06,
      "loss": 1.0183,
      "step": 7458
    },
    {
      "epoch": 0.5381479744597958,
      "grad_norm": 0.7981002961519909,
      "learning_rate": 1.8501181491899568e-06,
      "loss": 0.8377,
      "step": 7459
    },
    {
      "epoch": 0.5382201219292233,
      "grad_norm": 6.587308946766976,
      "learning_rate": 1.849652107702267e-06,
      "loss": 0.912,
      "step": 7460
    },
    {
      "epoch": 0.5382922693986508,
      "grad_norm": 5.966232647729716,
      "learning_rate": 1.849186074424518e-06,
      "loss": 0.9503,
      "step": 7461
    },
    {
      "epoch": 0.5383644168680783,
      "grad_norm": 3.765089701360321,
      "learning_rate": 1.8487200493821603e-06,
      "loss": 0.9476,
      "step": 7462
    },
    {
      "epoch": 0.5384365643375059,
      "grad_norm": 3.6513406773492707,
      "learning_rate": 1.8482540326006402e-06,
      "loss": 0.8577,
      "step": 7463
    },
    {
      "epoch": 0.5385087118069334,
      "grad_norm": 4.80049617904894,
      "learning_rate": 1.847788024105405e-06,
      "loss": 0.9321,
      "step": 7464
    },
    {
      "epoch": 0.5385808592763609,
      "grad_norm": 0.7855195301063477,
      "learning_rate": 1.847322023921903e-06,
      "loss": 0.8199,
      "step": 7465
    },
    {
      "epoch": 0.5386530067457884,
      "grad_norm": 4.261898439082484,
      "learning_rate": 1.8468560320755801e-06,
      "loss": 0.8013,
      "step": 7466
    },
    {
      "epoch": 0.5387251542152159,
      "grad_norm": 3.7308552646624307,
      "learning_rate": 1.8463900485918822e-06,
      "loss": 0.9015,
      "step": 7467
    },
    {
      "epoch": 0.5387973016846435,
      "grad_norm": 4.305676355342227,
      "learning_rate": 1.845924073496255e-06,
      "loss": 1.0873,
      "step": 7468
    },
    {
      "epoch": 0.5388694491540709,
      "grad_norm": 3.991130532168133,
      "learning_rate": 1.8454581068141442e-06,
      "loss": 0.8676,
      "step": 7469
    },
    {
      "epoch": 0.5389415966234984,
      "grad_norm": 3.3053944510158906,
      "learning_rate": 1.8449921485709936e-06,
      "loss": 1.0316,
      "step": 7470
    },
    {
      "epoch": 0.5390137440929259,
      "grad_norm": 4.063565158588663,
      "learning_rate": 1.844526198792248e-06,
      "loss": 0.9972,
      "step": 7471
    },
    {
      "epoch": 0.5390858915623534,
      "grad_norm": 8.995649345840484,
      "learning_rate": 1.8440602575033516e-06,
      "loss": 0.9634,
      "step": 7472
    },
    {
      "epoch": 0.539158039031781,
      "grad_norm": 4.666483444068419,
      "learning_rate": 1.8435943247297466e-06,
      "loss": 0.9714,
      "step": 7473
    },
    {
      "epoch": 0.5392301865012085,
      "grad_norm": 0.6933517455634932,
      "learning_rate": 1.8431284004968772e-06,
      "loss": 0.8241,
      "step": 7474
    },
    {
      "epoch": 0.539302333970636,
      "grad_norm": 5.968375788437082,
      "learning_rate": 1.842662484830185e-06,
      "loss": 1.0168,
      "step": 7475
    },
    {
      "epoch": 0.5393744814400635,
      "grad_norm": 5.917272545682074,
      "learning_rate": 1.8421965777551113e-06,
      "loss": 0.8762,
      "step": 7476
    },
    {
      "epoch": 0.539446628909491,
      "grad_norm": 4.747833159268758,
      "learning_rate": 1.8417306792970994e-06,
      "loss": 0.954,
      "step": 7477
    },
    {
      "epoch": 0.5395187763789185,
      "grad_norm": 3.625336203933745,
      "learning_rate": 1.8412647894815888e-06,
      "loss": 1.0166,
      "step": 7478
    },
    {
      "epoch": 0.5395909238483461,
      "grad_norm": 3.9278478847891347,
      "learning_rate": 1.8407989083340201e-06,
      "loss": 1.0557,
      "step": 7479
    },
    {
      "epoch": 0.5396630713177736,
      "grad_norm": 7.925290065715729,
      "learning_rate": 1.8403330358798342e-06,
      "loss": 1.0135,
      "step": 7480
    },
    {
      "epoch": 0.539735218787201,
      "grad_norm": 4.877200094805472,
      "learning_rate": 1.8398671721444696e-06,
      "loss": 0.955,
      "step": 7481
    },
    {
      "epoch": 0.5398073662566285,
      "grad_norm": 3.4582360188837153,
      "learning_rate": 1.8394013171533663e-06,
      "loss": 0.9881,
      "step": 7482
    },
    {
      "epoch": 0.539879513726056,
      "grad_norm": 2.9236541064300696,
      "learning_rate": 1.8389354709319622e-06,
      "loss": 0.9085,
      "step": 7483
    },
    {
      "epoch": 0.5399516611954835,
      "grad_norm": 3.222925237353682,
      "learning_rate": 1.8384696335056964e-06,
      "loss": 1.0129,
      "step": 7484
    },
    {
      "epoch": 0.5400238086649111,
      "grad_norm": 3.1651041459098255,
      "learning_rate": 1.8380038049000056e-06,
      "loss": 0.9961,
      "step": 7485
    },
    {
      "epoch": 0.5400959561343386,
      "grad_norm": 2.8484250719695963,
      "learning_rate": 1.8375379851403271e-06,
      "loss": 1.0074,
      "step": 7486
    },
    {
      "epoch": 0.5401681036037661,
      "grad_norm": 4.25022214421158,
      "learning_rate": 1.8370721742520984e-06,
      "loss": 0.8929,
      "step": 7487
    },
    {
      "epoch": 0.5402402510731936,
      "grad_norm": 2.961143949452624,
      "learning_rate": 1.8366063722607542e-06,
      "loss": 0.9395,
      "step": 7488
    },
    {
      "epoch": 0.5403123985426211,
      "grad_norm": 4.24754576785805,
      "learning_rate": 1.8361405791917318e-06,
      "loss": 1.0097,
      "step": 7489
    },
    {
      "epoch": 0.5403845460120487,
      "grad_norm": 3.872410555078074,
      "learning_rate": 1.835674795070466e-06,
      "loss": 0.9908,
      "step": 7490
    },
    {
      "epoch": 0.5404566934814762,
      "grad_norm": 4.63140672838085,
      "learning_rate": 1.8352090199223907e-06,
      "loss": 0.9306,
      "step": 7491
    },
    {
      "epoch": 0.5405288409509037,
      "grad_norm": 5.105699996556201,
      "learning_rate": 1.8347432537729416e-06,
      "loss": 0.9128,
      "step": 7492
    },
    {
      "epoch": 0.5406009884203311,
      "grad_norm": 3.1270308952500985,
      "learning_rate": 1.834277496647551e-06,
      "loss": 0.9235,
      "step": 7493
    },
    {
      "epoch": 0.5406731358897586,
      "grad_norm": 5.69922979404773,
      "learning_rate": 1.833811748571653e-06,
      "loss": 0.9135,
      "step": 7494
    },
    {
      "epoch": 0.5407452833591861,
      "grad_norm": 3.819063134473906,
      "learning_rate": 1.8333460095706798e-06,
      "loss": 0.8788,
      "step": 7495
    },
    {
      "epoch": 0.5408174308286137,
      "grad_norm": 5.4952215764753785,
      "learning_rate": 1.8328802796700649e-06,
      "loss": 0.9762,
      "step": 7496
    },
    {
      "epoch": 0.5408895782980412,
      "grad_norm": 4.780237626749163,
      "learning_rate": 1.8324145588952388e-06,
      "loss": 1.0556,
      "step": 7497
    },
    {
      "epoch": 0.5409617257674687,
      "grad_norm": 4.284080071179157,
      "learning_rate": 1.8319488472716331e-06,
      "loss": 0.9478,
      "step": 7498
    },
    {
      "epoch": 0.5410338732368962,
      "grad_norm": 3.54001778010844,
      "learning_rate": 1.8314831448246793e-06,
      "loss": 0.9548,
      "step": 7499
    },
    {
      "epoch": 0.5411060207063237,
      "grad_norm": 3.7216301031120285,
      "learning_rate": 1.8310174515798063e-06,
      "loss": 0.9754,
      "step": 7500
    },
    {
      "epoch": 0.5411781681757513,
      "grad_norm": 4.848187389063325,
      "learning_rate": 1.8305517675624453e-06,
      "loss": 0.9384,
      "step": 7501
    },
    {
      "epoch": 0.5412503156451788,
      "grad_norm": 4.918765099779773,
      "learning_rate": 1.8300860927980253e-06,
      "loss": 0.9445,
      "step": 7502
    },
    {
      "epoch": 0.5413224631146063,
      "grad_norm": 4.109949108096423,
      "learning_rate": 1.829620427311974e-06,
      "loss": 1.0942,
      "step": 7503
    },
    {
      "epoch": 0.5413946105840338,
      "grad_norm": 4.73576067799237,
      "learning_rate": 1.8291547711297211e-06,
      "loss": 1.0258,
      "step": 7504
    },
    {
      "epoch": 0.5414667580534612,
      "grad_norm": 3.252724166053317,
      "learning_rate": 1.8286891242766934e-06,
      "loss": 0.8454,
      "step": 7505
    },
    {
      "epoch": 0.5415389055228887,
      "grad_norm": 5.518115985952436,
      "learning_rate": 1.8282234867783182e-06,
      "loss": 0.9846,
      "step": 7506
    },
    {
      "epoch": 0.5416110529923163,
      "grad_norm": 4.652392946843155,
      "learning_rate": 1.827757858660023e-06,
      "loss": 0.987,
      "step": 7507
    },
    {
      "epoch": 0.5416832004617438,
      "grad_norm": 4.639738158697581,
      "learning_rate": 1.8272922399472333e-06,
      "loss": 0.83,
      "step": 7508
    },
    {
      "epoch": 0.5417553479311713,
      "grad_norm": 3.5484454337551665,
      "learning_rate": 1.8268266306653747e-06,
      "loss": 1.01,
      "step": 7509
    },
    {
      "epoch": 0.5418274954005988,
      "grad_norm": 3.559994268091281,
      "learning_rate": 1.826361030839873e-06,
      "loss": 0.9566,
      "step": 7510
    },
    {
      "epoch": 0.5418996428700263,
      "grad_norm": 3.5836055711343753,
      "learning_rate": 1.825895440496153e-06,
      "loss": 1.0192,
      "step": 7511
    },
    {
      "epoch": 0.5419717903394539,
      "grad_norm": 5.231582124255257,
      "learning_rate": 1.8254298596596376e-06,
      "loss": 0.9328,
      "step": 7512
    },
    {
      "epoch": 0.5420439378088814,
      "grad_norm": 3.9355537812051686,
      "learning_rate": 1.824964288355752e-06,
      "loss": 0.9162,
      "step": 7513
    },
    {
      "epoch": 0.5421160852783089,
      "grad_norm": 3.4991354555750487,
      "learning_rate": 1.824498726609919e-06,
      "loss": 0.9967,
      "step": 7514
    },
    {
      "epoch": 0.5421882327477364,
      "grad_norm": 0.7457432347936171,
      "learning_rate": 1.8240331744475598e-06,
      "loss": 0.7915,
      "step": 7515
    },
    {
      "epoch": 0.5422603802171638,
      "grad_norm": 6.256822448211032,
      "learning_rate": 1.8235676318940985e-06,
      "loss": 0.9349,
      "step": 7516
    },
    {
      "epoch": 0.5423325276865913,
      "grad_norm": 3.9754543596578564,
      "learning_rate": 1.8231020989749552e-06,
      "loss": 0.865,
      "step": 7517
    },
    {
      "epoch": 0.5424046751560189,
      "grad_norm": 5.677879496984362,
      "learning_rate": 1.8226365757155516e-06,
      "loss": 0.9802,
      "step": 7518
    },
    {
      "epoch": 0.5424768226254464,
      "grad_norm": 11.386933156214695,
      "learning_rate": 1.8221710621413081e-06,
      "loss": 1.0068,
      "step": 7519
    },
    {
      "epoch": 0.5425489700948739,
      "grad_norm": 6.548953652707621,
      "learning_rate": 1.8217055582776444e-06,
      "loss": 0.995,
      "step": 7520
    },
    {
      "epoch": 0.5426211175643014,
      "grad_norm": 19.011557025586026,
      "learning_rate": 1.8212400641499802e-06,
      "loss": 0.966,
      "step": 7521
    },
    {
      "epoch": 0.5426932650337289,
      "grad_norm": 4.725228214303103,
      "learning_rate": 1.8207745797837342e-06,
      "loss": 0.9323,
      "step": 7522
    },
    {
      "epoch": 0.5427654125031565,
      "grad_norm": 2.671600595404097,
      "learning_rate": 1.8203091052043254e-06,
      "loss": 0.9262,
      "step": 7523
    },
    {
      "epoch": 0.542837559972584,
      "grad_norm": 4.5209593630801495,
      "learning_rate": 1.8198436404371707e-06,
      "loss": 0.9906,
      "step": 7524
    },
    {
      "epoch": 0.5429097074420115,
      "grad_norm": 3.6194448327778077,
      "learning_rate": 1.8193781855076877e-06,
      "loss": 0.9136,
      "step": 7525
    },
    {
      "epoch": 0.542981854911439,
      "grad_norm": 3.1918084440954204,
      "learning_rate": 1.8189127404412941e-06,
      "loss": 0.9625,
      "step": 7526
    },
    {
      "epoch": 0.5430540023808665,
      "grad_norm": 0.9095452349226081,
      "learning_rate": 1.8184473052634044e-06,
      "loss": 0.8824,
      "step": 7527
    },
    {
      "epoch": 0.543126149850294,
      "grad_norm": 4.956162541160189,
      "learning_rate": 1.8179818799994362e-06,
      "loss": 0.9263,
      "step": 7528
    },
    {
      "epoch": 0.5431982973197215,
      "grad_norm": 3.576049069223688,
      "learning_rate": 1.8175164646748036e-06,
      "loss": 0.8652,
      "step": 7529
    },
    {
      "epoch": 0.543270444789149,
      "grad_norm": 6.982473776030884,
      "learning_rate": 1.8170510593149207e-06,
      "loss": 1.1257,
      "step": 7530
    },
    {
      "epoch": 0.5433425922585765,
      "grad_norm": 0.9041988904692642,
      "learning_rate": 1.8165856639452029e-06,
      "loss": 0.8835,
      "step": 7531
    },
    {
      "epoch": 0.543414739728004,
      "grad_norm": 0.6411950668615783,
      "learning_rate": 1.8161202785910625e-06,
      "loss": 0.8504,
      "step": 7532
    },
    {
      "epoch": 0.5434868871974315,
      "grad_norm": 0.645110199740588,
      "learning_rate": 1.8156549032779132e-06,
      "loss": 0.8595,
      "step": 7533
    },
    {
      "epoch": 0.5435590346668591,
      "grad_norm": 4.578002928060026,
      "learning_rate": 1.8151895380311667e-06,
      "loss": 0.9561,
      "step": 7534
    },
    {
      "epoch": 0.5436311821362866,
      "grad_norm": 3.633769772649351,
      "learning_rate": 1.8147241828762365e-06,
      "loss": 0.879,
      "step": 7535
    },
    {
      "epoch": 0.5437033296057141,
      "grad_norm": 6.374336357728924,
      "learning_rate": 1.814258837838532e-06,
      "loss": 0.9857,
      "step": 7536
    },
    {
      "epoch": 0.5437754770751416,
      "grad_norm": 2.770094369045128,
      "learning_rate": 1.8137935029434647e-06,
      "loss": 1.0329,
      "step": 7537
    },
    {
      "epoch": 0.5438476245445691,
      "grad_norm": 0.7791182234353268,
      "learning_rate": 1.8133281782164455e-06,
      "loss": 0.8251,
      "step": 7538
    },
    {
      "epoch": 0.5439197720139967,
      "grad_norm": 2.4202752521128645,
      "learning_rate": 1.8128628636828827e-06,
      "loss": 0.9211,
      "step": 7539
    },
    {
      "epoch": 0.5439919194834241,
      "grad_norm": 3.343827433892755,
      "learning_rate": 1.8123975593681867e-06,
      "loss": 0.8605,
      "step": 7540
    },
    {
      "epoch": 0.5440640669528516,
      "grad_norm": 5.31030658993448,
      "learning_rate": 1.8119322652977658e-06,
      "loss": 0.859,
      "step": 7541
    },
    {
      "epoch": 0.5441362144222791,
      "grad_norm": 0.806184357920689,
      "learning_rate": 1.8114669814970268e-06,
      "loss": 0.9636,
      "step": 7542
    },
    {
      "epoch": 0.5442083618917066,
      "grad_norm": 3.3062505249715293,
      "learning_rate": 1.8110017079913789e-06,
      "loss": 0.949,
      "step": 7543
    },
    {
      "epoch": 0.5442805093611341,
      "grad_norm": 5.241988654423108,
      "learning_rate": 1.8105364448062273e-06,
      "loss": 0.956,
      "step": 7544
    },
    {
      "epoch": 0.5443526568305617,
      "grad_norm": 3.7909479543000737,
      "learning_rate": 1.8100711919669794e-06,
      "loss": 1.0032,
      "step": 7545
    },
    {
      "epoch": 0.5444248042999892,
      "grad_norm": 2.554743774914159,
      "learning_rate": 1.809605949499041e-06,
      "loss": 0.9481,
      "step": 7546
    },
    {
      "epoch": 0.5444969517694167,
      "grad_norm": 0.6848711993255958,
      "learning_rate": 1.8091407174278164e-06,
      "loss": 0.7738,
      "step": 7547
    },
    {
      "epoch": 0.5445690992388442,
      "grad_norm": 4.245206935048834,
      "learning_rate": 1.8086754957787107e-06,
      "loss": 0.9978,
      "step": 7548
    },
    {
      "epoch": 0.5446412467082717,
      "grad_norm": 4.14641422566368,
      "learning_rate": 1.808210284577128e-06,
      "loss": 0.9282,
      "step": 7549
    },
    {
      "epoch": 0.5447133941776993,
      "grad_norm": 2.913931099636739,
      "learning_rate": 1.8077450838484719e-06,
      "loss": 0.9957,
      "step": 7550
    },
    {
      "epoch": 0.5447855416471268,
      "grad_norm": 5.925400140183704,
      "learning_rate": 1.8072798936181452e-06,
      "loss": 0.9365,
      "step": 7551
    },
    {
      "epoch": 0.5448576891165542,
      "grad_norm": 4.9558367601409214,
      "learning_rate": 1.806814713911549e-06,
      "loss": 0.9526,
      "step": 7552
    },
    {
      "epoch": 0.5449298365859817,
      "grad_norm": 3.6738899059465058,
      "learning_rate": 1.8063495447540873e-06,
      "loss": 0.9921,
      "step": 7553
    },
    {
      "epoch": 0.5450019840554092,
      "grad_norm": 5.7316189413181124,
      "learning_rate": 1.8058843861711592e-06,
      "loss": 0.8339,
      "step": 7554
    },
    {
      "epoch": 0.5450741315248367,
      "grad_norm": 3.90423104563219,
      "learning_rate": 1.805419238188167e-06,
      "loss": 1.0292,
      "step": 7555
    },
    {
      "epoch": 0.5451462789942643,
      "grad_norm": 0.6952791741774933,
      "learning_rate": 1.8049541008305093e-06,
      "loss": 0.7774,
      "step": 7556
    },
    {
      "epoch": 0.5452184264636918,
      "grad_norm": 5.39766880861589,
      "learning_rate": 1.8044889741235862e-06,
      "loss": 0.9204,
      "step": 7557
    },
    {
      "epoch": 0.5452905739331193,
      "grad_norm": 3.718594331448361,
      "learning_rate": 1.804023858092797e-06,
      "loss": 0.9962,
      "step": 7558
    },
    {
      "epoch": 0.5453627214025468,
      "grad_norm": 3.058921747658933,
      "learning_rate": 1.803558752763539e-06,
      "loss": 0.9872,
      "step": 7559
    },
    {
      "epoch": 0.5454348688719743,
      "grad_norm": 4.326280111160132,
      "learning_rate": 1.80309365816121e-06,
      "loss": 0.8311,
      "step": 7560
    },
    {
      "epoch": 0.5455070163414019,
      "grad_norm": 4.237432410677319,
      "learning_rate": 1.802628574311208e-06,
      "loss": 0.9705,
      "step": 7561
    },
    {
      "epoch": 0.5455791638108294,
      "grad_norm": 3.9379412691416547,
      "learning_rate": 1.8021635012389289e-06,
      "loss": 0.9057,
      "step": 7562
    },
    {
      "epoch": 0.5456513112802568,
      "grad_norm": 4.01924273138474,
      "learning_rate": 1.8016984389697688e-06,
      "loss": 0.9063,
      "step": 7563
    },
    {
      "epoch": 0.5457234587496843,
      "grad_norm": 4.356643750909508,
      "learning_rate": 1.8012333875291219e-06,
      "loss": 0.9475,
      "step": 7564
    },
    {
      "epoch": 0.5457956062191118,
      "grad_norm": 3.6665021830570095,
      "learning_rate": 1.8007683469423849e-06,
      "loss": 0.9417,
      "step": 7565
    },
    {
      "epoch": 0.5458677536885393,
      "grad_norm": 9.20520389826286,
      "learning_rate": 1.8003033172349498e-06,
      "loss": 0.9692,
      "step": 7566
    },
    {
      "epoch": 0.5459399011579669,
      "grad_norm": 3.4065372450917466,
      "learning_rate": 1.7998382984322125e-06,
      "loss": 0.9992,
      "step": 7567
    },
    {
      "epoch": 0.5460120486273944,
      "grad_norm": 0.7419295816501068,
      "learning_rate": 1.7993732905595643e-06,
      "loss": 0.8345,
      "step": 7568
    },
    {
      "epoch": 0.5460841960968219,
      "grad_norm": 3.644172609924685,
      "learning_rate": 1.7989082936423976e-06,
      "loss": 0.8969,
      "step": 7569
    },
    {
      "epoch": 0.5461563435662494,
      "grad_norm": 2.6473743926916242,
      "learning_rate": 1.7984433077061052e-06,
      "loss": 0.9487,
      "step": 7570
    },
    {
      "epoch": 0.5462284910356769,
      "grad_norm": 2.5934793721767386,
      "learning_rate": 1.7979783327760774e-06,
      "loss": 0.967,
      "step": 7571
    },
    {
      "epoch": 0.5463006385051045,
      "grad_norm": 5.709397636748707,
      "learning_rate": 1.7975133688777046e-06,
      "loss": 0.977,
      "step": 7572
    },
    {
      "epoch": 0.546372785974532,
      "grad_norm": 6.217515985938467,
      "learning_rate": 1.797048416036378e-06,
      "loss": 0.9919,
      "step": 7573
    },
    {
      "epoch": 0.5464449334439595,
      "grad_norm": 4.126410965482557,
      "learning_rate": 1.7965834742774855e-06,
      "loss": 1.0257,
      "step": 7574
    },
    {
      "epoch": 0.5465170809133869,
      "grad_norm": 3.5121042211878963,
      "learning_rate": 1.7961185436264167e-06,
      "loss": 0.8851,
      "step": 7575
    },
    {
      "epoch": 0.5465892283828144,
      "grad_norm": 0.7825982095556058,
      "learning_rate": 1.7956536241085588e-06,
      "loss": 0.8703,
      "step": 7576
    },
    {
      "epoch": 0.546661375852242,
      "grad_norm": 4.644212768004613,
      "learning_rate": 1.795188715749301e-06,
      "loss": 0.8719,
      "step": 7577
    },
    {
      "epoch": 0.5467335233216695,
      "grad_norm": 3.103163759848096,
      "learning_rate": 1.7947238185740282e-06,
      "loss": 0.9576,
      "step": 7578
    },
    {
      "epoch": 0.546805670791097,
      "grad_norm": 2.9040282586354293,
      "learning_rate": 1.7942589326081288e-06,
      "loss": 0.9854,
      "step": 7579
    },
    {
      "epoch": 0.5468778182605245,
      "grad_norm": 4.394093511184551,
      "learning_rate": 1.7937940578769872e-06,
      "loss": 0.8944,
      "step": 7580
    },
    {
      "epoch": 0.546949965729952,
      "grad_norm": 2.6985266056159314,
      "learning_rate": 1.793329194405988e-06,
      "loss": 0.9413,
      "step": 7581
    },
    {
      "epoch": 0.5470221131993795,
      "grad_norm": 2.7028482107995924,
      "learning_rate": 1.7928643422205175e-06,
      "loss": 0.9081,
      "step": 7582
    },
    {
      "epoch": 0.5470942606688071,
      "grad_norm": 6.053899268248329,
      "learning_rate": 1.7923995013459579e-06,
      "loss": 0.9419,
      "step": 7583
    },
    {
      "epoch": 0.5471664081382346,
      "grad_norm": 4.958323545472985,
      "learning_rate": 1.791934671807693e-06,
      "loss": 0.9887,
      "step": 7584
    },
    {
      "epoch": 0.5472385556076621,
      "grad_norm": 4.6036310128862,
      "learning_rate": 1.7914698536311058e-06,
      "loss": 0.9829,
      "step": 7585
    },
    {
      "epoch": 0.5473107030770896,
      "grad_norm": 3.8919565151408193,
      "learning_rate": 1.7910050468415776e-06,
      "loss": 0.9364,
      "step": 7586
    },
    {
      "epoch": 0.547382850546517,
      "grad_norm": 4.040035639690658,
      "learning_rate": 1.7905402514644905e-06,
      "loss": 1.0431,
      "step": 7587
    },
    {
      "epoch": 0.5474549980159445,
      "grad_norm": 4.03390841101796,
      "learning_rate": 1.790075467525224e-06,
      "loss": 0.9944,
      "step": 7588
    },
    {
      "epoch": 0.5475271454853721,
      "grad_norm": 4.690674380881808,
      "learning_rate": 1.78961069504916e-06,
      "loss": 1.0578,
      "step": 7589
    },
    {
      "epoch": 0.5475992929547996,
      "grad_norm": 5.3146390311107545,
      "learning_rate": 1.7891459340616773e-06,
      "loss": 0.8824,
      "step": 7590
    },
    {
      "epoch": 0.5476714404242271,
      "grad_norm": 3.7919016475316476,
      "learning_rate": 1.7886811845881533e-06,
      "loss": 0.9606,
      "step": 7591
    },
    {
      "epoch": 0.5477435878936546,
      "grad_norm": 4.630240487100404,
      "learning_rate": 1.7882164466539685e-06,
      "loss": 0.9539,
      "step": 7592
    },
    {
      "epoch": 0.5478157353630821,
      "grad_norm": 3.8104481412999993,
      "learning_rate": 1.7877517202844991e-06,
      "loss": 0.9707,
      "step": 7593
    },
    {
      "epoch": 0.5478878828325097,
      "grad_norm": 4.656507036935967,
      "learning_rate": 1.787287005505123e-06,
      "loss": 0.9582,
      "step": 7594
    },
    {
      "epoch": 0.5479600303019372,
      "grad_norm": 3.583382421349516,
      "learning_rate": 1.786822302341216e-06,
      "loss": 0.8679,
      "step": 7595
    },
    {
      "epoch": 0.5480321777713647,
      "grad_norm": 0.7000138860415127,
      "learning_rate": 1.7863576108181535e-06,
      "loss": 0.86,
      "step": 7596
    },
    {
      "epoch": 0.5481043252407922,
      "grad_norm": 6.403027426412343,
      "learning_rate": 1.7858929309613118e-06,
      "loss": 1.044,
      "step": 7597
    },
    {
      "epoch": 0.5481764727102197,
      "grad_norm": 4.038746332599624,
      "learning_rate": 1.785428262796064e-06,
      "loss": 0.9827,
      "step": 7598
    },
    {
      "epoch": 0.5482486201796472,
      "grad_norm": 3.5721667126392886,
      "learning_rate": 1.7849636063477843e-06,
      "loss": 0.9761,
      "step": 7599
    },
    {
      "epoch": 0.5483207676490747,
      "grad_norm": 3.4156740343879264,
      "learning_rate": 1.784498961641846e-06,
      "loss": 0.9619,
      "step": 7600
    },
    {
      "epoch": 0.5483929151185022,
      "grad_norm": 2.726770431328691,
      "learning_rate": 1.7840343287036222e-06,
      "loss": 0.9182,
      "step": 7601
    },
    {
      "epoch": 0.5484650625879297,
      "grad_norm": 5.164400075246594,
      "learning_rate": 1.7835697075584843e-06,
      "loss": 0.9499,
      "step": 7602
    },
    {
      "epoch": 0.5485372100573572,
      "grad_norm": 5.326447432537649,
      "learning_rate": 1.7831050982318026e-06,
      "loss": 1.0531,
      "step": 7603
    },
    {
      "epoch": 0.5486093575267847,
      "grad_norm": 4.950292505426546,
      "learning_rate": 1.7826405007489493e-06,
      "loss": 0.8527,
      "step": 7604
    },
    {
      "epoch": 0.5486815049962123,
      "grad_norm": 3.726225319889086,
      "learning_rate": 1.7821759151352928e-06,
      "loss": 1.034,
      "step": 7605
    },
    {
      "epoch": 0.5487536524656398,
      "grad_norm": 2.8033973893215416,
      "learning_rate": 1.7817113414162042e-06,
      "loss": 1.1042,
      "step": 7606
    },
    {
      "epoch": 0.5488257999350673,
      "grad_norm": 3.309499281151661,
      "learning_rate": 1.781246779617051e-06,
      "loss": 0.8823,
      "step": 7607
    },
    {
      "epoch": 0.5488979474044948,
      "grad_norm": 6.983673674939779,
      "learning_rate": 1.7807822297632004e-06,
      "loss": 0.9729,
      "step": 7608
    },
    {
      "epoch": 0.5489700948739223,
      "grad_norm": 2.959625517121908,
      "learning_rate": 1.7803176918800217e-06,
      "loss": 0.9795,
      "step": 7609
    },
    {
      "epoch": 0.5490422423433499,
      "grad_norm": 6.547527854251348,
      "learning_rate": 1.7798531659928803e-06,
      "loss": 0.9137,
      "step": 7610
    },
    {
      "epoch": 0.5491143898127773,
      "grad_norm": 7.165673327542956,
      "learning_rate": 1.7793886521271426e-06,
      "loss": 1.0323,
      "step": 7611
    },
    {
      "epoch": 0.5491865372822048,
      "grad_norm": 3.946196745935179,
      "learning_rate": 1.7789241503081743e-06,
      "loss": 1.0376,
      "step": 7612
    },
    {
      "epoch": 0.5492586847516323,
      "grad_norm": 2.8941543065964237,
      "learning_rate": 1.778459660561339e-06,
      "loss": 0.9474,
      "step": 7613
    },
    {
      "epoch": 0.5493308322210598,
      "grad_norm": 4.154762632932324,
      "learning_rate": 1.7779951829120023e-06,
      "loss": 0.9032,
      "step": 7614
    },
    {
      "epoch": 0.5494029796904873,
      "grad_norm": 0.9669403894428239,
      "learning_rate": 1.777530717385526e-06,
      "loss": 0.8476,
      "step": 7615
    },
    {
      "epoch": 0.5494751271599149,
      "grad_norm": 4.34695579733444,
      "learning_rate": 1.7770662640072744e-06,
      "loss": 0.9775,
      "step": 7616
    },
    {
      "epoch": 0.5495472746293424,
      "grad_norm": 4.108036882229374,
      "learning_rate": 1.776601822802608e-06,
      "loss": 0.9095,
      "step": 7617
    },
    {
      "epoch": 0.5496194220987699,
      "grad_norm": 4.220121485692551,
      "learning_rate": 1.77613739379689e-06,
      "loss": 0.8664,
      "step": 7618
    },
    {
      "epoch": 0.5496915695681974,
      "grad_norm": 5.124595253941879,
      "learning_rate": 1.77567297701548e-06,
      "loss": 1.0636,
      "step": 7619
    },
    {
      "epoch": 0.5497637170376249,
      "grad_norm": 3.2571704698410384,
      "learning_rate": 1.7752085724837375e-06,
      "loss": 0.9613,
      "step": 7620
    },
    {
      "epoch": 0.5498358645070525,
      "grad_norm": 5.3042898127346385,
      "learning_rate": 1.7747441802270238e-06,
      "loss": 1.0147,
      "step": 7621
    },
    {
      "epoch": 0.5499080119764799,
      "grad_norm": 3.92866348059673,
      "learning_rate": 1.774279800270696e-06,
      "loss": 0.9435,
      "step": 7622
    },
    {
      "epoch": 0.5499801594459074,
      "grad_norm": 4.062385674848687,
      "learning_rate": 1.7738154326401127e-06,
      "loss": 0.979,
      "step": 7623
    },
    {
      "epoch": 0.5500523069153349,
      "grad_norm": 4.008475384538025,
      "learning_rate": 1.7733510773606318e-06,
      "loss": 1.001,
      "step": 7624
    },
    {
      "epoch": 0.5501244543847624,
      "grad_norm": 3.8165474341368535,
      "learning_rate": 1.772886734457609e-06,
      "loss": 0.9547,
      "step": 7625
    },
    {
      "epoch": 0.55019660185419,
      "grad_norm": 4.373724397255663,
      "learning_rate": 1.7724224039564012e-06,
      "loss": 0.9594,
      "step": 7626
    },
    {
      "epoch": 0.5502687493236175,
      "grad_norm": 2.6733541577099666,
      "learning_rate": 1.771958085882363e-06,
      "loss": 0.9633,
      "step": 7627
    },
    {
      "epoch": 0.550340896793045,
      "grad_norm": 7.734764135570661,
      "learning_rate": 1.7714937802608501e-06,
      "loss": 0.8712,
      "step": 7628
    },
    {
      "epoch": 0.5504130442624725,
      "grad_norm": 4.603361700929546,
      "learning_rate": 1.771029487117216e-06,
      "loss": 0.9161,
      "step": 7629
    },
    {
      "epoch": 0.5504851917319,
      "grad_norm": 6.014953734361859,
      "learning_rate": 1.770565206476813e-06,
      "loss": 0.8826,
      "step": 7630
    },
    {
      "epoch": 0.5505573392013275,
      "grad_norm": 5.215343528544168,
      "learning_rate": 1.7701009383649958e-06,
      "loss": 0.8896,
      "step": 7631
    },
    {
      "epoch": 0.5506294866707551,
      "grad_norm": 3.4913969574567965,
      "learning_rate": 1.7696366828071144e-06,
      "loss": 1.0139,
      "step": 7632
    },
    {
      "epoch": 0.5507016341401826,
      "grad_norm": 4.890530618847685,
      "learning_rate": 1.769172439828522e-06,
      "loss": 1.0267,
      "step": 7633
    },
    {
      "epoch": 0.55077378160961,
      "grad_norm": 3.0692952341796813,
      "learning_rate": 1.7687082094545678e-06,
      "loss": 0.8234,
      "step": 7634
    },
    {
      "epoch": 0.5508459290790375,
      "grad_norm": 4.586056241475412,
      "learning_rate": 1.7682439917106014e-06,
      "loss": 0.9879,
      "step": 7635
    },
    {
      "epoch": 0.550918076548465,
      "grad_norm": 0.7518418266861298,
      "learning_rate": 1.7677797866219737e-06,
      "loss": 0.8433,
      "step": 7636
    },
    {
      "epoch": 0.5509902240178925,
      "grad_norm": 3.9821750925483426,
      "learning_rate": 1.7673155942140315e-06,
      "loss": 0.9916,
      "step": 7637
    },
    {
      "epoch": 0.5510623714873201,
      "grad_norm": 2.7589514436448304,
      "learning_rate": 1.7668514145121239e-06,
      "loss": 0.954,
      "step": 7638
    },
    {
      "epoch": 0.5511345189567476,
      "grad_norm": 4.117563190728063,
      "learning_rate": 1.7663872475415966e-06,
      "loss": 0.9669,
      "step": 7639
    },
    {
      "epoch": 0.5512066664261751,
      "grad_norm": 8.093421016178338,
      "learning_rate": 1.7659230933277979e-06,
      "loss": 0.9089,
      "step": 7640
    },
    {
      "epoch": 0.5512788138956026,
      "grad_norm": 4.031203158793997,
      "learning_rate": 1.7654589518960723e-06,
      "loss": 0.9998,
      "step": 7641
    },
    {
      "epoch": 0.5513509613650301,
      "grad_norm": 3.1557514061243577,
      "learning_rate": 1.7649948232717642e-06,
      "loss": 0.9332,
      "step": 7642
    },
    {
      "epoch": 0.5514231088344577,
      "grad_norm": 3.452172057247697,
      "learning_rate": 1.7645307074802197e-06,
      "loss": 1.0508,
      "step": 7643
    },
    {
      "epoch": 0.5514952563038852,
      "grad_norm": 2.470502110924868,
      "learning_rate": 1.7640666045467811e-06,
      "loss": 0.972,
      "step": 7644
    },
    {
      "epoch": 0.5515674037733127,
      "grad_norm": 3.159418158986322,
      "learning_rate": 1.7636025144967924e-06,
      "loss": 1.0084,
      "step": 7645
    },
    {
      "epoch": 0.5516395512427401,
      "grad_norm": 5.8268953309629685,
      "learning_rate": 1.7631384373555953e-06,
      "loss": 0.9283,
      "step": 7646
    },
    {
      "epoch": 0.5517116987121676,
      "grad_norm": 5.822049735493385,
      "learning_rate": 1.7626743731485306e-06,
      "loss": 0.9154,
      "step": 7647
    },
    {
      "epoch": 0.5517838461815952,
      "grad_norm": 4.172339481399719,
      "learning_rate": 1.7622103219009406e-06,
      "loss": 1.0,
      "step": 7648
    },
    {
      "epoch": 0.5518559936510227,
      "grad_norm": 6.169333560013353,
      "learning_rate": 1.761746283638164e-06,
      "loss": 0.9688,
      "step": 7649
    },
    {
      "epoch": 0.5519281411204502,
      "grad_norm": 3.229481408972412,
      "learning_rate": 1.7612822583855416e-06,
      "loss": 0.9861,
      "step": 7650
    },
    {
      "epoch": 0.5520002885898777,
      "grad_norm": 8.278022569056489,
      "learning_rate": 1.7608182461684107e-06,
      "loss": 0.9598,
      "step": 7651
    },
    {
      "epoch": 0.5520724360593052,
      "grad_norm": 6.578704301058204,
      "learning_rate": 1.7603542470121103e-06,
      "loss": 1.0053,
      "step": 7652
    },
    {
      "epoch": 0.5521445835287327,
      "grad_norm": 4.941307048781911,
      "learning_rate": 1.7598902609419772e-06,
      "loss": 0.979,
      "step": 7653
    },
    {
      "epoch": 0.5522167309981603,
      "grad_norm": 3.7632946860731056,
      "learning_rate": 1.7594262879833477e-06,
      "loss": 0.9874,
      "step": 7654
    },
    {
      "epoch": 0.5522888784675878,
      "grad_norm": 3.978323016516215,
      "learning_rate": 1.7589623281615586e-06,
      "loss": 0.8274,
      "step": 7655
    },
    {
      "epoch": 0.5523610259370153,
      "grad_norm": 2.718205452239532,
      "learning_rate": 1.7584983815019447e-06,
      "loss": 0.9055,
      "step": 7656
    },
    {
      "epoch": 0.5524331734064428,
      "grad_norm": 15.187751673744218,
      "learning_rate": 1.7580344480298392e-06,
      "loss": 0.973,
      "step": 7657
    },
    {
      "epoch": 0.5525053208758702,
      "grad_norm": 5.571012519840161,
      "learning_rate": 1.7575705277705779e-06,
      "loss": 0.8628,
      "step": 7658
    },
    {
      "epoch": 0.5525774683452978,
      "grad_norm": 4.388997601049311,
      "learning_rate": 1.7571066207494916e-06,
      "loss": 0.945,
      "step": 7659
    },
    {
      "epoch": 0.5526496158147253,
      "grad_norm": 4.38506358606219,
      "learning_rate": 1.7566427269919144e-06,
      "loss": 0.8953,
      "step": 7660
    },
    {
      "epoch": 0.5527217632841528,
      "grad_norm": 0.8335898401942707,
      "learning_rate": 1.7561788465231764e-06,
      "loss": 0.8044,
      "step": 7661
    },
    {
      "epoch": 0.5527939107535803,
      "grad_norm": 4.124987053128357,
      "learning_rate": 1.7557149793686097e-06,
      "loss": 0.9813,
      "step": 7662
    },
    {
      "epoch": 0.5528660582230078,
      "grad_norm": 3.1756956644948224,
      "learning_rate": 1.7552511255535432e-06,
      "loss": 0.9498,
      "step": 7663
    },
    {
      "epoch": 0.5529382056924353,
      "grad_norm": 6.214206588567089,
      "learning_rate": 1.7547872851033065e-06,
      "loss": 0.8764,
      "step": 7664
    },
    {
      "epoch": 0.5530103531618629,
      "grad_norm": 3.0011462564900495,
      "learning_rate": 1.7543234580432288e-06,
      "loss": 1.0659,
      "step": 7665
    },
    {
      "epoch": 0.5530825006312904,
      "grad_norm": 3.5651802554578147,
      "learning_rate": 1.7538596443986372e-06,
      "loss": 1.0624,
      "step": 7666
    },
    {
      "epoch": 0.5531546481007179,
      "grad_norm": 6.662941591325415,
      "learning_rate": 1.7533958441948597e-06,
      "loss": 0.9884,
      "step": 7667
    },
    {
      "epoch": 0.5532267955701454,
      "grad_norm": 4.927197875075294,
      "learning_rate": 1.7529320574572224e-06,
      "loss": 0.9566,
      "step": 7668
    },
    {
      "epoch": 0.5532989430395728,
      "grad_norm": 2.9510500445360877,
      "learning_rate": 1.7524682842110498e-06,
      "loss": 0.9209,
      "step": 7669
    },
    {
      "epoch": 0.5533710905090004,
      "grad_norm": 3.766581307065708,
      "learning_rate": 1.7520045244816691e-06,
      "loss": 1.0043,
      "step": 7670
    },
    {
      "epoch": 0.5534432379784279,
      "grad_norm": 3.282470185608573,
      "learning_rate": 1.751540778294402e-06,
      "loss": 0.866,
      "step": 7671
    },
    {
      "epoch": 0.5535153854478554,
      "grad_norm": 6.045337097007501,
      "learning_rate": 1.7510770456745742e-06,
      "loss": 1.0457,
      "step": 7672
    },
    {
      "epoch": 0.5535875329172829,
      "grad_norm": 3.9929322744310802,
      "learning_rate": 1.7506133266475076e-06,
      "loss": 0.8722,
      "step": 7673
    },
    {
      "epoch": 0.5536596803867104,
      "grad_norm": 5.74561773075012,
      "learning_rate": 1.7501496212385235e-06,
      "loss": 0.8586,
      "step": 7674
    },
    {
      "epoch": 0.553731827856138,
      "grad_norm": 5.120562982005548,
      "learning_rate": 1.7496859294729439e-06,
      "loss": 0.9675,
      "step": 7675
    },
    {
      "epoch": 0.5538039753255655,
      "grad_norm": 4.658248984313045,
      "learning_rate": 1.7492222513760888e-06,
      "loss": 0.9847,
      "step": 7676
    },
    {
      "epoch": 0.553876122794993,
      "grad_norm": 5.25354519989677,
      "learning_rate": 1.7487585869732786e-06,
      "loss": 1.0243,
      "step": 7677
    },
    {
      "epoch": 0.5539482702644205,
      "grad_norm": 9.382025769083043,
      "learning_rate": 1.7482949362898317e-06,
      "loss": 0.9151,
      "step": 7678
    },
    {
      "epoch": 0.554020417733848,
      "grad_norm": 11.969530842063174,
      "learning_rate": 1.7478312993510663e-06,
      "loss": 1.0143,
      "step": 7679
    },
    {
      "epoch": 0.5540925652032755,
      "grad_norm": 3.0694317901335135,
      "learning_rate": 1.7473676761823006e-06,
      "loss": 0.9224,
      "step": 7680
    },
    {
      "epoch": 0.554164712672703,
      "grad_norm": 4.161692103892663,
      "learning_rate": 1.74690406680885e-06,
      "loss": 0.9862,
      "step": 7681
    },
    {
      "epoch": 0.5542368601421305,
      "grad_norm": 3.4632132391229145,
      "learning_rate": 1.7464404712560325e-06,
      "loss": 1.0034,
      "step": 7682
    },
    {
      "epoch": 0.554309007611558,
      "grad_norm": 2.8005516564662556,
      "learning_rate": 1.7459768895491611e-06,
      "loss": 0.9646,
      "step": 7683
    },
    {
      "epoch": 0.5543811550809855,
      "grad_norm": 3.5616328204600656,
      "learning_rate": 1.7455133217135524e-06,
      "loss": 1.0405,
      "step": 7684
    },
    {
      "epoch": 0.554453302550413,
      "grad_norm": 0.7754368227785617,
      "learning_rate": 1.7450497677745193e-06,
      "loss": 0.8279,
      "step": 7685
    },
    {
      "epoch": 0.5545254500198405,
      "grad_norm": 4.257749112558555,
      "learning_rate": 1.7445862277573738e-06,
      "loss": 0.9781,
      "step": 7686
    },
    {
      "epoch": 0.5545975974892681,
      "grad_norm": 3.5650071808259827,
      "learning_rate": 1.744122701687429e-06,
      "loss": 0.9637,
      "step": 7687
    },
    {
      "epoch": 0.5546697449586956,
      "grad_norm": 5.319411684648629,
      "learning_rate": 1.7436591895899964e-06,
      "loss": 0.9404,
      "step": 7688
    },
    {
      "epoch": 0.5547418924281231,
      "grad_norm": 3.088417775508861,
      "learning_rate": 1.7431956914903868e-06,
      "loss": 0.9984,
      "step": 7689
    },
    {
      "epoch": 0.5548140398975506,
      "grad_norm": 4.122469501556047,
      "learning_rate": 1.7427322074139095e-06,
      "loss": 0.9741,
      "step": 7690
    },
    {
      "epoch": 0.5548861873669781,
      "grad_norm": 4.803675698532823,
      "learning_rate": 1.7422687373858737e-06,
      "loss": 0.9497,
      "step": 7691
    },
    {
      "epoch": 0.5549583348364057,
      "grad_norm": 6.135384922776628,
      "learning_rate": 1.741805281431589e-06,
      "loss": 0.9159,
      "step": 7692
    },
    {
      "epoch": 0.5550304823058331,
      "grad_norm": 4.672777171723689,
      "learning_rate": 1.7413418395763606e-06,
      "loss": 0.9924,
      "step": 7693
    },
    {
      "epoch": 0.5551026297752606,
      "grad_norm": 3.95668201468195,
      "learning_rate": 1.7408784118454982e-06,
      "loss": 1.0409,
      "step": 7694
    },
    {
      "epoch": 0.5551747772446881,
      "grad_norm": 6.974752945438451,
      "learning_rate": 1.7404149982643061e-06,
      "loss": 0.9455,
      "step": 7695
    },
    {
      "epoch": 0.5552469247141156,
      "grad_norm": 4.001046282305344,
      "learning_rate": 1.7399515988580895e-06,
      "loss": 0.9065,
      "step": 7696
    },
    {
      "epoch": 0.5553190721835431,
      "grad_norm": 3.6009412382755808,
      "learning_rate": 1.7394882136521541e-06,
      "loss": 0.9691,
      "step": 7697
    },
    {
      "epoch": 0.5553912196529707,
      "grad_norm": 3.6148671561877297,
      "learning_rate": 1.7390248426718027e-06,
      "loss": 0.8912,
      "step": 7698
    },
    {
      "epoch": 0.5554633671223982,
      "grad_norm": 6.789630545676056,
      "learning_rate": 1.738561485942338e-06,
      "loss": 0.8937,
      "step": 7699
    },
    {
      "epoch": 0.5555355145918257,
      "grad_norm": 4.005756289431171,
      "learning_rate": 1.7380981434890638e-06,
      "loss": 0.9708,
      "step": 7700
    },
    {
      "epoch": 0.5556076620612532,
      "grad_norm": 3.1401445770995786,
      "learning_rate": 1.7376348153372797e-06,
      "loss": 1.0031,
      "step": 7701
    },
    {
      "epoch": 0.5556798095306807,
      "grad_norm": 3.843217642289895,
      "learning_rate": 1.7371715015122871e-06,
      "loss": 0.9815,
      "step": 7702
    },
    {
      "epoch": 0.5557519570001083,
      "grad_norm": 3.897802878424297,
      "learning_rate": 1.7367082020393861e-06,
      "loss": 0.9212,
      "step": 7703
    },
    {
      "epoch": 0.5558241044695358,
      "grad_norm": 3.941247277130338,
      "learning_rate": 1.7362449169438758e-06,
      "loss": 0.9448,
      "step": 7704
    },
    {
      "epoch": 0.5558962519389632,
      "grad_norm": 7.876752870238382,
      "learning_rate": 1.7357816462510532e-06,
      "loss": 1.0579,
      "step": 7705
    },
    {
      "epoch": 0.5559683994083907,
      "grad_norm": 6.674220510395706,
      "learning_rate": 1.7353183899862179e-06,
      "loss": 0.9469,
      "step": 7706
    },
    {
      "epoch": 0.5560405468778182,
      "grad_norm": 6.646892148360284,
      "learning_rate": 1.7348551481746654e-06,
      "loss": 0.8526,
      "step": 7707
    },
    {
      "epoch": 0.5561126943472458,
      "grad_norm": 7.339872854519199,
      "learning_rate": 1.734391920841691e-06,
      "loss": 1.0858,
      "step": 7708
    },
    {
      "epoch": 0.5561848418166733,
      "grad_norm": 6.774146512596335,
      "learning_rate": 1.7339287080125911e-06,
      "loss": 0.9923,
      "step": 7709
    },
    {
      "epoch": 0.5562569892861008,
      "grad_norm": 11.963455503245081,
      "learning_rate": 1.7334655097126591e-06,
      "loss": 0.9352,
      "step": 7710
    },
    {
      "epoch": 0.5563291367555283,
      "grad_norm": 4.213676057549838,
      "learning_rate": 1.733002325967189e-06,
      "loss": 0.9814,
      "step": 7711
    },
    {
      "epoch": 0.5564012842249558,
      "grad_norm": 0.6485600413157934,
      "learning_rate": 1.732539156801474e-06,
      "loss": 0.8617,
      "step": 7712
    },
    {
      "epoch": 0.5564734316943833,
      "grad_norm": 4.43177275378724,
      "learning_rate": 1.7320760022408052e-06,
      "loss": 1.0004,
      "step": 7713
    },
    {
      "epoch": 0.5565455791638109,
      "grad_norm": 6.778059107599556,
      "learning_rate": 1.7316128623104737e-06,
      "loss": 0.954,
      "step": 7714
    },
    {
      "epoch": 0.5566177266332384,
      "grad_norm": 5.394257048889779,
      "learning_rate": 1.7311497370357708e-06,
      "loss": 0.8739,
      "step": 7715
    },
    {
      "epoch": 0.5566898741026658,
      "grad_norm": 3.863378510056048,
      "learning_rate": 1.7306866264419858e-06,
      "loss": 0.9489,
      "step": 7716
    },
    {
      "epoch": 0.5567620215720933,
      "grad_norm": 3.834181249522121,
      "learning_rate": 1.7302235305544063e-06,
      "loss": 0.979,
      "step": 7717
    },
    {
      "epoch": 0.5568341690415208,
      "grad_norm": 3.1753382825485166,
      "learning_rate": 1.7297604493983216e-06,
      "loss": 0.9444,
      "step": 7718
    },
    {
      "epoch": 0.5569063165109484,
      "grad_norm": 16.10711671853082,
      "learning_rate": 1.7292973829990185e-06,
      "loss": 0.9668,
      "step": 7719
    },
    {
      "epoch": 0.5569784639803759,
      "grad_norm": 4.037486611925727,
      "learning_rate": 1.7288343313817825e-06,
      "loss": 1.0122,
      "step": 7720
    },
    {
      "epoch": 0.5570506114498034,
      "grad_norm": 23.139316725340553,
      "learning_rate": 1.7283712945719007e-06,
      "loss": 0.9455,
      "step": 7721
    },
    {
      "epoch": 0.5571227589192309,
      "grad_norm": 7.039445364951885,
      "learning_rate": 1.727908272594657e-06,
      "loss": 1.0356,
      "step": 7722
    },
    {
      "epoch": 0.5571949063886584,
      "grad_norm": 6.7341903342575495,
      "learning_rate": 1.7274452654753345e-06,
      "loss": 0.9547,
      "step": 7723
    },
    {
      "epoch": 0.557267053858086,
      "grad_norm": 3.687646378020291,
      "learning_rate": 1.7269822732392177e-06,
      "loss": 1.0213,
      "step": 7724
    },
    {
      "epoch": 0.5573392013275135,
      "grad_norm": 6.904826712343347,
      "learning_rate": 1.726519295911588e-06,
      "loss": 0.9778,
      "step": 7725
    },
    {
      "epoch": 0.557411348796941,
      "grad_norm": 3.918182580195886,
      "learning_rate": 1.726056333517727e-06,
      "loss": 0.9011,
      "step": 7726
    },
    {
      "epoch": 0.5574834962663685,
      "grad_norm": 4.461941313832572,
      "learning_rate": 1.7255933860829156e-06,
      "loss": 0.9214,
      "step": 7727
    },
    {
      "epoch": 0.5575556437357959,
      "grad_norm": 7.542543420005998,
      "learning_rate": 1.725130453632434e-06,
      "loss": 0.8823,
      "step": 7728
    },
    {
      "epoch": 0.5576277912052234,
      "grad_norm": 3.6991308377060377,
      "learning_rate": 1.7246675361915605e-06,
      "loss": 0.8727,
      "step": 7729
    },
    {
      "epoch": 0.557699938674651,
      "grad_norm": 4.156049680543238,
      "learning_rate": 1.7242046337855736e-06,
      "loss": 0.9425,
      "step": 7730
    },
    {
      "epoch": 0.5577720861440785,
      "grad_norm": 3.2133040791152023,
      "learning_rate": 1.723741746439751e-06,
      "loss": 0.9557,
      "step": 7731
    },
    {
      "epoch": 0.557844233613506,
      "grad_norm": 3.365695985915022,
      "learning_rate": 1.723278874179368e-06,
      "loss": 0.9403,
      "step": 7732
    },
    {
      "epoch": 0.5579163810829335,
      "grad_norm": 2.249447860788456,
      "learning_rate": 1.7228160170297023e-06,
      "loss": 0.9594,
      "step": 7733
    },
    {
      "epoch": 0.557988528552361,
      "grad_norm": 0.7831633598145251,
      "learning_rate": 1.7223531750160279e-06,
      "loss": 0.8419,
      "step": 7734
    },
    {
      "epoch": 0.5580606760217885,
      "grad_norm": 4.7129088368639005,
      "learning_rate": 1.7218903481636179e-06,
      "loss": 0.8647,
      "step": 7735
    },
    {
      "epoch": 0.5581328234912161,
      "grad_norm": 3.850648349814382,
      "learning_rate": 1.7214275364977471e-06,
      "loss": 0.926,
      "step": 7736
    },
    {
      "epoch": 0.5582049709606436,
      "grad_norm": 3.2999117637166346,
      "learning_rate": 1.7209647400436871e-06,
      "loss": 1.01,
      "step": 7737
    },
    {
      "epoch": 0.5582771184300711,
      "grad_norm": 4.136470682776883,
      "learning_rate": 1.7205019588267097e-06,
      "loss": 0.8375,
      "step": 7738
    },
    {
      "epoch": 0.5583492658994986,
      "grad_norm": 3.4082953235648867,
      "learning_rate": 1.720039192872086e-06,
      "loss": 1.0059,
      "step": 7739
    },
    {
      "epoch": 0.558421413368926,
      "grad_norm": 3.582596296465694,
      "learning_rate": 1.719576442205085e-06,
      "loss": 0.9174,
      "step": 7740
    },
    {
      "epoch": 0.5584935608383536,
      "grad_norm": 5.47696735447433,
      "learning_rate": 1.7191137068509765e-06,
      "loss": 0.9362,
      "step": 7741
    },
    {
      "epoch": 0.5585657083077811,
      "grad_norm": 2.9633936697013405,
      "learning_rate": 1.7186509868350288e-06,
      "loss": 0.9191,
      "step": 7742
    },
    {
      "epoch": 0.5586378557772086,
      "grad_norm": 9.296548789576304,
      "learning_rate": 1.7181882821825095e-06,
      "loss": 0.9667,
      "step": 7743
    },
    {
      "epoch": 0.5587100032466361,
      "grad_norm": 1.0674679121068162,
      "learning_rate": 1.7177255929186845e-06,
      "loss": 0.868,
      "step": 7744
    },
    {
      "epoch": 0.5587821507160636,
      "grad_norm": 4.661700649108062,
      "learning_rate": 1.7172629190688199e-06,
      "loss": 0.9152,
      "step": 7745
    },
    {
      "epoch": 0.5588542981854911,
      "grad_norm": 3.884939919575454,
      "learning_rate": 1.7168002606581811e-06,
      "loss": 0.9516,
      "step": 7746
    },
    {
      "epoch": 0.5589264456549187,
      "grad_norm": 43.340129456302535,
      "learning_rate": 1.7163376177120307e-06,
      "loss": 0.9531,
      "step": 7747
    },
    {
      "epoch": 0.5589985931243462,
      "grad_norm": 3.3809737566854903,
      "learning_rate": 1.715874990255634e-06,
      "loss": 1.0125,
      "step": 7748
    },
    {
      "epoch": 0.5590707405937737,
      "grad_norm": 3.112680183312122,
      "learning_rate": 1.7154123783142517e-06,
      "loss": 0.9786,
      "step": 7749
    },
    {
      "epoch": 0.5591428880632012,
      "grad_norm": 7.478563318028257,
      "learning_rate": 1.714949781913146e-06,
      "loss": 0.9919,
      "step": 7750
    },
    {
      "epoch": 0.5592150355326287,
      "grad_norm": 3.49895666102901,
      "learning_rate": 1.714487201077578e-06,
      "loss": 0.8946,
      "step": 7751
    },
    {
      "epoch": 0.5592871830020562,
      "grad_norm": 7.1015625,
      "learning_rate": 1.7140246358328065e-06,
      "loss": 1.0653,
      "step": 7752
    },
    {
      "epoch": 0.5593593304714837,
      "grad_norm": 6.300857534023643,
      "learning_rate": 1.713562086204091e-06,
      "loss": 0.9522,
      "step": 7753
    },
    {
      "epoch": 0.5594314779409112,
      "grad_norm": 6.634094617599577,
      "learning_rate": 1.7130995522166893e-06,
      "loss": 0.9536,
      "step": 7754
    },
    {
      "epoch": 0.5595036254103387,
      "grad_norm": 0.7639125044861783,
      "learning_rate": 1.7126370338958598e-06,
      "loss": 0.8629,
      "step": 7755
    },
    {
      "epoch": 0.5595757728797662,
      "grad_norm": 5.151283230210379,
      "learning_rate": 1.7121745312668577e-06,
      "loss": 1.0446,
      "step": 7756
    },
    {
      "epoch": 0.5596479203491938,
      "grad_norm": 2.9413010374505557,
      "learning_rate": 1.7117120443549387e-06,
      "loss": 0.9892,
      "step": 7757
    },
    {
      "epoch": 0.5597200678186213,
      "grad_norm": 0.6639443685138094,
      "learning_rate": 1.7112495731853583e-06,
      "loss": 0.8331,
      "step": 7758
    },
    {
      "epoch": 0.5597922152880488,
      "grad_norm": 5.11609322299086,
      "learning_rate": 1.710787117783369e-06,
      "loss": 0.9232,
      "step": 7759
    },
    {
      "epoch": 0.5598643627574763,
      "grad_norm": 5.969855880147889,
      "learning_rate": 1.7103246781742253e-06,
      "loss": 0.9859,
      "step": 7760
    },
    {
      "epoch": 0.5599365102269038,
      "grad_norm": 4.480075383233314,
      "learning_rate": 1.7098622543831787e-06,
      "loss": 0.9235,
      "step": 7761
    },
    {
      "epoch": 0.5600086576963313,
      "grad_norm": 3.524314618947235,
      "learning_rate": 1.7093998464354793e-06,
      "loss": 0.7938,
      "step": 7762
    },
    {
      "epoch": 0.5600808051657589,
      "grad_norm": 3.133737023368173,
      "learning_rate": 1.7089374543563798e-06,
      "loss": 0.8397,
      "step": 7763
    },
    {
      "epoch": 0.5601529526351863,
      "grad_norm": 3.444093934208085,
      "learning_rate": 1.7084750781711274e-06,
      "loss": 0.9885,
      "step": 7764
    },
    {
      "epoch": 0.5602251001046138,
      "grad_norm": 7.526366110455515,
      "learning_rate": 1.7080127179049722e-06,
      "loss": 0.9566,
      "step": 7765
    },
    {
      "epoch": 0.5602972475740413,
      "grad_norm": 4.979837391048445,
      "learning_rate": 1.7075503735831621e-06,
      "loss": 0.9992,
      "step": 7766
    },
    {
      "epoch": 0.5603693950434688,
      "grad_norm": 4.15767536245384,
      "learning_rate": 1.7070880452309426e-06,
      "loss": 0.9859,
      "step": 7767
    },
    {
      "epoch": 0.5604415425128964,
      "grad_norm": 4.559977631095766,
      "learning_rate": 1.706625732873561e-06,
      "loss": 0.991,
      "step": 7768
    },
    {
      "epoch": 0.5605136899823239,
      "grad_norm": 0.6984327363058717,
      "learning_rate": 1.7061634365362618e-06,
      "loss": 0.8764,
      "step": 7769
    },
    {
      "epoch": 0.5605858374517514,
      "grad_norm": 3.190458233965112,
      "learning_rate": 1.7057011562442902e-06,
      "loss": 0.907,
      "step": 7770
    },
    {
      "epoch": 0.5606579849211789,
      "grad_norm": 2.647284062470378,
      "learning_rate": 1.7052388920228884e-06,
      "loss": 0.9442,
      "step": 7771
    },
    {
      "epoch": 0.5607301323906064,
      "grad_norm": 6.031013128343202,
      "learning_rate": 1.7047766438973002e-06,
      "loss": 0.9243,
      "step": 7772
    },
    {
      "epoch": 0.560802279860034,
      "grad_norm": 0.7566825071292841,
      "learning_rate": 1.7043144118927664e-06,
      "loss": 0.807,
      "step": 7773
    },
    {
      "epoch": 0.5608744273294615,
      "grad_norm": 6.6192060362104375,
      "learning_rate": 1.703852196034527e-06,
      "loss": 0.9111,
      "step": 7774
    },
    {
      "epoch": 0.5609465747988889,
      "grad_norm": 3.5104041094607257,
      "learning_rate": 1.7033899963478241e-06,
      "loss": 1.0852,
      "step": 7775
    },
    {
      "epoch": 0.5610187222683164,
      "grad_norm": 6.039756350118672,
      "learning_rate": 1.702927812857895e-06,
      "loss": 0.9483,
      "step": 7776
    },
    {
      "epoch": 0.5610908697377439,
      "grad_norm": 3.9280900685070486,
      "learning_rate": 1.7024656455899784e-06,
      "loss": 1.1098,
      "step": 7777
    },
    {
      "epoch": 0.5611630172071714,
      "grad_norm": 3.0480457110844568,
      "learning_rate": 1.7020034945693115e-06,
      "loss": 0.9552,
      "step": 7778
    },
    {
      "epoch": 0.561235164676599,
      "grad_norm": 15.306128597128534,
      "learning_rate": 1.7015413598211306e-06,
      "loss": 1.0483,
      "step": 7779
    },
    {
      "epoch": 0.5613073121460265,
      "grad_norm": 2.695059324896109,
      "learning_rate": 1.701079241370671e-06,
      "loss": 0.8806,
      "step": 7780
    },
    {
      "epoch": 0.561379459615454,
      "grad_norm": 2.8947187149356384,
      "learning_rate": 1.7006171392431674e-06,
      "loss": 0.99,
      "step": 7781
    },
    {
      "epoch": 0.5614516070848815,
      "grad_norm": 3.4716693861925227,
      "learning_rate": 1.7001550534638542e-06,
      "loss": 0.9771,
      "step": 7782
    },
    {
      "epoch": 0.561523754554309,
      "grad_norm": 5.644466877606552,
      "learning_rate": 1.6996929840579629e-06,
      "loss": 1.0784,
      "step": 7783
    },
    {
      "epoch": 0.5615959020237365,
      "grad_norm": 4.674559979461908,
      "learning_rate": 1.6992309310507263e-06,
      "loss": 1.0878,
      "step": 7784
    },
    {
      "epoch": 0.5616680494931641,
      "grad_norm": 4.497852872838132,
      "learning_rate": 1.6987688944673752e-06,
      "loss": 0.9285,
      "step": 7785
    },
    {
      "epoch": 0.5617401969625916,
      "grad_norm": 6.3767084562557805,
      "learning_rate": 1.698306874333139e-06,
      "loss": 0.9763,
      "step": 7786
    },
    {
      "epoch": 0.561812344432019,
      "grad_norm": 3.659064098412864,
      "learning_rate": 1.6978448706732485e-06,
      "loss": 0.9715,
      "step": 7787
    },
    {
      "epoch": 0.5618844919014465,
      "grad_norm": 5.166922511667574,
      "learning_rate": 1.6973828835129309e-06,
      "loss": 0.9863,
      "step": 7788
    },
    {
      "epoch": 0.561956639370874,
      "grad_norm": 4.551069180134823,
      "learning_rate": 1.696920912877413e-06,
      "loss": 1.0118,
      "step": 7789
    },
    {
      "epoch": 0.5620287868403016,
      "grad_norm": 3.3743171001002277,
      "learning_rate": 1.6964589587919228e-06,
      "loss": 0.9977,
      "step": 7790
    },
    {
      "epoch": 0.5621009343097291,
      "grad_norm": 4.064811635738267,
      "learning_rate": 1.695997021281685e-06,
      "loss": 1.0055,
      "step": 7791
    },
    {
      "epoch": 0.5621730817791566,
      "grad_norm": 3.3677277098473,
      "learning_rate": 1.6955351003719245e-06,
      "loss": 0.9448,
      "step": 7792
    },
    {
      "epoch": 0.5622452292485841,
      "grad_norm": 4.8386264406243615,
      "learning_rate": 1.695073196087865e-06,
      "loss": 0.9227,
      "step": 7793
    },
    {
      "epoch": 0.5623173767180116,
      "grad_norm": 3.635932222697744,
      "learning_rate": 1.6946113084547298e-06,
      "loss": 1.0102,
      "step": 7794
    },
    {
      "epoch": 0.5623895241874391,
      "grad_norm": 2.9853761601456386,
      "learning_rate": 1.6941494374977402e-06,
      "loss": 0.9304,
      "step": 7795
    },
    {
      "epoch": 0.5624616716568667,
      "grad_norm": 5.013405757509431,
      "learning_rate": 1.6936875832421173e-06,
      "loss": 0.8811,
      "step": 7796
    },
    {
      "epoch": 0.5625338191262942,
      "grad_norm": 2.9875399598836925,
      "learning_rate": 1.6932257457130826e-06,
      "loss": 1.001,
      "step": 7797
    },
    {
      "epoch": 0.5626059665957217,
      "grad_norm": 6.698303338058271,
      "learning_rate": 1.692763924935853e-06,
      "loss": 0.8405,
      "step": 7798
    },
    {
      "epoch": 0.5626781140651491,
      "grad_norm": 7.896891607728257,
      "learning_rate": 1.692302120935649e-06,
      "loss": 0.8515,
      "step": 7799
    },
    {
      "epoch": 0.5627502615345766,
      "grad_norm": 3.6154599111679686,
      "learning_rate": 1.6918403337376873e-06,
      "loss": 0.9727,
      "step": 7800
    },
    {
      "epoch": 0.5628224090040042,
      "grad_norm": 9.891064674765332,
      "learning_rate": 1.6913785633671831e-06,
      "loss": 0.8915,
      "step": 7801
    },
    {
      "epoch": 0.5628945564734317,
      "grad_norm": 3.76824998970667,
      "learning_rate": 1.690916809849354e-06,
      "loss": 0.9946,
      "step": 7802
    },
    {
      "epoch": 0.5629667039428592,
      "grad_norm": 4.854163610542001,
      "learning_rate": 1.6904550732094134e-06,
      "loss": 0.9208,
      "step": 7803
    },
    {
      "epoch": 0.5630388514122867,
      "grad_norm": 6.015227386413078,
      "learning_rate": 1.6899933534725752e-06,
      "loss": 1.0388,
      "step": 7804
    },
    {
      "epoch": 0.5631109988817142,
      "grad_norm": 6.113248801830818,
      "learning_rate": 1.689531650664053e-06,
      "loss": 0.894,
      "step": 7805
    },
    {
      "epoch": 0.5631831463511418,
      "grad_norm": 5.229221277850986,
      "learning_rate": 1.6890699648090574e-06,
      "loss": 0.9766,
      "step": 7806
    },
    {
      "epoch": 0.5632552938205693,
      "grad_norm": 3.7989492519651202,
      "learning_rate": 1.6886082959328e-06,
      "loss": 0.9611,
      "step": 7807
    },
    {
      "epoch": 0.5633274412899968,
      "grad_norm": 3.2609498356876916,
      "learning_rate": 1.688146644060491e-06,
      "loss": 0.8586,
      "step": 7808
    },
    {
      "epoch": 0.5633995887594243,
      "grad_norm": 3.1034046919089837,
      "learning_rate": 1.6876850092173399e-06,
      "loss": 0.9325,
      "step": 7809
    },
    {
      "epoch": 0.5634717362288518,
      "grad_norm": 9.662871525618328,
      "learning_rate": 1.687223391428554e-06,
      "loss": 0.8933,
      "step": 7810
    },
    {
      "epoch": 0.5635438836982792,
      "grad_norm": 3.587183916475786,
      "learning_rate": 1.68676179071934e-06,
      "loss": 0.9301,
      "step": 7811
    },
    {
      "epoch": 0.5636160311677068,
      "grad_norm": 5.041520150738054,
      "learning_rate": 1.686300207114906e-06,
      "loss": 1.0044,
      "step": 7812
    },
    {
      "epoch": 0.5636881786371343,
      "grad_norm": 4.909696109791616,
      "learning_rate": 1.6858386406404554e-06,
      "loss": 0.963,
      "step": 7813
    },
    {
      "epoch": 0.5637603261065618,
      "grad_norm": 3.794416722844443,
      "learning_rate": 1.6853770913211947e-06,
      "loss": 0.9881,
      "step": 7814
    },
    {
      "epoch": 0.5638324735759893,
      "grad_norm": 3.197823356401976,
      "learning_rate": 1.684915559182326e-06,
      "loss": 0.9489,
      "step": 7815
    },
    {
      "epoch": 0.5639046210454168,
      "grad_norm": 3.8774571165643983,
      "learning_rate": 1.6844540442490517e-06,
      "loss": 0.97,
      "step": 7816
    },
    {
      "epoch": 0.5639767685148444,
      "grad_norm": 4.83679014217017,
      "learning_rate": 1.6839925465465746e-06,
      "loss": 0.8515,
      "step": 7817
    },
    {
      "epoch": 0.5640489159842719,
      "grad_norm": 4.665532632821766,
      "learning_rate": 1.6835310661000946e-06,
      "loss": 0.9099,
      "step": 7818
    },
    {
      "epoch": 0.5641210634536994,
      "grad_norm": 3.1355422154296075,
      "learning_rate": 1.6830696029348113e-06,
      "loss": 0.9406,
      "step": 7819
    },
    {
      "epoch": 0.5641932109231269,
      "grad_norm": 2.6083299748538233,
      "learning_rate": 1.6826081570759238e-06,
      "loss": 0.8349,
      "step": 7820
    },
    {
      "epoch": 0.5642653583925544,
      "grad_norm": 3.012422274144791,
      "learning_rate": 1.6821467285486303e-06,
      "loss": 0.9727,
      "step": 7821
    },
    {
      "epoch": 0.5643375058619818,
      "grad_norm": 4.21110576795934,
      "learning_rate": 1.6816853173781273e-06,
      "loss": 0.9708,
      "step": 7822
    },
    {
      "epoch": 0.5644096533314094,
      "grad_norm": 3.2231297000705017,
      "learning_rate": 1.68122392358961e-06,
      "loss": 0.8869,
      "step": 7823
    },
    {
      "epoch": 0.5644818008008369,
      "grad_norm": 3.0018978473684204,
      "learning_rate": 1.6807625472082749e-06,
      "loss": 1.0606,
      "step": 7824
    },
    {
      "epoch": 0.5645539482702644,
      "grad_norm": 0.6953133358039547,
      "learning_rate": 1.6803011882593146e-06,
      "loss": 0.7732,
      "step": 7825
    },
    {
      "epoch": 0.5646260957396919,
      "grad_norm": 4.097883627285958,
      "learning_rate": 1.6798398467679236e-06,
      "loss": 0.9414,
      "step": 7826
    },
    {
      "epoch": 0.5646982432091194,
      "grad_norm": 4.960025252923976,
      "learning_rate": 1.6793785227592934e-06,
      "loss": 0.9864,
      "step": 7827
    },
    {
      "epoch": 0.564770390678547,
      "grad_norm": 5.704635461900597,
      "learning_rate": 1.6789172162586145e-06,
      "loss": 0.9195,
      "step": 7828
    },
    {
      "epoch": 0.5648425381479745,
      "grad_norm": 8.806283909317477,
      "learning_rate": 1.6784559272910786e-06,
      "loss": 0.9388,
      "step": 7829
    },
    {
      "epoch": 0.564914685617402,
      "grad_norm": 8.567271441574224,
      "learning_rate": 1.6779946558818736e-06,
      "loss": 1.0512,
      "step": 7830
    },
    {
      "epoch": 0.5649868330868295,
      "grad_norm": 2.817919024606978,
      "learning_rate": 1.6775334020561886e-06,
      "loss": 1.0586,
      "step": 7831
    },
    {
      "epoch": 0.565058980556257,
      "grad_norm": 20.239942525548337,
      "learning_rate": 1.6770721658392112e-06,
      "loss": 1.017,
      "step": 7832
    },
    {
      "epoch": 0.5651311280256845,
      "grad_norm": 9.13288308957429,
      "learning_rate": 1.6766109472561266e-06,
      "loss": 0.8397,
      "step": 7833
    },
    {
      "epoch": 0.565203275495112,
      "grad_norm": 3.1959542746871574,
      "learning_rate": 1.676149746332122e-06,
      "loss": 0.9394,
      "step": 7834
    },
    {
      "epoch": 0.5652754229645395,
      "grad_norm": 4.661963931604894,
      "learning_rate": 1.6756885630923798e-06,
      "loss": 0.9494,
      "step": 7835
    },
    {
      "epoch": 0.565347570433967,
      "grad_norm": 5.788532351210094,
      "learning_rate": 1.6752273975620854e-06,
      "loss": 0.9572,
      "step": 7836
    },
    {
      "epoch": 0.5654197179033945,
      "grad_norm": 2.9088864417179554,
      "learning_rate": 1.6747662497664199e-06,
      "loss": 0.933,
      "step": 7837
    },
    {
      "epoch": 0.565491865372822,
      "grad_norm": 5.027526425330598,
      "learning_rate": 1.6743051197305662e-06,
      "loss": 0.9614,
      "step": 7838
    },
    {
      "epoch": 0.5655640128422496,
      "grad_norm": 8.644347232423454,
      "learning_rate": 1.6738440074797042e-06,
      "loss": 0.9727,
      "step": 7839
    },
    {
      "epoch": 0.5656361603116771,
      "grad_norm": 12.384454987676966,
      "learning_rate": 1.6733829130390127e-06,
      "loss": 0.9156,
      "step": 7840
    },
    {
      "epoch": 0.5657083077811046,
      "grad_norm": 7.4412680946000656,
      "learning_rate": 1.672921836433672e-06,
      "loss": 0.9314,
      "step": 7841
    },
    {
      "epoch": 0.5657804552505321,
      "grad_norm": 2.9064939663120746,
      "learning_rate": 1.672460777688859e-06,
      "loss": 1.039,
      "step": 7842
    },
    {
      "epoch": 0.5658526027199596,
      "grad_norm": 10.353282949599738,
      "learning_rate": 1.67199973682975e-06,
      "loss": 0.9925,
      "step": 7843
    },
    {
      "epoch": 0.5659247501893871,
      "grad_norm": 5.4176144088684195,
      "learning_rate": 1.6715387138815217e-06,
      "loss": 1.0893,
      "step": 7844
    },
    {
      "epoch": 0.5659968976588147,
      "grad_norm": 5.575553921264744,
      "learning_rate": 1.6710777088693482e-06,
      "loss": 1.0257,
      "step": 7845
    },
    {
      "epoch": 0.5660690451282421,
      "grad_norm": 0.7440247933655841,
      "learning_rate": 1.6706167218184036e-06,
      "loss": 0.907,
      "step": 7846
    },
    {
      "epoch": 0.5661411925976696,
      "grad_norm": 0.7673773660755357,
      "learning_rate": 1.6701557527538594e-06,
      "loss": 0.7932,
      "step": 7847
    },
    {
      "epoch": 0.5662133400670971,
      "grad_norm": 4.6374430938272795,
      "learning_rate": 1.6696948017008897e-06,
      "loss": 0.9193,
      "step": 7848
    },
    {
      "epoch": 0.5662854875365246,
      "grad_norm": 5.427401581534768,
      "learning_rate": 1.6692338686846638e-06,
      "loss": 0.9722,
      "step": 7849
    },
    {
      "epoch": 0.5663576350059522,
      "grad_norm": 3.4210346182839295,
      "learning_rate": 1.6687729537303512e-06,
      "loss": 0.8887,
      "step": 7850
    },
    {
      "epoch": 0.5664297824753797,
      "grad_norm": 11.031790830846408,
      "learning_rate": 1.6683120568631224e-06,
      "loss": 0.9217,
      "step": 7851
    },
    {
      "epoch": 0.5665019299448072,
      "grad_norm": 3.9467618489601355,
      "learning_rate": 1.6678511781081433e-06,
      "loss": 1.0898,
      "step": 7852
    },
    {
      "epoch": 0.5665740774142347,
      "grad_norm": 4.040637065400345,
      "learning_rate": 1.6673903174905825e-06,
      "loss": 0.9269,
      "step": 7853
    },
    {
      "epoch": 0.5666462248836622,
      "grad_norm": 5.791087755136794,
      "learning_rate": 1.6669294750356056e-06,
      "loss": 0.9514,
      "step": 7854
    },
    {
      "epoch": 0.5667183723530897,
      "grad_norm": 3.8842714205348456,
      "learning_rate": 1.6664686507683758e-06,
      "loss": 0.9629,
      "step": 7855
    },
    {
      "epoch": 0.5667905198225173,
      "grad_norm": 5.214809917668811,
      "learning_rate": 1.6660078447140595e-06,
      "loss": 0.9133,
      "step": 7856
    },
    {
      "epoch": 0.5668626672919448,
      "grad_norm": 3.1645677704686874,
      "learning_rate": 1.6655470568978177e-06,
      "loss": 1.0186,
      "step": 7857
    },
    {
      "epoch": 0.5669348147613722,
      "grad_norm": 7.761054000456255,
      "learning_rate": 1.6650862873448136e-06,
      "loss": 1.0673,
      "step": 7858
    },
    {
      "epoch": 0.5670069622307997,
      "grad_norm": 4.476693089895505,
      "learning_rate": 1.6646255360802066e-06,
      "loss": 1.0526,
      "step": 7859
    },
    {
      "epoch": 0.5670791097002272,
      "grad_norm": 4.502887647039978,
      "learning_rate": 1.6641648031291585e-06,
      "loss": 0.942,
      "step": 7860
    },
    {
      "epoch": 0.5671512571696548,
      "grad_norm": 6.05344997433669,
      "learning_rate": 1.6637040885168272e-06,
      "loss": 1.0398,
      "step": 7861
    },
    {
      "epoch": 0.5672234046390823,
      "grad_norm": 4.1994351779255545,
      "learning_rate": 1.66324339226837e-06,
      "loss": 0.9716,
      "step": 7862
    },
    {
      "epoch": 0.5672955521085098,
      "grad_norm": 3.717897461690482,
      "learning_rate": 1.6627827144089451e-06,
      "loss": 0.9641,
      "step": 7863
    },
    {
      "epoch": 0.5673676995779373,
      "grad_norm": 3.5081005360497746,
      "learning_rate": 1.6623220549637074e-06,
      "loss": 0.9561,
      "step": 7864
    },
    {
      "epoch": 0.5674398470473648,
      "grad_norm": 3.0475219039818278,
      "learning_rate": 1.6618614139578126e-06,
      "loss": 1.005,
      "step": 7865
    },
    {
      "epoch": 0.5675119945167924,
      "grad_norm": 2.8638181092027573,
      "learning_rate": 1.6614007914164146e-06,
      "loss": 0.8364,
      "step": 7866
    },
    {
      "epoch": 0.5675841419862199,
      "grad_norm": 3.3647865887514765,
      "learning_rate": 1.6609401873646649e-06,
      "loss": 1.028,
      "step": 7867
    },
    {
      "epoch": 0.5676562894556474,
      "grad_norm": 3.6293581054042026,
      "learning_rate": 1.6604796018277175e-06,
      "loss": 0.9637,
      "step": 7868
    },
    {
      "epoch": 0.5677284369250748,
      "grad_norm": 3.09609634356952,
      "learning_rate": 1.6600190348307214e-06,
      "loss": 0.9871,
      "step": 7869
    },
    {
      "epoch": 0.5678005843945023,
      "grad_norm": 4.057118299091396,
      "learning_rate": 1.659558486398828e-06,
      "loss": 0.8833,
      "step": 7870
    },
    {
      "epoch": 0.5678727318639298,
      "grad_norm": 51.54240849093953,
      "learning_rate": 1.6590979565571845e-06,
      "loss": 0.9137,
      "step": 7871
    },
    {
      "epoch": 0.5679448793333574,
      "grad_norm": 6.5583767971487354,
      "learning_rate": 1.6586374453309399e-06,
      "loss": 1.0544,
      "step": 7872
    },
    {
      "epoch": 0.5680170268027849,
      "grad_norm": 0.7772566013270573,
      "learning_rate": 1.6581769527452413e-06,
      "loss": 0.804,
      "step": 7873
    },
    {
      "epoch": 0.5680891742722124,
      "grad_norm": 5.95997653879758,
      "learning_rate": 1.657716478825233e-06,
      "loss": 0.9395,
      "step": 7874
    },
    {
      "epoch": 0.5681613217416399,
      "grad_norm": 4.863749114668586,
      "learning_rate": 1.6572560235960614e-06,
      "loss": 0.9221,
      "step": 7875
    },
    {
      "epoch": 0.5682334692110674,
      "grad_norm": 5.979883804468228,
      "learning_rate": 1.6567955870828696e-06,
      "loss": 0.8881,
      "step": 7876
    },
    {
      "epoch": 0.568305616680495,
      "grad_norm": 3.4070503624456756,
      "learning_rate": 1.6563351693107994e-06,
      "loss": 0.9198,
      "step": 7877
    },
    {
      "epoch": 0.5683777641499225,
      "grad_norm": 2.82361415865583,
      "learning_rate": 1.6558747703049944e-06,
      "loss": 1.0387,
      "step": 7878
    },
    {
      "epoch": 0.56844991161935,
      "grad_norm": 6.422851488259274,
      "learning_rate": 1.6554143900905935e-06,
      "loss": 1.0801,
      "step": 7879
    },
    {
      "epoch": 0.5685220590887775,
      "grad_norm": 3.6850007202893074,
      "learning_rate": 1.654954028692738e-06,
      "loss": 0.9482,
      "step": 7880
    },
    {
      "epoch": 0.5685942065582049,
      "grad_norm": 3.17753509478093,
      "learning_rate": 1.6544936861365648e-06,
      "loss": 0.8834,
      "step": 7881
    },
    {
      "epoch": 0.5686663540276324,
      "grad_norm": 6.21363075450419,
      "learning_rate": 1.6540333624472131e-06,
      "loss": 0.9791,
      "step": 7882
    },
    {
      "epoch": 0.56873850149706,
      "grad_norm": 5.754418044108995,
      "learning_rate": 1.6535730576498184e-06,
      "loss": 0.873,
      "step": 7883
    },
    {
      "epoch": 0.5688106489664875,
      "grad_norm": 4.060018162804177,
      "learning_rate": 1.6531127717695167e-06,
      "loss": 0.9263,
      "step": 7884
    },
    {
      "epoch": 0.568882796435915,
      "grad_norm": 3.3317133940841646,
      "learning_rate": 1.6526525048314426e-06,
      "loss": 1.0334,
      "step": 7885
    },
    {
      "epoch": 0.5689549439053425,
      "grad_norm": 5.475908226732144,
      "learning_rate": 1.6521922568607285e-06,
      "loss": 0.9659,
      "step": 7886
    },
    {
      "epoch": 0.56902709137477,
      "grad_norm": 3.560759537498526,
      "learning_rate": 1.6517320278825086e-06,
      "loss": 0.9011,
      "step": 7887
    },
    {
      "epoch": 0.5690992388441976,
      "grad_norm": 10.482462087390262,
      "learning_rate": 1.6512718179219135e-06,
      "loss": 1.0197,
      "step": 7888
    },
    {
      "epoch": 0.5691713863136251,
      "grad_norm": 7.581912008818927,
      "learning_rate": 1.6508116270040726e-06,
      "loss": 0.8806,
      "step": 7889
    },
    {
      "epoch": 0.5692435337830526,
      "grad_norm": 3.489100514882611,
      "learning_rate": 1.650351455154117e-06,
      "loss": 0.9759,
      "step": 7890
    },
    {
      "epoch": 0.5693156812524801,
      "grad_norm": 4.889677010695174,
      "learning_rate": 1.6498913023971732e-06,
      "loss": 1.0054,
      "step": 7891
    },
    {
      "epoch": 0.5693878287219076,
      "grad_norm": 0.7689222840573571,
      "learning_rate": 1.6494311687583702e-06,
      "loss": 0.8944,
      "step": 7892
    },
    {
      "epoch": 0.569459976191335,
      "grad_norm": 0.7729105840342532,
      "learning_rate": 1.6489710542628333e-06,
      "loss": 0.8242,
      "step": 7893
    },
    {
      "epoch": 0.5695321236607626,
      "grad_norm": 7.2099772730454585,
      "learning_rate": 1.6485109589356871e-06,
      "loss": 0.9097,
      "step": 7894
    },
    {
      "epoch": 0.5696042711301901,
      "grad_norm": 4.0140658071585245,
      "learning_rate": 1.6480508828020562e-06,
      "loss": 1.0204,
      "step": 7895
    },
    {
      "epoch": 0.5696764185996176,
      "grad_norm": 3.7466647256957035,
      "learning_rate": 1.6475908258870641e-06,
      "loss": 1.0149,
      "step": 7896
    },
    {
      "epoch": 0.5697485660690451,
      "grad_norm": 4.3697266541632205,
      "learning_rate": 1.6471307882158324e-06,
      "loss": 1.1066,
      "step": 7897
    },
    {
      "epoch": 0.5698207135384726,
      "grad_norm": 4.626895413052912,
      "learning_rate": 1.646670769813482e-06,
      "loss": 0.9626,
      "step": 7898
    },
    {
      "epoch": 0.5698928610079002,
      "grad_norm": 6.208496228156949,
      "learning_rate": 1.646210770705133e-06,
      "loss": 1.0293,
      "step": 7899
    },
    {
      "epoch": 0.5699650084773277,
      "grad_norm": 5.410395097108115,
      "learning_rate": 1.6457507909159041e-06,
      "loss": 0.8476,
      "step": 7900
    },
    {
      "epoch": 0.5700371559467552,
      "grad_norm": 5.326547696850305,
      "learning_rate": 1.6452908304709126e-06,
      "loss": 0.9472,
      "step": 7901
    },
    {
      "epoch": 0.5701093034161827,
      "grad_norm": 5.364676564218189,
      "learning_rate": 1.6448308893952764e-06,
      "loss": 0.9844,
      "step": 7902
    },
    {
      "epoch": 0.5701814508856102,
      "grad_norm": 5.564172728983078,
      "learning_rate": 1.6443709677141097e-06,
      "loss": 0.9822,
      "step": 7903
    },
    {
      "epoch": 0.5702535983550377,
      "grad_norm": 4.881956565604334,
      "learning_rate": 1.643911065452529e-06,
      "loss": 0.9851,
      "step": 7904
    },
    {
      "epoch": 0.5703257458244652,
      "grad_norm": 3.1354658729181497,
      "learning_rate": 1.6434511826356467e-06,
      "loss": 0.9601,
      "step": 7905
    },
    {
      "epoch": 0.5703978932938927,
      "grad_norm": 3.082057551476019,
      "learning_rate": 1.642991319288575e-06,
      "loss": 0.9231,
      "step": 7906
    },
    {
      "epoch": 0.5704700407633202,
      "grad_norm": 2.9919516368363928,
      "learning_rate": 1.6425314754364258e-06,
      "loss": 0.9381,
      "step": 7907
    },
    {
      "epoch": 0.5705421882327477,
      "grad_norm": 4.221191829860086,
      "learning_rate": 1.6420716511043092e-06,
      "loss": 0.9746,
      "step": 7908
    },
    {
      "epoch": 0.5706143357021752,
      "grad_norm": 6.7786818181090425,
      "learning_rate": 1.6416118463173355e-06,
      "loss": 0.9235,
      "step": 7909
    },
    {
      "epoch": 0.5706864831716028,
      "grad_norm": 4.734850365470744,
      "learning_rate": 1.6411520611006115e-06,
      "loss": 1.0305,
      "step": 7910
    },
    {
      "epoch": 0.5707586306410303,
      "grad_norm": 5.515560160571879,
      "learning_rate": 1.640692295479245e-06,
      "loss": 0.9294,
      "step": 7911
    },
    {
      "epoch": 0.5708307781104578,
      "grad_norm": 3.6072159291680275,
      "learning_rate": 1.6402325494783428e-06,
      "loss": 0.8754,
      "step": 7912
    },
    {
      "epoch": 0.5709029255798853,
      "grad_norm": 3.973942521223389,
      "learning_rate": 1.6397728231230084e-06,
      "loss": 1.0287,
      "step": 7913
    },
    {
      "epoch": 0.5709750730493128,
      "grad_norm": 4.783541261271926,
      "learning_rate": 1.6393131164383474e-06,
      "loss": 0.9244,
      "step": 7914
    },
    {
      "epoch": 0.5710472205187404,
      "grad_norm": 4.442427142935276,
      "learning_rate": 1.6388534294494618e-06,
      "loss": 0.9684,
      "step": 7915
    },
    {
      "epoch": 0.5711193679881679,
      "grad_norm": 4.044326274202791,
      "learning_rate": 1.6383937621814525e-06,
      "loss": 0.8466,
      "step": 7916
    },
    {
      "epoch": 0.5711915154575953,
      "grad_norm": 4.098180804574442,
      "learning_rate": 1.6379341146594222e-06,
      "loss": 0.9912,
      "step": 7917
    },
    {
      "epoch": 0.5712636629270228,
      "grad_norm": 5.575553408127732,
      "learning_rate": 1.6374744869084689e-06,
      "loss": 0.9125,
      "step": 7918
    },
    {
      "epoch": 0.5713358103964503,
      "grad_norm": 5.14083333158542,
      "learning_rate": 1.6370148789536925e-06,
      "loss": 1.0006,
      "step": 7919
    },
    {
      "epoch": 0.5714079578658778,
      "grad_norm": 4.931989661225236,
      "learning_rate": 1.63655529082019e-06,
      "loss": 1.0003,
      "step": 7920
    },
    {
      "epoch": 0.5714801053353054,
      "grad_norm": 5.006270482167838,
      "learning_rate": 1.6360957225330571e-06,
      "loss": 1.0807,
      "step": 7921
    },
    {
      "epoch": 0.5715522528047329,
      "grad_norm": 2.849402877092224,
      "learning_rate": 1.6356361741173897e-06,
      "loss": 1.0513,
      "step": 7922
    },
    {
      "epoch": 0.5716244002741604,
      "grad_norm": 3.402786453824217,
      "learning_rate": 1.6351766455982821e-06,
      "loss": 0.9863,
      "step": 7923
    },
    {
      "epoch": 0.5716965477435879,
      "grad_norm": 4.2913178731171575,
      "learning_rate": 1.6347171370008278e-06,
      "loss": 1.0472,
      "step": 7924
    },
    {
      "epoch": 0.5717686952130154,
      "grad_norm": 4.293397474184971,
      "learning_rate": 1.6342576483501173e-06,
      "loss": 1.0395,
      "step": 7925
    },
    {
      "epoch": 0.571840842682443,
      "grad_norm": 3.6971357497081465,
      "learning_rate": 1.6337981796712437e-06,
      "loss": 0.9175,
      "step": 7926
    },
    {
      "epoch": 0.5719129901518705,
      "grad_norm": 6.8544119778440304,
      "learning_rate": 1.633338730989296e-06,
      "loss": 0.9883,
      "step": 7927
    },
    {
      "epoch": 0.5719851376212979,
      "grad_norm": 4.0156026757951935,
      "learning_rate": 1.6328793023293617e-06,
      "loss": 0.9827,
      "step": 7928
    },
    {
      "epoch": 0.5720572850907254,
      "grad_norm": 0.7945109061878396,
      "learning_rate": 1.6324198937165307e-06,
      "loss": 0.8499,
      "step": 7929
    },
    {
      "epoch": 0.5721294325601529,
      "grad_norm": 5.412675864377746,
      "learning_rate": 1.6319605051758879e-06,
      "loss": 1.0015,
      "step": 7930
    },
    {
      "epoch": 0.5722015800295804,
      "grad_norm": 3.710635138792245,
      "learning_rate": 1.63150113673252e-06,
      "loss": 0.875,
      "step": 7931
    },
    {
      "epoch": 0.572273727499008,
      "grad_norm": 12.2906169550889,
      "learning_rate": 1.6310417884115112e-06,
      "loss": 0.9776,
      "step": 7932
    },
    {
      "epoch": 0.5723458749684355,
      "grad_norm": 10.6176118183468,
      "learning_rate": 1.6305824602379436e-06,
      "loss": 0.9142,
      "step": 7933
    },
    {
      "epoch": 0.572418022437863,
      "grad_norm": 2.9925038143124087,
      "learning_rate": 1.6301231522369007e-06,
      "loss": 0.9656,
      "step": 7934
    },
    {
      "epoch": 0.5724901699072905,
      "grad_norm": 3.0943918814571356,
      "learning_rate": 1.6296638644334632e-06,
      "loss": 0.8936,
      "step": 7935
    },
    {
      "epoch": 0.572562317376718,
      "grad_norm": 3.7539016771239058,
      "learning_rate": 1.6292045968527114e-06,
      "loss": 0.8531,
      "step": 7936
    },
    {
      "epoch": 0.5726344648461456,
      "grad_norm": 3.558693843568651,
      "learning_rate": 1.6287453495197236e-06,
      "loss": 0.9354,
      "step": 7937
    },
    {
      "epoch": 0.5727066123155731,
      "grad_norm": 5.779486639859331,
      "learning_rate": 1.6282861224595778e-06,
      "loss": 0.9253,
      "step": 7938
    },
    {
      "epoch": 0.5727787597850006,
      "grad_norm": 5.11993939900297,
      "learning_rate": 1.6278269156973514e-06,
      "loss": 0.9206,
      "step": 7939
    },
    {
      "epoch": 0.572850907254428,
      "grad_norm": 3.458554931036738,
      "learning_rate": 1.6273677292581185e-06,
      "loss": 0.8689,
      "step": 7940
    },
    {
      "epoch": 0.5729230547238555,
      "grad_norm": 5.191225116173166,
      "learning_rate": 1.6269085631669554e-06,
      "loss": 0.8757,
      "step": 7941
    },
    {
      "epoch": 0.572995202193283,
      "grad_norm": 3.6558246854238234,
      "learning_rate": 1.6264494174489345e-06,
      "loss": 0.919,
      "step": 7942
    },
    {
      "epoch": 0.5730673496627106,
      "grad_norm": 5.442346627173129,
      "learning_rate": 1.6259902921291271e-06,
      "loss": 0.8581,
      "step": 7943
    },
    {
      "epoch": 0.5731394971321381,
      "grad_norm": 3.2746834580128668,
      "learning_rate": 1.6255311872326064e-06,
      "loss": 0.992,
      "step": 7944
    },
    {
      "epoch": 0.5732116446015656,
      "grad_norm": 2.677693862664225,
      "learning_rate": 1.6250721027844405e-06,
      "loss": 0.9719,
      "step": 7945
    },
    {
      "epoch": 0.5732837920709931,
      "grad_norm": 6.4718178371260855,
      "learning_rate": 1.6246130388096995e-06,
      "loss": 0.9274,
      "step": 7946
    },
    {
      "epoch": 0.5733559395404206,
      "grad_norm": 7.726259601699993,
      "learning_rate": 1.6241539953334508e-06,
      "loss": 1.0956,
      "step": 7947
    },
    {
      "epoch": 0.5734280870098482,
      "grad_norm": 3.1288742463836123,
      "learning_rate": 1.6236949723807612e-06,
      "loss": 1.039,
      "step": 7948
    },
    {
      "epoch": 0.5735002344792757,
      "grad_norm": 4.680998298590925,
      "learning_rate": 1.623235969976696e-06,
      "loss": 0.9142,
      "step": 7949
    },
    {
      "epoch": 0.5735723819487032,
      "grad_norm": 2.426766359862966,
      "learning_rate": 1.6227769881463196e-06,
      "loss": 0.9307,
      "step": 7950
    },
    {
      "epoch": 0.5736445294181307,
      "grad_norm": 4.657552652107469,
      "learning_rate": 1.622318026914696e-06,
      "loss": 1.0273,
      "step": 7951
    },
    {
      "epoch": 0.5737166768875581,
      "grad_norm": 3.091019629730519,
      "learning_rate": 1.6218590863068857e-06,
      "loss": 1.0231,
      "step": 7952
    },
    {
      "epoch": 0.5737888243569856,
      "grad_norm": 6.609383657463302,
      "learning_rate": 1.621400166347952e-06,
      "loss": 0.8393,
      "step": 7953
    },
    {
      "epoch": 0.5738609718264132,
      "grad_norm": 9.18220336202281,
      "learning_rate": 1.6209412670629535e-06,
      "loss": 0.909,
      "step": 7954
    },
    {
      "epoch": 0.5739331192958407,
      "grad_norm": 0.6427908548170413,
      "learning_rate": 1.6204823884769482e-06,
      "loss": 0.7969,
      "step": 7955
    },
    {
      "epoch": 0.5740052667652682,
      "grad_norm": 4.700324051419699,
      "learning_rate": 1.620023530614996e-06,
      "loss": 0.9075,
      "step": 7956
    },
    {
      "epoch": 0.5740774142346957,
      "grad_norm": 3.814214477261635,
      "learning_rate": 1.6195646935021517e-06,
      "loss": 0.9986,
      "step": 7957
    },
    {
      "epoch": 0.5741495617041232,
      "grad_norm": 0.678833962686598,
      "learning_rate": 1.619105877163471e-06,
      "loss": 0.8089,
      "step": 7958
    },
    {
      "epoch": 0.5742217091735508,
      "grad_norm": 4.141962950846089,
      "learning_rate": 1.6186470816240086e-06,
      "loss": 1.0111,
      "step": 7959
    },
    {
      "epoch": 0.5742938566429783,
      "grad_norm": 7.017487888196625,
      "learning_rate": 1.6181883069088172e-06,
      "loss": 1.0265,
      "step": 7960
    },
    {
      "epoch": 0.5743660041124058,
      "grad_norm": 3.541183408790398,
      "learning_rate": 1.6177295530429487e-06,
      "loss": 0.9222,
      "step": 7961
    },
    {
      "epoch": 0.5744381515818333,
      "grad_norm": 4.36823032328042,
      "learning_rate": 1.6172708200514543e-06,
      "loss": 0.9532,
      "step": 7962
    },
    {
      "epoch": 0.5745102990512608,
      "grad_norm": 4.747295814530539,
      "learning_rate": 1.616812107959384e-06,
      "loss": 0.8696,
      "step": 7963
    },
    {
      "epoch": 0.5745824465206882,
      "grad_norm": 4.269366178261934,
      "learning_rate": 1.6163534167917855e-06,
      "loss": 0.9746,
      "step": 7964
    },
    {
      "epoch": 0.5746545939901158,
      "grad_norm": 2.7786717841173862,
      "learning_rate": 1.615894746573707e-06,
      "loss": 0.9318,
      "step": 7965
    },
    {
      "epoch": 0.5747267414595433,
      "grad_norm": 2.571174639319334,
      "learning_rate": 1.6154360973301946e-06,
      "loss": 0.9002,
      "step": 7966
    },
    {
      "epoch": 0.5747988889289708,
      "grad_norm": 2.65398094649328,
      "learning_rate": 1.6149774690862924e-06,
      "loss": 0.9036,
      "step": 7967
    },
    {
      "epoch": 0.5748710363983983,
      "grad_norm": 5.481235050890033,
      "learning_rate": 1.6145188618670465e-06,
      "loss": 1.053,
      "step": 7968
    },
    {
      "epoch": 0.5749431838678258,
      "grad_norm": 0.7012881126858463,
      "learning_rate": 1.6140602756974979e-06,
      "loss": 0.7866,
      "step": 7969
    },
    {
      "epoch": 0.5750153313372534,
      "grad_norm": 3.878973984294997,
      "learning_rate": 1.613601710602689e-06,
      "loss": 0.8906,
      "step": 7970
    },
    {
      "epoch": 0.5750874788066809,
      "grad_norm": 2.895965756103214,
      "learning_rate": 1.6131431666076607e-06,
      "loss": 0.9529,
      "step": 7971
    },
    {
      "epoch": 0.5751596262761084,
      "grad_norm": 3.380657258319746,
      "learning_rate": 1.6126846437374516e-06,
      "loss": 0.8627,
      "step": 7972
    },
    {
      "epoch": 0.5752317737455359,
      "grad_norm": 7.153727124204426,
      "learning_rate": 1.6122261420171e-06,
      "loss": 0.9292,
      "step": 7973
    },
    {
      "epoch": 0.5753039212149634,
      "grad_norm": 3.8658273482038235,
      "learning_rate": 1.6117676614716437e-06,
      "loss": 0.9074,
      "step": 7974
    },
    {
      "epoch": 0.5753760686843908,
      "grad_norm": 3.356857989205439,
      "learning_rate": 1.6113092021261186e-06,
      "loss": 1.0182,
      "step": 7975
    },
    {
      "epoch": 0.5754482161538184,
      "grad_norm": 6.453404011825367,
      "learning_rate": 1.6108507640055587e-06,
      "loss": 0.9717,
      "step": 7976
    },
    {
      "epoch": 0.5755203636232459,
      "grad_norm": 4.840154088302431,
      "learning_rate": 1.6103923471349975e-06,
      "loss": 0.9504,
      "step": 7977
    },
    {
      "epoch": 0.5755925110926734,
      "grad_norm": 8.045089493959674,
      "learning_rate": 1.609933951539469e-06,
      "loss": 0.9094,
      "step": 7978
    },
    {
      "epoch": 0.5756646585621009,
      "grad_norm": 3.597303673514709,
      "learning_rate": 1.6094755772440023e-06,
      "loss": 1.0351,
      "step": 7979
    },
    {
      "epoch": 0.5757368060315284,
      "grad_norm": 4.661984388052821,
      "learning_rate": 1.6090172242736295e-06,
      "loss": 0.9175,
      "step": 7980
    },
    {
      "epoch": 0.575808953500956,
      "grad_norm": 3.072167843028324,
      "learning_rate": 1.6085588926533787e-06,
      "loss": 0.9998,
      "step": 7981
    },
    {
      "epoch": 0.5758811009703835,
      "grad_norm": 0.802786461098629,
      "learning_rate": 1.608100582408277e-06,
      "loss": 0.8984,
      "step": 7982
    },
    {
      "epoch": 0.575953248439811,
      "grad_norm": 2.1607405853472432,
      "learning_rate": 1.6076422935633524e-06,
      "loss": 1.031,
      "step": 7983
    },
    {
      "epoch": 0.5760253959092385,
      "grad_norm": 3.889971697392997,
      "learning_rate": 1.6071840261436294e-06,
      "loss": 0.9743,
      "step": 7984
    },
    {
      "epoch": 0.576097543378666,
      "grad_norm": 3.8568351511858596,
      "learning_rate": 1.6067257801741325e-06,
      "loss": 0.9504,
      "step": 7985
    },
    {
      "epoch": 0.5761696908480936,
      "grad_norm": 3.5434837766081912,
      "learning_rate": 1.6062675556798853e-06,
      "loss": 0.9997,
      "step": 7986
    },
    {
      "epoch": 0.576241838317521,
      "grad_norm": 3.2828588991730796,
      "learning_rate": 1.6058093526859088e-06,
      "loss": 0.9662,
      "step": 7987
    },
    {
      "epoch": 0.5763139857869485,
      "grad_norm": 5.318389500236016,
      "learning_rate": 1.6053511712172245e-06,
      "loss": 0.9895,
      "step": 7988
    },
    {
      "epoch": 0.576386133256376,
      "grad_norm": 9.826666646445839,
      "learning_rate": 1.604893011298852e-06,
      "loss": 0.9362,
      "step": 7989
    },
    {
      "epoch": 0.5764582807258035,
      "grad_norm": 4.419706318277193,
      "learning_rate": 1.6044348729558095e-06,
      "loss": 0.994,
      "step": 7990
    },
    {
      "epoch": 0.576530428195231,
      "grad_norm": 3.5979698284722565,
      "learning_rate": 1.6039767562131137e-06,
      "loss": 1.0126,
      "step": 7991
    },
    {
      "epoch": 0.5766025756646586,
      "grad_norm": 0.7434129736736301,
      "learning_rate": 1.6035186610957819e-06,
      "loss": 0.7501,
      "step": 7992
    },
    {
      "epoch": 0.5766747231340861,
      "grad_norm": 3.196030664299251,
      "learning_rate": 1.6030605876288284e-06,
      "loss": 0.975,
      "step": 7993
    },
    {
      "epoch": 0.5767468706035136,
      "grad_norm": 0.7813369702568079,
      "learning_rate": 1.602602535837266e-06,
      "loss": 0.8105,
      "step": 7994
    },
    {
      "epoch": 0.5768190180729411,
      "grad_norm": 5.451312531775425,
      "learning_rate": 1.6021445057461084e-06,
      "loss": 0.9814,
      "step": 7995
    },
    {
      "epoch": 0.5768911655423686,
      "grad_norm": 0.7004269438909994,
      "learning_rate": 1.6016864973803666e-06,
      "loss": 0.8357,
      "step": 7996
    },
    {
      "epoch": 0.5769633130117962,
      "grad_norm": 7.921099201565093,
      "learning_rate": 1.6012285107650504e-06,
      "loss": 1.0194,
      "step": 7997
    },
    {
      "epoch": 0.5770354604812237,
      "grad_norm": 3.013495132890476,
      "learning_rate": 1.6007705459251696e-06,
      "loss": 0.9769,
      "step": 7998
    },
    {
      "epoch": 0.5771076079506511,
      "grad_norm": 4.4363866134594305,
      "learning_rate": 1.6003126028857313e-06,
      "loss": 0.8723,
      "step": 7999
    },
    {
      "epoch": 0.5771797554200786,
      "grad_norm": 3.557959490634978,
      "learning_rate": 1.5998546816717416e-06,
      "loss": 1.0503,
      "step": 8000
    },
    {
      "epoch": 0.5772519028895061,
      "grad_norm": 3.14295485889571,
      "learning_rate": 1.5993967823082066e-06,
      "loss": 0.9947,
      "step": 8001
    },
    {
      "epoch": 0.5773240503589336,
      "grad_norm": 3.3189194573137417,
      "learning_rate": 1.598938904820131e-06,
      "loss": 0.9399,
      "step": 8002
    },
    {
      "epoch": 0.5773961978283612,
      "grad_norm": 3.7526317262751583,
      "learning_rate": 1.5984810492325165e-06,
      "loss": 0.8819,
      "step": 8003
    },
    {
      "epoch": 0.5774683452977887,
      "grad_norm": 3.2085422761595073,
      "learning_rate": 1.5980232155703655e-06,
      "loss": 0.901,
      "step": 8004
    },
    {
      "epoch": 0.5775404927672162,
      "grad_norm": 4.288915079220232,
      "learning_rate": 1.597565403858679e-06,
      "loss": 0.9427,
      "step": 8005
    },
    {
      "epoch": 0.5776126402366437,
      "grad_norm": 3.618458468960139,
      "learning_rate": 1.5971076141224554e-06,
      "loss": 0.9982,
      "step": 8006
    },
    {
      "epoch": 0.5776847877060712,
      "grad_norm": 3.789866191142523,
      "learning_rate": 1.5966498463866941e-06,
      "loss": 0.912,
      "step": 8007
    },
    {
      "epoch": 0.5777569351754988,
      "grad_norm": 14.438434785247727,
      "learning_rate": 1.5961921006763918e-06,
      "loss": 0.9537,
      "step": 8008
    },
    {
      "epoch": 0.5778290826449263,
      "grad_norm": 2.194936699790877,
      "learning_rate": 1.595734377016543e-06,
      "loss": 0.9772,
      "step": 8009
    },
    {
      "epoch": 0.5779012301143538,
      "grad_norm": 4.722344450521901,
      "learning_rate": 1.5952766754321445e-06,
      "loss": 0.9394,
      "step": 8010
    },
    {
      "epoch": 0.5779733775837812,
      "grad_norm": 4.172666096064142,
      "learning_rate": 1.5948189959481879e-06,
      "loss": 0.8896,
      "step": 8011
    },
    {
      "epoch": 0.5780455250532087,
      "grad_norm": 3.9852962849940043,
      "learning_rate": 1.5943613385896659e-06,
      "loss": 1.0262,
      "step": 8012
    },
    {
      "epoch": 0.5781176725226362,
      "grad_norm": 0.7328079223489836,
      "learning_rate": 1.5939037033815696e-06,
      "loss": 0.8337,
      "step": 8013
    },
    {
      "epoch": 0.5781898199920638,
      "grad_norm": 2.1169561843914724,
      "learning_rate": 1.5934460903488892e-06,
      "loss": 0.9798,
      "step": 8014
    },
    {
      "epoch": 0.5782619674614913,
      "grad_norm": 4.049526686149375,
      "learning_rate": 1.5929884995166125e-06,
      "loss": 1.0157,
      "step": 8015
    },
    {
      "epoch": 0.5783341149309188,
      "grad_norm": 4.7452278257772065,
      "learning_rate": 1.5925309309097271e-06,
      "loss": 1.0018,
      "step": 8016
    },
    {
      "epoch": 0.5784062624003463,
      "grad_norm": 4.959075914869963,
      "learning_rate": 1.5920733845532194e-06,
      "loss": 0.9748,
      "step": 8017
    },
    {
      "epoch": 0.5784784098697738,
      "grad_norm": 2.837782582735584,
      "learning_rate": 1.5916158604720733e-06,
      "loss": 0.8824,
      "step": 8018
    },
    {
      "epoch": 0.5785505573392014,
      "grad_norm": 3.8253316847659264,
      "learning_rate": 1.5911583586912742e-06,
      "loss": 1.1321,
      "step": 8019
    },
    {
      "epoch": 0.5786227048086289,
      "grad_norm": 4.367287731897076,
      "learning_rate": 1.5907008792358036e-06,
      "loss": 0.9436,
      "step": 8020
    },
    {
      "epoch": 0.5786948522780564,
      "grad_norm": 3.8948450449921,
      "learning_rate": 1.590243422130642e-06,
      "loss": 0.9315,
      "step": 8021
    },
    {
      "epoch": 0.5787669997474839,
      "grad_norm": 3.6105141431560255,
      "learning_rate": 1.589785987400771e-06,
      "loss": 0.9777,
      "step": 8022
    },
    {
      "epoch": 0.5788391472169113,
      "grad_norm": 3.7858466590502955,
      "learning_rate": 1.5893285750711685e-06,
      "loss": 0.8855,
      "step": 8023
    },
    {
      "epoch": 0.5789112946863388,
      "grad_norm": 5.59183354287445,
      "learning_rate": 1.588871185166812e-06,
      "loss": 0.9657,
      "step": 8024
    },
    {
      "epoch": 0.5789834421557664,
      "grad_norm": 5.2627301960884605,
      "learning_rate": 1.5884138177126786e-06,
      "loss": 1.0156,
      "step": 8025
    },
    {
      "epoch": 0.5790555896251939,
      "grad_norm": 0.7626361115713299,
      "learning_rate": 1.5879564727337427e-06,
      "loss": 0.8324,
      "step": 8026
    },
    {
      "epoch": 0.5791277370946214,
      "grad_norm": 3.8757090996468566,
      "learning_rate": 1.5874991502549784e-06,
      "loss": 0.8399,
      "step": 8027
    },
    {
      "epoch": 0.5791998845640489,
      "grad_norm": 2.9443759390421183,
      "learning_rate": 1.5870418503013582e-06,
      "loss": 0.899,
      "step": 8028
    },
    {
      "epoch": 0.5792720320334764,
      "grad_norm": 2.411392463679772,
      "learning_rate": 1.5865845728978546e-06,
      "loss": 0.9108,
      "step": 8029
    },
    {
      "epoch": 0.579344179502904,
      "grad_norm": 2.657781630599732,
      "learning_rate": 1.5861273180694367e-06,
      "loss": 0.9679,
      "step": 8030
    },
    {
      "epoch": 0.5794163269723315,
      "grad_norm": 3.786992780003138,
      "learning_rate": 1.5856700858410736e-06,
      "loss": 0.9906,
      "step": 8031
    },
    {
      "epoch": 0.579488474441759,
      "grad_norm": 4.410809045394793,
      "learning_rate": 1.585212876237734e-06,
      "loss": 0.8494,
      "step": 8032
    },
    {
      "epoch": 0.5795606219111865,
      "grad_norm": 0.72485886054993,
      "learning_rate": 1.5847556892843828e-06,
      "loss": 0.8361,
      "step": 8033
    },
    {
      "epoch": 0.5796327693806139,
      "grad_norm": 2.944386789569155,
      "learning_rate": 1.5842985250059871e-06,
      "loss": 0.9681,
      "step": 8034
    },
    {
      "epoch": 0.5797049168500414,
      "grad_norm": 13.852888204150053,
      "learning_rate": 1.5838413834275092e-06,
      "loss": 0.9234,
      "step": 8035
    },
    {
      "epoch": 0.579777064319469,
      "grad_norm": 5.2723217497313675,
      "learning_rate": 1.5833842645739134e-06,
      "loss": 0.8784,
      "step": 8036
    },
    {
      "epoch": 0.5798492117888965,
      "grad_norm": 0.833057397616571,
      "learning_rate": 1.5829271684701609e-06,
      "loss": 0.8132,
      "step": 8037
    },
    {
      "epoch": 0.579921359258324,
      "grad_norm": 3.252238309801832,
      "learning_rate": 1.5824700951412111e-06,
      "loss": 0.9065,
      "step": 8038
    },
    {
      "epoch": 0.5799935067277515,
      "grad_norm": 0.6811193752172081,
      "learning_rate": 1.582013044612024e-06,
      "loss": 0.8384,
      "step": 8039
    },
    {
      "epoch": 0.580065654197179,
      "grad_norm": 3.5660319972803722,
      "learning_rate": 1.5815560169075572e-06,
      "loss": 0.886,
      "step": 8040
    },
    {
      "epoch": 0.5801378016666066,
      "grad_norm": 0.9085728046761551,
      "learning_rate": 1.5810990120527676e-06,
      "loss": 0.8294,
      "step": 8041
    },
    {
      "epoch": 0.5802099491360341,
      "grad_norm": 3.2030385075498495,
      "learning_rate": 1.58064203007261e-06,
      "loss": 1.0708,
      "step": 8042
    },
    {
      "epoch": 0.5802820966054616,
      "grad_norm": 0.7749451479420193,
      "learning_rate": 1.5801850709920386e-06,
      "loss": 0.8229,
      "step": 8043
    },
    {
      "epoch": 0.5803542440748891,
      "grad_norm": 2.9683961356228985,
      "learning_rate": 1.5797281348360068e-06,
      "loss": 1.0598,
      "step": 8044
    },
    {
      "epoch": 0.5804263915443166,
      "grad_norm": 3.445633715406301,
      "learning_rate": 1.5792712216294649e-06,
      "loss": 1.0429,
      "step": 8045
    },
    {
      "epoch": 0.580498539013744,
      "grad_norm": 2.789462790093841,
      "learning_rate": 1.578814331397365e-06,
      "loss": 0.8961,
      "step": 8046
    },
    {
      "epoch": 0.5805706864831716,
      "grad_norm": 0.9221662934003366,
      "learning_rate": 1.578357464164655e-06,
      "loss": 0.9483,
      "step": 8047
    },
    {
      "epoch": 0.5806428339525991,
      "grad_norm": 3.5550833345456145,
      "learning_rate": 1.5779006199562826e-06,
      "loss": 0.9237,
      "step": 8048
    },
    {
      "epoch": 0.5807149814220266,
      "grad_norm": 8.906225157585967,
      "learning_rate": 1.5774437987971954e-06,
      "loss": 0.9993,
      "step": 8049
    },
    {
      "epoch": 0.5807871288914541,
      "grad_norm": 8.013938681809387,
      "learning_rate": 1.5769870007123377e-06,
      "loss": 0.9821,
      "step": 8050
    },
    {
      "epoch": 0.5808592763608816,
      "grad_norm": 4.569802280213994,
      "learning_rate": 1.576530225726654e-06,
      "loss": 0.9911,
      "step": 8051
    },
    {
      "epoch": 0.5809314238303092,
      "grad_norm": 3.131993821782319,
      "learning_rate": 1.5760734738650873e-06,
      "loss": 0.8678,
      "step": 8052
    },
    {
      "epoch": 0.5810035712997367,
      "grad_norm": 5.089950173964132,
      "learning_rate": 1.5756167451525789e-06,
      "loss": 0.9439,
      "step": 8053
    },
    {
      "epoch": 0.5810757187691642,
      "grad_norm": 3.663894356649691,
      "learning_rate": 1.5751600396140686e-06,
      "loss": 0.9669,
      "step": 8054
    },
    {
      "epoch": 0.5811478662385917,
      "grad_norm": 3.244294954656191,
      "learning_rate": 1.5747033572744958e-06,
      "loss": 0.8216,
      "step": 8055
    },
    {
      "epoch": 0.5812200137080192,
      "grad_norm": 2.9931941716980717,
      "learning_rate": 1.5742466981587989e-06,
      "loss": 0.9779,
      "step": 8056
    },
    {
      "epoch": 0.5812921611774468,
      "grad_norm": 3.763525919912538,
      "learning_rate": 1.573790062291913e-06,
      "loss": 1.0662,
      "step": 8057
    },
    {
      "epoch": 0.5813643086468742,
      "grad_norm": 4.052176872422464,
      "learning_rate": 1.5733334496987744e-06,
      "loss": 0.9604,
      "step": 8058
    },
    {
      "epoch": 0.5814364561163017,
      "grad_norm": 2.6076498210580183,
      "learning_rate": 1.572876860404317e-06,
      "loss": 0.9405,
      "step": 8059
    },
    {
      "epoch": 0.5815086035857292,
      "grad_norm": 5.0512532709154065,
      "learning_rate": 1.5724202944334724e-06,
      "loss": 0.8788,
      "step": 8060
    },
    {
      "epoch": 0.5815807510551567,
      "grad_norm": 8.949292066482741,
      "learning_rate": 1.5719637518111732e-06,
      "loss": 0.9418,
      "step": 8061
    },
    {
      "epoch": 0.5816528985245842,
      "grad_norm": 0.9301252857028895,
      "learning_rate": 1.5715072325623487e-06,
      "loss": 0.8627,
      "step": 8062
    },
    {
      "epoch": 0.5817250459940118,
      "grad_norm": 4.54651695493471,
      "learning_rate": 1.571050736711928e-06,
      "loss": 0.9484,
      "step": 8063
    },
    {
      "epoch": 0.5817971934634393,
      "grad_norm": 6.285998319425974,
      "learning_rate": 1.5705942642848392e-06,
      "loss": 0.9416,
      "step": 8064
    },
    {
      "epoch": 0.5818693409328668,
      "grad_norm": 5.1014428584745755,
      "learning_rate": 1.5701378153060072e-06,
      "loss": 0.887,
      "step": 8065
    },
    {
      "epoch": 0.5819414884022943,
      "grad_norm": 3.6436956332879937,
      "learning_rate": 1.569681389800358e-06,
      "loss": 1.0786,
      "step": 8066
    },
    {
      "epoch": 0.5820136358717218,
      "grad_norm": 4.336337124391053,
      "learning_rate": 1.5692249877928153e-06,
      "loss": 0.9483,
      "step": 8067
    },
    {
      "epoch": 0.5820857833411494,
      "grad_norm": 4.5487943707823435,
      "learning_rate": 1.5687686093083016e-06,
      "loss": 0.9392,
      "step": 8068
    },
    {
      "epoch": 0.5821579308105769,
      "grad_norm": 0.8220095564200864,
      "learning_rate": 1.568312254371738e-06,
      "loss": 0.8962,
      "step": 8069
    },
    {
      "epoch": 0.5822300782800043,
      "grad_norm": 3.291048514997001,
      "learning_rate": 1.567855923008043e-06,
      "loss": 1.0097,
      "step": 8070
    },
    {
      "epoch": 0.5823022257494318,
      "grad_norm": 6.600660418211421,
      "learning_rate": 1.567399615242137e-06,
      "loss": 0.8893,
      "step": 8071
    },
    {
      "epoch": 0.5823743732188593,
      "grad_norm": 3.980328109098432,
      "learning_rate": 1.5669433310989361e-06,
      "loss": 0.9224,
      "step": 8072
    },
    {
      "epoch": 0.5824465206882868,
      "grad_norm": 3.3493995356101527,
      "learning_rate": 1.5664870706033577e-06,
      "loss": 0.9385,
      "step": 8073
    },
    {
      "epoch": 0.5825186681577144,
      "grad_norm": 3.237095522560336,
      "learning_rate": 1.5660308337803152e-06,
      "loss": 0.9903,
      "step": 8074
    },
    {
      "epoch": 0.5825908156271419,
      "grad_norm": 5.739663494494193,
      "learning_rate": 1.5655746206547219e-06,
      "loss": 0.9454,
      "step": 8075
    },
    {
      "epoch": 0.5826629630965694,
      "grad_norm": 4.332061850024365,
      "learning_rate": 1.565118431251491e-06,
      "loss": 0.9982,
      "step": 8076
    },
    {
      "epoch": 0.5827351105659969,
      "grad_norm": 4.547478599728671,
      "learning_rate": 1.5646622655955322e-06,
      "loss": 0.991,
      "step": 8077
    },
    {
      "epoch": 0.5828072580354244,
      "grad_norm": 3.994919530752771,
      "learning_rate": 1.5642061237117555e-06,
      "loss": 0.9332,
      "step": 8078
    },
    {
      "epoch": 0.582879405504852,
      "grad_norm": 2.4451968619697313,
      "learning_rate": 1.5637500056250692e-06,
      "loss": 0.9474,
      "step": 8079
    },
    {
      "epoch": 0.5829515529742795,
      "grad_norm": 3.4915249257842977,
      "learning_rate": 1.5632939113603807e-06,
      "loss": 0.9658,
      "step": 8080
    },
    {
      "epoch": 0.5830237004437069,
      "grad_norm": 4.781401189271161,
      "learning_rate": 1.5628378409425951e-06,
      "loss": 1.0135,
      "step": 8081
    },
    {
      "epoch": 0.5830958479131344,
      "grad_norm": 4.068187312582093,
      "learning_rate": 1.5623817943966156e-06,
      "loss": 0.912,
      "step": 8082
    },
    {
      "epoch": 0.5831679953825619,
      "grad_norm": 3.0783652584593897,
      "learning_rate": 1.5619257717473474e-06,
      "loss": 0.934,
      "step": 8083
    },
    {
      "epoch": 0.5832401428519894,
      "grad_norm": 3.9410562350392437,
      "learning_rate": 1.56146977301969e-06,
      "loss": 0.9775,
      "step": 8084
    },
    {
      "epoch": 0.583312290321417,
      "grad_norm": 11.149636410656795,
      "learning_rate": 1.561013798238546e-06,
      "loss": 1.0015,
      "step": 8085
    },
    {
      "epoch": 0.5833844377908445,
      "grad_norm": 2.860949807007766,
      "learning_rate": 1.560557847428813e-06,
      "loss": 1.0662,
      "step": 8086
    },
    {
      "epoch": 0.583456585260272,
      "grad_norm": 6.8707045574469765,
      "learning_rate": 1.5601019206153888e-06,
      "loss": 0.9514,
      "step": 8087
    },
    {
      "epoch": 0.5835287327296995,
      "grad_norm": 2.9539144762555716,
      "learning_rate": 1.559646017823171e-06,
      "loss": 0.9248,
      "step": 8088
    },
    {
      "epoch": 0.583600880199127,
      "grad_norm": 5.068491463235449,
      "learning_rate": 1.5591901390770535e-06,
      "loss": 1.0337,
      "step": 8089
    },
    {
      "epoch": 0.5836730276685546,
      "grad_norm": 5.988050323234098,
      "learning_rate": 1.5587342844019308e-06,
      "loss": 0.8663,
      "step": 8090
    },
    {
      "epoch": 0.5837451751379821,
      "grad_norm": 5.452780289124964,
      "learning_rate": 1.5582784538226956e-06,
      "loss": 0.9837,
      "step": 8091
    },
    {
      "epoch": 0.5838173226074096,
      "grad_norm": 3.1894150011595515,
      "learning_rate": 1.5578226473642384e-06,
      "loss": 0.9514,
      "step": 8092
    },
    {
      "epoch": 0.583889470076837,
      "grad_norm": 2.5319693214402017,
      "learning_rate": 1.5573668650514501e-06,
      "loss": 0.9614,
      "step": 8093
    },
    {
      "epoch": 0.5839616175462645,
      "grad_norm": 5.1183905432205625,
      "learning_rate": 1.5569111069092178e-06,
      "loss": 0.9533,
      "step": 8094
    },
    {
      "epoch": 0.584033765015692,
      "grad_norm": 3.26306037036581,
      "learning_rate": 1.5564553729624308e-06,
      "loss": 0.9156,
      "step": 8095
    },
    {
      "epoch": 0.5841059124851196,
      "grad_norm": 4.470859056572389,
      "learning_rate": 1.5559996632359736e-06,
      "loss": 0.8884,
      "step": 8096
    },
    {
      "epoch": 0.5841780599545471,
      "grad_norm": 5.094178164220725,
      "learning_rate": 1.5555439777547306e-06,
      "loss": 0.9218,
      "step": 8097
    },
    {
      "epoch": 0.5842502074239746,
      "grad_norm": 0.8184315142097661,
      "learning_rate": 1.5550883165435863e-06,
      "loss": 0.8937,
      "step": 8098
    },
    {
      "epoch": 0.5843223548934021,
      "grad_norm": 3.1792176295470793,
      "learning_rate": 1.5546326796274212e-06,
      "loss": 1.0777,
      "step": 8099
    },
    {
      "epoch": 0.5843945023628296,
      "grad_norm": 5.014957942158747,
      "learning_rate": 1.5541770670311177e-06,
      "loss": 1.0205,
      "step": 8100
    },
    {
      "epoch": 0.5844666498322572,
      "grad_norm": 3.066189003799186,
      "learning_rate": 1.5537214787795537e-06,
      "loss": 0.8831,
      "step": 8101
    },
    {
      "epoch": 0.5845387973016847,
      "grad_norm": 6.929052760353446,
      "learning_rate": 1.5532659148976074e-06,
      "loss": 1.0103,
      "step": 8102
    },
    {
      "epoch": 0.5846109447711122,
      "grad_norm": 4.132628352023354,
      "learning_rate": 1.5528103754101565e-06,
      "loss": 0.9945,
      "step": 8103
    },
    {
      "epoch": 0.5846830922405397,
      "grad_norm": 2.8092031865521867,
      "learning_rate": 1.552354860342075e-06,
      "loss": 0.8602,
      "step": 8104
    },
    {
      "epoch": 0.5847552397099671,
      "grad_norm": 4.3220496108611615,
      "learning_rate": 1.551899369718238e-06,
      "loss": 0.9086,
      "step": 8105
    },
    {
      "epoch": 0.5848273871793946,
      "grad_norm": 11.089807631919873,
      "learning_rate": 1.5514439035635168e-06,
      "loss": 0.9741,
      "step": 8106
    },
    {
      "epoch": 0.5848995346488222,
      "grad_norm": 3.643774806571259,
      "learning_rate": 1.5509884619027843e-06,
      "loss": 1.0307,
      "step": 8107
    },
    {
      "epoch": 0.5849716821182497,
      "grad_norm": 4.393648999981016,
      "learning_rate": 1.5505330447609096e-06,
      "loss": 0.9676,
      "step": 8108
    },
    {
      "epoch": 0.5850438295876772,
      "grad_norm": 3.5067570719167285,
      "learning_rate": 1.5500776521627608e-06,
      "loss": 0.869,
      "step": 8109
    },
    {
      "epoch": 0.5851159770571047,
      "grad_norm": 0.8194441786368706,
      "learning_rate": 1.5496222841332064e-06,
      "loss": 0.9175,
      "step": 8110
    },
    {
      "epoch": 0.5851881245265322,
      "grad_norm": 4.38297699808277,
      "learning_rate": 1.5491669406971114e-06,
      "loss": 1.1056,
      "step": 8111
    },
    {
      "epoch": 0.5852602719959598,
      "grad_norm": 2.8282170570523326,
      "learning_rate": 1.5487116218793415e-06,
      "loss": 0.9049,
      "step": 8112
    },
    {
      "epoch": 0.5853324194653873,
      "grad_norm": 3.292201051109317,
      "learning_rate": 1.548256327704759e-06,
      "loss": 0.9968,
      "step": 8113
    },
    {
      "epoch": 0.5854045669348148,
      "grad_norm": 4.0597729259562225,
      "learning_rate": 1.5478010581982259e-06,
      "loss": 0.9548,
      "step": 8114
    },
    {
      "epoch": 0.5854767144042423,
      "grad_norm": 9.647349381962709,
      "learning_rate": 1.5473458133846036e-06,
      "loss": 0.9382,
      "step": 8115
    },
    {
      "epoch": 0.5855488618736698,
      "grad_norm": 2.719997602069022,
      "learning_rate": 1.5468905932887502e-06,
      "loss": 0.9519,
      "step": 8116
    },
    {
      "epoch": 0.5856210093430972,
      "grad_norm": 3.7153997636933678,
      "learning_rate": 1.5464353979355247e-06,
      "loss": 1.012,
      "step": 8117
    },
    {
      "epoch": 0.5856931568125248,
      "grad_norm": 3.303955862881959,
      "learning_rate": 1.545980227349782e-06,
      "loss": 1.018,
      "step": 8118
    },
    {
      "epoch": 0.5857653042819523,
      "grad_norm": 3.7848067776783996,
      "learning_rate": 1.5455250815563796e-06,
      "loss": 1.0984,
      "step": 8119
    },
    {
      "epoch": 0.5858374517513798,
      "grad_norm": 3.3446259108943237,
      "learning_rate": 1.5450699605801698e-06,
      "loss": 0.8736,
      "step": 8120
    },
    {
      "epoch": 0.5859095992208073,
      "grad_norm": 3.7146890730400024,
      "learning_rate": 1.5446148644460047e-06,
      "loss": 0.9928,
      "step": 8121
    },
    {
      "epoch": 0.5859817466902348,
      "grad_norm": 5.087371764570314,
      "learning_rate": 1.544159793178737e-06,
      "loss": 1.112,
      "step": 8122
    },
    {
      "epoch": 0.5860538941596624,
      "grad_norm": 3.7934131303069347,
      "learning_rate": 1.5437047468032142e-06,
      "loss": 0.9774,
      "step": 8123
    },
    {
      "epoch": 0.5861260416290899,
      "grad_norm": 4.083225145009624,
      "learning_rate": 1.5432497253442873e-06,
      "loss": 0.9451,
      "step": 8124
    },
    {
      "epoch": 0.5861981890985174,
      "grad_norm": 0.674659790735143,
      "learning_rate": 1.542794728826802e-06,
      "loss": 0.8341,
      "step": 8125
    },
    {
      "epoch": 0.5862703365679449,
      "grad_norm": 5.240879492588525,
      "learning_rate": 1.542339757275603e-06,
      "loss": 0.9913,
      "step": 8126
    },
    {
      "epoch": 0.5863424840373724,
      "grad_norm": 4.399723928633859,
      "learning_rate": 1.541884810715537e-06,
      "loss": 0.9827,
      "step": 8127
    },
    {
      "epoch": 0.5864146315067998,
      "grad_norm": 6.4976315217982625,
      "learning_rate": 1.5414298891714448e-06,
      "loss": 0.9942,
      "step": 8128
    },
    {
      "epoch": 0.5864867789762274,
      "grad_norm": 4.1119212105620555,
      "learning_rate": 1.5409749926681693e-06,
      "loss": 1.0263,
      "step": 8129
    },
    {
      "epoch": 0.5865589264456549,
      "grad_norm": 4.527309835601511,
      "learning_rate": 1.5405201212305496e-06,
      "loss": 0.9892,
      "step": 8130
    },
    {
      "epoch": 0.5866310739150824,
      "grad_norm": 4.120946800179205,
      "learning_rate": 1.5400652748834254e-06,
      "loss": 0.9962,
      "step": 8131
    },
    {
      "epoch": 0.5867032213845099,
      "grad_norm": 3.9568090348204965,
      "learning_rate": 1.539610453651634e-06,
      "loss": 0.98,
      "step": 8132
    },
    {
      "epoch": 0.5867753688539374,
      "grad_norm": 26.55026027745832,
      "learning_rate": 1.539155657560011e-06,
      "loss": 0.8382,
      "step": 8133
    },
    {
      "epoch": 0.586847516323365,
      "grad_norm": 5.370296993460756,
      "learning_rate": 1.538700886633392e-06,
      "loss": 0.9749,
      "step": 8134
    },
    {
      "epoch": 0.5869196637927925,
      "grad_norm": 4.17803477473051,
      "learning_rate": 1.53824614089661e-06,
      "loss": 0.9095,
      "step": 8135
    },
    {
      "epoch": 0.58699181126222,
      "grad_norm": 14.339596880536392,
      "learning_rate": 1.5377914203744965e-06,
      "loss": 0.8601,
      "step": 8136
    },
    {
      "epoch": 0.5870639587316475,
      "grad_norm": 3.0915823384289225,
      "learning_rate": 1.5373367250918828e-06,
      "loss": 0.9819,
      "step": 8137
    },
    {
      "epoch": 0.587136106201075,
      "grad_norm": 3.8349330853489483,
      "learning_rate": 1.536882055073597e-06,
      "loss": 0.9296,
      "step": 8138
    },
    {
      "epoch": 0.5872082536705026,
      "grad_norm": 6.349223965060305,
      "learning_rate": 1.5364274103444687e-06,
      "loss": 0.9731,
      "step": 8139
    },
    {
      "epoch": 0.58728040113993,
      "grad_norm": 3.585319249663728,
      "learning_rate": 1.5359727909293232e-06,
      "loss": 0.9172,
      "step": 8140
    },
    {
      "epoch": 0.5873525486093575,
      "grad_norm": 3.7920696477570055,
      "learning_rate": 1.5355181968529862e-06,
      "loss": 0.9202,
      "step": 8141
    },
    {
      "epoch": 0.587424696078785,
      "grad_norm": 4.3057259693397665,
      "learning_rate": 1.5350636281402804e-06,
      "loss": 0.938,
      "step": 8142
    },
    {
      "epoch": 0.5874968435482125,
      "grad_norm": 3.155321796770389,
      "learning_rate": 1.534609084816029e-06,
      "loss": 0.9383,
      "step": 8143
    },
    {
      "epoch": 0.58756899101764,
      "grad_norm": 3.229768578001193,
      "learning_rate": 1.5341545669050528e-06,
      "loss": 0.9905,
      "step": 8144
    },
    {
      "epoch": 0.5876411384870676,
      "grad_norm": 4.982699408553168,
      "learning_rate": 1.5337000744321707e-06,
      "loss": 1.0099,
      "step": 8145
    },
    {
      "epoch": 0.5877132859564951,
      "grad_norm": 3.4586284159199843,
      "learning_rate": 1.5332456074222018e-06,
      "loss": 0.9836,
      "step": 8146
    },
    {
      "epoch": 0.5877854334259226,
      "grad_norm": 4.693079565043939,
      "learning_rate": 1.5327911658999625e-06,
      "loss": 0.9404,
      "step": 8147
    },
    {
      "epoch": 0.5878575808953501,
      "grad_norm": 5.498052078705353,
      "learning_rate": 1.5323367498902676e-06,
      "loss": 1.0021,
      "step": 8148
    },
    {
      "epoch": 0.5879297283647776,
      "grad_norm": 3.147184894232286,
      "learning_rate": 1.531882359417932e-06,
      "loss": 0.9224,
      "step": 8149
    },
    {
      "epoch": 0.5880018758342052,
      "grad_norm": 2.918535414982292,
      "learning_rate": 1.531427994507767e-06,
      "loss": 0.8184,
      "step": 8150
    },
    {
      "epoch": 0.5880740233036327,
      "grad_norm": 0.8486972516597558,
      "learning_rate": 1.5309736551845857e-06,
      "loss": 0.8066,
      "step": 8151
    },
    {
      "epoch": 0.5881461707730601,
      "grad_norm": 4.524011552453574,
      "learning_rate": 1.5305193414731968e-06,
      "loss": 0.9759,
      "step": 8152
    },
    {
      "epoch": 0.5882183182424876,
      "grad_norm": 6.176754338561276,
      "learning_rate": 1.5300650533984082e-06,
      "loss": 0.9073,
      "step": 8153
    },
    {
      "epoch": 0.5882904657119151,
      "grad_norm": 0.7573527878587394,
      "learning_rate": 1.5296107909850272e-06,
      "loss": 0.9248,
      "step": 8154
    },
    {
      "epoch": 0.5883626131813426,
      "grad_norm": 3.636622901907245,
      "learning_rate": 1.52915655425786e-06,
      "loss": 0.9684,
      "step": 8155
    },
    {
      "epoch": 0.5884347606507702,
      "grad_norm": 0.709539065104819,
      "learning_rate": 1.5287023432417106e-06,
      "loss": 0.7512,
      "step": 8156
    },
    {
      "epoch": 0.5885069081201977,
      "grad_norm": 4.841423017156701,
      "learning_rate": 1.528248157961381e-06,
      "loss": 0.977,
      "step": 8157
    },
    {
      "epoch": 0.5885790555896252,
      "grad_norm": 3.105287558589897,
      "learning_rate": 1.5277939984416733e-06,
      "loss": 0.9502,
      "step": 8158
    },
    {
      "epoch": 0.5886512030590527,
      "grad_norm": 4.671184578086037,
      "learning_rate": 1.5273398647073877e-06,
      "loss": 1.0029,
      "step": 8159
    },
    {
      "epoch": 0.5887233505284802,
      "grad_norm": 6.706734065835173,
      "learning_rate": 1.5268857567833214e-06,
      "loss": 0.9889,
      "step": 8160
    },
    {
      "epoch": 0.5887954979979078,
      "grad_norm": 4.373262332789618,
      "learning_rate": 1.5264316746942733e-06,
      "loss": 0.8785,
      "step": 8161
    },
    {
      "epoch": 0.5888676454673353,
      "grad_norm": 24.4739565321965,
      "learning_rate": 1.5259776184650376e-06,
      "loss": 1.0343,
      "step": 8162
    },
    {
      "epoch": 0.5889397929367628,
      "grad_norm": 4.350732912605766,
      "learning_rate": 1.5255235881204102e-06,
      "loss": 0.9386,
      "step": 8163
    },
    {
      "epoch": 0.5890119404061902,
      "grad_norm": 3.3763943370372553,
      "learning_rate": 1.5250695836851835e-06,
      "loss": 0.987,
      "step": 8164
    },
    {
      "epoch": 0.5890840878756177,
      "grad_norm": 5.970594510583887,
      "learning_rate": 1.5246156051841479e-06,
      "loss": 0.9201,
      "step": 8165
    },
    {
      "epoch": 0.5891562353450452,
      "grad_norm": 4.0859223287092865,
      "learning_rate": 1.5241616526420944e-06,
      "loss": 0.9804,
      "step": 8166
    },
    {
      "epoch": 0.5892283828144728,
      "grad_norm": 7.769681488932035,
      "learning_rate": 1.5237077260838116e-06,
      "loss": 0.9916,
      "step": 8167
    },
    {
      "epoch": 0.5893005302839003,
      "grad_norm": 4.155288240829977,
      "learning_rate": 1.5232538255340873e-06,
      "loss": 0.869,
      "step": 8168
    },
    {
      "epoch": 0.5893726777533278,
      "grad_norm": 3.052199655514295,
      "learning_rate": 1.5227999510177061e-06,
      "loss": 0.8821,
      "step": 8169
    },
    {
      "epoch": 0.5894448252227553,
      "grad_norm": 4.962070029438398,
      "learning_rate": 1.5223461025594532e-06,
      "loss": 1.005,
      "step": 8170
    },
    {
      "epoch": 0.5895169726921828,
      "grad_norm": 6.493445980235932,
      "learning_rate": 1.5218922801841117e-06,
      "loss": 1.0248,
      "step": 8171
    },
    {
      "epoch": 0.5895891201616104,
      "grad_norm": 3.2656664685988464,
      "learning_rate": 1.5214384839164618e-06,
      "loss": 1.008,
      "step": 8172
    },
    {
      "epoch": 0.5896612676310379,
      "grad_norm": 3.2092013155138917,
      "learning_rate": 1.5209847137812859e-06,
      "loss": 0.997,
      "step": 8173
    },
    {
      "epoch": 0.5897334151004654,
      "grad_norm": 3.6015765340866754,
      "learning_rate": 1.5205309698033615e-06,
      "loss": 0.947,
      "step": 8174
    },
    {
      "epoch": 0.5898055625698929,
      "grad_norm": 7.921001920422683,
      "learning_rate": 1.520077252007465e-06,
      "loss": 1.0467,
      "step": 8175
    },
    {
      "epoch": 0.5898777100393203,
      "grad_norm": 3.3240794635643613,
      "learning_rate": 1.519623560418374e-06,
      "loss": 1.0407,
      "step": 8176
    },
    {
      "epoch": 0.5899498575087478,
      "grad_norm": 5.010697936518939,
      "learning_rate": 1.5191698950608616e-06,
      "loss": 0.9057,
      "step": 8177
    },
    {
      "epoch": 0.5900220049781754,
      "grad_norm": 4.0939608221150205,
      "learning_rate": 1.5187162559597014e-06,
      "loss": 0.9057,
      "step": 8178
    },
    {
      "epoch": 0.5900941524476029,
      "grad_norm": 2.8451432388798374,
      "learning_rate": 1.518262643139665e-06,
      "loss": 0.9344,
      "step": 8179
    },
    {
      "epoch": 0.5901662999170304,
      "grad_norm": 0.6474872888694517,
      "learning_rate": 1.5178090566255221e-06,
      "loss": 0.8193,
      "step": 8180
    },
    {
      "epoch": 0.5902384473864579,
      "grad_norm": 0.6627067863984347,
      "learning_rate": 1.5173554964420417e-06,
      "loss": 0.8062,
      "step": 8181
    },
    {
      "epoch": 0.5903105948558854,
      "grad_norm": 16.36573587685612,
      "learning_rate": 1.5169019626139906e-06,
      "loss": 0.9292,
      "step": 8182
    },
    {
      "epoch": 0.590382742325313,
      "grad_norm": 6.1051613406295155,
      "learning_rate": 1.5164484551661357e-06,
      "loss": 0.8556,
      "step": 8183
    },
    {
      "epoch": 0.5904548897947405,
      "grad_norm": 2.9521935018165677,
      "learning_rate": 1.5159949741232396e-06,
      "loss": 1.0047,
      "step": 8184
    },
    {
      "epoch": 0.590527037264168,
      "grad_norm": 3.872254414168428,
      "learning_rate": 1.5155415195100672e-06,
      "loss": 1.0038,
      "step": 8185
    },
    {
      "epoch": 0.5905991847335955,
      "grad_norm": 13.21375773785366,
      "learning_rate": 1.515088091351379e-06,
      "loss": 1.0187,
      "step": 8186
    },
    {
      "epoch": 0.5906713322030229,
      "grad_norm": 2.977332788956224,
      "learning_rate": 1.514634689671934e-06,
      "loss": 0.9199,
      "step": 8187
    },
    {
      "epoch": 0.5907434796724504,
      "grad_norm": 3.4752109600772463,
      "learning_rate": 1.514181314496493e-06,
      "loss": 1.0276,
      "step": 8188
    },
    {
      "epoch": 0.590815627141878,
      "grad_norm": 3.765956626630167,
      "learning_rate": 1.5137279658498113e-06,
      "loss": 0.9801,
      "step": 8189
    },
    {
      "epoch": 0.5908877746113055,
      "grad_norm": 9.66183557039528,
      "learning_rate": 1.5132746437566454e-06,
      "loss": 1.0001,
      "step": 8190
    },
    {
      "epoch": 0.590959922080733,
      "grad_norm": 5.079649523261896,
      "learning_rate": 1.5128213482417499e-06,
      "loss": 0.972,
      "step": 8191
    },
    {
      "epoch": 0.5910320695501605,
      "grad_norm": 4.125198359488541,
      "learning_rate": 1.5123680793298766e-06,
      "loss": 0.9696,
      "step": 8192
    },
    {
      "epoch": 0.591104217019588,
      "grad_norm": 3.433121406779691,
      "learning_rate": 1.5119148370457773e-06,
      "loss": 0.9634,
      "step": 8193
    },
    {
      "epoch": 0.5911763644890156,
      "grad_norm": 4.303572550142025,
      "learning_rate": 1.5114616214142017e-06,
      "loss": 0.9335,
      "step": 8194
    },
    {
      "epoch": 0.5912485119584431,
      "grad_norm": 4.114245411319727,
      "learning_rate": 1.5110084324598992e-06,
      "loss": 0.8223,
      "step": 8195
    },
    {
      "epoch": 0.5913206594278706,
      "grad_norm": 2.752589220615737,
      "learning_rate": 1.5105552702076153e-06,
      "loss": 1.0546,
      "step": 8196
    },
    {
      "epoch": 0.5913928068972981,
      "grad_norm": 8.698924066068548,
      "learning_rate": 1.5101021346820966e-06,
      "loss": 1.0061,
      "step": 8197
    },
    {
      "epoch": 0.5914649543667256,
      "grad_norm": 6.272951936279525,
      "learning_rate": 1.5096490259080873e-06,
      "loss": 1.0157,
      "step": 8198
    },
    {
      "epoch": 0.591537101836153,
      "grad_norm": 3.3040003140982788,
      "learning_rate": 1.5091959439103282e-06,
      "loss": 1.0583,
      "step": 8199
    },
    {
      "epoch": 0.5916092493055806,
      "grad_norm": 3.866225737610596,
      "learning_rate": 1.5087428887135628e-06,
      "loss": 1.0312,
      "step": 8200
    },
    {
      "epoch": 0.5916813967750081,
      "grad_norm": 3.9332094582485757,
      "learning_rate": 1.5082898603425294e-06,
      "loss": 1.0039,
      "step": 8201
    },
    {
      "epoch": 0.5917535442444356,
      "grad_norm": 9.527920701190979,
      "learning_rate": 1.507836858821966e-06,
      "loss": 0.9706,
      "step": 8202
    },
    {
      "epoch": 0.5918256917138631,
      "grad_norm": 4.334040266229293,
      "learning_rate": 1.5073838841766105e-06,
      "loss": 0.9571,
      "step": 8203
    },
    {
      "epoch": 0.5918978391832906,
      "grad_norm": 6.051698796130868,
      "learning_rate": 1.5069309364311969e-06,
      "loss": 0.9984,
      "step": 8204
    },
    {
      "epoch": 0.5919699866527182,
      "grad_norm": 3.554941424009291,
      "learning_rate": 1.5064780156104597e-06,
      "loss": 1.0603,
      "step": 8205
    },
    {
      "epoch": 0.5920421341221457,
      "grad_norm": 4.581915942576204,
      "learning_rate": 1.5060251217391311e-06,
      "loss": 0.9317,
      "step": 8206
    },
    {
      "epoch": 0.5921142815915732,
      "grad_norm": 5.233188232710485,
      "learning_rate": 1.5055722548419425e-06,
      "loss": 0.9309,
      "step": 8207
    },
    {
      "epoch": 0.5921864290610007,
      "grad_norm": 12.715942372851785,
      "learning_rate": 1.5051194149436221e-06,
      "loss": 0.9322,
      "step": 8208
    },
    {
      "epoch": 0.5922585765304282,
      "grad_norm": 4.322587971274753,
      "learning_rate": 1.5046666020688987e-06,
      "loss": 1.0311,
      "step": 8209
    },
    {
      "epoch": 0.5923307239998558,
      "grad_norm": 4.8926845516119295,
      "learning_rate": 1.5042138162424988e-06,
      "loss": 1.0074,
      "step": 8210
    },
    {
      "epoch": 0.5924028714692832,
      "grad_norm": 4.316519812561542,
      "learning_rate": 1.5037610574891465e-06,
      "loss": 0.9142,
      "step": 8211
    },
    {
      "epoch": 0.5924750189387107,
      "grad_norm": 3.852800831767401,
      "learning_rate": 1.5033083258335663e-06,
      "loss": 0.9705,
      "step": 8212
    },
    {
      "epoch": 0.5925471664081382,
      "grad_norm": 3.8443441784621193,
      "learning_rate": 1.50285562130048e-06,
      "loss": 1.0221,
      "step": 8213
    },
    {
      "epoch": 0.5926193138775657,
      "grad_norm": 2.8667121513957348,
      "learning_rate": 1.502402943914607e-06,
      "loss": 0.8232,
      "step": 8214
    },
    {
      "epoch": 0.5926914613469932,
      "grad_norm": 4.203891840188554,
      "learning_rate": 1.5019502937006677e-06,
      "loss": 0.9016,
      "step": 8215
    },
    {
      "epoch": 0.5927636088164208,
      "grad_norm": 5.469253080933965,
      "learning_rate": 1.5014976706833792e-06,
      "loss": 1.0247,
      "step": 8216
    },
    {
      "epoch": 0.5928357562858483,
      "grad_norm": 3.6897289199294137,
      "learning_rate": 1.501045074887457e-06,
      "loss": 1.0466,
      "step": 8217
    },
    {
      "epoch": 0.5929079037552758,
      "grad_norm": 4.502281246515981,
      "learning_rate": 1.500592506337617e-06,
      "loss": 0.9481,
      "step": 8218
    },
    {
      "epoch": 0.5929800512247033,
      "grad_norm": 5.369949807158022,
      "learning_rate": 1.500139965058571e-06,
      "loss": 1.0147,
      "step": 8219
    },
    {
      "epoch": 0.5930521986941308,
      "grad_norm": 3.6959351867838843,
      "learning_rate": 1.4996874510750307e-06,
      "loss": 0.9991,
      "step": 8220
    },
    {
      "epoch": 0.5931243461635584,
      "grad_norm": 3.4563978056664175,
      "learning_rate": 1.4992349644117073e-06,
      "loss": 0.9608,
      "step": 8221
    },
    {
      "epoch": 0.5931964936329859,
      "grad_norm": 4.458157021907418,
      "learning_rate": 1.4987825050933086e-06,
      "loss": 1.0136,
      "step": 8222
    },
    {
      "epoch": 0.5932686411024133,
      "grad_norm": 4.230160125154631,
      "learning_rate": 1.4983300731445416e-06,
      "loss": 0.9941,
      "step": 8223
    },
    {
      "epoch": 0.5933407885718408,
      "grad_norm": 3.055125173448619,
      "learning_rate": 1.4978776685901122e-06,
      "loss": 0.9242,
      "step": 8224
    },
    {
      "epoch": 0.5934129360412683,
      "grad_norm": 0.816457025874568,
      "learning_rate": 1.497425291454725e-06,
      "loss": 0.8866,
      "step": 8225
    },
    {
      "epoch": 0.5934850835106958,
      "grad_norm": 2.9768510012847633,
      "learning_rate": 1.4969729417630812e-06,
      "loss": 1.028,
      "step": 8226
    },
    {
      "epoch": 0.5935572309801234,
      "grad_norm": 0.7055681380412354,
      "learning_rate": 1.496520619539884e-06,
      "loss": 0.8366,
      "step": 8227
    },
    {
      "epoch": 0.5936293784495509,
      "grad_norm": 4.053498845443853,
      "learning_rate": 1.496068324809831e-06,
      "loss": 0.8913,
      "step": 8228
    },
    {
      "epoch": 0.5937015259189784,
      "grad_norm": 4.4382695350163335,
      "learning_rate": 1.4956160575976218e-06,
      "loss": 0.9893,
      "step": 8229
    },
    {
      "epoch": 0.5937736733884059,
      "grad_norm": 4.828687122793729,
      "learning_rate": 1.4951638179279528e-06,
      "loss": 0.9955,
      "step": 8230
    },
    {
      "epoch": 0.5938458208578334,
      "grad_norm": 3.526690484911766,
      "learning_rate": 1.4947116058255184e-06,
      "loss": 0.9383,
      "step": 8231
    },
    {
      "epoch": 0.593917968327261,
      "grad_norm": 4.190915209615849,
      "learning_rate": 1.4942594213150127e-06,
      "loss": 0.8982,
      "step": 8232
    },
    {
      "epoch": 0.5939901157966885,
      "grad_norm": 6.785381693325815,
      "learning_rate": 1.493807264421128e-06,
      "loss": 0.9871,
      "step": 8233
    },
    {
      "epoch": 0.5940622632661159,
      "grad_norm": 3.7713513659546867,
      "learning_rate": 1.4933551351685549e-06,
      "loss": 0.9987,
      "step": 8234
    },
    {
      "epoch": 0.5941344107355434,
      "grad_norm": 3.3933670980076442,
      "learning_rate": 1.4929030335819822e-06,
      "loss": 0.952,
      "step": 8235
    },
    {
      "epoch": 0.5942065582049709,
      "grad_norm": 4.523670671478434,
      "learning_rate": 1.4924509596860973e-06,
      "loss": 0.9659,
      "step": 8236
    },
    {
      "epoch": 0.5942787056743984,
      "grad_norm": 0.6885699702537439,
      "learning_rate": 1.4919989135055872e-06,
      "loss": 0.8068,
      "step": 8237
    },
    {
      "epoch": 0.594350853143826,
      "grad_norm": 5.173799001310066,
      "learning_rate": 1.491546895065135e-06,
      "loss": 1.0204,
      "step": 8238
    },
    {
      "epoch": 0.5944230006132535,
      "grad_norm": 5.020330485677691,
      "learning_rate": 1.4910949043894255e-06,
      "loss": 0.8891,
      "step": 8239
    },
    {
      "epoch": 0.594495148082681,
      "grad_norm": 4.17062530659349,
      "learning_rate": 1.4906429415031393e-06,
      "loss": 1.0506,
      "step": 8240
    },
    {
      "epoch": 0.5945672955521085,
      "grad_norm": 5.203188967383162,
      "learning_rate": 1.4901910064309555e-06,
      "loss": 0.8639,
      "step": 8241
    },
    {
      "epoch": 0.594639443021536,
      "grad_norm": 21.318620535648378,
      "learning_rate": 1.4897390991975547e-06,
      "loss": 0.9397,
      "step": 8242
    },
    {
      "epoch": 0.5947115904909636,
      "grad_norm": 2.9900953184518606,
      "learning_rate": 1.4892872198276124e-06,
      "loss": 0.9589,
      "step": 8243
    },
    {
      "epoch": 0.5947837379603911,
      "grad_norm": 4.862981800833647,
      "learning_rate": 1.4888353683458042e-06,
      "loss": 1.0466,
      "step": 8244
    },
    {
      "epoch": 0.5948558854298186,
      "grad_norm": 3.037380508144695,
      "learning_rate": 1.4883835447768046e-06,
      "loss": 0.9344,
      "step": 8245
    },
    {
      "epoch": 0.594928032899246,
      "grad_norm": 3.903764468021803,
      "learning_rate": 1.4879317491452857e-06,
      "loss": 0.9488,
      "step": 8246
    },
    {
      "epoch": 0.5950001803686735,
      "grad_norm": 3.9731411408232584,
      "learning_rate": 1.487479981475918e-06,
      "loss": 0.995,
      "step": 8247
    },
    {
      "epoch": 0.595072327838101,
      "grad_norm": 3.498683273044384,
      "learning_rate": 1.4870282417933715e-06,
      "loss": 0.8065,
      "step": 8248
    },
    {
      "epoch": 0.5951444753075286,
      "grad_norm": 0.8574140196420842,
      "learning_rate": 1.486576530122314e-06,
      "loss": 0.8781,
      "step": 8249
    },
    {
      "epoch": 0.5952166227769561,
      "grad_norm": 3.61450953046371,
      "learning_rate": 1.4861248464874105e-06,
      "loss": 1.0072,
      "step": 8250
    },
    {
      "epoch": 0.5952887702463836,
      "grad_norm": 2.659684149124066,
      "learning_rate": 1.485673190913328e-06,
      "loss": 0.8784,
      "step": 8251
    },
    {
      "epoch": 0.5953609177158111,
      "grad_norm": 3.3029090659189353,
      "learning_rate": 1.4852215634247285e-06,
      "loss": 1.0223,
      "step": 8252
    },
    {
      "epoch": 0.5954330651852386,
      "grad_norm": 3.4807495962903157,
      "learning_rate": 1.484769964046273e-06,
      "loss": 0.8667,
      "step": 8253
    },
    {
      "epoch": 0.5955052126546662,
      "grad_norm": 3.484598195407192,
      "learning_rate": 1.4843183928026234e-06,
      "loss": 1.071,
      "step": 8254
    },
    {
      "epoch": 0.5955773601240937,
      "grad_norm": 3.9432533253678876,
      "learning_rate": 1.4838668497184364e-06,
      "loss": 0.934,
      "step": 8255
    },
    {
      "epoch": 0.5956495075935212,
      "grad_norm": 3.0253086657717025,
      "learning_rate": 1.4834153348183706e-06,
      "loss": 0.8445,
      "step": 8256
    },
    {
      "epoch": 0.5957216550629487,
      "grad_norm": 2.665321835521549,
      "learning_rate": 1.4829638481270812e-06,
      "loss": 0.919,
      "step": 8257
    },
    {
      "epoch": 0.5957938025323761,
      "grad_norm": 5.21036298750364,
      "learning_rate": 1.4825123896692218e-06,
      "loss": 0.8603,
      "step": 8258
    },
    {
      "epoch": 0.5958659500018036,
      "grad_norm": 5.042285450355445,
      "learning_rate": 1.482060959469445e-06,
      "loss": 1.0202,
      "step": 8259
    },
    {
      "epoch": 0.5959380974712312,
      "grad_norm": 3.3995449322607816,
      "learning_rate": 1.4816095575524018e-06,
      "loss": 1.0214,
      "step": 8260
    },
    {
      "epoch": 0.5960102449406587,
      "grad_norm": 3.557904676098574,
      "learning_rate": 1.4811581839427425e-06,
      "loss": 0.8794,
      "step": 8261
    },
    {
      "epoch": 0.5960823924100862,
      "grad_norm": 5.492222228286362,
      "learning_rate": 1.4807068386651135e-06,
      "loss": 0.9297,
      "step": 8262
    },
    {
      "epoch": 0.5961545398795137,
      "grad_norm": 3.5083875970577942,
      "learning_rate": 1.4802555217441612e-06,
      "loss": 1.0177,
      "step": 8263
    },
    {
      "epoch": 0.5962266873489412,
      "grad_norm": 4.505902340158722,
      "learning_rate": 1.4798042332045318e-06,
      "loss": 0.9679,
      "step": 8264
    },
    {
      "epoch": 0.5962988348183688,
      "grad_norm": 4.141100355968583,
      "learning_rate": 1.4793529730708663e-06,
      "loss": 0.9712,
      "step": 8265
    },
    {
      "epoch": 0.5963709822877963,
      "grad_norm": 3.4254591473966522,
      "learning_rate": 1.4789017413678084e-06,
      "loss": 1.0422,
      "step": 8266
    },
    {
      "epoch": 0.5964431297572238,
      "grad_norm": 3.646226364022518,
      "learning_rate": 1.478450538119998e-06,
      "loss": 0.9312,
      "step": 8267
    },
    {
      "epoch": 0.5965152772266513,
      "grad_norm": 3.544733283120454,
      "learning_rate": 1.4779993633520716e-06,
      "loss": 0.8326,
      "step": 8268
    },
    {
      "epoch": 0.5965874246960788,
      "grad_norm": 3.0562304724321034,
      "learning_rate": 1.4775482170886684e-06,
      "loss": 0.9733,
      "step": 8269
    },
    {
      "epoch": 0.5966595721655062,
      "grad_norm": 6.547485323114247,
      "learning_rate": 1.4770970993544227e-06,
      "loss": 0.9969,
      "step": 8270
    },
    {
      "epoch": 0.5967317196349338,
      "grad_norm": 3.493288963074915,
      "learning_rate": 1.476646010173969e-06,
      "loss": 1.0104,
      "step": 8271
    },
    {
      "epoch": 0.5968038671043613,
      "grad_norm": 7.746102860638238,
      "learning_rate": 1.4761949495719388e-06,
      "loss": 1.1031,
      "step": 8272
    },
    {
      "epoch": 0.5968760145737888,
      "grad_norm": 3.478861051069227,
      "learning_rate": 1.4757439175729639e-06,
      "loss": 0.9749,
      "step": 8273
    },
    {
      "epoch": 0.5969481620432163,
      "grad_norm": 15.640878650540625,
      "learning_rate": 1.4752929142016726e-06,
      "loss": 1.0273,
      "step": 8274
    },
    {
      "epoch": 0.5970203095126438,
      "grad_norm": 3.5237260180168417,
      "learning_rate": 1.474841939482693e-06,
      "loss": 0.9001,
      "step": 8275
    },
    {
      "epoch": 0.5970924569820714,
      "grad_norm": 3.0059402782119204,
      "learning_rate": 1.4743909934406512e-06,
      "loss": 0.8935,
      "step": 8276
    },
    {
      "epoch": 0.5971646044514989,
      "grad_norm": 3.8576305323312923,
      "learning_rate": 1.4739400761001708e-06,
      "loss": 0.947,
      "step": 8277
    },
    {
      "epoch": 0.5972367519209264,
      "grad_norm": 5.3337961035864305,
      "learning_rate": 1.4734891874858761e-06,
      "loss": 1.0598,
      "step": 8278
    },
    {
      "epoch": 0.5973088993903539,
      "grad_norm": 2.747576946538247,
      "learning_rate": 1.4730383276223879e-06,
      "loss": 0.9966,
      "step": 8279
    },
    {
      "epoch": 0.5973810468597814,
      "grad_norm": 3.792487855513358,
      "learning_rate": 1.472587496534325e-06,
      "loss": 1.0508,
      "step": 8280
    },
    {
      "epoch": 0.5974531943292088,
      "grad_norm": 7.18350135034376,
      "learning_rate": 1.4721366942463074e-06,
      "loss": 0.9831,
      "step": 8281
    },
    {
      "epoch": 0.5975253417986364,
      "grad_norm": 6.46915806759393,
      "learning_rate": 1.4716859207829507e-06,
      "loss": 1.0316,
      "step": 8282
    },
    {
      "epoch": 0.5975974892680639,
      "grad_norm": 4.554385140853864,
      "learning_rate": 1.4712351761688697e-06,
      "loss": 0.9892,
      "step": 8283
    },
    {
      "epoch": 0.5976696367374914,
      "grad_norm": 14.535611324675383,
      "learning_rate": 1.4707844604286788e-06,
      "loss": 1.0095,
      "step": 8284
    },
    {
      "epoch": 0.5977417842069189,
      "grad_norm": 3.530170241835418,
      "learning_rate": 1.4703337735869891e-06,
      "loss": 0.9547,
      "step": 8285
    },
    {
      "epoch": 0.5978139316763464,
      "grad_norm": 7.481175513283115,
      "learning_rate": 1.4698831156684112e-06,
      "loss": 0.8812,
      "step": 8286
    },
    {
      "epoch": 0.597886079145774,
      "grad_norm": 4.234149997265192,
      "learning_rate": 1.469432486697554e-06,
      "loss": 0.8363,
      "step": 8287
    },
    {
      "epoch": 0.5979582266152015,
      "grad_norm": 4.488549180495579,
      "learning_rate": 1.4689818866990251e-06,
      "loss": 0.9614,
      "step": 8288
    },
    {
      "epoch": 0.598030374084629,
      "grad_norm": 4.729929029323434,
      "learning_rate": 1.4685313156974296e-06,
      "loss": 1.0496,
      "step": 8289
    },
    {
      "epoch": 0.5981025215540565,
      "grad_norm": 2.32906941965504,
      "learning_rate": 1.4680807737173704e-06,
      "loss": 0.9547,
      "step": 8290
    },
    {
      "epoch": 0.598174669023484,
      "grad_norm": 2.555768732034553,
      "learning_rate": 1.4676302607834521e-06,
      "loss": 0.8888,
      "step": 8291
    },
    {
      "epoch": 0.5982468164929116,
      "grad_norm": 3.0840200913760647,
      "learning_rate": 1.4671797769202736e-06,
      "loss": 0.9572,
      "step": 8292
    },
    {
      "epoch": 0.598318963962339,
      "grad_norm": 2.7643487431665412,
      "learning_rate": 1.466729322152436e-06,
      "loss": 0.9194,
      "step": 8293
    },
    {
      "epoch": 0.5983911114317665,
      "grad_norm": 6.948025803135388,
      "learning_rate": 1.4662788965045353e-06,
      "loss": 0.9536,
      "step": 8294
    },
    {
      "epoch": 0.598463258901194,
      "grad_norm": 3.3479745862627497,
      "learning_rate": 1.4658285000011685e-06,
      "loss": 0.9775,
      "step": 8295
    },
    {
      "epoch": 0.5985354063706215,
      "grad_norm": 4.688176220755302,
      "learning_rate": 1.4653781326669305e-06,
      "loss": 0.934,
      "step": 8296
    },
    {
      "epoch": 0.598607553840049,
      "grad_norm": 3.7635116028326205,
      "learning_rate": 1.464927794526413e-06,
      "loss": 1.0173,
      "step": 8297
    },
    {
      "epoch": 0.5986797013094766,
      "grad_norm": 5.838949751282164,
      "learning_rate": 1.464477485604208e-06,
      "loss": 0.9988,
      "step": 8298
    },
    {
      "epoch": 0.5987518487789041,
      "grad_norm": 2.892004889973454,
      "learning_rate": 1.464027205924905e-06,
      "loss": 1.0016,
      "step": 8299
    },
    {
      "epoch": 0.5988239962483316,
      "grad_norm": 0.7694445720983125,
      "learning_rate": 1.4635769555130923e-06,
      "loss": 0.8504,
      "step": 8300
    },
    {
      "epoch": 0.5988961437177591,
      "grad_norm": 5.750392402859585,
      "learning_rate": 1.4631267343933564e-06,
      "loss": 0.8749,
      "step": 8301
    },
    {
      "epoch": 0.5989682911871866,
      "grad_norm": 5.492865530082398,
      "learning_rate": 1.4626765425902814e-06,
      "loss": 0.868,
      "step": 8302
    },
    {
      "epoch": 0.5990404386566142,
      "grad_norm": 6.6798862360786595,
      "learning_rate": 1.462226380128452e-06,
      "loss": 0.9828,
      "step": 8303
    },
    {
      "epoch": 0.5991125861260417,
      "grad_norm": 3.3509710185665513,
      "learning_rate": 1.4617762470324486e-06,
      "loss": 1.012,
      "step": 8304
    },
    {
      "epoch": 0.5991847335954691,
      "grad_norm": 3.446439320895516,
      "learning_rate": 1.4613261433268526e-06,
      "loss": 1.0168,
      "step": 8305
    },
    {
      "epoch": 0.5992568810648966,
      "grad_norm": 4.123050575851016,
      "learning_rate": 1.4608760690362418e-06,
      "loss": 0.9427,
      "step": 8306
    },
    {
      "epoch": 0.5993290285343241,
      "grad_norm": 4.39217514964579,
      "learning_rate": 1.4604260241851925e-06,
      "loss": 0.9373,
      "step": 8307
    },
    {
      "epoch": 0.5994011760037516,
      "grad_norm": 3.597111995256875,
      "learning_rate": 1.459976008798281e-06,
      "loss": 0.8923,
      "step": 8308
    },
    {
      "epoch": 0.5994733234731792,
      "grad_norm": 8.010516883279424,
      "learning_rate": 1.4595260229000804e-06,
      "loss": 0.9072,
      "step": 8309
    },
    {
      "epoch": 0.5995454709426067,
      "grad_norm": 4.4002061882258845,
      "learning_rate": 1.4590760665151628e-06,
      "loss": 0.9114,
      "step": 8310
    },
    {
      "epoch": 0.5996176184120342,
      "grad_norm": 5.503640617203758,
      "learning_rate": 1.4586261396680995e-06,
      "loss": 0.9885,
      "step": 8311
    },
    {
      "epoch": 0.5996897658814617,
      "grad_norm": 3.531073101642829,
      "learning_rate": 1.4581762423834578e-06,
      "loss": 0.969,
      "step": 8312
    },
    {
      "epoch": 0.5997619133508892,
      "grad_norm": 3.663055867270272,
      "learning_rate": 1.4577263746858064e-06,
      "loss": 0.9397,
      "step": 8313
    },
    {
      "epoch": 0.5998340608203168,
      "grad_norm": 0.875849379680745,
      "learning_rate": 1.4572765365997093e-06,
      "loss": 0.8928,
      "step": 8314
    },
    {
      "epoch": 0.5999062082897443,
      "grad_norm": 3.867223011200592,
      "learning_rate": 1.4568267281497325e-06,
      "loss": 0.9196,
      "step": 8315
    },
    {
      "epoch": 0.5999783557591718,
      "grad_norm": 0.6852721617551487,
      "learning_rate": 1.456376949360436e-06,
      "loss": 0.8719,
      "step": 8316
    },
    {
      "epoch": 0.6000505032285992,
      "grad_norm": 4.6292842919862105,
      "learning_rate": 1.4559272002563832e-06,
      "loss": 0.979,
      "step": 8317
    },
    {
      "epoch": 0.6001226506980267,
      "grad_norm": 3.2046909877213166,
      "learning_rate": 1.4554774808621317e-06,
      "loss": 0.8261,
      "step": 8318
    },
    {
      "epoch": 0.6001947981674542,
      "grad_norm": 0.7565701713314896,
      "learning_rate": 1.4550277912022384e-06,
      "loss": 0.8145,
      "step": 8319
    },
    {
      "epoch": 0.6002669456368818,
      "grad_norm": 11.200740885071475,
      "learning_rate": 1.4545781313012608e-06,
      "loss": 0.9618,
      "step": 8320
    },
    {
      "epoch": 0.6003390931063093,
      "grad_norm": 5.381396368962503,
      "learning_rate": 1.454128501183752e-06,
      "loss": 0.9007,
      "step": 8321
    },
    {
      "epoch": 0.6004112405757368,
      "grad_norm": 3.8040585477797952,
      "learning_rate": 1.453678900874265e-06,
      "loss": 0.9293,
      "step": 8322
    },
    {
      "epoch": 0.6004833880451643,
      "grad_norm": 5.88702719828934,
      "learning_rate": 1.4532293303973512e-06,
      "loss": 0.9002,
      "step": 8323
    },
    {
      "epoch": 0.6005555355145918,
      "grad_norm": 2.603739517784171,
      "learning_rate": 1.4527797897775592e-06,
      "loss": 0.9298,
      "step": 8324
    },
    {
      "epoch": 0.6006276829840194,
      "grad_norm": 5.013826326993772,
      "learning_rate": 1.4523302790394375e-06,
      "loss": 0.9977,
      "step": 8325
    },
    {
      "epoch": 0.6006998304534469,
      "grad_norm": 4.199430408909921,
      "learning_rate": 1.4518807982075312e-06,
      "loss": 1.0064,
      "step": 8326
    },
    {
      "epoch": 0.6007719779228744,
      "grad_norm": 2.6455547656697562,
      "learning_rate": 1.451431347306386e-06,
      "loss": 0.9657,
      "step": 8327
    },
    {
      "epoch": 0.6008441253923019,
      "grad_norm": 2.979675107309048,
      "learning_rate": 1.4509819263605445e-06,
      "loss": 0.9623,
      "step": 8328
    },
    {
      "epoch": 0.6009162728617293,
      "grad_norm": 4.214352046084304,
      "learning_rate": 1.4505325353945468e-06,
      "loss": 0.8692,
      "step": 8329
    },
    {
      "epoch": 0.6009884203311568,
      "grad_norm": 2.78526256727522,
      "learning_rate": 1.4500831744329338e-06,
      "loss": 0.972,
      "step": 8330
    },
    {
      "epoch": 0.6010605678005844,
      "grad_norm": 4.020470215735373,
      "learning_rate": 1.449633843500242e-06,
      "loss": 1.1064,
      "step": 8331
    },
    {
      "epoch": 0.6011327152700119,
      "grad_norm": 3.316233420389877,
      "learning_rate": 1.44918454262101e-06,
      "loss": 0.9038,
      "step": 8332
    },
    {
      "epoch": 0.6012048627394394,
      "grad_norm": 5.402787174498889,
      "learning_rate": 1.4487352718197707e-06,
      "loss": 0.8502,
      "step": 8333
    },
    {
      "epoch": 0.6012770102088669,
      "grad_norm": 3.5968585294848516,
      "learning_rate": 1.4482860311210567e-06,
      "loss": 0.9742,
      "step": 8334
    },
    {
      "epoch": 0.6013491576782944,
      "grad_norm": 3.0002698776925483,
      "learning_rate": 1.447836820549401e-06,
      "loss": 1.0182,
      "step": 8335
    },
    {
      "epoch": 0.601421305147722,
      "grad_norm": 5.593641738270173,
      "learning_rate": 1.4473876401293321e-06,
      "loss": 0.9255,
      "step": 8336
    },
    {
      "epoch": 0.6014934526171495,
      "grad_norm": 3.5669057272775175,
      "learning_rate": 1.4469384898853784e-06,
      "loss": 0.962,
      "step": 8337
    },
    {
      "epoch": 0.601565600086577,
      "grad_norm": 4.106557830466511,
      "learning_rate": 1.4464893698420663e-06,
      "loss": 0.9223,
      "step": 8338
    },
    {
      "epoch": 0.6016377475560045,
      "grad_norm": 4.563609994137193,
      "learning_rate": 1.4460402800239215e-06,
      "loss": 0.9613,
      "step": 8339
    },
    {
      "epoch": 0.6017098950254319,
      "grad_norm": 8.99749466675698,
      "learning_rate": 1.445591220455466e-06,
      "loss": 1.0437,
      "step": 8340
    },
    {
      "epoch": 0.6017820424948594,
      "grad_norm": 4.609174303761051,
      "learning_rate": 1.4451421911612209e-06,
      "loss": 0.8872,
      "step": 8341
    },
    {
      "epoch": 0.601854189964287,
      "grad_norm": 3.584334669742662,
      "learning_rate": 1.4446931921657078e-06,
      "loss": 1.0042,
      "step": 8342
    },
    {
      "epoch": 0.6019263374337145,
      "grad_norm": 3.1789977429175846,
      "learning_rate": 1.4442442234934425e-06,
      "loss": 1.0094,
      "step": 8343
    },
    {
      "epoch": 0.601998484903142,
      "grad_norm": 2.8087069895786754,
      "learning_rate": 1.4437952851689437e-06,
      "loss": 0.9873,
      "step": 8344
    },
    {
      "epoch": 0.6020706323725695,
      "grad_norm": 3.452418052148279,
      "learning_rate": 1.4433463772167257e-06,
      "loss": 1.0076,
      "step": 8345
    },
    {
      "epoch": 0.602142779841997,
      "grad_norm": 5.669188144158491,
      "learning_rate": 1.4428974996613001e-06,
      "loss": 0.8676,
      "step": 8346
    },
    {
      "epoch": 0.6022149273114246,
      "grad_norm": 4.50657173944601,
      "learning_rate": 1.4424486525271808e-06,
      "loss": 0.9701,
      "step": 8347
    },
    {
      "epoch": 0.6022870747808521,
      "grad_norm": 0.7087895990171528,
      "learning_rate": 1.4419998358388763e-06,
      "loss": 0.8439,
      "step": 8348
    },
    {
      "epoch": 0.6023592222502796,
      "grad_norm": 4.744609384004874,
      "learning_rate": 1.4415510496208951e-06,
      "loss": 0.9885,
      "step": 8349
    },
    {
      "epoch": 0.6024313697197071,
      "grad_norm": 4.11219209486537,
      "learning_rate": 1.4411022938977442e-06,
      "loss": 0.8911,
      "step": 8350
    },
    {
      "epoch": 0.6025035171891346,
      "grad_norm": 6.0817207742285895,
      "learning_rate": 1.4406535686939277e-06,
      "loss": 1.0077,
      "step": 8351
    },
    {
      "epoch": 0.602575664658562,
      "grad_norm": 0.7837645309415358,
      "learning_rate": 1.4402048740339492e-06,
      "loss": 0.8362,
      "step": 8352
    },
    {
      "epoch": 0.6026478121279896,
      "grad_norm": 4.164685630337076,
      "learning_rate": 1.4397562099423098e-06,
      "loss": 0.9861,
      "step": 8353
    },
    {
      "epoch": 0.6027199595974171,
      "grad_norm": 4.939352919171887,
      "learning_rate": 1.4393075764435106e-06,
      "loss": 0.9885,
      "step": 8354
    },
    {
      "epoch": 0.6027921070668446,
      "grad_norm": 2.8899256504619206,
      "learning_rate": 1.438858973562049e-06,
      "loss": 0.8891,
      "step": 8355
    },
    {
      "epoch": 0.6028642545362721,
      "grad_norm": 3.078677829728251,
      "learning_rate": 1.4384104013224203e-06,
      "loss": 0.9811,
      "step": 8356
    },
    {
      "epoch": 0.6029364020056996,
      "grad_norm": 3.437886441790291,
      "learning_rate": 1.4379618597491221e-06,
      "loss": 0.8576,
      "step": 8357
    },
    {
      "epoch": 0.6030085494751272,
      "grad_norm": 3.3800997880730597,
      "learning_rate": 1.4375133488666449e-06,
      "loss": 0.9654,
      "step": 8358
    },
    {
      "epoch": 0.6030806969445547,
      "grad_norm": 4.970108902043188,
      "learning_rate": 1.4370648686994826e-06,
      "loss": 0.8884,
      "step": 8359
    },
    {
      "epoch": 0.6031528444139822,
      "grad_norm": 3.219420872374254,
      "learning_rate": 1.4366164192721236e-06,
      "loss": 0.8975,
      "step": 8360
    },
    {
      "epoch": 0.6032249918834097,
      "grad_norm": 8.280605535801762,
      "learning_rate": 1.436168000609056e-06,
      "loss": 0.9009,
      "step": 8361
    },
    {
      "epoch": 0.6032971393528372,
      "grad_norm": 4.640118195573954,
      "learning_rate": 1.435719612734767e-06,
      "loss": 0.9448,
      "step": 8362
    },
    {
      "epoch": 0.6033692868222648,
      "grad_norm": 2.7660956548009668,
      "learning_rate": 1.4352712556737409e-06,
      "loss": 0.9539,
      "step": 8363
    },
    {
      "epoch": 0.6034414342916922,
      "grad_norm": 2.42502757971113,
      "learning_rate": 1.4348229294504615e-06,
      "loss": 0.9679,
      "step": 8364
    },
    {
      "epoch": 0.6035135817611197,
      "grad_norm": 2.912441426617291,
      "learning_rate": 1.4343746340894086e-06,
      "loss": 0.9565,
      "step": 8365
    },
    {
      "epoch": 0.6035857292305472,
      "grad_norm": 8.842288054821195,
      "learning_rate": 1.433926369615064e-06,
      "loss": 0.9882,
      "step": 8366
    },
    {
      "epoch": 0.6036578766999747,
      "grad_norm": 4.2735990671583135,
      "learning_rate": 1.433478136051905e-06,
      "loss": 1.0139,
      "step": 8367
    },
    {
      "epoch": 0.6037300241694022,
      "grad_norm": 3.5684509098567028,
      "learning_rate": 1.4330299334244068e-06,
      "loss": 0.8441,
      "step": 8368
    },
    {
      "epoch": 0.6038021716388298,
      "grad_norm": 5.092925584226587,
      "learning_rate": 1.432581761757046e-06,
      "loss": 0.9216,
      "step": 8369
    },
    {
      "epoch": 0.6038743191082573,
      "grad_norm": 2.3489218044330604,
      "learning_rate": 1.4321336210742936e-06,
      "loss": 0.9655,
      "step": 8370
    },
    {
      "epoch": 0.6039464665776848,
      "grad_norm": 4.3557590815509615,
      "learning_rate": 1.431685511400623e-06,
      "loss": 0.8348,
      "step": 8371
    },
    {
      "epoch": 0.6040186140471123,
      "grad_norm": 5.058080180252512,
      "learning_rate": 1.4312374327605028e-06,
      "loss": 1.0522,
      "step": 8372
    },
    {
      "epoch": 0.6040907615165398,
      "grad_norm": 3.997138788199187,
      "learning_rate": 1.4307893851784e-06,
      "loss": 1.1007,
      "step": 8373
    },
    {
      "epoch": 0.6041629089859674,
      "grad_norm": 2.8241984892945458,
      "learning_rate": 1.4303413686787826e-06,
      "loss": 0.9737,
      "step": 8374
    },
    {
      "epoch": 0.6042350564553949,
      "grad_norm": 4.020624158619894,
      "learning_rate": 1.429893383286114e-06,
      "loss": 0.8305,
      "step": 8375
    },
    {
      "epoch": 0.6043072039248223,
      "grad_norm": 4.41615491577371,
      "learning_rate": 1.4294454290248576e-06,
      "loss": 1.0156,
      "step": 8376
    },
    {
      "epoch": 0.6043793513942498,
      "grad_norm": 4.321107318741691,
      "learning_rate": 1.4289975059194737e-06,
      "loss": 1.01,
      "step": 8377
    },
    {
      "epoch": 0.6044514988636773,
      "grad_norm": 3.0200316494426436,
      "learning_rate": 1.4285496139944222e-06,
      "loss": 0.9922,
      "step": 8378
    },
    {
      "epoch": 0.6045236463331048,
      "grad_norm": 0.6959087301473281,
      "learning_rate": 1.4281017532741613e-06,
      "loss": 0.7787,
      "step": 8379
    },
    {
      "epoch": 0.6045957938025324,
      "grad_norm": 3.090807970453241,
      "learning_rate": 1.4276539237831458e-06,
      "loss": 1.034,
      "step": 8380
    },
    {
      "epoch": 0.6046679412719599,
      "grad_norm": 3.568261423421281,
      "learning_rate": 1.4272061255458314e-06,
      "loss": 1.0243,
      "step": 8381
    },
    {
      "epoch": 0.6047400887413874,
      "grad_norm": 5.189737779464825,
      "learning_rate": 1.4267583585866696e-06,
      "loss": 0.9937,
      "step": 8382
    },
    {
      "epoch": 0.6048122362108149,
      "grad_norm": 4.700856621778166,
      "learning_rate": 1.4263106229301123e-06,
      "loss": 0.9316,
      "step": 8383
    },
    {
      "epoch": 0.6048843836802424,
      "grad_norm": 3.6285218855497585,
      "learning_rate": 1.4258629186006081e-06,
      "loss": 1.0016,
      "step": 8384
    },
    {
      "epoch": 0.60495653114967,
      "grad_norm": 3.3484382917765485,
      "learning_rate": 1.425415245622604e-06,
      "loss": 0.891,
      "step": 8385
    },
    {
      "epoch": 0.6050286786190975,
      "grad_norm": 3.9091772479226714,
      "learning_rate": 1.424967604020547e-06,
      "loss": 0.9759,
      "step": 8386
    },
    {
      "epoch": 0.6051008260885249,
      "grad_norm": 2.8987581224808885,
      "learning_rate": 1.42451999381888e-06,
      "loss": 0.9837,
      "step": 8387
    },
    {
      "epoch": 0.6051729735579524,
      "grad_norm": 3.2968797728879107,
      "learning_rate": 1.4240724150420461e-06,
      "loss": 0.9293,
      "step": 8388
    },
    {
      "epoch": 0.6052451210273799,
      "grad_norm": 4.965476055084072,
      "learning_rate": 1.4236248677144853e-06,
      "loss": 0.9594,
      "step": 8389
    },
    {
      "epoch": 0.6053172684968074,
      "grad_norm": 3.6527092898627997,
      "learning_rate": 1.423177351860637e-06,
      "loss": 0.841,
      "step": 8390
    },
    {
      "epoch": 0.605389415966235,
      "grad_norm": 2.398480144705626,
      "learning_rate": 1.4227298675049381e-06,
      "loss": 1.0228,
      "step": 8391
    },
    {
      "epoch": 0.6054615634356625,
      "grad_norm": 4.340479435961975,
      "learning_rate": 1.4222824146718236e-06,
      "loss": 0.8547,
      "step": 8392
    },
    {
      "epoch": 0.60553371090509,
      "grad_norm": 2.6418817001413033,
      "learning_rate": 1.4218349933857287e-06,
      "loss": 0.9623,
      "step": 8393
    },
    {
      "epoch": 0.6056058583745175,
      "grad_norm": 11.853191065183012,
      "learning_rate": 1.4213876036710846e-06,
      "loss": 0.9988,
      "step": 8394
    },
    {
      "epoch": 0.605678005843945,
      "grad_norm": 21.57575806736905,
      "learning_rate": 1.4209402455523205e-06,
      "loss": 0.9897,
      "step": 8395
    },
    {
      "epoch": 0.6057501533133726,
      "grad_norm": 3.2776297807058863,
      "learning_rate": 1.4204929190538669e-06,
      "loss": 0.9988,
      "step": 8396
    },
    {
      "epoch": 0.6058223007828001,
      "grad_norm": 5.827884127059966,
      "learning_rate": 1.4200456242001489e-06,
      "loss": 0.9457,
      "step": 8397
    },
    {
      "epoch": 0.6058944482522276,
      "grad_norm": 3.9047417742623436,
      "learning_rate": 1.419598361015593e-06,
      "loss": 0.9218,
      "step": 8398
    },
    {
      "epoch": 0.605966595721655,
      "grad_norm": 3.2345194761285763,
      "learning_rate": 1.4191511295246223e-06,
      "loss": 0.9328,
      "step": 8399
    },
    {
      "epoch": 0.6060387431910825,
      "grad_norm": 3.719708872037219,
      "learning_rate": 1.418703929751658e-06,
      "loss": 1.0005,
      "step": 8400
    },
    {
      "epoch": 0.60611089066051,
      "grad_norm": 3.4951551147006152,
      "learning_rate": 1.4182567617211198e-06,
      "loss": 0.8811,
      "step": 8401
    },
    {
      "epoch": 0.6061830381299376,
      "grad_norm": 5.069282979730848,
      "learning_rate": 1.4178096254574263e-06,
      "loss": 0.9305,
      "step": 8402
    },
    {
      "epoch": 0.6062551855993651,
      "grad_norm": 4.754775608034902,
      "learning_rate": 1.4173625209849949e-06,
      "loss": 0.8889,
      "step": 8403
    },
    {
      "epoch": 0.6063273330687926,
      "grad_norm": 3.145294968640986,
      "learning_rate": 1.416915448328238e-06,
      "loss": 0.9409,
      "step": 8404
    },
    {
      "epoch": 0.6063994805382201,
      "grad_norm": 3.100527343196348,
      "learning_rate": 1.416468407511571e-06,
      "loss": 1.002,
      "step": 8405
    },
    {
      "epoch": 0.6064716280076476,
      "grad_norm": 3.6618070187682195,
      "learning_rate": 1.416021398559404e-06,
      "loss": 0.8913,
      "step": 8406
    },
    {
      "epoch": 0.6065437754770752,
      "grad_norm": 4.344379845930549,
      "learning_rate": 1.4155744214961459e-06,
      "loss": 0.9619,
      "step": 8407
    },
    {
      "epoch": 0.6066159229465027,
      "grad_norm": 6.145288613527821,
      "learning_rate": 1.415127476346206e-06,
      "loss": 0.878,
      "step": 8408
    },
    {
      "epoch": 0.6066880704159302,
      "grad_norm": 3.4085261588047624,
      "learning_rate": 1.4146805631339886e-06,
      "loss": 0.9724,
      "step": 8409
    },
    {
      "epoch": 0.6067602178853577,
      "grad_norm": 3.7919383667793984,
      "learning_rate": 1.4142336818838999e-06,
      "loss": 0.9567,
      "step": 8410
    },
    {
      "epoch": 0.6068323653547851,
      "grad_norm": 0.723055136626706,
      "learning_rate": 1.413786832620341e-06,
      "loss": 0.8162,
      "step": 8411
    },
    {
      "epoch": 0.6069045128242126,
      "grad_norm": 5.031776353302412,
      "learning_rate": 1.4133400153677132e-06,
      "loss": 0.8881,
      "step": 8412
    },
    {
      "epoch": 0.6069766602936402,
      "grad_norm": 0.5954998732726045,
      "learning_rate": 1.412893230150415e-06,
      "loss": 0.793,
      "step": 8413
    },
    {
      "epoch": 0.6070488077630677,
      "grad_norm": 3.330248088358747,
      "learning_rate": 1.4124464769928442e-06,
      "loss": 0.8707,
      "step": 8414
    },
    {
      "epoch": 0.6071209552324952,
      "grad_norm": 3.1990358628757716,
      "learning_rate": 1.4119997559193965e-06,
      "loss": 0.97,
      "step": 8415
    },
    {
      "epoch": 0.6071931027019227,
      "grad_norm": 2.8630675773466496,
      "learning_rate": 1.4115530669544653e-06,
      "loss": 0.9232,
      "step": 8416
    },
    {
      "epoch": 0.6072652501713502,
      "grad_norm": 4.80083945404353,
      "learning_rate": 1.4111064101224425e-06,
      "loss": 0.909,
      "step": 8417
    },
    {
      "epoch": 0.6073373976407778,
      "grad_norm": 4.603086986501325,
      "learning_rate": 1.4106597854477194e-06,
      "loss": 1.0478,
      "step": 8418
    },
    {
      "epoch": 0.6074095451102053,
      "grad_norm": 4.311188650471706,
      "learning_rate": 1.4102131929546826e-06,
      "loss": 0.9349,
      "step": 8419
    },
    {
      "epoch": 0.6074816925796328,
      "grad_norm": 6.6355052644652535,
      "learning_rate": 1.409766632667721e-06,
      "loss": 1.054,
      "step": 8420
    },
    {
      "epoch": 0.6075538400490603,
      "grad_norm": 7.109484728291259,
      "learning_rate": 1.4093201046112184e-06,
      "loss": 1.0039,
      "step": 8421
    },
    {
      "epoch": 0.6076259875184878,
      "grad_norm": 9.099871590515285,
      "learning_rate": 1.4088736088095576e-06,
      "loss": 0.9291,
      "step": 8422
    },
    {
      "epoch": 0.6076981349879153,
      "grad_norm": 3.072167222180466,
      "learning_rate": 1.408427145287122e-06,
      "loss": 0.9933,
      "step": 8423
    },
    {
      "epoch": 0.6077702824573428,
      "grad_norm": 6.285502801780571,
      "learning_rate": 1.407980714068289e-06,
      "loss": 0.9038,
      "step": 8424
    },
    {
      "epoch": 0.6078424299267703,
      "grad_norm": 3.4550915119953345,
      "learning_rate": 1.4075343151774379e-06,
      "loss": 1.0489,
      "step": 8425
    },
    {
      "epoch": 0.6079145773961978,
      "grad_norm": 3.335513085697737,
      "learning_rate": 1.4070879486389446e-06,
      "loss": 0.9782,
      "step": 8426
    },
    {
      "epoch": 0.6079867248656253,
      "grad_norm": 3.746889095582132,
      "learning_rate": 1.406641614477184e-06,
      "loss": 1.0234,
      "step": 8427
    },
    {
      "epoch": 0.6080588723350528,
      "grad_norm": 3.4857032878443155,
      "learning_rate": 1.4061953127165277e-06,
      "loss": 0.8951,
      "step": 8428
    },
    {
      "epoch": 0.6081310198044804,
      "grad_norm": 4.91889810302812,
      "learning_rate": 1.4057490433813474e-06,
      "loss": 0.9464,
      "step": 8429
    },
    {
      "epoch": 0.6082031672739079,
      "grad_norm": 3.063775244694036,
      "learning_rate": 1.4053028064960125e-06,
      "loss": 0.9781,
      "step": 8430
    },
    {
      "epoch": 0.6082753147433354,
      "grad_norm": 0.6969441118738341,
      "learning_rate": 1.4048566020848886e-06,
      "loss": 0.846,
      "step": 8431
    },
    {
      "epoch": 0.6083474622127629,
      "grad_norm": 3.2305943604188108,
      "learning_rate": 1.4044104301723437e-06,
      "loss": 1.0255,
      "step": 8432
    },
    {
      "epoch": 0.6084196096821904,
      "grad_norm": 0.7909462604014947,
      "learning_rate": 1.4039642907827403e-06,
      "loss": 0.8423,
      "step": 8433
    },
    {
      "epoch": 0.608491757151618,
      "grad_norm": 3.854016495262047,
      "learning_rate": 1.4035181839404397e-06,
      "loss": 0.9112,
      "step": 8434
    },
    {
      "epoch": 0.6085639046210454,
      "grad_norm": 3.8533642870133855,
      "learning_rate": 1.4030721096698037e-06,
      "loss": 1.0405,
      "step": 8435
    },
    {
      "epoch": 0.6086360520904729,
      "grad_norm": 2.6885485710480066,
      "learning_rate": 1.4026260679951897e-06,
      "loss": 0.9788,
      "step": 8436
    },
    {
      "epoch": 0.6087081995599004,
      "grad_norm": 4.770206837838044,
      "learning_rate": 1.4021800589409547e-06,
      "loss": 0.9122,
      "step": 8437
    },
    {
      "epoch": 0.6087803470293279,
      "grad_norm": 3.457047939799049,
      "learning_rate": 1.4017340825314543e-06,
      "loss": 1.0544,
      "step": 8438
    },
    {
      "epoch": 0.6088524944987554,
      "grad_norm": 3.364545524707548,
      "learning_rate": 1.4012881387910402e-06,
      "loss": 1.0216,
      "step": 8439
    },
    {
      "epoch": 0.608924641968183,
      "grad_norm": 0.8038414887348021,
      "learning_rate": 1.4008422277440648e-06,
      "loss": 0.827,
      "step": 8440
    },
    {
      "epoch": 0.6089967894376105,
      "grad_norm": 3.8582627396411233,
      "learning_rate": 1.4003963494148771e-06,
      "loss": 0.9035,
      "step": 8441
    },
    {
      "epoch": 0.609068936907038,
      "grad_norm": 2.3708000189703795,
      "learning_rate": 1.3999505038278255e-06,
      "loss": 0.9366,
      "step": 8442
    },
    {
      "epoch": 0.6091410843764655,
      "grad_norm": 4.33300638188082,
      "learning_rate": 1.3995046910072551e-06,
      "loss": 0.7512,
      "step": 8443
    },
    {
      "epoch": 0.609213231845893,
      "grad_norm": 2.8068108556804163,
      "learning_rate": 1.3990589109775109e-06,
      "loss": 0.7908,
      "step": 8444
    },
    {
      "epoch": 0.6092853793153206,
      "grad_norm": 4.671759052177409,
      "learning_rate": 1.3986131637629351e-06,
      "loss": 1.0432,
      "step": 8445
    },
    {
      "epoch": 0.609357526784748,
      "grad_norm": 3.3849515781291903,
      "learning_rate": 1.3981674493878672e-06,
      "loss": 0.9974,
      "step": 8446
    },
    {
      "epoch": 0.6094296742541755,
      "grad_norm": 7.533878864802096,
      "learning_rate": 1.3977217678766482e-06,
      "loss": 0.9618,
      "step": 8447
    },
    {
      "epoch": 0.609501821723603,
      "grad_norm": 7.00393402723856,
      "learning_rate": 1.3972761192536133e-06,
      "loss": 1.0914,
      "step": 8448
    },
    {
      "epoch": 0.6095739691930305,
      "grad_norm": 3.9188008817688287,
      "learning_rate": 1.3968305035430984e-06,
      "loss": 0.9099,
      "step": 8449
    },
    {
      "epoch": 0.609646116662458,
      "grad_norm": 3.8604671770727887,
      "learning_rate": 1.3963849207694371e-06,
      "loss": 1.0042,
      "step": 8450
    },
    {
      "epoch": 0.6097182641318856,
      "grad_norm": 4.69903325628806,
      "learning_rate": 1.3959393709569604e-06,
      "loss": 0.9178,
      "step": 8451
    },
    {
      "epoch": 0.6097904116013131,
      "grad_norm": 5.108141455643077,
      "learning_rate": 1.3954938541299988e-06,
      "loss": 1.0116,
      "step": 8452
    },
    {
      "epoch": 0.6098625590707406,
      "grad_norm": 2.8990469649090778,
      "learning_rate": 1.3950483703128795e-06,
      "loss": 0.9334,
      "step": 8453
    },
    {
      "epoch": 0.6099347065401681,
      "grad_norm": 4.826177605533323,
      "learning_rate": 1.3946029195299301e-06,
      "loss": 0.9652,
      "step": 8454
    },
    {
      "epoch": 0.6100068540095956,
      "grad_norm": 5.288353470985277,
      "learning_rate": 1.3941575018054735e-06,
      "loss": 0.9252,
      "step": 8455
    },
    {
      "epoch": 0.6100790014790232,
      "grad_norm": 4.9689793563785525,
      "learning_rate": 1.393712117163833e-06,
      "loss": 0.9747,
      "step": 8456
    },
    {
      "epoch": 0.6101511489484507,
      "grad_norm": 10.371706152469978,
      "learning_rate": 1.3932667656293297e-06,
      "loss": 0.9817,
      "step": 8457
    },
    {
      "epoch": 0.6102232964178781,
      "grad_norm": 4.623475931980561,
      "learning_rate": 1.392821447226282e-06,
      "loss": 0.9435,
      "step": 8458
    },
    {
      "epoch": 0.6102954438873056,
      "grad_norm": 3.2132715805208725,
      "learning_rate": 1.3923761619790075e-06,
      "loss": 0.9731,
      "step": 8459
    },
    {
      "epoch": 0.6103675913567331,
      "grad_norm": 9.062056747646121,
      "learning_rate": 1.3919309099118216e-06,
      "loss": 0.8462,
      "step": 8460
    },
    {
      "epoch": 0.6104397388261606,
      "grad_norm": 3.870182734667607,
      "learning_rate": 1.391485691049037e-06,
      "loss": 0.9932,
      "step": 8461
    },
    {
      "epoch": 0.6105118862955882,
      "grad_norm": 5.345530430851386,
      "learning_rate": 1.391040505414967e-06,
      "loss": 0.9263,
      "step": 8462
    },
    {
      "epoch": 0.6105840337650157,
      "grad_norm": 3.9185608010500332,
      "learning_rate": 1.3905953530339205e-06,
      "loss": 0.976,
      "step": 8463
    },
    {
      "epoch": 0.6106561812344432,
      "grad_norm": 4.8281374440449465,
      "learning_rate": 1.3901502339302055e-06,
      "loss": 0.977,
      "step": 8464
    },
    {
      "epoch": 0.6107283287038707,
      "grad_norm": 3.2918214077531593,
      "learning_rate": 1.3897051481281295e-06,
      "loss": 0.9414,
      "step": 8465
    },
    {
      "epoch": 0.6108004761732982,
      "grad_norm": 5.731084477188237,
      "learning_rate": 1.3892600956519953e-06,
      "loss": 0.9658,
      "step": 8466
    },
    {
      "epoch": 0.6108726236427258,
      "grad_norm": 4.348855510417457,
      "learning_rate": 1.3888150765261063e-06,
      "loss": 0.9273,
      "step": 8467
    },
    {
      "epoch": 0.6109447711121533,
      "grad_norm": 5.582254371962192,
      "learning_rate": 1.3883700907747637e-06,
      "loss": 0.9257,
      "step": 8468
    },
    {
      "epoch": 0.6110169185815808,
      "grad_norm": 5.777130473793237,
      "learning_rate": 1.3879251384222665e-06,
      "loss": 1.0809,
      "step": 8469
    },
    {
      "epoch": 0.6110890660510082,
      "grad_norm": 5.049351701575279,
      "learning_rate": 1.387480219492911e-06,
      "loss": 1.028,
      "step": 8470
    },
    {
      "epoch": 0.6111612135204357,
      "grad_norm": 4.989798056084327,
      "learning_rate": 1.3870353340109941e-06,
      "loss": 1.0498,
      "step": 8471
    },
    {
      "epoch": 0.6112333609898633,
      "grad_norm": 5.268302440526588,
      "learning_rate": 1.3865904820008085e-06,
      "loss": 0.9245,
      "step": 8472
    },
    {
      "epoch": 0.6113055084592908,
      "grad_norm": 3.412716109031738,
      "learning_rate": 1.386145663486645e-06,
      "loss": 1.0183,
      "step": 8473
    },
    {
      "epoch": 0.6113776559287183,
      "grad_norm": 3.4425235454388514,
      "learning_rate": 1.3857008784927952e-06,
      "loss": 0.8788,
      "step": 8474
    },
    {
      "epoch": 0.6114498033981458,
      "grad_norm": 2.892728627762963,
      "learning_rate": 1.3852561270435458e-06,
      "loss": 0.9715,
      "step": 8475
    },
    {
      "epoch": 0.6115219508675733,
      "grad_norm": 3.3153915740169366,
      "learning_rate": 1.384811409163184e-06,
      "loss": 0.908,
      "step": 8476
    },
    {
      "epoch": 0.6115940983370008,
      "grad_norm": 4.171440187576155,
      "learning_rate": 1.384366724875994e-06,
      "loss": 1.0029,
      "step": 8477
    },
    {
      "epoch": 0.6116662458064284,
      "grad_norm": 3.807487194017201,
      "learning_rate": 1.3839220742062576e-06,
      "loss": 0.9455,
      "step": 8478
    },
    {
      "epoch": 0.6117383932758559,
      "grad_norm": 4.047141990544189,
      "learning_rate": 1.3834774571782564e-06,
      "loss": 1.0526,
      "step": 8479
    },
    {
      "epoch": 0.6118105407452834,
      "grad_norm": 3.2854642417493265,
      "learning_rate": 1.3830328738162686e-06,
      "loss": 0.8182,
      "step": 8480
    },
    {
      "epoch": 0.6118826882147109,
      "grad_norm": 3.25321918391518,
      "learning_rate": 1.3825883241445725e-06,
      "loss": 0.9735,
      "step": 8481
    },
    {
      "epoch": 0.6119548356841383,
      "grad_norm": 4.165758542598153,
      "learning_rate": 1.3821438081874418e-06,
      "loss": 0.9843,
      "step": 8482
    },
    {
      "epoch": 0.6120269831535659,
      "grad_norm": 4.43748194731143,
      "learning_rate": 1.3816993259691506e-06,
      "loss": 1.0841,
      "step": 8483
    },
    {
      "epoch": 0.6120991306229934,
      "grad_norm": 3.7656743849685674,
      "learning_rate": 1.3812548775139706e-06,
      "loss": 0.9181,
      "step": 8484
    },
    {
      "epoch": 0.6121712780924209,
      "grad_norm": 6.0852194522852425,
      "learning_rate": 1.3808104628461703e-06,
      "loss": 0.9372,
      "step": 8485
    },
    {
      "epoch": 0.6122434255618484,
      "grad_norm": 2.9010657555466355,
      "learning_rate": 1.3803660819900194e-06,
      "loss": 0.9711,
      "step": 8486
    },
    {
      "epoch": 0.6123155730312759,
      "grad_norm": 4.718132938687084,
      "learning_rate": 1.3799217349697832e-06,
      "loss": 0.9656,
      "step": 8487
    },
    {
      "epoch": 0.6123877205007034,
      "grad_norm": 3.6491183640154405,
      "learning_rate": 1.3794774218097246e-06,
      "loss": 0.8646,
      "step": 8488
    },
    {
      "epoch": 0.612459867970131,
      "grad_norm": 0.779269110392975,
      "learning_rate": 1.3790331425341072e-06,
      "loss": 0.822,
      "step": 8489
    },
    {
      "epoch": 0.6125320154395585,
      "grad_norm": 4.359309589072998,
      "learning_rate": 1.3785888971671913e-06,
      "loss": 0.8956,
      "step": 8490
    },
    {
      "epoch": 0.612604162908986,
      "grad_norm": 4.760057542917906,
      "learning_rate": 1.3781446857332353e-06,
      "loss": 0.9792,
      "step": 8491
    },
    {
      "epoch": 0.6126763103784135,
      "grad_norm": 0.8801380934519464,
      "learning_rate": 1.3777005082564959e-06,
      "loss": 0.8073,
      "step": 8492
    },
    {
      "epoch": 0.6127484578478409,
      "grad_norm": 2.873925381386636,
      "learning_rate": 1.3772563647612288e-06,
      "loss": 0.9552,
      "step": 8493
    },
    {
      "epoch": 0.6128206053172685,
      "grad_norm": 3.733982847684134,
      "learning_rate": 1.3768122552716854e-06,
      "loss": 0.9947,
      "step": 8494
    },
    {
      "epoch": 0.612892752786696,
      "grad_norm": 6.6913438249605655,
      "learning_rate": 1.376368179812118e-06,
      "loss": 0.9432,
      "step": 8495
    },
    {
      "epoch": 0.6129649002561235,
      "grad_norm": 2.5833547550256664,
      "learning_rate": 1.375924138406776e-06,
      "loss": 0.9436,
      "step": 8496
    },
    {
      "epoch": 0.613037047725551,
      "grad_norm": 4.990760850102421,
      "learning_rate": 1.375480131079906e-06,
      "loss": 0.9484,
      "step": 8497
    },
    {
      "epoch": 0.6131091951949785,
      "grad_norm": 3.189631927206443,
      "learning_rate": 1.3750361578557548e-06,
      "loss": 0.9638,
      "step": 8498
    },
    {
      "epoch": 0.613181342664406,
      "grad_norm": 3.8275113606549076,
      "learning_rate": 1.3745922187585655e-06,
      "loss": 1.0558,
      "step": 8499
    },
    {
      "epoch": 0.6132534901338336,
      "grad_norm": 3.40457826240438,
      "learning_rate": 1.374148313812579e-06,
      "loss": 0.7865,
      "step": 8500
    },
    {
      "epoch": 0.6133256376032611,
      "grad_norm": 0.6445400815416483,
      "learning_rate": 1.3737044430420372e-06,
      "loss": 0.8401,
      "step": 8501
    },
    {
      "epoch": 0.6133977850726886,
      "grad_norm": 4.065022551038814,
      "learning_rate": 1.373260606471177e-06,
      "loss": 0.833,
      "step": 8502
    },
    {
      "epoch": 0.6134699325421161,
      "grad_norm": 3.8604154226867573,
      "learning_rate": 1.3728168041242347e-06,
      "loss": 0.9853,
      "step": 8503
    },
    {
      "epoch": 0.6135420800115436,
      "grad_norm": 3.29522967752242,
      "learning_rate": 1.3723730360254455e-06,
      "loss": 0.9729,
      "step": 8504
    },
    {
      "epoch": 0.613614227480971,
      "grad_norm": 3.8656292482813788,
      "learning_rate": 1.371929302199041e-06,
      "loss": 1.0294,
      "step": 8505
    },
    {
      "epoch": 0.6136863749503986,
      "grad_norm": 5.080392372417198,
      "learning_rate": 1.3714856026692526e-06,
      "loss": 1.0073,
      "step": 8506
    },
    {
      "epoch": 0.6137585224198261,
      "grad_norm": 10.306057113776436,
      "learning_rate": 1.3710419374603083e-06,
      "loss": 0.927,
      "step": 8507
    },
    {
      "epoch": 0.6138306698892536,
      "grad_norm": 4.471365849948639,
      "learning_rate": 1.370598306596436e-06,
      "loss": 1.0303,
      "step": 8508
    },
    {
      "epoch": 0.6139028173586811,
      "grad_norm": 4.005788429558257,
      "learning_rate": 1.3701547101018598e-06,
      "loss": 1.0079,
      "step": 8509
    },
    {
      "epoch": 0.6139749648281086,
      "grad_norm": 3.0627774482700887,
      "learning_rate": 1.3697111480008032e-06,
      "loss": 0.9765,
      "step": 8510
    },
    {
      "epoch": 0.6140471122975362,
      "grad_norm": 2.811264953709204,
      "learning_rate": 1.3692676203174883e-06,
      "loss": 0.9002,
      "step": 8511
    },
    {
      "epoch": 0.6141192597669637,
      "grad_norm": 3.454830111178686,
      "learning_rate": 1.3688241270761327e-06,
      "loss": 0.9467,
      "step": 8512
    },
    {
      "epoch": 0.6141914072363912,
      "grad_norm": 4.817863942975399,
      "learning_rate": 1.3683806683009555e-06,
      "loss": 1.016,
      "step": 8513
    },
    {
      "epoch": 0.6142635547058187,
      "grad_norm": 0.7529716034736401,
      "learning_rate": 1.367937244016172e-06,
      "loss": 0.8408,
      "step": 8514
    },
    {
      "epoch": 0.6143357021752462,
      "grad_norm": 3.807327012899509,
      "learning_rate": 1.3674938542459955e-06,
      "loss": 0.9604,
      "step": 8515
    },
    {
      "epoch": 0.6144078496446738,
      "grad_norm": 2.9710151463356524,
      "learning_rate": 1.3670504990146388e-06,
      "loss": 0.9551,
      "step": 8516
    },
    {
      "epoch": 0.6144799971141012,
      "grad_norm": 3.197451595986904,
      "learning_rate": 1.3666071783463104e-06,
      "loss": 1.0792,
      "step": 8517
    },
    {
      "epoch": 0.6145521445835287,
      "grad_norm": 4.01473858620052,
      "learning_rate": 1.3661638922652196e-06,
      "loss": 0.9391,
      "step": 8518
    },
    {
      "epoch": 0.6146242920529562,
      "grad_norm": 3.933190788229237,
      "learning_rate": 1.3657206407955721e-06,
      "loss": 0.8189,
      "step": 8519
    },
    {
      "epoch": 0.6146964395223837,
      "grad_norm": 8.502687365994804,
      "learning_rate": 1.3652774239615728e-06,
      "loss": 1.0171,
      "step": 8520
    },
    {
      "epoch": 0.6147685869918113,
      "grad_norm": 3.866984660676295,
      "learning_rate": 1.3648342417874235e-06,
      "loss": 0.9362,
      "step": 8521
    },
    {
      "epoch": 0.6148407344612388,
      "grad_norm": 3.740520703384261,
      "learning_rate": 1.3643910942973248e-06,
      "loss": 0.9771,
      "step": 8522
    },
    {
      "epoch": 0.6149128819306663,
      "grad_norm": 4.360935601808414,
      "learning_rate": 1.363947981515476e-06,
      "loss": 1.0453,
      "step": 8523
    },
    {
      "epoch": 0.6149850294000938,
      "grad_norm": 4.125122993253978,
      "learning_rate": 1.3635049034660723e-06,
      "loss": 0.9084,
      "step": 8524
    },
    {
      "epoch": 0.6150571768695213,
      "grad_norm": 4.736695781477795,
      "learning_rate": 1.3630618601733106e-06,
      "loss": 0.9559,
      "step": 8525
    },
    {
      "epoch": 0.6151293243389488,
      "grad_norm": 3.2411860746353574,
      "learning_rate": 1.3626188516613828e-06,
      "loss": 0.9325,
      "step": 8526
    },
    {
      "epoch": 0.6152014718083764,
      "grad_norm": 4.282858887358238,
      "learning_rate": 1.3621758779544792e-06,
      "loss": 1.0165,
      "step": 8527
    },
    {
      "epoch": 0.6152736192778039,
      "grad_norm": 3.9225480311451415,
      "learning_rate": 1.3617329390767905e-06,
      "loss": 0.9222,
      "step": 8528
    },
    {
      "epoch": 0.6153457667472313,
      "grad_norm": 5.034607898101109,
      "learning_rate": 1.3612900350525029e-06,
      "loss": 0.8645,
      "step": 8529
    },
    {
      "epoch": 0.6154179142166588,
      "grad_norm": 3.7418842232601484,
      "learning_rate": 1.360847165905802e-06,
      "loss": 0.9782,
      "step": 8530
    },
    {
      "epoch": 0.6154900616860863,
      "grad_norm": 5.074391280626326,
      "learning_rate": 1.360404331660872e-06,
      "loss": 1.0605,
      "step": 8531
    },
    {
      "epoch": 0.6155622091555139,
      "grad_norm": 3.7430190912443173,
      "learning_rate": 1.3599615323418929e-06,
      "loss": 0.9779,
      "step": 8532
    },
    {
      "epoch": 0.6156343566249414,
      "grad_norm": 9.959057727041595,
      "learning_rate": 1.3595187679730455e-06,
      "loss": 0.9313,
      "step": 8533
    },
    {
      "epoch": 0.6157065040943689,
      "grad_norm": 4.866195161178771,
      "learning_rate": 1.359076038578507e-06,
      "loss": 1.0721,
      "step": 8534
    },
    {
      "epoch": 0.6157786515637964,
      "grad_norm": 3.6900437196865625,
      "learning_rate": 1.3586333441824537e-06,
      "loss": 0.9942,
      "step": 8535
    },
    {
      "epoch": 0.6158507990332239,
      "grad_norm": 4.542643567400411,
      "learning_rate": 1.3581906848090588e-06,
      "loss": 0.9374,
      "step": 8536
    },
    {
      "epoch": 0.6159229465026514,
      "grad_norm": 2.7926442631381163,
      "learning_rate": 1.3577480604824954e-06,
      "loss": 1.0004,
      "step": 8537
    },
    {
      "epoch": 0.615995093972079,
      "grad_norm": 15.986928845807078,
      "learning_rate": 1.3573054712269328e-06,
      "loss": 0.9365,
      "step": 8538
    },
    {
      "epoch": 0.6160672414415065,
      "grad_norm": 6.044840783482742,
      "learning_rate": 1.3568629170665382e-06,
      "loss": 1.0366,
      "step": 8539
    },
    {
      "epoch": 0.6161393889109339,
      "grad_norm": 4.688033010060613,
      "learning_rate": 1.35642039802548e-06,
      "loss": 0.9352,
      "step": 8540
    },
    {
      "epoch": 0.6162115363803614,
      "grad_norm": 5.418521954622819,
      "learning_rate": 1.355977914127921e-06,
      "loss": 0.9071,
      "step": 8541
    },
    {
      "epoch": 0.6162836838497889,
      "grad_norm": 3.9507666025009023,
      "learning_rate": 1.355535465398024e-06,
      "loss": 0.9514,
      "step": 8542
    },
    {
      "epoch": 0.6163558313192165,
      "grad_norm": 19.026025715871814,
      "learning_rate": 1.35509305185995e-06,
      "loss": 0.997,
      "step": 8543
    },
    {
      "epoch": 0.616427978788644,
      "grad_norm": 4.633341715596985,
      "learning_rate": 1.3546506735378564e-06,
      "loss": 0.9145,
      "step": 8544
    },
    {
      "epoch": 0.6165001262580715,
      "grad_norm": 2.7885315119820056,
      "learning_rate": 1.3542083304559007e-06,
      "loss": 0.9598,
      "step": 8545
    },
    {
      "epoch": 0.616572273727499,
      "grad_norm": 5.5393252290256365,
      "learning_rate": 1.3537660226382373e-06,
      "loss": 0.9841,
      "step": 8546
    },
    {
      "epoch": 0.6166444211969265,
      "grad_norm": 10.724647612217849,
      "learning_rate": 1.3533237501090197e-06,
      "loss": 0.9297,
      "step": 8547
    },
    {
      "epoch": 0.616716568666354,
      "grad_norm": 7.644486338971295,
      "learning_rate": 1.3528815128923982e-06,
      "loss": 0.9466,
      "step": 8548
    },
    {
      "epoch": 0.6167887161357816,
      "grad_norm": 6.261888412490158,
      "learning_rate": 1.3524393110125208e-06,
      "loss": 0.9075,
      "step": 8549
    },
    {
      "epoch": 0.6168608636052091,
      "grad_norm": 8.912135271097046,
      "learning_rate": 1.3519971444935364e-06,
      "loss": 0.8917,
      "step": 8550
    },
    {
      "epoch": 0.6169330110746366,
      "grad_norm": 2.658220525119851,
      "learning_rate": 1.351555013359588e-06,
      "loss": 0.9611,
      "step": 8551
    },
    {
      "epoch": 0.617005158544064,
      "grad_norm": 2.5681414957786277,
      "learning_rate": 1.3511129176348212e-06,
      "loss": 0.987,
      "step": 8552
    },
    {
      "epoch": 0.6170773060134915,
      "grad_norm": 3.8947351033110555,
      "learning_rate": 1.3506708573433754e-06,
      "loss": 0.8498,
      "step": 8553
    },
    {
      "epoch": 0.617149453482919,
      "grad_norm": 4.561140092909179,
      "learning_rate": 1.3502288325093895e-06,
      "loss": 0.9579,
      "step": 8554
    },
    {
      "epoch": 0.6172216009523466,
      "grad_norm": 2.9199992804330437,
      "learning_rate": 1.3497868431570029e-06,
      "loss": 1.0158,
      "step": 8555
    },
    {
      "epoch": 0.6172937484217741,
      "grad_norm": 3.6931240061256516,
      "learning_rate": 1.3493448893103489e-06,
      "loss": 0.9308,
      "step": 8556
    },
    {
      "epoch": 0.6173658958912016,
      "grad_norm": 4.118126894136573,
      "learning_rate": 1.3489029709935621e-06,
      "loss": 0.9753,
      "step": 8557
    },
    {
      "epoch": 0.6174380433606291,
      "grad_norm": 2.6198210717757267,
      "learning_rate": 1.3484610882307739e-06,
      "loss": 0.9495,
      "step": 8558
    },
    {
      "epoch": 0.6175101908300566,
      "grad_norm": 3.89523691654132,
      "learning_rate": 1.348019241046114e-06,
      "loss": 0.8606,
      "step": 8559
    },
    {
      "epoch": 0.6175823382994842,
      "grad_norm": 4.458849416554324,
      "learning_rate": 1.3475774294637097e-06,
      "loss": 0.8535,
      "step": 8560
    },
    {
      "epoch": 0.6176544857689117,
      "grad_norm": 4.789904173527603,
      "learning_rate": 1.3471356535076864e-06,
      "loss": 1.0571,
      "step": 8561
    },
    {
      "epoch": 0.6177266332383392,
      "grad_norm": 3.7430953993099028,
      "learning_rate": 1.3466939132021686e-06,
      "loss": 0.8982,
      "step": 8562
    },
    {
      "epoch": 0.6177987807077667,
      "grad_norm": 0.7973670188503185,
      "learning_rate": 1.3462522085712771e-06,
      "loss": 0.8881,
      "step": 8563
    },
    {
      "epoch": 0.6178709281771941,
      "grad_norm": 4.519590754929147,
      "learning_rate": 1.345810539639133e-06,
      "loss": 0.8326,
      "step": 8564
    },
    {
      "epoch": 0.6179430756466217,
      "grad_norm": 6.486633642778433,
      "learning_rate": 1.345368906429854e-06,
      "loss": 0.8718,
      "step": 8565
    },
    {
      "epoch": 0.6180152231160492,
      "grad_norm": 2.9773616168632926,
      "learning_rate": 1.3449273089675547e-06,
      "loss": 0.9375,
      "step": 8566
    },
    {
      "epoch": 0.6180873705854767,
      "grad_norm": 3.5313916473336557,
      "learning_rate": 1.3444857472763507e-06,
      "loss": 0.9657,
      "step": 8567
    },
    {
      "epoch": 0.6181595180549042,
      "grad_norm": 2.8605434344254195,
      "learning_rate": 1.3440442213803533e-06,
      "loss": 0.9353,
      "step": 8568
    },
    {
      "epoch": 0.6182316655243317,
      "grad_norm": 3.022869046811311,
      "learning_rate": 1.3436027313036723e-06,
      "loss": 0.9302,
      "step": 8569
    },
    {
      "epoch": 0.6183038129937592,
      "grad_norm": 3.0792932762232397,
      "learning_rate": 1.343161277070417e-06,
      "loss": 1.0229,
      "step": 8570
    },
    {
      "epoch": 0.6183759604631868,
      "grad_norm": 3.0666892534612833,
      "learning_rate": 1.3427198587046924e-06,
      "loss": 1.0714,
      "step": 8571
    },
    {
      "epoch": 0.6184481079326143,
      "grad_norm": 4.062199508117493,
      "learning_rate": 1.3422784762306036e-06,
      "loss": 0.9927,
      "step": 8572
    },
    {
      "epoch": 0.6185202554020418,
      "grad_norm": 2.9109858821630077,
      "learning_rate": 1.3418371296722514e-06,
      "loss": 0.8841,
      "step": 8573
    },
    {
      "epoch": 0.6185924028714693,
      "grad_norm": 8.5141183622743,
      "learning_rate": 1.341395819053738e-06,
      "loss": 0.9358,
      "step": 8574
    },
    {
      "epoch": 0.6186645503408968,
      "grad_norm": 3.158350632619326,
      "learning_rate": 1.3409545443991607e-06,
      "loss": 0.9975,
      "step": 8575
    },
    {
      "epoch": 0.6187366978103243,
      "grad_norm": 3.2723268796213323,
      "learning_rate": 1.3405133057326155e-06,
      "loss": 0.9728,
      "step": 8576
    },
    {
      "epoch": 0.6188088452797518,
      "grad_norm": 2.9218625971077,
      "learning_rate": 1.340072103078198e-06,
      "loss": 0.9574,
      "step": 8577
    },
    {
      "epoch": 0.6188809927491793,
      "grad_norm": 4.361998066268136,
      "learning_rate": 1.339630936459999e-06,
      "loss": 0.8971,
      "step": 8578
    },
    {
      "epoch": 0.6189531402186068,
      "grad_norm": 3.0448772434489606,
      "learning_rate": 1.339189805902111e-06,
      "loss": 1.0607,
      "step": 8579
    },
    {
      "epoch": 0.6190252876880343,
      "grad_norm": 2.700469989843646,
      "learning_rate": 1.338748711428621e-06,
      "loss": 0.9443,
      "step": 8580
    },
    {
      "epoch": 0.6190974351574619,
      "grad_norm": 3.8044507464117023,
      "learning_rate": 1.3383076530636159e-06,
      "loss": 0.94,
      "step": 8581
    },
    {
      "epoch": 0.6191695826268894,
      "grad_norm": 2.73350214514431,
      "learning_rate": 1.337866630831181e-06,
      "loss": 0.9206,
      "step": 8582
    },
    {
      "epoch": 0.6192417300963169,
      "grad_norm": 0.6946520293768812,
      "learning_rate": 1.3374256447553977e-06,
      "loss": 0.8448,
      "step": 8583
    },
    {
      "epoch": 0.6193138775657444,
      "grad_norm": 9.229377646578563,
      "learning_rate": 1.3369846948603475e-06,
      "loss": 0.997,
      "step": 8584
    },
    {
      "epoch": 0.6193860250351719,
      "grad_norm": 4.651925965839352,
      "learning_rate": 1.3365437811701083e-06,
      "loss": 0.9861,
      "step": 8585
    },
    {
      "epoch": 0.6194581725045994,
      "grad_norm": 8.71265222753041,
      "learning_rate": 1.3361029037087574e-06,
      "loss": 0.9123,
      "step": 8586
    },
    {
      "epoch": 0.619530319974027,
      "grad_norm": 2.8375031324192856,
      "learning_rate": 1.3356620625003699e-06,
      "loss": 1.0028,
      "step": 8587
    },
    {
      "epoch": 0.6196024674434544,
      "grad_norm": 3.5904872512622923,
      "learning_rate": 1.3352212575690165e-06,
      "loss": 0.904,
      "step": 8588
    },
    {
      "epoch": 0.6196746149128819,
      "grad_norm": 3.4412216847145447,
      "learning_rate": 1.3347804889387705e-06,
      "loss": 0.9569,
      "step": 8589
    },
    {
      "epoch": 0.6197467623823094,
      "grad_norm": 3.2618095339630195,
      "learning_rate": 1.3343397566336983e-06,
      "loss": 0.9338,
      "step": 8590
    },
    {
      "epoch": 0.6198189098517369,
      "grad_norm": 3.5308491977325946,
      "learning_rate": 1.3338990606778689e-06,
      "loss": 0.9963,
      "step": 8591
    },
    {
      "epoch": 0.6198910573211645,
      "grad_norm": 3.619735446557381,
      "learning_rate": 1.3334584010953458e-06,
      "loss": 1.0121,
      "step": 8592
    },
    {
      "epoch": 0.619963204790592,
      "grad_norm": 4.3284374358405975,
      "learning_rate": 1.3330177779101913e-06,
      "loss": 0.9776,
      "step": 8593
    },
    {
      "epoch": 0.6200353522600195,
      "grad_norm": 4.355431526606633,
      "learning_rate": 1.3325771911464676e-06,
      "loss": 0.9367,
      "step": 8594
    },
    {
      "epoch": 0.620107499729447,
      "grad_norm": 2.8035776816828575,
      "learning_rate": 1.3321366408282325e-06,
      "loss": 0.9369,
      "step": 8595
    },
    {
      "epoch": 0.6201796471988745,
      "grad_norm": 11.630043822424893,
      "learning_rate": 1.3316961269795432e-06,
      "loss": 0.9839,
      "step": 8596
    },
    {
      "epoch": 0.620251794668302,
      "grad_norm": 3.026921436017423,
      "learning_rate": 1.3312556496244543e-06,
      "loss": 1.0083,
      "step": 8597
    },
    {
      "epoch": 0.6203239421377296,
      "grad_norm": 2.4984500848907656,
      "learning_rate": 1.3308152087870187e-06,
      "loss": 0.8636,
      "step": 8598
    },
    {
      "epoch": 0.620396089607157,
      "grad_norm": 2.2436642983733477,
      "learning_rate": 1.3303748044912875e-06,
      "loss": 0.9814,
      "step": 8599
    },
    {
      "epoch": 0.6204682370765845,
      "grad_norm": 2.838967630232048,
      "learning_rate": 1.3299344367613085e-06,
      "loss": 0.9984,
      "step": 8600
    },
    {
      "epoch": 0.620540384546012,
      "grad_norm": 4.29948229667354,
      "learning_rate": 1.3294941056211306e-06,
      "loss": 0.9685,
      "step": 8601
    },
    {
      "epoch": 0.6206125320154395,
      "grad_norm": 7.92714730904053,
      "learning_rate": 1.3290538110947965e-06,
      "loss": 0.9573,
      "step": 8602
    },
    {
      "epoch": 0.620684679484867,
      "grad_norm": 2.828276403600103,
      "learning_rate": 1.3286135532063508e-06,
      "loss": 0.9124,
      "step": 8603
    },
    {
      "epoch": 0.6207568269542946,
      "grad_norm": 3.4429659302217153,
      "learning_rate": 1.3281733319798337e-06,
      "loss": 0.9405,
      "step": 8604
    },
    {
      "epoch": 0.6208289744237221,
      "grad_norm": 3.5334093361603656,
      "learning_rate": 1.3277331474392828e-06,
      "loss": 0.8916,
      "step": 8605
    },
    {
      "epoch": 0.6209011218931496,
      "grad_norm": 8.508106181082498,
      "learning_rate": 1.3272929996087373e-06,
      "loss": 0.9532,
      "step": 8606
    },
    {
      "epoch": 0.6209732693625771,
      "grad_norm": 2.600913026862533,
      "learning_rate": 1.3268528885122302e-06,
      "loss": 0.8856,
      "step": 8607
    },
    {
      "epoch": 0.6210454168320046,
      "grad_norm": 4.910969207884633,
      "learning_rate": 1.3264128141737953e-06,
      "loss": 1.0506,
      "step": 8608
    },
    {
      "epoch": 0.6211175643014322,
      "grad_norm": 3.6546375599058933,
      "learning_rate": 1.3259727766174627e-06,
      "loss": 1.0106,
      "step": 8609
    },
    {
      "epoch": 0.6211897117708597,
      "grad_norm": 3.074431241167224,
      "learning_rate": 1.3255327758672614e-06,
      "loss": 0.9583,
      "step": 8610
    },
    {
      "epoch": 0.6212618592402871,
      "grad_norm": 4.762406259650473,
      "learning_rate": 1.3250928119472193e-06,
      "loss": 0.8991,
      "step": 8611
    },
    {
      "epoch": 0.6213340067097146,
      "grad_norm": 2.892402391163309,
      "learning_rate": 1.3246528848813591e-06,
      "loss": 1.0217,
      "step": 8612
    },
    {
      "epoch": 0.6214061541791421,
      "grad_norm": 4.293699776745825,
      "learning_rate": 1.3242129946937057e-06,
      "loss": 0.9159,
      "step": 8613
    },
    {
      "epoch": 0.6214783016485697,
      "grad_norm": 3.3340704738626354,
      "learning_rate": 1.3237731414082794e-06,
      "loss": 0.9131,
      "step": 8614
    },
    {
      "epoch": 0.6215504491179972,
      "grad_norm": 5.246449496093856,
      "learning_rate": 1.3233333250490972e-06,
      "loss": 1.0237,
      "step": 8615
    },
    {
      "epoch": 0.6216225965874247,
      "grad_norm": 3.428078241298876,
      "learning_rate": 1.3228935456401785e-06,
      "loss": 0.9352,
      "step": 8616
    },
    {
      "epoch": 0.6216947440568522,
      "grad_norm": 4.667348448586986,
      "learning_rate": 1.3224538032055352e-06,
      "loss": 0.8727,
      "step": 8617
    },
    {
      "epoch": 0.6217668915262797,
      "grad_norm": 4.318248731833931,
      "learning_rate": 1.3220140977691828e-06,
      "loss": 0.971,
      "step": 8618
    },
    {
      "epoch": 0.6218390389957072,
      "grad_norm": 3.7147712259202845,
      "learning_rate": 1.3215744293551303e-06,
      "loss": 1.0083,
      "step": 8619
    },
    {
      "epoch": 0.6219111864651348,
      "grad_norm": 3.7279968700150214,
      "learning_rate": 1.321134797987387e-06,
      "loss": 0.9349,
      "step": 8620
    },
    {
      "epoch": 0.6219833339345623,
      "grad_norm": 4.373912349107021,
      "learning_rate": 1.320695203689959e-06,
      "loss": 1.029,
      "step": 8621
    },
    {
      "epoch": 0.6220554814039898,
      "grad_norm": 13.935766274770794,
      "learning_rate": 1.3202556464868514e-06,
      "loss": 0.9441,
      "step": 8622
    },
    {
      "epoch": 0.6221276288734172,
      "grad_norm": 5.154626764203891,
      "learning_rate": 1.319816126402067e-06,
      "loss": 0.9071,
      "step": 8623
    },
    {
      "epoch": 0.6221997763428447,
      "grad_norm": 3.473640499798213,
      "learning_rate": 1.3193766434596048e-06,
      "loss": 0.94,
      "step": 8624
    },
    {
      "epoch": 0.6222719238122723,
      "grad_norm": 6.12780292759831,
      "learning_rate": 1.3189371976834655e-06,
      "loss": 0.8154,
      "step": 8625
    },
    {
      "epoch": 0.6223440712816998,
      "grad_norm": 4.682875119167056,
      "learning_rate": 1.3184977890976448e-06,
      "loss": 0.8897,
      "step": 8626
    },
    {
      "epoch": 0.6224162187511273,
      "grad_norm": 6.117750275936466,
      "learning_rate": 1.3180584177261361e-06,
      "loss": 1.0322,
      "step": 8627
    },
    {
      "epoch": 0.6224883662205548,
      "grad_norm": 4.405226784799245,
      "learning_rate": 1.3176190835929334e-06,
      "loss": 0.9807,
      "step": 8628
    },
    {
      "epoch": 0.6225605136899823,
      "grad_norm": 0.81102420585906,
      "learning_rate": 1.3171797867220255e-06,
      "loss": 0.8684,
      "step": 8629
    },
    {
      "epoch": 0.6226326611594099,
      "grad_norm": 4.23912991935191,
      "learning_rate": 1.3167405271374031e-06,
      "loss": 0.9973,
      "step": 8630
    },
    {
      "epoch": 0.6227048086288374,
      "grad_norm": 2.659081240431155,
      "learning_rate": 1.3163013048630508e-06,
      "loss": 0.933,
      "step": 8631
    },
    {
      "epoch": 0.6227769560982649,
      "grad_norm": 4.350658384462736,
      "learning_rate": 1.315862119922953e-06,
      "loss": 0.9178,
      "step": 8632
    },
    {
      "epoch": 0.6228491035676924,
      "grad_norm": 4.132324189902,
      "learning_rate": 1.3154229723410922e-06,
      "loss": 0.9484,
      "step": 8633
    },
    {
      "epoch": 0.6229212510371199,
      "grad_norm": 2.5954561252343216,
      "learning_rate": 1.3149838621414486e-06,
      "loss": 0.9207,
      "step": 8634
    },
    {
      "epoch": 0.6229933985065473,
      "grad_norm": 4.189340215553088,
      "learning_rate": 1.3145447893480008e-06,
      "loss": 1.0048,
      "step": 8635
    },
    {
      "epoch": 0.6230655459759749,
      "grad_norm": 5.239395148333989,
      "learning_rate": 1.3141057539847244e-06,
      "loss": 0.8543,
      "step": 8636
    },
    {
      "epoch": 0.6231376934454024,
      "grad_norm": 4.451640206137772,
      "learning_rate": 1.3136667560755936e-06,
      "loss": 0.8529,
      "step": 8637
    },
    {
      "epoch": 0.6232098409148299,
      "grad_norm": 3.6805111880397328,
      "learning_rate": 1.3132277956445808e-06,
      "loss": 0.8905,
      "step": 8638
    },
    {
      "epoch": 0.6232819883842574,
      "grad_norm": 5.087241666266317,
      "learning_rate": 1.312788872715655e-06,
      "loss": 0.9436,
      "step": 8639
    },
    {
      "epoch": 0.6233541358536849,
      "grad_norm": 15.400627301803732,
      "learning_rate": 1.3123499873127858e-06,
      "loss": 1.0249,
      "step": 8640
    },
    {
      "epoch": 0.6234262833231125,
      "grad_norm": 3.4034845167524894,
      "learning_rate": 1.3119111394599374e-06,
      "loss": 0.9544,
      "step": 8641
    },
    {
      "epoch": 0.62349843079254,
      "grad_norm": 2.2316172193424597,
      "learning_rate": 1.3114723291810753e-06,
      "loss": 0.9817,
      "step": 8642
    },
    {
      "epoch": 0.6235705782619675,
      "grad_norm": 3.04042771768742,
      "learning_rate": 1.3110335565001602e-06,
      "loss": 1.0086,
      "step": 8643
    },
    {
      "epoch": 0.623642725731395,
      "grad_norm": 4.726698490424872,
      "learning_rate": 1.3105948214411519e-06,
      "loss": 0.8699,
      "step": 8644
    },
    {
      "epoch": 0.6237148732008225,
      "grad_norm": 2.9239602221537324,
      "learning_rate": 1.3101561240280083e-06,
      "loss": 1.0579,
      "step": 8645
    },
    {
      "epoch": 0.6237870206702499,
      "grad_norm": 3.1750111977687623,
      "learning_rate": 1.309717464284685e-06,
      "loss": 1.0626,
      "step": 8646
    },
    {
      "epoch": 0.6238591681396775,
      "grad_norm": 3.57043643794731,
      "learning_rate": 1.3092788422351362e-06,
      "loss": 0.8595,
      "step": 8647
    },
    {
      "epoch": 0.623931315609105,
      "grad_norm": 4.044042590255666,
      "learning_rate": 1.3088402579033122e-06,
      "loss": 0.8527,
      "step": 8648
    },
    {
      "epoch": 0.6240034630785325,
      "grad_norm": 0.8019139907883709,
      "learning_rate": 1.3084017113131633e-06,
      "loss": 0.8865,
      "step": 8649
    },
    {
      "epoch": 0.62407561054796,
      "grad_norm": 2.869941905958917,
      "learning_rate": 1.307963202488637e-06,
      "loss": 0.905,
      "step": 8650
    },
    {
      "epoch": 0.6241477580173875,
      "grad_norm": 2.4954228462180517,
      "learning_rate": 1.3075247314536776e-06,
      "loss": 0.8604,
      "step": 8651
    },
    {
      "epoch": 0.624219905486815,
      "grad_norm": 5.524935636533859,
      "learning_rate": 1.3070862982322302e-06,
      "loss": 0.9562,
      "step": 8652
    },
    {
      "epoch": 0.6242920529562426,
      "grad_norm": 4.650150272033177,
      "learning_rate": 1.3066479028482347e-06,
      "loss": 0.9385,
      "step": 8653
    },
    {
      "epoch": 0.6243642004256701,
      "grad_norm": 4.682675332831124,
      "learning_rate": 1.3062095453256298e-06,
      "loss": 0.9159,
      "step": 8654
    },
    {
      "epoch": 0.6244363478950976,
      "grad_norm": 4.511983809178298,
      "learning_rate": 1.3057712256883542e-06,
      "loss": 0.9296,
      "step": 8655
    },
    {
      "epoch": 0.6245084953645251,
      "grad_norm": 0.7933399853574401,
      "learning_rate": 1.3053329439603412e-06,
      "loss": 0.8511,
      "step": 8656
    },
    {
      "epoch": 0.6245806428339526,
      "grad_norm": 3.6411063076677386,
      "learning_rate": 1.3048947001655253e-06,
      "loss": 0.9469,
      "step": 8657
    },
    {
      "epoch": 0.6246527903033801,
      "grad_norm": 4.218932649402609,
      "learning_rate": 1.3044564943278364e-06,
      "loss": 0.9561,
      "step": 8658
    },
    {
      "epoch": 0.6247249377728076,
      "grad_norm": 3.318569596554409,
      "learning_rate": 1.3040183264712035e-06,
      "loss": 0.9725,
      "step": 8659
    },
    {
      "epoch": 0.6247970852422351,
      "grad_norm": 3.9420046999442206,
      "learning_rate": 1.3035801966195532e-06,
      "loss": 0.9823,
      "step": 8660
    },
    {
      "epoch": 0.6248692327116626,
      "grad_norm": 3.6303634427983433,
      "learning_rate": 1.3031421047968106e-06,
      "loss": 0.9786,
      "step": 8661
    },
    {
      "epoch": 0.6249413801810901,
      "grad_norm": 4.0035208465425995,
      "learning_rate": 1.302704051026898e-06,
      "loss": 0.9571,
      "step": 8662
    },
    {
      "epoch": 0.6250135276505177,
      "grad_norm": 6.056655426154923,
      "learning_rate": 1.3022660353337352e-06,
      "loss": 1.0451,
      "step": 8663
    },
    {
      "epoch": 0.6250856751199452,
      "grad_norm": 3.3189470423018945,
      "learning_rate": 1.301828057741242e-06,
      "loss": 0.9256,
      "step": 8664
    },
    {
      "epoch": 0.6251578225893727,
      "grad_norm": 6.724633767745478,
      "learning_rate": 1.3013901182733342e-06,
      "loss": 0.9796,
      "step": 8665
    },
    {
      "epoch": 0.6252299700588002,
      "grad_norm": 3.167589722335873,
      "learning_rate": 1.300952216953925e-06,
      "loss": 0.9075,
      "step": 8666
    },
    {
      "epoch": 0.6253021175282277,
      "grad_norm": 6.575157239070855,
      "learning_rate": 1.300514353806928e-06,
      "loss": 0.9349,
      "step": 8667
    },
    {
      "epoch": 0.6253742649976552,
      "grad_norm": 3.4016737688866696,
      "learning_rate": 1.3000765288562523e-06,
      "loss": 0.8632,
      "step": 8668
    },
    {
      "epoch": 0.6254464124670828,
      "grad_norm": 4.495861905878938,
      "learning_rate": 1.2996387421258068e-06,
      "loss": 1.0109,
      "step": 8669
    },
    {
      "epoch": 0.6255185599365102,
      "grad_norm": 1.7947873136704124,
      "learning_rate": 1.2992009936394973e-06,
      "loss": 1.0179,
      "step": 8670
    },
    {
      "epoch": 0.6255907074059377,
      "grad_norm": 3.450942925121876,
      "learning_rate": 1.2987632834212266e-06,
      "loss": 0.9385,
      "step": 8671
    },
    {
      "epoch": 0.6256628548753652,
      "grad_norm": 3.9326410740443554,
      "learning_rate": 1.2983256114948972e-06,
      "loss": 0.9751,
      "step": 8672
    },
    {
      "epoch": 0.6257350023447927,
      "grad_norm": 6.01354183441731,
      "learning_rate": 1.2978879778844087e-06,
      "loss": 1.0442,
      "step": 8673
    },
    {
      "epoch": 0.6258071498142203,
      "grad_norm": 0.6906738928494469,
      "learning_rate": 1.297450382613659e-06,
      "loss": 0.8152,
      "step": 8674
    },
    {
      "epoch": 0.6258792972836478,
      "grad_norm": 3.657655494706485,
      "learning_rate": 1.2970128257065427e-06,
      "loss": 1.0438,
      "step": 8675
    },
    {
      "epoch": 0.6259514447530753,
      "grad_norm": 4.3335508634303,
      "learning_rate": 1.2965753071869534e-06,
      "loss": 1.0224,
      "step": 8676
    },
    {
      "epoch": 0.6260235922225028,
      "grad_norm": 3.070500569492443,
      "learning_rate": 1.2961378270787832e-06,
      "loss": 0.8835,
      "step": 8677
    },
    {
      "epoch": 0.6260957396919303,
      "grad_norm": 2.4330806283067465,
      "learning_rate": 1.2957003854059202e-06,
      "loss": 0.9633,
      "step": 8678
    },
    {
      "epoch": 0.6261678871613579,
      "grad_norm": 3.318140805182194,
      "learning_rate": 1.295262982192252e-06,
      "loss": 0.9444,
      "step": 8679
    },
    {
      "epoch": 0.6262400346307854,
      "grad_norm": 2.775509342497432,
      "learning_rate": 1.294825617461664e-06,
      "loss": 1.0085,
      "step": 8680
    },
    {
      "epoch": 0.6263121821002129,
      "grad_norm": 5.285924885210709,
      "learning_rate": 1.2943882912380377e-06,
      "loss": 1.0097,
      "step": 8681
    },
    {
      "epoch": 0.6263843295696403,
      "grad_norm": 2.681754335575079,
      "learning_rate": 1.2939510035452555e-06,
      "loss": 0.9037,
      "step": 8682
    },
    {
      "epoch": 0.6264564770390678,
      "grad_norm": 3.938962876581959,
      "learning_rate": 1.293513754407195e-06,
      "loss": 0.9083,
      "step": 8683
    },
    {
      "epoch": 0.6265286245084953,
      "grad_norm": 1.179799914530214,
      "learning_rate": 1.2930765438477327e-06,
      "loss": 0.8277,
      "step": 8684
    },
    {
      "epoch": 0.6266007719779229,
      "grad_norm": 3.121342464081148,
      "learning_rate": 1.2926393718907436e-06,
      "loss": 0.9,
      "step": 8685
    },
    {
      "epoch": 0.6266729194473504,
      "grad_norm": 5.176433110483941,
      "learning_rate": 1.2922022385601e-06,
      "loss": 1.0256,
      "step": 8686
    },
    {
      "epoch": 0.6267450669167779,
      "grad_norm": 2.833401641770264,
      "learning_rate": 1.291765143879672e-06,
      "loss": 0.8573,
      "step": 8687
    },
    {
      "epoch": 0.6268172143862054,
      "grad_norm": 6.0000486371930055,
      "learning_rate": 1.2913280878733278e-06,
      "loss": 0.9882,
      "step": 8688
    },
    {
      "epoch": 0.6268893618556329,
      "grad_norm": 3.8764975330547156,
      "learning_rate": 1.2908910705649333e-06,
      "loss": 0.9536,
      "step": 8689
    },
    {
      "epoch": 0.6269615093250605,
      "grad_norm": 3.314768032743902,
      "learning_rate": 1.2904540919783516e-06,
      "loss": 0.9123,
      "step": 8690
    },
    {
      "epoch": 0.627033656794488,
      "grad_norm": 3.117120192334109,
      "learning_rate": 1.2900171521374462e-06,
      "loss": 0.8828,
      "step": 8691
    },
    {
      "epoch": 0.6271058042639155,
      "grad_norm": 3.778256200590077,
      "learning_rate": 1.2895802510660761e-06,
      "loss": 0.9206,
      "step": 8692
    },
    {
      "epoch": 0.6271779517333429,
      "grad_norm": 5.525182639853216,
      "learning_rate": 1.2891433887880974e-06,
      "loss": 0.8967,
      "step": 8693
    },
    {
      "epoch": 0.6272500992027704,
      "grad_norm": 2.9312383590499627,
      "learning_rate": 1.288706565327368e-06,
      "loss": 0.9745,
      "step": 8694
    },
    {
      "epoch": 0.6273222466721979,
      "grad_norm": 3.093836773512807,
      "learning_rate": 1.2882697807077389e-06,
      "loss": 0.9994,
      "step": 8695
    },
    {
      "epoch": 0.6273943941416255,
      "grad_norm": 0.7473750272152072,
      "learning_rate": 1.2878330349530628e-06,
      "loss": 0.8292,
      "step": 8696
    },
    {
      "epoch": 0.627466541611053,
      "grad_norm": 3.6483347712886154,
      "learning_rate": 1.2873963280871887e-06,
      "loss": 1.0344,
      "step": 8697
    },
    {
      "epoch": 0.6275386890804805,
      "grad_norm": 3.602118525747752,
      "learning_rate": 1.2869596601339628e-06,
      "loss": 1.0503,
      "step": 8698
    },
    {
      "epoch": 0.627610836549908,
      "grad_norm": 3.055603982560925,
      "learning_rate": 1.2865230311172299e-06,
      "loss": 0.9843,
      "step": 8699
    },
    {
      "epoch": 0.6276829840193355,
      "grad_norm": 0.6312457339454248,
      "learning_rate": 1.2860864410608334e-06,
      "loss": 0.8044,
      "step": 8700
    },
    {
      "epoch": 0.627755131488763,
      "grad_norm": 0.735504094043076,
      "learning_rate": 1.285649889988614e-06,
      "loss": 0.832,
      "step": 8701
    },
    {
      "epoch": 0.6278272789581906,
      "grad_norm": 3.0432550298081553,
      "learning_rate": 1.2852133779244091e-06,
      "loss": 0.9242,
      "step": 8702
    },
    {
      "epoch": 0.6278994264276181,
      "grad_norm": 2.5591957374179515,
      "learning_rate": 1.2847769048920555e-06,
      "loss": 0.9734,
      "step": 8703
    },
    {
      "epoch": 0.6279715738970456,
      "grad_norm": 3.673981666905447,
      "learning_rate": 1.284340470915388e-06,
      "loss": 0.9916,
      "step": 8704
    },
    {
      "epoch": 0.628043721366473,
      "grad_norm": 3.05500124517687,
      "learning_rate": 1.2839040760182371e-06,
      "loss": 0.9577,
      "step": 8705
    },
    {
      "epoch": 0.6281158688359005,
      "grad_norm": 2.927959614157311,
      "learning_rate": 1.283467720224435e-06,
      "loss": 0.8978,
      "step": 8706
    },
    {
      "epoch": 0.6281880163053281,
      "grad_norm": 4.347393895105345,
      "learning_rate": 1.2830314035578073e-06,
      "loss": 0.8889,
      "step": 8707
    },
    {
      "epoch": 0.6282601637747556,
      "grad_norm": 3.4266381404545294,
      "learning_rate": 1.2825951260421809e-06,
      "loss": 0.8563,
      "step": 8708
    },
    {
      "epoch": 0.6283323112441831,
      "grad_norm": 5.200595916934458,
      "learning_rate": 1.2821588877013789e-06,
      "loss": 1.0357,
      "step": 8709
    },
    {
      "epoch": 0.6284044587136106,
      "grad_norm": 12.98671043132742,
      "learning_rate": 1.2817226885592227e-06,
      "loss": 0.9726,
      "step": 8710
    },
    {
      "epoch": 0.6284766061830381,
      "grad_norm": 4.520596522913058,
      "learning_rate": 1.2812865286395314e-06,
      "loss": 0.9597,
      "step": 8711
    },
    {
      "epoch": 0.6285487536524657,
      "grad_norm": 3.392503912184841,
      "learning_rate": 1.2808504079661218e-06,
      "loss": 0.9494,
      "step": 8712
    },
    {
      "epoch": 0.6286209011218932,
      "grad_norm": 5.154799378687931,
      "learning_rate": 1.28041432656281e-06,
      "loss": 0.9331,
      "step": 8713
    },
    {
      "epoch": 0.6286930485913207,
      "grad_norm": 2.6774201434824048,
      "learning_rate": 1.2799782844534077e-06,
      "loss": 0.8844,
      "step": 8714
    },
    {
      "epoch": 0.6287651960607482,
      "grad_norm": 2.8461198259489477,
      "learning_rate": 1.2795422816617256e-06,
      "loss": 0.9366,
      "step": 8715
    },
    {
      "epoch": 0.6288373435301757,
      "grad_norm": 2.338853833632285,
      "learning_rate": 1.279106318211573e-06,
      "loss": 0.9519,
      "step": 8716
    },
    {
      "epoch": 0.6289094909996031,
      "grad_norm": 4.015119587600645,
      "learning_rate": 1.2786703941267548e-06,
      "loss": 0.9773,
      "step": 8717
    },
    {
      "epoch": 0.6289816384690307,
      "grad_norm": 2.5997011783247803,
      "learning_rate": 1.278234509431077e-06,
      "loss": 0.8722,
      "step": 8718
    },
    {
      "epoch": 0.6290537859384582,
      "grad_norm": 3.313464474171141,
      "learning_rate": 1.2777986641483406e-06,
      "loss": 0.9475,
      "step": 8719
    },
    {
      "epoch": 0.6291259334078857,
      "grad_norm": 4.849400921818808,
      "learning_rate": 1.277362858302345e-06,
      "loss": 1.0145,
      "step": 8720
    },
    {
      "epoch": 0.6291980808773132,
      "grad_norm": 12.400344942278824,
      "learning_rate": 1.2769270919168894e-06,
      "loss": 0.9926,
      "step": 8721
    },
    {
      "epoch": 0.6292702283467407,
      "grad_norm": 3.777009462004857,
      "learning_rate": 1.276491365015768e-06,
      "loss": 0.83,
      "step": 8722
    },
    {
      "epoch": 0.6293423758161683,
      "grad_norm": 2.9576425059175184,
      "learning_rate": 1.276055677622775e-06,
      "loss": 0.923,
      "step": 8723
    },
    {
      "epoch": 0.6294145232855958,
      "grad_norm": 3.2564792603602957,
      "learning_rate": 1.2756200297617018e-06,
      "loss": 1.0886,
      "step": 8724
    },
    {
      "epoch": 0.6294866707550233,
      "grad_norm": 2.5577752820792767,
      "learning_rate": 1.2751844214563369e-06,
      "loss": 0.8848,
      "step": 8725
    },
    {
      "epoch": 0.6295588182244508,
      "grad_norm": 5.506598069515728,
      "learning_rate": 1.2747488527304677e-06,
      "loss": 0.9841,
      "step": 8726
    },
    {
      "epoch": 0.6296309656938783,
      "grad_norm": 2.6653540380754133,
      "learning_rate": 1.2743133236078785e-06,
      "loss": 0.9411,
      "step": 8727
    },
    {
      "epoch": 0.6297031131633058,
      "grad_norm": 3.869518463772914,
      "learning_rate": 1.273877834112353e-06,
      "loss": 0.988,
      "step": 8728
    },
    {
      "epoch": 0.6297752606327333,
      "grad_norm": 3.999597767633196,
      "learning_rate": 1.2734423842676702e-06,
      "loss": 0.8954,
      "step": 8729
    },
    {
      "epoch": 0.6298474081021608,
      "grad_norm": 4.2141909230727075,
      "learning_rate": 1.2730069740976098e-06,
      "loss": 0.9385,
      "step": 8730
    },
    {
      "epoch": 0.6299195555715883,
      "grad_norm": 4.632033284161617,
      "learning_rate": 1.2725716036259473e-06,
      "loss": 0.8701,
      "step": 8731
    },
    {
      "epoch": 0.6299917030410158,
      "grad_norm": 3.3975057091850944,
      "learning_rate": 1.2721362728764558e-06,
      "loss": 0.8839,
      "step": 8732
    },
    {
      "epoch": 0.6300638505104433,
      "grad_norm": 2.9101875457104986,
      "learning_rate": 1.2717009818729093e-06,
      "loss": 1.0136,
      "step": 8733
    },
    {
      "epoch": 0.6301359979798709,
      "grad_norm": 4.024296878033333,
      "learning_rate": 1.2712657306390751e-06,
      "loss": 1.0615,
      "step": 8734
    },
    {
      "epoch": 0.6302081454492984,
      "grad_norm": 3.650046539009945,
      "learning_rate": 1.2708305191987218e-06,
      "loss": 1.0347,
      "step": 8735
    },
    {
      "epoch": 0.6302802929187259,
      "grad_norm": 3.82206040164283,
      "learning_rate": 1.2703953475756151e-06,
      "loss": 0.9188,
      "step": 8736
    },
    {
      "epoch": 0.6303524403881534,
      "grad_norm": 2.703532899278588,
      "learning_rate": 1.2699602157935169e-06,
      "loss": 0.9814,
      "step": 8737
    },
    {
      "epoch": 0.6304245878575809,
      "grad_norm": 11.379957523099067,
      "learning_rate": 1.269525123876189e-06,
      "loss": 0.8905,
      "step": 8738
    },
    {
      "epoch": 0.6304967353270085,
      "grad_norm": 3.3236759159458193,
      "learning_rate": 1.2690900718473901e-06,
      "loss": 1.0126,
      "step": 8739
    },
    {
      "epoch": 0.630568882796436,
      "grad_norm": 3.792927514902016,
      "learning_rate": 1.268655059730877e-06,
      "loss": 0.9416,
      "step": 8740
    },
    {
      "epoch": 0.6306410302658634,
      "grad_norm": 2.917840303711611,
      "learning_rate": 1.2682200875504032e-06,
      "loss": 0.9319,
      "step": 8741
    },
    {
      "epoch": 0.6307131777352909,
      "grad_norm": 3.203422718633589,
      "learning_rate": 1.2677851553297213e-06,
      "loss": 1.0215,
      "step": 8742
    },
    {
      "epoch": 0.6307853252047184,
      "grad_norm": 6.438835227785859,
      "learning_rate": 1.2673502630925822e-06,
      "loss": 1.0025,
      "step": 8743
    },
    {
      "epoch": 0.6308574726741459,
      "grad_norm": 3.779575071519607,
      "learning_rate": 1.2669154108627322e-06,
      "loss": 1.0052,
      "step": 8744
    },
    {
      "epoch": 0.6309296201435735,
      "grad_norm": 2.5326755418406908,
      "learning_rate": 1.2664805986639186e-06,
      "loss": 1.0382,
      "step": 8745
    },
    {
      "epoch": 0.631001767613001,
      "grad_norm": 3.624234875139273,
      "learning_rate": 1.266045826519884e-06,
      "loss": 0.9505,
      "step": 8746
    },
    {
      "epoch": 0.6310739150824285,
      "grad_norm": 5.461052409042294,
      "learning_rate": 1.2656110944543695e-06,
      "loss": 0.9538,
      "step": 8747
    },
    {
      "epoch": 0.631146062551856,
      "grad_norm": 2.7643268361769837,
      "learning_rate": 1.2651764024911147e-06,
      "loss": 0.9764,
      "step": 8748
    },
    {
      "epoch": 0.6312182100212835,
      "grad_norm": 6.606864434344676,
      "learning_rate": 1.2647417506538563e-06,
      "loss": 0.9019,
      "step": 8749
    },
    {
      "epoch": 0.631290357490711,
      "grad_norm": 3.4084391427501743,
      "learning_rate": 1.2643071389663286e-06,
      "loss": 0.9219,
      "step": 8750
    },
    {
      "epoch": 0.6313625049601386,
      "grad_norm": 7.4569816287964015,
      "learning_rate": 1.2638725674522652e-06,
      "loss": 0.9641,
      "step": 8751
    },
    {
      "epoch": 0.631434652429566,
      "grad_norm": 4.745655080357572,
      "learning_rate": 1.2634380361353958e-06,
      "loss": 0.8987,
      "step": 8752
    },
    {
      "epoch": 0.6315067998989935,
      "grad_norm": 3.345007829186553,
      "learning_rate": 1.2630035450394485e-06,
      "loss": 0.9201,
      "step": 8753
    },
    {
      "epoch": 0.631578947368421,
      "grad_norm": 3.2783763109139734,
      "learning_rate": 1.2625690941881492e-06,
      "loss": 0.9549,
      "step": 8754
    },
    {
      "epoch": 0.6316510948378485,
      "grad_norm": 3.0709439082188976,
      "learning_rate": 1.2621346836052223e-06,
      "loss": 0.8875,
      "step": 8755
    },
    {
      "epoch": 0.6317232423072761,
      "grad_norm": 4.4927377380404145,
      "learning_rate": 1.2617003133143878e-06,
      "loss": 0.9216,
      "step": 8756
    },
    {
      "epoch": 0.6317953897767036,
      "grad_norm": 3.1338806614456627,
      "learning_rate": 1.261265983339367e-06,
      "loss": 1.0343,
      "step": 8757
    },
    {
      "epoch": 0.6318675372461311,
      "grad_norm": 6.291104457906654,
      "learning_rate": 1.260831693703876e-06,
      "loss": 0.9898,
      "step": 8758
    },
    {
      "epoch": 0.6319396847155586,
      "grad_norm": 4.194418813754853,
      "learning_rate": 1.2603974444316294e-06,
      "loss": 0.8978,
      "step": 8759
    },
    {
      "epoch": 0.6320118321849861,
      "grad_norm": 5.157270480812099,
      "learning_rate": 1.2599632355463413e-06,
      "loss": 0.9604,
      "step": 8760
    },
    {
      "epoch": 0.6320839796544137,
      "grad_norm": 3.7207996265782737,
      "learning_rate": 1.2595290670717207e-06,
      "loss": 0.9489,
      "step": 8761
    },
    {
      "epoch": 0.6321561271238412,
      "grad_norm": 3.038678220954856,
      "learning_rate": 1.2590949390314767e-06,
      "loss": 0.9439,
      "step": 8762
    },
    {
      "epoch": 0.6322282745932687,
      "grad_norm": 3.883155608032762,
      "learning_rate": 1.2586608514493157e-06,
      "loss": 0.971,
      "step": 8763
    },
    {
      "epoch": 0.6323004220626961,
      "grad_norm": 2.5834205571698,
      "learning_rate": 1.258226804348941e-06,
      "loss": 0.9359,
      "step": 8764
    },
    {
      "epoch": 0.6323725695321236,
      "grad_norm": 3.6493970764924475,
      "learning_rate": 1.2577927977540546e-06,
      "loss": 1.0725,
      "step": 8765
    },
    {
      "epoch": 0.6324447170015511,
      "grad_norm": 3.1140073075597696,
      "learning_rate": 1.2573588316883556e-06,
      "loss": 1.0383,
      "step": 8766
    },
    {
      "epoch": 0.6325168644709787,
      "grad_norm": 0.8200137729767418,
      "learning_rate": 1.2569249061755422e-06,
      "loss": 0.8577,
      "step": 8767
    },
    {
      "epoch": 0.6325890119404062,
      "grad_norm": 3.1554694390821543,
      "learning_rate": 1.256491021239309e-06,
      "loss": 1.0426,
      "step": 8768
    },
    {
      "epoch": 0.6326611594098337,
      "grad_norm": 3.737107302853638,
      "learning_rate": 1.2560571769033482e-06,
      "loss": 0.8975,
      "step": 8769
    },
    {
      "epoch": 0.6327333068792612,
      "grad_norm": 1.0363692176629011,
      "learning_rate": 1.2556233731913514e-06,
      "loss": 0.8336,
      "step": 8770
    },
    {
      "epoch": 0.6328054543486887,
      "grad_norm": 3.345951817094337,
      "learning_rate": 1.2551896101270059e-06,
      "loss": 0.8953,
      "step": 8771
    },
    {
      "epoch": 0.6328776018181163,
      "grad_norm": 8.786717569129467,
      "learning_rate": 1.2547558877339994e-06,
      "loss": 0.9398,
      "step": 8772
    },
    {
      "epoch": 0.6329497492875438,
      "grad_norm": 3.48193314867521,
      "learning_rate": 1.2543222060360149e-06,
      "loss": 0.9749,
      "step": 8773
    },
    {
      "epoch": 0.6330218967569713,
      "grad_norm": 3.983191939096287,
      "learning_rate": 1.2538885650567342e-06,
      "loss": 1.1134,
      "step": 8774
    },
    {
      "epoch": 0.6330940442263988,
      "grad_norm": 2.854760119887843,
      "learning_rate": 1.253454964819837e-06,
      "loss": 1.0365,
      "step": 8775
    },
    {
      "epoch": 0.6331661916958262,
      "grad_norm": 2.5499985152595537,
      "learning_rate": 1.2530214053490004e-06,
      "loss": 0.9294,
      "step": 8776
    },
    {
      "epoch": 0.6332383391652537,
      "grad_norm": 3.5583813600775787,
      "learning_rate": 1.2525878866678999e-06,
      "loss": 0.8985,
      "step": 8777
    },
    {
      "epoch": 0.6333104866346813,
      "grad_norm": 2.385075529490014,
      "learning_rate": 1.2521544088002077e-06,
      "loss": 0.8708,
      "step": 8778
    },
    {
      "epoch": 0.6333826341041088,
      "grad_norm": 3.321372475567888,
      "learning_rate": 1.2517209717695955e-06,
      "loss": 0.8736,
      "step": 8779
    },
    {
      "epoch": 0.6334547815735363,
      "grad_norm": 14.62783551238336,
      "learning_rate": 1.2512875755997306e-06,
      "loss": 0.9748,
      "step": 8780
    },
    {
      "epoch": 0.6335269290429638,
      "grad_norm": 4.235751558202792,
      "learning_rate": 1.2508542203142792e-06,
      "loss": 0.9529,
      "step": 8781
    },
    {
      "epoch": 0.6335990765123913,
      "grad_norm": 4.6884036401132265,
      "learning_rate": 1.2504209059369062e-06,
      "loss": 0.952,
      "step": 8782
    },
    {
      "epoch": 0.6336712239818189,
      "grad_norm": 3.633939047437131,
      "learning_rate": 1.2499876324912716e-06,
      "loss": 1.0222,
      "step": 8783
    },
    {
      "epoch": 0.6337433714512464,
      "grad_norm": 4.203449676339489,
      "learning_rate": 1.2495544000010374e-06,
      "loss": 0.8913,
      "step": 8784
    },
    {
      "epoch": 0.6338155189206739,
      "grad_norm": 4.55735007403048,
      "learning_rate": 1.2491212084898587e-06,
      "loss": 1.0289,
      "step": 8785
    },
    {
      "epoch": 0.6338876663901014,
      "grad_norm": 2.8888592249822658,
      "learning_rate": 1.2486880579813908e-06,
      "loss": 0.8739,
      "step": 8786
    },
    {
      "epoch": 0.6339598138595289,
      "grad_norm": 3.7958605003571178,
      "learning_rate": 1.2482549484992872e-06,
      "loss": 1.033,
      "step": 8787
    },
    {
      "epoch": 0.6340319613289563,
      "grad_norm": 7.921462068950053,
      "learning_rate": 1.247821880067198e-06,
      "loss": 1.0055,
      "step": 8788
    },
    {
      "epoch": 0.6341041087983839,
      "grad_norm": 4.4072056809346325,
      "learning_rate": 1.2473888527087712e-06,
      "loss": 0.8459,
      "step": 8789
    },
    {
      "epoch": 0.6341762562678114,
      "grad_norm": 4.217287615463637,
      "learning_rate": 1.2469558664476536e-06,
      "loss": 1.0423,
      "step": 8790
    },
    {
      "epoch": 0.6342484037372389,
      "grad_norm": 4.977172814138742,
      "learning_rate": 1.246522921307488e-06,
      "loss": 1.0312,
      "step": 8791
    },
    {
      "epoch": 0.6343205512066664,
      "grad_norm": 5.0693040500285385,
      "learning_rate": 1.2460900173119164e-06,
      "loss": 0.979,
      "step": 8792
    },
    {
      "epoch": 0.6343926986760939,
      "grad_norm": 2.880408300226408,
      "learning_rate": 1.2456571544845782e-06,
      "loss": 0.9853,
      "step": 8793
    },
    {
      "epoch": 0.6344648461455215,
      "grad_norm": 5.235328701305943,
      "learning_rate": 1.2452243328491106e-06,
      "loss": 1.0398,
      "step": 8794
    },
    {
      "epoch": 0.634536993614949,
      "grad_norm": 8.619041264227311,
      "learning_rate": 1.2447915524291473e-06,
      "loss": 0.9568,
      "step": 8795
    },
    {
      "epoch": 0.6346091410843765,
      "grad_norm": 2.9654410141368848,
      "learning_rate": 1.2443588132483224e-06,
      "loss": 0.9317,
      "step": 8796
    },
    {
      "epoch": 0.634681288553804,
      "grad_norm": 2.9592604102373525,
      "learning_rate": 1.2439261153302654e-06,
      "loss": 1.0025,
      "step": 8797
    },
    {
      "epoch": 0.6347534360232315,
      "grad_norm": 0.69432621558743,
      "learning_rate": 1.2434934586986036e-06,
      "loss": 0.8298,
      "step": 8798
    },
    {
      "epoch": 0.6348255834926589,
      "grad_norm": 11.795345384069607,
      "learning_rate": 1.2430608433769646e-06,
      "loss": 0.8964,
      "step": 8799
    },
    {
      "epoch": 0.6348977309620865,
      "grad_norm": 3.106769486091537,
      "learning_rate": 1.24262826938897e-06,
      "loss": 0.9165,
      "step": 8800
    },
    {
      "epoch": 0.634969878431514,
      "grad_norm": 3.7927158008150244,
      "learning_rate": 1.242195736758242e-06,
      "loss": 0.9811,
      "step": 8801
    },
    {
      "epoch": 0.6350420259009415,
      "grad_norm": 3.2934167450857967,
      "learning_rate": 1.2417632455083999e-06,
      "loss": 0.8977,
      "step": 8802
    },
    {
      "epoch": 0.635114173370369,
      "grad_norm": 3.0968609206171185,
      "learning_rate": 1.2413307956630598e-06,
      "loss": 1.126,
      "step": 8803
    },
    {
      "epoch": 0.6351863208397965,
      "grad_norm": 3.8065461743509457,
      "learning_rate": 1.2408983872458367e-06,
      "loss": 0.8914,
      "step": 8804
    },
    {
      "epoch": 0.6352584683092241,
      "grad_norm": 6.493233313028408,
      "learning_rate": 1.2404660202803423e-06,
      "loss": 1.1394,
      "step": 8805
    },
    {
      "epoch": 0.6353306157786516,
      "grad_norm": 0.670815230849954,
      "learning_rate": 1.2400336947901871e-06,
      "loss": 0.7713,
      "step": 8806
    },
    {
      "epoch": 0.6354027632480791,
      "grad_norm": 5.685501250461048,
      "learning_rate": 1.2396014107989786e-06,
      "loss": 0.9568,
      "step": 8807
    },
    {
      "epoch": 0.6354749107175066,
      "grad_norm": 3.6532014054008646,
      "learning_rate": 1.2391691683303216e-06,
      "loss": 1.1529,
      "step": 8808
    },
    {
      "epoch": 0.6355470581869341,
      "grad_norm": 3.8067799166408656,
      "learning_rate": 1.2387369674078204e-06,
      "loss": 1.0464,
      "step": 8809
    },
    {
      "epoch": 0.6356192056563617,
      "grad_norm": 2.8830897086824314,
      "learning_rate": 1.2383048080550747e-06,
      "loss": 1.0353,
      "step": 8810
    },
    {
      "epoch": 0.6356913531257891,
      "grad_norm": 4.621079922910949,
      "learning_rate": 1.2378726902956842e-06,
      "loss": 1.0829,
      "step": 8811
    },
    {
      "epoch": 0.6357635005952166,
      "grad_norm": 3.654058989993998,
      "learning_rate": 1.237440614153245e-06,
      "loss": 0.9564,
      "step": 8812
    },
    {
      "epoch": 0.6358356480646441,
      "grad_norm": 3.924082222321115,
      "learning_rate": 1.23700857965135e-06,
      "loss": 1.0375,
      "step": 8813
    },
    {
      "epoch": 0.6359077955340716,
      "grad_norm": 3.812561034667695,
      "learning_rate": 1.236576586813593e-06,
      "loss": 1.0024,
      "step": 8814
    },
    {
      "epoch": 0.6359799430034991,
      "grad_norm": 3.6632407106648697,
      "learning_rate": 1.2361446356635622e-06,
      "loss": 0.9555,
      "step": 8815
    },
    {
      "epoch": 0.6360520904729267,
      "grad_norm": 3.5541386777991417,
      "learning_rate": 1.2357127262248448e-06,
      "loss": 0.914,
      "step": 8816
    },
    {
      "epoch": 0.6361242379423542,
      "grad_norm": 3.833721818450343,
      "learning_rate": 1.2352808585210261e-06,
      "loss": 0.9309,
      "step": 8817
    },
    {
      "epoch": 0.6361963854117817,
      "grad_norm": 3.1042001880312178,
      "learning_rate": 1.2348490325756898e-06,
      "loss": 0.8006,
      "step": 8818
    },
    {
      "epoch": 0.6362685328812092,
      "grad_norm": 3.3570311420485934,
      "learning_rate": 1.2344172484124147e-06,
      "loss": 0.9456,
      "step": 8819
    },
    {
      "epoch": 0.6363406803506367,
      "grad_norm": 0.8640134304247862,
      "learning_rate": 1.233985506054779e-06,
      "loss": 0.8868,
      "step": 8820
    },
    {
      "epoch": 0.6364128278200643,
      "grad_norm": 3.089346169243254,
      "learning_rate": 1.23355380552636e-06,
      "loss": 0.9234,
      "step": 8821
    },
    {
      "epoch": 0.6364849752894918,
      "grad_norm": 3.1863541881098154,
      "learning_rate": 1.2331221468507296e-06,
      "loss": 0.9805,
      "step": 8822
    },
    {
      "epoch": 0.6365571227589192,
      "grad_norm": 3.0548587370582863,
      "learning_rate": 1.2326905300514605e-06,
      "loss": 0.8526,
      "step": 8823
    },
    {
      "epoch": 0.6366292702283467,
      "grad_norm": 3.33898286487973,
      "learning_rate": 1.232258955152121e-06,
      "loss": 0.9228,
      "step": 8824
    },
    {
      "epoch": 0.6367014176977742,
      "grad_norm": 3.570033490320403,
      "learning_rate": 1.2318274221762775e-06,
      "loss": 0.9583,
      "step": 8825
    },
    {
      "epoch": 0.6367735651672017,
      "grad_norm": 4.024591906218155,
      "learning_rate": 1.2313959311474953e-06,
      "loss": 0.954,
      "step": 8826
    },
    {
      "epoch": 0.6368457126366293,
      "grad_norm": 3.768465482432492,
      "learning_rate": 1.2309644820893356e-06,
      "loss": 1.026,
      "step": 8827
    },
    {
      "epoch": 0.6369178601060568,
      "grad_norm": 3.164959966276788,
      "learning_rate": 1.2305330750253588e-06,
      "loss": 1.0339,
      "step": 8828
    },
    {
      "epoch": 0.6369900075754843,
      "grad_norm": 2.8696670828268154,
      "learning_rate": 1.230101709979123e-06,
      "loss": 1.0227,
      "step": 8829
    },
    {
      "epoch": 0.6370621550449118,
      "grad_norm": 4.851571542236944,
      "learning_rate": 1.2296703869741815e-06,
      "loss": 0.8942,
      "step": 8830
    },
    {
      "epoch": 0.6371343025143393,
      "grad_norm": 4.779139034032826,
      "learning_rate": 1.2292391060340897e-06,
      "loss": 0.8581,
      "step": 8831
    },
    {
      "epoch": 0.6372064499837669,
      "grad_norm": 3.876472439502021,
      "learning_rate": 1.2288078671823959e-06,
      "loss": 0.9507,
      "step": 8832
    },
    {
      "epoch": 0.6372785974531944,
      "grad_norm": 3.891966316618424,
      "learning_rate": 1.2283766704426506e-06,
      "loss": 0.8497,
      "step": 8833
    },
    {
      "epoch": 0.6373507449226219,
      "grad_norm": 3.3344021196343974,
      "learning_rate": 1.2279455158383987e-06,
      "loss": 0.887,
      "step": 8834
    },
    {
      "epoch": 0.6374228923920493,
      "grad_norm": 3.3488750219982633,
      "learning_rate": 1.2275144033931835e-06,
      "loss": 1.0399,
      "step": 8835
    },
    {
      "epoch": 0.6374950398614768,
      "grad_norm": 3.6319782570763675,
      "learning_rate": 1.2270833331305475e-06,
      "loss": 1.0149,
      "step": 8836
    },
    {
      "epoch": 0.6375671873309043,
      "grad_norm": 4.492644762732897,
      "learning_rate": 1.2266523050740291e-06,
      "loss": 0.9652,
      "step": 8837
    },
    {
      "epoch": 0.6376393348003319,
      "grad_norm": 4.470297805457979,
      "learning_rate": 1.226221319247166e-06,
      "loss": 0.944,
      "step": 8838
    },
    {
      "epoch": 0.6377114822697594,
      "grad_norm": 0.7493561921929134,
      "learning_rate": 1.2257903756734919e-06,
      "loss": 0.8441,
      "step": 8839
    },
    {
      "epoch": 0.6377836297391869,
      "grad_norm": 5.048758046237534,
      "learning_rate": 1.2253594743765395e-06,
      "loss": 1.0815,
      "step": 8840
    },
    {
      "epoch": 0.6378557772086144,
      "grad_norm": 28.96171199564608,
      "learning_rate": 1.2249286153798389e-06,
      "loss": 0.9254,
      "step": 8841
    },
    {
      "epoch": 0.6379279246780419,
      "grad_norm": 3.4188487985342038,
      "learning_rate": 1.224497798706917e-06,
      "loss": 0.9478,
      "step": 8842
    },
    {
      "epoch": 0.6380000721474695,
      "grad_norm": 5.587706129807724,
      "learning_rate": 1.2240670243812998e-06,
      "loss": 0.9427,
      "step": 8843
    },
    {
      "epoch": 0.638072219616897,
      "grad_norm": 3.998165186634882,
      "learning_rate": 1.2236362924265092e-06,
      "loss": 0.8956,
      "step": 8844
    },
    {
      "epoch": 0.6381443670863245,
      "grad_norm": 3.3954482689091847,
      "learning_rate": 1.2232056028660675e-06,
      "loss": 0.9168,
      "step": 8845
    },
    {
      "epoch": 0.638216514555752,
      "grad_norm": 2.585531980866698,
      "learning_rate": 1.2227749557234924e-06,
      "loss": 0.8968,
      "step": 8846
    },
    {
      "epoch": 0.6382886620251794,
      "grad_norm": 3.9715733135189337,
      "learning_rate": 1.2223443510222988e-06,
      "loss": 0.9384,
      "step": 8847
    },
    {
      "epoch": 0.6383608094946069,
      "grad_norm": 2.919846590532395,
      "learning_rate": 1.221913788786002e-06,
      "loss": 0.9572,
      "step": 8848
    },
    {
      "epoch": 0.6384329569640345,
      "grad_norm": 3.2231163852377542,
      "learning_rate": 1.2214832690381125e-06,
      "loss": 0.8836,
      "step": 8849
    },
    {
      "epoch": 0.638505104433462,
      "grad_norm": 3.859673986552582,
      "learning_rate": 1.2210527918021402e-06,
      "loss": 0.9385,
      "step": 8850
    },
    {
      "epoch": 0.6385772519028895,
      "grad_norm": 4.133319902910952,
      "learning_rate": 1.2206223571015916e-06,
      "loss": 1.0289,
      "step": 8851
    },
    {
      "epoch": 0.638649399372317,
      "grad_norm": 6.421685209045385,
      "learning_rate": 1.2201919649599698e-06,
      "loss": 0.9561,
      "step": 8852
    },
    {
      "epoch": 0.6387215468417445,
      "grad_norm": 2.6316800271247334,
      "learning_rate": 1.219761615400779e-06,
      "loss": 0.9421,
      "step": 8853
    },
    {
      "epoch": 0.6387936943111721,
      "grad_norm": 5.52906126694689,
      "learning_rate": 1.2193313084475176e-06,
      "loss": 0.9845,
      "step": 8854
    },
    {
      "epoch": 0.6388658417805996,
      "grad_norm": 2.624977565851168,
      "learning_rate": 1.218901044123684e-06,
      "loss": 0.8782,
      "step": 8855
    },
    {
      "epoch": 0.6389379892500271,
      "grad_norm": 3.122015939763474,
      "learning_rate": 1.2184708224527725e-06,
      "loss": 0.9842,
      "step": 8856
    },
    {
      "epoch": 0.6390101367194546,
      "grad_norm": 3.966751797080467,
      "learning_rate": 1.2180406434582761e-06,
      "loss": 0.9287,
      "step": 8857
    },
    {
      "epoch": 0.639082284188882,
      "grad_norm": 5.143068793649462,
      "learning_rate": 1.2176105071636863e-06,
      "loss": 0.8948,
      "step": 8858
    },
    {
      "epoch": 0.6391544316583095,
      "grad_norm": 3.059484749092834,
      "learning_rate": 1.217180413592489e-06,
      "loss": 0.9526,
      "step": 8859
    },
    {
      "epoch": 0.6392265791277371,
      "grad_norm": 3.720927009955386,
      "learning_rate": 1.2167503627681726e-06,
      "loss": 0.9149,
      "step": 8860
    },
    {
      "epoch": 0.6392987265971646,
      "grad_norm": 2.7781612947408125,
      "learning_rate": 1.2163203547142182e-06,
      "loss": 0.992,
      "step": 8861
    },
    {
      "epoch": 0.6393708740665921,
      "grad_norm": 2.7258930422817778,
      "learning_rate": 1.2158903894541092e-06,
      "loss": 0.9726,
      "step": 8862
    },
    {
      "epoch": 0.6394430215360196,
      "grad_norm": 8.408820344288463,
      "learning_rate": 1.2154604670113232e-06,
      "loss": 0.939,
      "step": 8863
    },
    {
      "epoch": 0.6395151690054471,
      "grad_norm": 6.64462913517332,
      "learning_rate": 1.2150305874093362e-06,
      "loss": 0.9221,
      "step": 8864
    },
    {
      "epoch": 0.6395873164748747,
      "grad_norm": 2.8016831822525976,
      "learning_rate": 1.2146007506716238e-06,
      "loss": 0.9466,
      "step": 8865
    },
    {
      "epoch": 0.6396594639443022,
      "grad_norm": 7.845683676382417,
      "learning_rate": 1.2141709568216567e-06,
      "loss": 0.9362,
      "step": 8866
    },
    {
      "epoch": 0.6397316114137297,
      "grad_norm": 2.904147792701926,
      "learning_rate": 1.2137412058829046e-06,
      "loss": 0.832,
      "step": 8867
    },
    {
      "epoch": 0.6398037588831572,
      "grad_norm": 4.821390634215929,
      "learning_rate": 1.2133114978788346e-06,
      "loss": 0.8616,
      "step": 8868
    },
    {
      "epoch": 0.6398759063525847,
      "grad_norm": 2.8132746159571442,
      "learning_rate": 1.212881832832911e-06,
      "loss": 0.9964,
      "step": 8869
    },
    {
      "epoch": 0.6399480538220121,
      "grad_norm": 2.617657155271474,
      "learning_rate": 1.2124522107685976e-06,
      "loss": 0.9925,
      "step": 8870
    },
    {
      "epoch": 0.6400202012914397,
      "grad_norm": 4.126033306747375,
      "learning_rate": 1.2120226317093524e-06,
      "loss": 1.0043,
      "step": 8871
    },
    {
      "epoch": 0.6400923487608672,
      "grad_norm": 5.6825536661093565,
      "learning_rate": 1.2115930956786354e-06,
      "loss": 1.064,
      "step": 8872
    },
    {
      "epoch": 0.6401644962302947,
      "grad_norm": 4.8585557645695125,
      "learning_rate": 1.2111636026999006e-06,
      "loss": 0.9112,
      "step": 8873
    },
    {
      "epoch": 0.6402366436997222,
      "grad_norm": 0.7581193440376502,
      "learning_rate": 1.2107341527966007e-06,
      "loss": 0.8485,
      "step": 8874
    },
    {
      "epoch": 0.6403087911691497,
      "grad_norm": 3.894103062085297,
      "learning_rate": 1.2103047459921878e-06,
      "loss": 0.934,
      "step": 8875
    },
    {
      "epoch": 0.6403809386385773,
      "grad_norm": 3.4220048714295968,
      "learning_rate": 1.2098753823101082e-06,
      "loss": 0.8883,
      "step": 8876
    },
    {
      "epoch": 0.6404530861080048,
      "grad_norm": 3.6903752900611804,
      "learning_rate": 1.2094460617738103e-06,
      "loss": 1.0491,
      "step": 8877
    },
    {
      "epoch": 0.6405252335774323,
      "grad_norm": 3.79930420076797,
      "learning_rate": 1.209016784406736e-06,
      "loss": 0.8602,
      "step": 8878
    },
    {
      "epoch": 0.6405973810468598,
      "grad_norm": 3.821162029897192,
      "learning_rate": 1.2085875502323268e-06,
      "loss": 0.9539,
      "step": 8879
    },
    {
      "epoch": 0.6406695285162873,
      "grad_norm": 3.4188875718131957,
      "learning_rate": 1.2081583592740217e-06,
      "loss": 0.8813,
      "step": 8880
    },
    {
      "epoch": 0.6407416759857149,
      "grad_norm": 3.5590125966975936,
      "learning_rate": 1.2077292115552573e-06,
      "loss": 0.9909,
      "step": 8881
    },
    {
      "epoch": 0.6408138234551423,
      "grad_norm": 4.209641860625029,
      "learning_rate": 1.2073001070994685e-06,
      "loss": 1.0228,
      "step": 8882
    },
    {
      "epoch": 0.6408859709245698,
      "grad_norm": 7.3444571945892845,
      "learning_rate": 1.206871045930085e-06,
      "loss": 0.999,
      "step": 8883
    },
    {
      "epoch": 0.6409581183939973,
      "grad_norm": 2.8756139348527534,
      "learning_rate": 1.2064420280705388e-06,
      "loss": 0.9029,
      "step": 8884
    },
    {
      "epoch": 0.6410302658634248,
      "grad_norm": 3.1430926141835926,
      "learning_rate": 1.2060130535442557e-06,
      "loss": 0.8449,
      "step": 8885
    },
    {
      "epoch": 0.6411024133328523,
      "grad_norm": 2.97635343523301,
      "learning_rate": 1.2055841223746598e-06,
      "loss": 1.0485,
      "step": 8886
    },
    {
      "epoch": 0.6411745608022799,
      "grad_norm": 9.118068623071395,
      "learning_rate": 1.2051552345851748e-06,
      "loss": 0.9226,
      "step": 8887
    },
    {
      "epoch": 0.6412467082717074,
      "grad_norm": 5.152960820587118,
      "learning_rate": 1.2047263901992193e-06,
      "loss": 1.0144,
      "step": 8888
    },
    {
      "epoch": 0.6413188557411349,
      "grad_norm": 24.456810075764814,
      "learning_rate": 1.2042975892402125e-06,
      "loss": 0.9256,
      "step": 8889
    },
    {
      "epoch": 0.6413910032105624,
      "grad_norm": 2.8840896547960946,
      "learning_rate": 1.2038688317315685e-06,
      "loss": 0.9752,
      "step": 8890
    },
    {
      "epoch": 0.6414631506799899,
      "grad_norm": 5.335972351585946,
      "learning_rate": 1.2034401176967004e-06,
      "loss": 0.99,
      "step": 8891
    },
    {
      "epoch": 0.6415352981494175,
      "grad_norm": 4.280026448961337,
      "learning_rate": 1.2030114471590185e-06,
      "loss": 1.0528,
      "step": 8892
    },
    {
      "epoch": 0.641607445618845,
      "grad_norm": 3.8953096306366035,
      "learning_rate": 1.202582820141931e-06,
      "loss": 0.9715,
      "step": 8893
    },
    {
      "epoch": 0.6416795930882724,
      "grad_norm": 2.704270842141869,
      "learning_rate": 1.2021542366688444e-06,
      "loss": 1.0373,
      "step": 8894
    },
    {
      "epoch": 0.6417517405576999,
      "grad_norm": 0.5870201027740595,
      "learning_rate": 1.201725696763161e-06,
      "loss": 0.7928,
      "step": 8895
    },
    {
      "epoch": 0.6418238880271274,
      "grad_norm": 2.488066419972118,
      "learning_rate": 1.201297200448282e-06,
      "loss": 0.8946,
      "step": 8896
    },
    {
      "epoch": 0.6418960354965549,
      "grad_norm": 3.058408690172961,
      "learning_rate": 1.2008687477476068e-06,
      "loss": 0.9033,
      "step": 8897
    },
    {
      "epoch": 0.6419681829659825,
      "grad_norm": 3.173420646919491,
      "learning_rate": 1.2004403386845299e-06,
      "loss": 0.9802,
      "step": 8898
    },
    {
      "epoch": 0.64204033043541,
      "grad_norm": 4.252071324651876,
      "learning_rate": 1.2000119732824472e-06,
      "loss": 0.9361,
      "step": 8899
    },
    {
      "epoch": 0.6421124779048375,
      "grad_norm": 0.7057635708313896,
      "learning_rate": 1.1995836515647493e-06,
      "loss": 0.8566,
      "step": 8900
    },
    {
      "epoch": 0.642184625374265,
      "grad_norm": 0.8707269486479186,
      "learning_rate": 1.199155373554824e-06,
      "loss": 0.8266,
      "step": 8901
    },
    {
      "epoch": 0.6422567728436925,
      "grad_norm": 3.475115322696857,
      "learning_rate": 1.1987271392760603e-06,
      "loss": 0.9409,
      "step": 8902
    },
    {
      "epoch": 0.64232892031312,
      "grad_norm": 0.8162739112588075,
      "learning_rate": 1.1982989487518409e-06,
      "loss": 0.939,
      "step": 8903
    },
    {
      "epoch": 0.6424010677825476,
      "grad_norm": 2.4619724579981903,
      "learning_rate": 1.1978708020055478e-06,
      "loss": 0.889,
      "step": 8904
    },
    {
      "epoch": 0.642473215251975,
      "grad_norm": 2.9062972526400537,
      "learning_rate": 1.1974426990605611e-06,
      "loss": 1.0351,
      "step": 8905
    },
    {
      "epoch": 0.6425453627214025,
      "grad_norm": 5.593143877830281,
      "learning_rate": 1.197014639940258e-06,
      "loss": 0.9165,
      "step": 8906
    },
    {
      "epoch": 0.64261751019083,
      "grad_norm": 0.700404413941743,
      "learning_rate": 1.1965866246680126e-06,
      "loss": 0.8654,
      "step": 8907
    },
    {
      "epoch": 0.6426896576602575,
      "grad_norm": 5.896874469540749,
      "learning_rate": 1.1961586532671973e-06,
      "loss": 0.9289,
      "step": 8908
    },
    {
      "epoch": 0.6427618051296851,
      "grad_norm": 3.2473921950421096,
      "learning_rate": 1.1957307257611824e-06,
      "loss": 0.9811,
      "step": 8909
    },
    {
      "epoch": 0.6428339525991126,
      "grad_norm": 4.358119113731594,
      "learning_rate": 1.1953028421733349e-06,
      "loss": 0.9981,
      "step": 8910
    },
    {
      "epoch": 0.6429061000685401,
      "grad_norm": 3.0358208870002157,
      "learning_rate": 1.1948750025270209e-06,
      "loss": 1.0312,
      "step": 8911
    },
    {
      "epoch": 0.6429782475379676,
      "grad_norm": 0.7950575429256411,
      "learning_rate": 1.1944472068456026e-06,
      "loss": 0.8844,
      "step": 8912
    },
    {
      "epoch": 0.6430503950073951,
      "grad_norm": 2.210237965264859,
      "learning_rate": 1.1940194551524391e-06,
      "loss": 0.9397,
      "step": 8913
    },
    {
      "epoch": 0.6431225424768227,
      "grad_norm": 5.924007464623755,
      "learning_rate": 1.193591747470891e-06,
      "loss": 0.8946,
      "step": 8914
    },
    {
      "epoch": 0.6431946899462502,
      "grad_norm": 3.4317092671528417,
      "learning_rate": 1.1931640838243114e-06,
      "loss": 1.0367,
      "step": 8915
    },
    {
      "epoch": 0.6432668374156777,
      "grad_norm": 2.96327395087504,
      "learning_rate": 1.1927364642360547e-06,
      "loss": 1.0124,
      "step": 8916
    },
    {
      "epoch": 0.6433389848851051,
      "grad_norm": 9.31037130924517,
      "learning_rate": 1.1923088887294714e-06,
      "loss": 1.0019,
      "step": 8917
    },
    {
      "epoch": 0.6434111323545326,
      "grad_norm": 3.907488573171122,
      "learning_rate": 1.1918813573279095e-06,
      "loss": 0.8831,
      "step": 8918
    },
    {
      "epoch": 0.6434832798239601,
      "grad_norm": 3.643642370293113,
      "learning_rate": 1.1914538700547153e-06,
      "loss": 0.9064,
      "step": 8919
    },
    {
      "epoch": 0.6435554272933877,
      "grad_norm": 0.6831536320360079,
      "learning_rate": 1.1910264269332317e-06,
      "loss": 0.848,
      "step": 8920
    },
    {
      "epoch": 0.6436275747628152,
      "grad_norm": 5.775437190694081,
      "learning_rate": 1.190599027986801e-06,
      "loss": 0.9769,
      "step": 8921
    },
    {
      "epoch": 0.6436997222322427,
      "grad_norm": 4.091046787603312,
      "learning_rate": 1.1901716732387605e-06,
      "loss": 0.8995,
      "step": 8922
    },
    {
      "epoch": 0.6437718697016702,
      "grad_norm": 4.191321380125727,
      "learning_rate": 1.189744362712447e-06,
      "loss": 1.0438,
      "step": 8923
    },
    {
      "epoch": 0.6438440171710977,
      "grad_norm": 4.494999332279415,
      "learning_rate": 1.189317096431195e-06,
      "loss": 0.8795,
      "step": 8924
    },
    {
      "epoch": 0.6439161646405253,
      "grad_norm": 6.879449410253151,
      "learning_rate": 1.1888898744183342e-06,
      "loss": 0.9516,
      "step": 8925
    },
    {
      "epoch": 0.6439883121099528,
      "grad_norm": 3.57446595051063,
      "learning_rate": 1.1884626966971957e-06,
      "loss": 0.9709,
      "step": 8926
    },
    {
      "epoch": 0.6440604595793803,
      "grad_norm": 3.112783892450003,
      "learning_rate": 1.188035563291105e-06,
      "loss": 1.0397,
      "step": 8927
    },
    {
      "epoch": 0.6441326070488078,
      "grad_norm": 4.014851655226699,
      "learning_rate": 1.1876084742233858e-06,
      "loss": 0.8549,
      "step": 8928
    },
    {
      "epoch": 0.6442047545182352,
      "grad_norm": 7.533376812879732,
      "learning_rate": 1.1871814295173615e-06,
      "loss": 0.9318,
      "step": 8929
    },
    {
      "epoch": 0.6442769019876627,
      "grad_norm": 4.373790246735264,
      "learning_rate": 1.1867544291963496e-06,
      "loss": 0.9919,
      "step": 8930
    },
    {
      "epoch": 0.6443490494570903,
      "grad_norm": 4.476204370150507,
      "learning_rate": 1.1863274732836677e-06,
      "loss": 0.9819,
      "step": 8931
    },
    {
      "epoch": 0.6444211969265178,
      "grad_norm": 4.5150733267573955,
      "learning_rate": 1.1859005618026308e-06,
      "loss": 0.9281,
      "step": 8932
    },
    {
      "epoch": 0.6444933443959453,
      "grad_norm": 5.692325096341688,
      "learning_rate": 1.1854736947765505e-06,
      "loss": 0.9081,
      "step": 8933
    },
    {
      "epoch": 0.6445654918653728,
      "grad_norm": 3.7856800201181398,
      "learning_rate": 1.185046872228736e-06,
      "loss": 0.9023,
      "step": 8934
    },
    {
      "epoch": 0.6446376393348003,
      "grad_norm": 2.859579777289934,
      "learning_rate": 1.1846200941824953e-06,
      "loss": 1.0019,
      "step": 8935
    },
    {
      "epoch": 0.6447097868042279,
      "grad_norm": 3.57345769732213,
      "learning_rate": 1.1841933606611332e-06,
      "loss": 0.9065,
      "step": 8936
    },
    {
      "epoch": 0.6447819342736554,
      "grad_norm": 3.755447626977014,
      "learning_rate": 1.1837666716879505e-06,
      "loss": 0.9457,
      "step": 8937
    },
    {
      "epoch": 0.6448540817430829,
      "grad_norm": 6.684925475065353,
      "learning_rate": 1.1833400272862495e-06,
      "loss": 0.8966,
      "step": 8938
    },
    {
      "epoch": 0.6449262292125104,
      "grad_norm": 2.9146873479192594,
      "learning_rate": 1.1829134274793262e-06,
      "loss": 0.9408,
      "step": 8939
    },
    {
      "epoch": 0.6449983766819379,
      "grad_norm": 3.188847126246078,
      "learning_rate": 1.1824868722904753e-06,
      "loss": 0.9914,
      "step": 8940
    },
    {
      "epoch": 0.6450705241513653,
      "grad_norm": 3.8293309394763417,
      "learning_rate": 1.1820603617429906e-06,
      "loss": 0.9689,
      "step": 8941
    },
    {
      "epoch": 0.6451426716207929,
      "grad_norm": 4.301591871267325,
      "learning_rate": 1.1816338958601612e-06,
      "loss": 1.0787,
      "step": 8942
    },
    {
      "epoch": 0.6452148190902204,
      "grad_norm": 3.1266891491968893,
      "learning_rate": 1.1812074746652752e-06,
      "loss": 1.0357,
      "step": 8943
    },
    {
      "epoch": 0.6452869665596479,
      "grad_norm": 2.7310919534797464,
      "learning_rate": 1.1807810981816185e-06,
      "loss": 1.0039,
      "step": 8944
    },
    {
      "epoch": 0.6453591140290754,
      "grad_norm": 3.598403173855221,
      "learning_rate": 1.180354766432473e-06,
      "loss": 0.9312,
      "step": 8945
    },
    {
      "epoch": 0.6454312614985029,
      "grad_norm": 0.6735503135414801,
      "learning_rate": 1.1799284794411193e-06,
      "loss": 0.8375,
      "step": 8946
    },
    {
      "epoch": 0.6455034089679305,
      "grad_norm": 5.6102333368472355,
      "learning_rate": 1.1795022372308356e-06,
      "loss": 1.0591,
      "step": 8947
    },
    {
      "epoch": 0.645575556437358,
      "grad_norm": 2.4953024599450195,
      "learning_rate": 1.1790760398248977e-06,
      "loss": 1.0383,
      "step": 8948
    },
    {
      "epoch": 0.6456477039067855,
      "grad_norm": 7.953310861270661,
      "learning_rate": 1.1786498872465775e-06,
      "loss": 0.9904,
      "step": 8949
    },
    {
      "epoch": 0.645719851376213,
      "grad_norm": 4.208588066826611,
      "learning_rate": 1.1782237795191475e-06,
      "loss": 1.0342,
      "step": 8950
    },
    {
      "epoch": 0.6457919988456405,
      "grad_norm": 3.516442233572971,
      "learning_rate": 1.1777977166658746e-06,
      "loss": 0.9303,
      "step": 8951
    },
    {
      "epoch": 0.6458641463150679,
      "grad_norm": 4.085420243908323,
      "learning_rate": 1.1773716987100239e-06,
      "loss": 1.0439,
      "step": 8952
    },
    {
      "epoch": 0.6459362937844955,
      "grad_norm": 3.4292705169439968,
      "learning_rate": 1.1769457256748602e-06,
      "loss": 0.9822,
      "step": 8953
    },
    {
      "epoch": 0.646008441253923,
      "grad_norm": 2.68145871361554,
      "learning_rate": 1.1765197975836434e-06,
      "loss": 0.9274,
      "step": 8954
    },
    {
      "epoch": 0.6460805887233505,
      "grad_norm": 3.673359672211953,
      "learning_rate": 1.1760939144596318e-06,
      "loss": 1.0991,
      "step": 8955
    },
    {
      "epoch": 0.646152736192778,
      "grad_norm": 3.3058163854307048,
      "learning_rate": 1.1756680763260823e-06,
      "loss": 0.777,
      "step": 8956
    },
    {
      "epoch": 0.6462248836622055,
      "grad_norm": 3.4797970929678557,
      "learning_rate": 1.1752422832062468e-06,
      "loss": 1.0456,
      "step": 8957
    },
    {
      "epoch": 0.6462970311316331,
      "grad_norm": 3.309702429634648,
      "learning_rate": 1.174816535123377e-06,
      "loss": 0.9318,
      "step": 8958
    },
    {
      "epoch": 0.6463691786010606,
      "grad_norm": 3.662223045110867,
      "learning_rate": 1.1743908321007219e-06,
      "loss": 0.9912,
      "step": 8959
    },
    {
      "epoch": 0.6464413260704881,
      "grad_norm": 4.089400681886249,
      "learning_rate": 1.1739651741615273e-06,
      "loss": 1.0818,
      "step": 8960
    },
    {
      "epoch": 0.6465134735399156,
      "grad_norm": 3.394115988876524,
      "learning_rate": 1.1735395613290362e-06,
      "loss": 0.9569,
      "step": 8961
    },
    {
      "epoch": 0.6465856210093431,
      "grad_norm": 2.5360999092373686,
      "learning_rate": 1.1731139936264902e-06,
      "loss": 1.0839,
      "step": 8962
    },
    {
      "epoch": 0.6466577684787707,
      "grad_norm": 3.1990208081316656,
      "learning_rate": 1.1726884710771285e-06,
      "loss": 0.997,
      "step": 8963
    },
    {
      "epoch": 0.6467299159481981,
      "grad_norm": 2.4467134697653306,
      "learning_rate": 1.1722629937041858e-06,
      "loss": 0.852,
      "step": 8964
    },
    {
      "epoch": 0.6468020634176256,
      "grad_norm": 3.321058481227029,
      "learning_rate": 1.171837561530898e-06,
      "loss": 0.904,
      "step": 8965
    },
    {
      "epoch": 0.6468742108870531,
      "grad_norm": 3.861537435644296,
      "learning_rate": 1.171412174580495e-06,
      "loss": 0.9565,
      "step": 8966
    },
    {
      "epoch": 0.6469463583564806,
      "grad_norm": 3.570627010858371,
      "learning_rate": 1.170986832876205e-06,
      "loss": 1.0396,
      "step": 8967
    },
    {
      "epoch": 0.6470185058259081,
      "grad_norm": 4.145470199144733,
      "learning_rate": 1.170561536441256e-06,
      "loss": 1.0265,
      "step": 8968
    },
    {
      "epoch": 0.6470906532953357,
      "grad_norm": 0.6344813722772444,
      "learning_rate": 1.1701362852988705e-06,
      "loss": 0.8331,
      "step": 8969
    },
    {
      "epoch": 0.6471628007647632,
      "grad_norm": 2.672175608003895,
      "learning_rate": 1.1697110794722707e-06,
      "loss": 1.0422,
      "step": 8970
    },
    {
      "epoch": 0.6472349482341907,
      "grad_norm": 5.328801713406651,
      "learning_rate": 1.1692859189846749e-06,
      "loss": 0.9488,
      "step": 8971
    },
    {
      "epoch": 0.6473070957036182,
      "grad_norm": 3.096088950966584,
      "learning_rate": 1.1688608038593005e-06,
      "loss": 0.8664,
      "step": 8972
    },
    {
      "epoch": 0.6473792431730457,
      "grad_norm": 2.70851083809591,
      "learning_rate": 1.1684357341193604e-06,
      "loss": 0.9688,
      "step": 8973
    },
    {
      "epoch": 0.6474513906424733,
      "grad_norm": 3.567475040725405,
      "learning_rate": 1.1680107097880662e-06,
      "loss": 1.0467,
      "step": 8974
    },
    {
      "epoch": 0.6475235381119008,
      "grad_norm": 2.5804019005778653,
      "learning_rate": 1.1675857308886282e-06,
      "loss": 1.011,
      "step": 8975
    },
    {
      "epoch": 0.6475956855813282,
      "grad_norm": 3.5520786647076097,
      "learning_rate": 1.167160797444251e-06,
      "loss": 0.9023,
      "step": 8976
    },
    {
      "epoch": 0.6476678330507557,
      "grad_norm": 3.4951312397122516,
      "learning_rate": 1.1667359094781398e-06,
      "loss": 0.9164,
      "step": 8977
    },
    {
      "epoch": 0.6477399805201832,
      "grad_norm": 6.18799966663441,
      "learning_rate": 1.166311067013496e-06,
      "loss": 0.9962,
      "step": 8978
    },
    {
      "epoch": 0.6478121279896107,
      "grad_norm": 3.816250316671089,
      "learning_rate": 1.1658862700735185e-06,
      "loss": 0.9175,
      "step": 8979
    },
    {
      "epoch": 0.6478842754590383,
      "grad_norm": 10.930898537937296,
      "learning_rate": 1.165461518681405e-06,
      "loss": 1.0034,
      "step": 8980
    },
    {
      "epoch": 0.6479564229284658,
      "grad_norm": 3.1734170406852185,
      "learning_rate": 1.1650368128603468e-06,
      "loss": 0.9762,
      "step": 8981
    },
    {
      "epoch": 0.6480285703978933,
      "grad_norm": 5.396482431064163,
      "learning_rate": 1.164612152633539e-06,
      "loss": 0.8453,
      "step": 8982
    },
    {
      "epoch": 0.6481007178673208,
      "grad_norm": 2.6974072263813045,
      "learning_rate": 1.1641875380241683e-06,
      "loss": 0.9853,
      "step": 8983
    },
    {
      "epoch": 0.6481728653367483,
      "grad_norm": 4.054863427963215,
      "learning_rate": 1.163762969055422e-06,
      "loss": 0.9851,
      "step": 8984
    },
    {
      "epoch": 0.6482450128061759,
      "grad_norm": 3.8231260708340775,
      "learning_rate": 1.1633384457504845e-06,
      "loss": 0.8754,
      "step": 8985
    },
    {
      "epoch": 0.6483171602756034,
      "grad_norm": 3.0191626478002256,
      "learning_rate": 1.162913968132537e-06,
      "loss": 1.0163,
      "step": 8986
    },
    {
      "epoch": 0.6483893077450309,
      "grad_norm": 4.553336150772169,
      "learning_rate": 1.1624895362247592e-06,
      "loss": 1.0406,
      "step": 8987
    },
    {
      "epoch": 0.6484614552144583,
      "grad_norm": 3.2105414153230285,
      "learning_rate": 1.1620651500503278e-06,
      "loss": 0.9466,
      "step": 8988
    },
    {
      "epoch": 0.6485336026838858,
      "grad_norm": 8.538000991889144,
      "learning_rate": 1.1616408096324161e-06,
      "loss": 0.9487,
      "step": 8989
    },
    {
      "epoch": 0.6486057501533133,
      "grad_norm": 3.8850548024282867,
      "learning_rate": 1.1612165149941964e-06,
      "loss": 0.9175,
      "step": 8990
    },
    {
      "epoch": 0.6486778976227409,
      "grad_norm": 2.6239211272271845,
      "learning_rate": 1.1607922661588376e-06,
      "loss": 0.9713,
      "step": 8991
    },
    {
      "epoch": 0.6487500450921684,
      "grad_norm": 2.560609818186131,
      "learning_rate": 1.1603680631495066e-06,
      "loss": 0.8782,
      "step": 8992
    },
    {
      "epoch": 0.6488221925615959,
      "grad_norm": 7.574020815973559,
      "learning_rate": 1.1599439059893673e-06,
      "loss": 0.7582,
      "step": 8993
    },
    {
      "epoch": 0.6488943400310234,
      "grad_norm": 2.0705276089605213,
      "learning_rate": 1.1595197947015818e-06,
      "loss": 0.9799,
      "step": 8994
    },
    {
      "epoch": 0.6489664875004509,
      "grad_norm": 5.146663192161481,
      "learning_rate": 1.1590957293093094e-06,
      "loss": 0.9932,
      "step": 8995
    },
    {
      "epoch": 0.6490386349698785,
      "grad_norm": 2.6638121725029253,
      "learning_rate": 1.1586717098357051e-06,
      "loss": 0.9572,
      "step": 8996
    },
    {
      "epoch": 0.649110782439306,
      "grad_norm": 3.7704822664894766,
      "learning_rate": 1.1582477363039255e-06,
      "loss": 0.9269,
      "step": 8997
    },
    {
      "epoch": 0.6491829299087335,
      "grad_norm": 3.1368987339396512,
      "learning_rate": 1.1578238087371195e-06,
      "loss": 0.9142,
      "step": 8998
    },
    {
      "epoch": 0.649255077378161,
      "grad_norm": 0.8088374116287632,
      "learning_rate": 1.1573999271584392e-06,
      "loss": 0.8719,
      "step": 8999
    },
    {
      "epoch": 0.6493272248475884,
      "grad_norm": 36.03651441376857,
      "learning_rate": 1.1569760915910292e-06,
      "loss": 0.9888,
      "step": 9000
    },
    {
      "epoch": 0.6493993723170159,
      "grad_norm": 2.541880384954897,
      "learning_rate": 1.1565523020580337e-06,
      "loss": 0.9835,
      "step": 9001
    },
    {
      "epoch": 0.6494715197864435,
      "grad_norm": 3.186627586502859,
      "learning_rate": 1.1561285585825946e-06,
      "loss": 0.885,
      "step": 9002
    },
    {
      "epoch": 0.649543667255871,
      "grad_norm": 3.231773325232386,
      "learning_rate": 1.1557048611878513e-06,
      "loss": 0.9461,
      "step": 9003
    },
    {
      "epoch": 0.6496158147252985,
      "grad_norm": 2.5734749212827372,
      "learning_rate": 1.1552812098969407e-06,
      "loss": 0.8741,
      "step": 9004
    },
    {
      "epoch": 0.649687962194726,
      "grad_norm": 4.686307017787654,
      "learning_rate": 1.1548576047329955e-06,
      "loss": 0.9807,
      "step": 9005
    },
    {
      "epoch": 0.6497601096641535,
      "grad_norm": 4.229308077447757,
      "learning_rate": 1.1544340457191478e-06,
      "loss": 0.9788,
      "step": 9006
    },
    {
      "epoch": 0.6498322571335811,
      "grad_norm": 3.7612000104752243,
      "learning_rate": 1.1540105328785267e-06,
      "loss": 0.9929,
      "step": 9007
    },
    {
      "epoch": 0.6499044046030086,
      "grad_norm": 6.92208772422844,
      "learning_rate": 1.1535870662342583e-06,
      "loss": 1.0262,
      "step": 9008
    },
    {
      "epoch": 0.6499765520724361,
      "grad_norm": 4.149416475750946,
      "learning_rate": 1.1531636458094674e-06,
      "loss": 1.0153,
      "step": 9009
    },
    {
      "epoch": 0.6500486995418636,
      "grad_norm": 5.667848949888309,
      "learning_rate": 1.1527402716272756e-06,
      "loss": 0.8903,
      "step": 9010
    },
    {
      "epoch": 0.650120847011291,
      "grad_norm": 3.4465518028652906,
      "learning_rate": 1.1523169437107997e-06,
      "loss": 1.0705,
      "step": 9011
    },
    {
      "epoch": 0.6501929944807185,
      "grad_norm": 3.8811815240443224,
      "learning_rate": 1.1518936620831588e-06,
      "loss": 0.8997,
      "step": 9012
    },
    {
      "epoch": 0.6502651419501461,
      "grad_norm": 12.573999104947118,
      "learning_rate": 1.1514704267674646e-06,
      "loss": 1.0254,
      "step": 9013
    },
    {
      "epoch": 0.6503372894195736,
      "grad_norm": 3.1975133353664362,
      "learning_rate": 1.1510472377868295e-06,
      "loss": 0.8624,
      "step": 9014
    },
    {
      "epoch": 0.6504094368890011,
      "grad_norm": 3.9070498448695803,
      "learning_rate": 1.1506240951643618e-06,
      "loss": 0.8785,
      "step": 9015
    },
    {
      "epoch": 0.6504815843584286,
      "grad_norm": 7.611765899089321,
      "learning_rate": 1.1502009989231681e-06,
      "loss": 0.9089,
      "step": 9016
    },
    {
      "epoch": 0.6505537318278561,
      "grad_norm": 3.0012175950124482,
      "learning_rate": 1.149777949086352e-06,
      "loss": 0.9782,
      "step": 9017
    },
    {
      "epoch": 0.6506258792972837,
      "grad_norm": 8.184250980470242,
      "learning_rate": 1.1493549456770145e-06,
      "loss": 1.0066,
      "step": 9018
    },
    {
      "epoch": 0.6506980267667112,
      "grad_norm": 2.6376938662759826,
      "learning_rate": 1.1489319887182553e-06,
      "loss": 0.9347,
      "step": 9019
    },
    {
      "epoch": 0.6507701742361387,
      "grad_norm": 3.862785778152279,
      "learning_rate": 1.148509078233168e-06,
      "loss": 0.9923,
      "step": 9020
    },
    {
      "epoch": 0.6508423217055662,
      "grad_norm": 3.185653001805532,
      "learning_rate": 1.1480862142448497e-06,
      "loss": 0.8527,
      "step": 9021
    },
    {
      "epoch": 0.6509144691749937,
      "grad_norm": 4.214023683543882,
      "learning_rate": 1.1476633967763887e-06,
      "loss": 0.9384,
      "step": 9022
    },
    {
      "epoch": 0.6509866166444211,
      "grad_norm": 3.7576320549354376,
      "learning_rate": 1.1472406258508741e-06,
      "loss": 0.9167,
      "step": 9023
    },
    {
      "epoch": 0.6510587641138487,
      "grad_norm": 2.434081590474509,
      "learning_rate": 1.1468179014913926e-06,
      "loss": 0.9552,
      "step": 9024
    },
    {
      "epoch": 0.6511309115832762,
      "grad_norm": 4.180380391238828,
      "learning_rate": 1.1463952237210265e-06,
      "loss": 0.9807,
      "step": 9025
    },
    {
      "epoch": 0.6512030590527037,
      "grad_norm": 8.686456631365505,
      "learning_rate": 1.1459725925628579e-06,
      "loss": 0.9381,
      "step": 9026
    },
    {
      "epoch": 0.6512752065221312,
      "grad_norm": 3.2218210328667025,
      "learning_rate": 1.145550008039965e-06,
      "loss": 0.9795,
      "step": 9027
    },
    {
      "epoch": 0.6513473539915587,
      "grad_norm": 3.9964745005482913,
      "learning_rate": 1.1451274701754222e-06,
      "loss": 0.8988,
      "step": 9028
    },
    {
      "epoch": 0.6514195014609863,
      "grad_norm": 11.716306385851663,
      "learning_rate": 1.1447049789923038e-06,
      "loss": 0.9346,
      "step": 9029
    },
    {
      "epoch": 0.6514916489304138,
      "grad_norm": 4.688499445860371,
      "learning_rate": 1.1442825345136803e-06,
      "loss": 0.8559,
      "step": 9030
    },
    {
      "epoch": 0.6515637963998413,
      "grad_norm": 4.332324032740187,
      "learning_rate": 1.1438601367626198e-06,
      "loss": 1.0404,
      "step": 9031
    },
    {
      "epoch": 0.6516359438692688,
      "grad_norm": 4.007504575920705,
      "learning_rate": 1.1434377857621877e-06,
      "loss": 0.8977,
      "step": 9032
    },
    {
      "epoch": 0.6517080913386963,
      "grad_norm": 7.369658523992404,
      "learning_rate": 1.1430154815354475e-06,
      "loss": 0.94,
      "step": 9033
    },
    {
      "epoch": 0.6517802388081239,
      "grad_norm": 2.9122310334140695,
      "learning_rate": 1.1425932241054601e-06,
      "loss": 0.9146,
      "step": 9034
    },
    {
      "epoch": 0.6518523862775513,
      "grad_norm": 6.619216986041535,
      "learning_rate": 1.1421710134952814e-06,
      "loss": 0.9052,
      "step": 9035
    },
    {
      "epoch": 0.6519245337469788,
      "grad_norm": 3.814800506415448,
      "learning_rate": 1.1417488497279694e-06,
      "loss": 0.9501,
      "step": 9036
    },
    {
      "epoch": 0.6519966812164063,
      "grad_norm": 4.367568980361704,
      "learning_rate": 1.1413267328265739e-06,
      "loss": 0.9781,
      "step": 9037
    },
    {
      "epoch": 0.6520688286858338,
      "grad_norm": 2.8907214999200628,
      "learning_rate": 1.1409046628141485e-06,
      "loss": 0.9892,
      "step": 9038
    },
    {
      "epoch": 0.6521409761552613,
      "grad_norm": 3.017527558233274,
      "learning_rate": 1.1404826397137385e-06,
      "loss": 0.955,
      "step": 9039
    },
    {
      "epoch": 0.6522131236246889,
      "grad_norm": 2.5776438871046614,
      "learning_rate": 1.1400606635483895e-06,
      "loss": 1.0401,
      "step": 9040
    },
    {
      "epoch": 0.6522852710941164,
      "grad_norm": 3.0454971719395822,
      "learning_rate": 1.1396387343411446e-06,
      "loss": 0.8437,
      "step": 9041
    },
    {
      "epoch": 0.6523574185635439,
      "grad_norm": 3.72747471678645,
      "learning_rate": 1.1392168521150433e-06,
      "loss": 0.8987,
      "step": 9042
    },
    {
      "epoch": 0.6524295660329714,
      "grad_norm": 2.8753161049260427,
      "learning_rate": 1.1387950168931242e-06,
      "loss": 1.0067,
      "step": 9043
    },
    {
      "epoch": 0.6525017135023989,
      "grad_norm": 4.035048476900556,
      "learning_rate": 1.13837322869842e-06,
      "loss": 1.0156,
      "step": 9044
    },
    {
      "epoch": 0.6525738609718265,
      "grad_norm": 2.694649965984166,
      "learning_rate": 1.1379514875539645e-06,
      "loss": 0.9803,
      "step": 9045
    },
    {
      "epoch": 0.652646008441254,
      "grad_norm": 4.800157449047326,
      "learning_rate": 1.1375297934827868e-06,
      "loss": 0.8839,
      "step": 9046
    },
    {
      "epoch": 0.6527181559106814,
      "grad_norm": 3.0624011276789136,
      "learning_rate": 1.1371081465079142e-06,
      "loss": 0.9481,
      "step": 9047
    },
    {
      "epoch": 0.6527903033801089,
      "grad_norm": 5.641128475301153,
      "learning_rate": 1.1366865466523718e-06,
      "loss": 0.9244,
      "step": 9048
    },
    {
      "epoch": 0.6528624508495364,
      "grad_norm": 4.425501761571766,
      "learning_rate": 1.1362649939391817e-06,
      "loss": 1.0124,
      "step": 9049
    },
    {
      "epoch": 0.6529345983189639,
      "grad_norm": 3.9026462869150063,
      "learning_rate": 1.1358434883913614e-06,
      "loss": 0.9238,
      "step": 9050
    },
    {
      "epoch": 0.6530067457883915,
      "grad_norm": 3.381956348924793,
      "learning_rate": 1.1354220300319309e-06,
      "loss": 0.9441,
      "step": 9051
    },
    {
      "epoch": 0.653078893257819,
      "grad_norm": 4.012146389314825,
      "learning_rate": 1.1350006188839016e-06,
      "loss": 1.1253,
      "step": 9052
    },
    {
      "epoch": 0.6531510407272465,
      "grad_norm": 2.638270846711518,
      "learning_rate": 1.1345792549702866e-06,
      "loss": 0.9883,
      "step": 9053
    },
    {
      "epoch": 0.653223188196674,
      "grad_norm": 2.4481244479254967,
      "learning_rate": 1.1341579383140948e-06,
      "loss": 1.0093,
      "step": 9054
    },
    {
      "epoch": 0.6532953356661015,
      "grad_norm": 4.655549067025126,
      "learning_rate": 1.133736668938333e-06,
      "loss": 0.9449,
      "step": 9055
    },
    {
      "epoch": 0.6533674831355291,
      "grad_norm": 3.6112701592917573,
      "learning_rate": 1.1333154468660047e-06,
      "loss": 0.9899,
      "step": 9056
    },
    {
      "epoch": 0.6534396306049566,
      "grad_norm": 2.376715391960578,
      "learning_rate": 1.1328942721201116e-06,
      "loss": 1.0195,
      "step": 9057
    },
    {
      "epoch": 0.653511778074384,
      "grad_norm": 3.824335581599245,
      "learning_rate": 1.132473144723653e-06,
      "loss": 1.0456,
      "step": 9058
    },
    {
      "epoch": 0.6535839255438115,
      "grad_norm": 3.0063120877205036,
      "learning_rate": 1.1320520646996228e-06,
      "loss": 0.9724,
      "step": 9059
    },
    {
      "epoch": 0.653656073013239,
      "grad_norm": 3.492755203698631,
      "learning_rate": 1.1316310320710182e-06,
      "loss": 0.8314,
      "step": 9060
    },
    {
      "epoch": 0.6537282204826665,
      "grad_norm": 2.9384035384744953,
      "learning_rate": 1.1312100468608274e-06,
      "loss": 0.8902,
      "step": 9061
    },
    {
      "epoch": 0.6538003679520941,
      "grad_norm": 2.418180736807466,
      "learning_rate": 1.13078910909204e-06,
      "loss": 0.9819,
      "step": 9062
    },
    {
      "epoch": 0.6538725154215216,
      "grad_norm": 6.33423219961139,
      "learning_rate": 1.130368218787641e-06,
      "loss": 0.9395,
      "step": 9063
    },
    {
      "epoch": 0.6539446628909491,
      "grad_norm": 3.753353399716587,
      "learning_rate": 1.129947375970615e-06,
      "loss": 1.0252,
      "step": 9064
    },
    {
      "epoch": 0.6540168103603766,
      "grad_norm": 0.7273855981967544,
      "learning_rate": 1.1295265806639417e-06,
      "loss": 0.8306,
      "step": 9065
    },
    {
      "epoch": 0.6540889578298041,
      "grad_norm": 3.111573941756733,
      "learning_rate": 1.1291058328906004e-06,
      "loss": 0.9846,
      "step": 9066
    },
    {
      "epoch": 0.6541611052992317,
      "grad_norm": 0.7666910107175312,
      "learning_rate": 1.1286851326735648e-06,
      "loss": 0.8831,
      "step": 9067
    },
    {
      "epoch": 0.6542332527686592,
      "grad_norm": 2.959634539493152,
      "learning_rate": 1.1282644800358085e-06,
      "loss": 0.8808,
      "step": 9068
    },
    {
      "epoch": 0.6543054002380867,
      "grad_norm": 3.656046152545596,
      "learning_rate": 1.1278438750003018e-06,
      "loss": 0.9778,
      "step": 9069
    },
    {
      "epoch": 0.6543775477075141,
      "grad_norm": 4.957993097679808,
      "learning_rate": 1.1274233175900129e-06,
      "loss": 0.9506,
      "step": 9070
    },
    {
      "epoch": 0.6544496951769416,
      "grad_norm": 3.4945371774870084,
      "learning_rate": 1.127002807827906e-06,
      "loss": 0.9592,
      "step": 9071
    },
    {
      "epoch": 0.6545218426463691,
      "grad_norm": 2.983524540859868,
      "learning_rate": 1.1265823457369438e-06,
      "loss": 0.9671,
      "step": 9072
    },
    {
      "epoch": 0.6545939901157967,
      "grad_norm": 3.8244266005424934,
      "learning_rate": 1.126161931340088e-06,
      "loss": 0.9507,
      "step": 9073
    },
    {
      "epoch": 0.6546661375852242,
      "grad_norm": 4.6516094252808164,
      "learning_rate": 1.1257415646602921e-06,
      "loss": 0.9513,
      "step": 9074
    },
    {
      "epoch": 0.6547382850546517,
      "grad_norm": 2.925406051230358,
      "learning_rate": 1.1253212457205147e-06,
      "loss": 0.9399,
      "step": 9075
    },
    {
      "epoch": 0.6548104325240792,
      "grad_norm": 4.713230163509814,
      "learning_rate": 1.1249009745437055e-06,
      "loss": 0.9557,
      "step": 9076
    },
    {
      "epoch": 0.6548825799935067,
      "grad_norm": 4.094036529064971,
      "learning_rate": 1.1244807511528142e-06,
      "loss": 0.9994,
      "step": 9077
    },
    {
      "epoch": 0.6549547274629343,
      "grad_norm": 3.9357991253681917,
      "learning_rate": 1.1240605755707884e-06,
      "loss": 0.8624,
      "step": 9078
    },
    {
      "epoch": 0.6550268749323618,
      "grad_norm": 4.911489034201447,
      "learning_rate": 1.1236404478205719e-06,
      "loss": 0.9842,
      "step": 9079
    },
    {
      "epoch": 0.6550990224017893,
      "grad_norm": 3.239890293571664,
      "learning_rate": 1.1232203679251063e-06,
      "loss": 0.9967,
      "step": 9080
    },
    {
      "epoch": 0.6551711698712168,
      "grad_norm": 2.6626972716413886,
      "learning_rate": 1.1228003359073306e-06,
      "loss": 0.9691,
      "step": 9081
    },
    {
      "epoch": 0.6552433173406442,
      "grad_norm": 2.2941699025659803,
      "learning_rate": 1.1223803517901822e-06,
      "loss": 0.9113,
      "step": 9082
    },
    {
      "epoch": 0.6553154648100717,
      "grad_norm": 3.570971271665327,
      "learning_rate": 1.1219604155965931e-06,
      "loss": 0.9424,
      "step": 9083
    },
    {
      "epoch": 0.6553876122794993,
      "grad_norm": 5.930358768308171,
      "learning_rate": 1.1215405273494952e-06,
      "loss": 0.9358,
      "step": 9084
    },
    {
      "epoch": 0.6554597597489268,
      "grad_norm": 3.5420114386797628,
      "learning_rate": 1.1211206870718175e-06,
      "loss": 0.8771,
      "step": 9085
    },
    {
      "epoch": 0.6555319072183543,
      "grad_norm": 2.8045914402400007,
      "learning_rate": 1.1207008947864851e-06,
      "loss": 0.9716,
      "step": 9086
    },
    {
      "epoch": 0.6556040546877818,
      "grad_norm": 6.758731422349399,
      "learning_rate": 1.1202811505164223e-06,
      "loss": 0.9914,
      "step": 9087
    },
    {
      "epoch": 0.6556762021572093,
      "grad_norm": 4.825092832630877,
      "learning_rate": 1.1198614542845497e-06,
      "loss": 0.9391,
      "step": 9088
    },
    {
      "epoch": 0.6557483496266369,
      "grad_norm": 2.651360713044581,
      "learning_rate": 1.1194418061137833e-06,
      "loss": 0.9276,
      "step": 9089
    },
    {
      "epoch": 0.6558204970960644,
      "grad_norm": 34.54862225624471,
      "learning_rate": 1.1190222060270418e-06,
      "loss": 1.1055,
      "step": 9090
    },
    {
      "epoch": 0.6558926445654919,
      "grad_norm": 3.2616412673244537,
      "learning_rate": 1.1186026540472357e-06,
      "loss": 0.9357,
      "step": 9091
    },
    {
      "epoch": 0.6559647920349194,
      "grad_norm": 4.548440041208819,
      "learning_rate": 1.118183150197276e-06,
      "loss": 0.9184,
      "step": 9092
    },
    {
      "epoch": 0.6560369395043469,
      "grad_norm": 3.4121137072006698,
      "learning_rate": 1.1177636945000698e-06,
      "loss": 0.8842,
      "step": 9093
    },
    {
      "epoch": 0.6561090869737743,
      "grad_norm": 4.691399541718247,
      "learning_rate": 1.1173442869785223e-06,
      "loss": 0.9843,
      "step": 9094
    },
    {
      "epoch": 0.6561812344432019,
      "grad_norm": 4.0180734969727965,
      "learning_rate": 1.116924927655536e-06,
      "loss": 0.8901,
      "step": 9095
    },
    {
      "epoch": 0.6562533819126294,
      "grad_norm": 3.0591493299286214,
      "learning_rate": 1.11650561655401e-06,
      "loss": 0.8649,
      "step": 9096
    },
    {
      "epoch": 0.6563255293820569,
      "grad_norm": 2.9914398454486744,
      "learning_rate": 1.1160863536968427e-06,
      "loss": 0.8931,
      "step": 9097
    },
    {
      "epoch": 0.6563976768514844,
      "grad_norm": 3.617885843512005,
      "learning_rate": 1.1156671391069268e-06,
      "loss": 0.9166,
      "step": 9098
    },
    {
      "epoch": 0.6564698243209119,
      "grad_norm": 3.665981026325147,
      "learning_rate": 1.1152479728071547e-06,
      "loss": 1.0426,
      "step": 9099
    },
    {
      "epoch": 0.6565419717903395,
      "grad_norm": 3.5894411211999335,
      "learning_rate": 1.1148288548204155e-06,
      "loss": 0.9075,
      "step": 9100
    },
    {
      "epoch": 0.656614119259767,
      "grad_norm": 4.313149444771717,
      "learning_rate": 1.1144097851695958e-06,
      "loss": 0.9955,
      "step": 9101
    },
    {
      "epoch": 0.6566862667291945,
      "grad_norm": 4.600078557214425,
      "learning_rate": 1.1139907638775794e-06,
      "loss": 0.9103,
      "step": 9102
    },
    {
      "epoch": 0.656758414198622,
      "grad_norm": 3.1474941156042964,
      "learning_rate": 1.1135717909672474e-06,
      "loss": 0.9365,
      "step": 9103
    },
    {
      "epoch": 0.6568305616680495,
      "grad_norm": 4.0448905173669845,
      "learning_rate": 1.1131528664614787e-06,
      "loss": 1.0259,
      "step": 9104
    },
    {
      "epoch": 0.656902709137477,
      "grad_norm": 2.8972989954112496,
      "learning_rate": 1.1127339903831496e-06,
      "loss": 1.01,
      "step": 9105
    },
    {
      "epoch": 0.6569748566069045,
      "grad_norm": 3.370593974350693,
      "learning_rate": 1.1123151627551316e-06,
      "loss": 1.0344,
      "step": 9106
    },
    {
      "epoch": 0.657047004076332,
      "grad_norm": 4.176907022739592,
      "learning_rate": 1.1118963836002971e-06,
      "loss": 0.9821,
      "step": 9107
    },
    {
      "epoch": 0.6571191515457595,
      "grad_norm": 3.169029425164524,
      "learning_rate": 1.111477652941513e-06,
      "loss": 1.017,
      "step": 9108
    },
    {
      "epoch": 0.657191299015187,
      "grad_norm": 2.965221677473308,
      "learning_rate": 1.1110589708016454e-06,
      "loss": 0.8824,
      "step": 9109
    },
    {
      "epoch": 0.6572634464846145,
      "grad_norm": 9.416457002269269,
      "learning_rate": 1.1106403372035563e-06,
      "loss": 0.9766,
      "step": 9110
    },
    {
      "epoch": 0.6573355939540421,
      "grad_norm": 3.1391265482845734,
      "learning_rate": 1.1102217521701063e-06,
      "loss": 1.0141,
      "step": 9111
    },
    {
      "epoch": 0.6574077414234696,
      "grad_norm": 4.16889578637278,
      "learning_rate": 1.1098032157241535e-06,
      "loss": 0.9488,
      "step": 9112
    },
    {
      "epoch": 0.6574798888928971,
      "grad_norm": 4.39915945520886,
      "learning_rate": 1.1093847278885497e-06,
      "loss": 0.9965,
      "step": 9113
    },
    {
      "epoch": 0.6575520363623246,
      "grad_norm": 6.19180672009264,
      "learning_rate": 1.1089662886861509e-06,
      "loss": 0.9432,
      "step": 9114
    },
    {
      "epoch": 0.6576241838317521,
      "grad_norm": 3.5038776715494246,
      "learning_rate": 1.1085478981398036e-06,
      "loss": 0.9637,
      "step": 9115
    },
    {
      "epoch": 0.6576963313011797,
      "grad_norm": 4.5673536846609695,
      "learning_rate": 1.1081295562723557e-06,
      "loss": 0.9095,
      "step": 9116
    },
    {
      "epoch": 0.6577684787706071,
      "grad_norm": 3.044281780309169,
      "learning_rate": 1.107711263106651e-06,
      "loss": 1.0418,
      "step": 9117
    },
    {
      "epoch": 0.6578406262400346,
      "grad_norm": 3.2051520655795365,
      "learning_rate": 1.1072930186655312e-06,
      "loss": 0.9076,
      "step": 9118
    },
    {
      "epoch": 0.6579127737094621,
      "grad_norm": 9.839348671919176,
      "learning_rate": 1.1068748229718349e-06,
      "loss": 0.9307,
      "step": 9119
    },
    {
      "epoch": 0.6579849211788896,
      "grad_norm": 3.5221981736360655,
      "learning_rate": 1.1064566760483983e-06,
      "loss": 1.0314,
      "step": 9120
    },
    {
      "epoch": 0.6580570686483171,
      "grad_norm": 0.7517931879550366,
      "learning_rate": 1.1060385779180553e-06,
      "loss": 0.8074,
      "step": 9121
    },
    {
      "epoch": 0.6581292161177447,
      "grad_norm": 3.351043020684314,
      "learning_rate": 1.1056205286036359e-06,
      "loss": 0.9661,
      "step": 9122
    },
    {
      "epoch": 0.6582013635871722,
      "grad_norm": 2.9949594113460623,
      "learning_rate": 1.1052025281279683e-06,
      "loss": 0.9054,
      "step": 9123
    },
    {
      "epoch": 0.6582735110565997,
      "grad_norm": 16.13044398473059,
      "learning_rate": 1.1047845765138783e-06,
      "loss": 1.0212,
      "step": 9124
    },
    {
      "epoch": 0.6583456585260272,
      "grad_norm": 13.569328030820385,
      "learning_rate": 1.1043666737841887e-06,
      "loss": 1.0026,
      "step": 9125
    },
    {
      "epoch": 0.6584178059954547,
      "grad_norm": 5.101110464635334,
      "learning_rate": 1.1039488199617196e-06,
      "loss": 1.0665,
      "step": 9126
    },
    {
      "epoch": 0.6584899534648823,
      "grad_norm": 4.683336163358184,
      "learning_rate": 1.1035310150692888e-06,
      "loss": 0.9258,
      "step": 9127
    },
    {
      "epoch": 0.6585621009343098,
      "grad_norm": 7.184846869358651,
      "learning_rate": 1.1031132591297091e-06,
      "loss": 1.0168,
      "step": 9128
    },
    {
      "epoch": 0.6586342484037372,
      "grad_norm": 4.894994179524438,
      "learning_rate": 1.1026955521657961e-06,
      "loss": 1.0348,
      "step": 9129
    },
    {
      "epoch": 0.6587063958731647,
      "grad_norm": 3.245106166989638,
      "learning_rate": 1.1022778942003564e-06,
      "loss": 0.9178,
      "step": 9130
    },
    {
      "epoch": 0.6587785433425922,
      "grad_norm": 4.09889747868515,
      "learning_rate": 1.1018602852561976e-06,
      "loss": 0.974,
      "step": 9131
    },
    {
      "epoch": 0.6588506908120197,
      "grad_norm": 6.923863385841102,
      "learning_rate": 1.1014427253561236e-06,
      "loss": 0.8187,
      "step": 9132
    },
    {
      "epoch": 0.6589228382814473,
      "grad_norm": 4.066273741161743,
      "learning_rate": 1.1010252145229362e-06,
      "loss": 0.9262,
      "step": 9133
    },
    {
      "epoch": 0.6589949857508748,
      "grad_norm": 4.8131455818254345,
      "learning_rate": 1.100607752779434e-06,
      "loss": 1.0526,
      "step": 9134
    },
    {
      "epoch": 0.6590671332203023,
      "grad_norm": 2.4280791525028773,
      "learning_rate": 1.1001903401484128e-06,
      "loss": 0.9513,
      "step": 9135
    },
    {
      "epoch": 0.6591392806897298,
      "grad_norm": 2.8748693021837832,
      "learning_rate": 1.0997729766526673e-06,
      "loss": 0.9835,
      "step": 9136
    },
    {
      "epoch": 0.6592114281591573,
      "grad_norm": 0.6888536218833029,
      "learning_rate": 1.099355662314986e-06,
      "loss": 0.8481,
      "step": 9137
    },
    {
      "epoch": 0.6592835756285849,
      "grad_norm": 3.589722341782003,
      "learning_rate": 1.0989383971581578e-06,
      "loss": 1.001,
      "step": 9138
    },
    {
      "epoch": 0.6593557230980124,
      "grad_norm": 4.395713382669316,
      "learning_rate": 1.0985211812049682e-06,
      "loss": 0.8861,
      "step": 9139
    },
    {
      "epoch": 0.6594278705674399,
      "grad_norm": 3.1550931415623205,
      "learning_rate": 1.0981040144781997e-06,
      "loss": 0.8522,
      "step": 9140
    },
    {
      "epoch": 0.6595000180368673,
      "grad_norm": 3.3984961493682038,
      "learning_rate": 1.0976868970006324e-06,
      "loss": 0.9116,
      "step": 9141
    },
    {
      "epoch": 0.6595721655062948,
      "grad_norm": 4.131672404476202,
      "learning_rate": 1.0972698287950432e-06,
      "loss": 0.9873,
      "step": 9142
    },
    {
      "epoch": 0.6596443129757223,
      "grad_norm": 2.4257285685201397,
      "learning_rate": 1.096852809884207e-06,
      "loss": 0.9381,
      "step": 9143
    },
    {
      "epoch": 0.6597164604451499,
      "grad_norm": 2.87112430893601,
      "learning_rate": 1.0964358402908965e-06,
      "loss": 1.0157,
      "step": 9144
    },
    {
      "epoch": 0.6597886079145774,
      "grad_norm": 3.2733490779469565,
      "learning_rate": 1.0960189200378787e-06,
      "loss": 1.0177,
      "step": 9145
    },
    {
      "epoch": 0.6598607553840049,
      "grad_norm": 3.0184311534492205,
      "learning_rate": 1.0956020491479215e-06,
      "loss": 1.0205,
      "step": 9146
    },
    {
      "epoch": 0.6599329028534324,
      "grad_norm": 3.3684665879195927,
      "learning_rate": 1.0951852276437882e-06,
      "loss": 0.9614,
      "step": 9147
    },
    {
      "epoch": 0.6600050503228599,
      "grad_norm": 0.7899158131734025,
      "learning_rate": 1.09476845554824e-06,
      "loss": 0.8252,
      "step": 9148
    },
    {
      "epoch": 0.6600771977922875,
      "grad_norm": 3.7477148086842793,
      "learning_rate": 1.0943517328840359e-06,
      "loss": 0.9122,
      "step": 9149
    },
    {
      "epoch": 0.660149345261715,
      "grad_norm": 0.8635572104594702,
      "learning_rate": 1.0939350596739308e-06,
      "loss": 0.9365,
      "step": 9150
    },
    {
      "epoch": 0.6602214927311425,
      "grad_norm": 5.551963342034415,
      "learning_rate": 1.0935184359406788e-06,
      "loss": 0.9309,
      "step": 9151
    },
    {
      "epoch": 0.66029364020057,
      "grad_norm": 4.147610281960612,
      "learning_rate": 1.0931018617070278e-06,
      "loss": 0.9274,
      "step": 9152
    },
    {
      "epoch": 0.6603657876699974,
      "grad_norm": 4.959461671472094,
      "learning_rate": 1.0926853369957284e-06,
      "loss": 0.9218,
      "step": 9153
    },
    {
      "epoch": 0.660437935139425,
      "grad_norm": 3.944065614576223,
      "learning_rate": 1.092268861829523e-06,
      "loss": 0.9469,
      "step": 9154
    },
    {
      "epoch": 0.6605100826088525,
      "grad_norm": 2.7955586836224096,
      "learning_rate": 1.091852436231155e-06,
      "loss": 0.9093,
      "step": 9155
    },
    {
      "epoch": 0.66058223007828,
      "grad_norm": 2.911813967872277,
      "learning_rate": 1.0914360602233636e-06,
      "loss": 0.9181,
      "step": 9156
    },
    {
      "epoch": 0.6606543775477075,
      "grad_norm": 5.696509299622321,
      "learning_rate": 1.0910197338288856e-06,
      "loss": 0.9366,
      "step": 9157
    },
    {
      "epoch": 0.660726525017135,
      "grad_norm": 6.229655302991635,
      "learning_rate": 1.090603457070455e-06,
      "loss": 0.9727,
      "step": 9158
    },
    {
      "epoch": 0.6607986724865625,
      "grad_norm": 3.5918969725890437,
      "learning_rate": 1.0901872299708038e-06,
      "loss": 0.8699,
      "step": 9159
    },
    {
      "epoch": 0.6608708199559901,
      "grad_norm": 4.456105730415252,
      "learning_rate": 1.0897710525526593e-06,
      "loss": 1.027,
      "step": 9160
    },
    {
      "epoch": 0.6609429674254176,
      "grad_norm": 3.818232744463537,
      "learning_rate": 1.089354924838748e-06,
      "loss": 1.0225,
      "step": 9161
    },
    {
      "epoch": 0.6610151148948451,
      "grad_norm": 6.981521194094719,
      "learning_rate": 1.0889388468517932e-06,
      "loss": 1.0416,
      "step": 9162
    },
    {
      "epoch": 0.6610872623642726,
      "grad_norm": 0.6744692861847125,
      "learning_rate": 1.0885228186145152e-06,
      "loss": 0.8401,
      "step": 9163
    },
    {
      "epoch": 0.6611594098337,
      "grad_norm": 2.8326484375969616,
      "learning_rate": 1.088106840149632e-06,
      "loss": 0.8927,
      "step": 9164
    },
    {
      "epoch": 0.6612315573031275,
      "grad_norm": 7.13330932090259,
      "learning_rate": 1.0876909114798585e-06,
      "loss": 0.9743,
      "step": 9165
    },
    {
      "epoch": 0.6613037047725551,
      "grad_norm": 4.842400935178396,
      "learning_rate": 1.0872750326279079e-06,
      "loss": 0.8881,
      "step": 9166
    },
    {
      "epoch": 0.6613758522419826,
      "grad_norm": 9.24011527117543,
      "learning_rate": 1.0868592036164876e-06,
      "loss": 1.0142,
      "step": 9167
    },
    {
      "epoch": 0.6614479997114101,
      "grad_norm": 4.109654406197547,
      "learning_rate": 1.0864434244683072e-06,
      "loss": 0.9788,
      "step": 9168
    },
    {
      "epoch": 0.6615201471808376,
      "grad_norm": 3.970275585689029,
      "learning_rate": 1.0860276952060688e-06,
      "loss": 0.9112,
      "step": 9169
    },
    {
      "epoch": 0.6615922946502651,
      "grad_norm": 4.048793264442303,
      "learning_rate": 1.0856120158524744e-06,
      "loss": 0.9297,
      "step": 9170
    },
    {
      "epoch": 0.6616644421196927,
      "grad_norm": 5.231861570061009,
      "learning_rate": 1.085196386430223e-06,
      "loss": 0.9738,
      "step": 9171
    },
    {
      "epoch": 0.6617365895891202,
      "grad_norm": 3.4906954383402615,
      "learning_rate": 1.0847808069620103e-06,
      "loss": 0.9308,
      "step": 9172
    },
    {
      "epoch": 0.6618087370585477,
      "grad_norm": 3.2456697813720843,
      "learning_rate": 1.08436527747053e-06,
      "loss": 0.9436,
      "step": 9173
    },
    {
      "epoch": 0.6618808845279752,
      "grad_norm": 4.135374951412451,
      "learning_rate": 1.0839497979784722e-06,
      "loss": 0.99,
      "step": 9174
    },
    {
      "epoch": 0.6619530319974027,
      "grad_norm": 3.3105473071072726,
      "learning_rate": 1.0835343685085256e-06,
      "loss": 0.9333,
      "step": 9175
    },
    {
      "epoch": 0.6620251794668301,
      "grad_norm": 3.7853018749399405,
      "learning_rate": 1.0831189890833737e-06,
      "loss": 0.945,
      "step": 9176
    },
    {
      "epoch": 0.6620973269362577,
      "grad_norm": 0.8101139245155398,
      "learning_rate": 1.0827036597256998e-06,
      "loss": 0.9006,
      "step": 9177
    },
    {
      "epoch": 0.6621694744056852,
      "grad_norm": 7.185623587437073,
      "learning_rate": 1.082288380458183e-06,
      "loss": 1.0224,
      "step": 9178
    },
    {
      "epoch": 0.6622416218751127,
      "grad_norm": 3.0294665566396093,
      "learning_rate": 1.0818731513035007e-06,
      "loss": 1.0186,
      "step": 9179
    },
    {
      "epoch": 0.6623137693445402,
      "grad_norm": 7.296973805708186,
      "learning_rate": 1.0814579722843267e-06,
      "loss": 0.9866,
      "step": 9180
    },
    {
      "epoch": 0.6623859168139677,
      "grad_norm": 0.8115715444102708,
      "learning_rate": 1.0810428434233335e-06,
      "loss": 0.8235,
      "step": 9181
    },
    {
      "epoch": 0.6624580642833953,
      "grad_norm": 6.6377840315398355,
      "learning_rate": 1.080627764743187e-06,
      "loss": 0.7932,
      "step": 9182
    },
    {
      "epoch": 0.6625302117528228,
      "grad_norm": 5.142188486651474,
      "learning_rate": 1.0802127362665565e-06,
      "loss": 0.9132,
      "step": 9183
    },
    {
      "epoch": 0.6626023592222503,
      "grad_norm": 12.040819043660921,
      "learning_rate": 1.0797977580161029e-06,
      "loss": 0.9353,
      "step": 9184
    },
    {
      "epoch": 0.6626745066916778,
      "grad_norm": 4.6483277380029575,
      "learning_rate": 1.079382830014487e-06,
      "loss": 0.9644,
      "step": 9185
    },
    {
      "epoch": 0.6627466541611053,
      "grad_norm": 2.957065274686598,
      "learning_rate": 1.078967952284367e-06,
      "loss": 1.0203,
      "step": 9186
    },
    {
      "epoch": 0.6628188016305329,
      "grad_norm": 3.1953668193350544,
      "learning_rate": 1.0785531248483972e-06,
      "loss": 0.934,
      "step": 9187
    },
    {
      "epoch": 0.6628909490999603,
      "grad_norm": 3.554794544662285,
      "learning_rate": 1.0781383477292312e-06,
      "loss": 0.8899,
      "step": 9188
    },
    {
      "epoch": 0.6629630965693878,
      "grad_norm": 2.947805629476748,
      "learning_rate": 1.0777236209495159e-06,
      "loss": 0.9,
      "step": 9189
    },
    {
      "epoch": 0.6630352440388153,
      "grad_norm": 3.7415439633737297,
      "learning_rate": 1.077308944531901e-06,
      "loss": 0.9175,
      "step": 9190
    },
    {
      "epoch": 0.6631073915082428,
      "grad_norm": 6.496117239133721,
      "learning_rate": 1.0768943184990273e-06,
      "loss": 0.9507,
      "step": 9191
    },
    {
      "epoch": 0.6631795389776703,
      "grad_norm": 3.245745147826032,
      "learning_rate": 1.0764797428735396e-06,
      "loss": 0.9186,
      "step": 9192
    },
    {
      "epoch": 0.6632516864470979,
      "grad_norm": 2.7136811985060523,
      "learning_rate": 1.076065217678073e-06,
      "loss": 0.9544,
      "step": 9193
    },
    {
      "epoch": 0.6633238339165254,
      "grad_norm": 6.064086077358507,
      "learning_rate": 1.075650742935265e-06,
      "loss": 1.0324,
      "step": 9194
    },
    {
      "epoch": 0.6633959813859529,
      "grad_norm": 0.7263814639078121,
      "learning_rate": 1.075236318667748e-06,
      "loss": 0.8232,
      "step": 9195
    },
    {
      "epoch": 0.6634681288553804,
      "grad_norm": 5.4786406096483,
      "learning_rate": 1.0748219448981523e-06,
      "loss": 0.9715,
      "step": 9196
    },
    {
      "epoch": 0.6635402763248079,
      "grad_norm": 3.7338657430940354,
      "learning_rate": 1.0744076216491053e-06,
      "loss": 0.8645,
      "step": 9197
    },
    {
      "epoch": 0.6636124237942355,
      "grad_norm": 0.8058112732459195,
      "learning_rate": 1.0739933489432324e-06,
      "loss": 0.8573,
      "step": 9198
    },
    {
      "epoch": 0.663684571263663,
      "grad_norm": 2.7209680213324843,
      "learning_rate": 1.073579126803154e-06,
      "loss": 1.0298,
      "step": 9199
    },
    {
      "epoch": 0.6637567187330904,
      "grad_norm": 3.815602275458922,
      "learning_rate": 1.07316495525149e-06,
      "loss": 0.9508,
      "step": 9200
    },
    {
      "epoch": 0.6638288662025179,
      "grad_norm": 0.6894834690062539,
      "learning_rate": 1.072750834310857e-06,
      "loss": 0.8354,
      "step": 9201
    },
    {
      "epoch": 0.6639010136719454,
      "grad_norm": 3.0586762204019635,
      "learning_rate": 1.0723367640038678e-06,
      "loss": 0.9297,
      "step": 9202
    },
    {
      "epoch": 0.663973161141373,
      "grad_norm": 19.0691626740109,
      "learning_rate": 1.071922744353135e-06,
      "loss": 0.9513,
      "step": 9203
    },
    {
      "epoch": 0.6640453086108005,
      "grad_norm": 2.9312713817475426,
      "learning_rate": 1.0715087753812637e-06,
      "loss": 0.857,
      "step": 9204
    },
    {
      "epoch": 0.664117456080228,
      "grad_norm": 11.823969686838403,
      "learning_rate": 1.071094857110863e-06,
      "loss": 0.908,
      "step": 9205
    },
    {
      "epoch": 0.6641896035496555,
      "grad_norm": 5.303877890917427,
      "learning_rate": 1.0706809895645315e-06,
      "loss": 0.9285,
      "step": 9206
    },
    {
      "epoch": 0.664261751019083,
      "grad_norm": 3.376483449932918,
      "learning_rate": 1.070267172764873e-06,
      "loss": 1.0034,
      "step": 9207
    },
    {
      "epoch": 0.6643338984885105,
      "grad_norm": 3.293359988898798,
      "learning_rate": 1.0698534067344812e-06,
      "loss": 0.9143,
      "step": 9208
    },
    {
      "epoch": 0.6644060459579381,
      "grad_norm": 3.209975792386528,
      "learning_rate": 1.0694396914959515e-06,
      "loss": 1.0329,
      "step": 9209
    },
    {
      "epoch": 0.6644781934273656,
      "grad_norm": 3.0082836228029777,
      "learning_rate": 1.0690260270718757e-06,
      "loss": 1.0017,
      "step": 9210
    },
    {
      "epoch": 0.664550340896793,
      "grad_norm": 3.2784739058635535,
      "learning_rate": 1.0686124134848421e-06,
      "loss": 0.9853,
      "step": 9211
    },
    {
      "epoch": 0.6646224883662205,
      "grad_norm": 3.832394332872315,
      "learning_rate": 1.0681988507574375e-06,
      "loss": 0.9207,
      "step": 9212
    },
    {
      "epoch": 0.664694635835648,
      "grad_norm": 3.0280349828294204,
      "learning_rate": 1.0677853389122428e-06,
      "loss": 0.8892,
      "step": 9213
    },
    {
      "epoch": 0.6647667833050755,
      "grad_norm": 4.779950133506477,
      "learning_rate": 1.0673718779718414e-06,
      "loss": 0.8662,
      "step": 9214
    },
    {
      "epoch": 0.6648389307745031,
      "grad_norm": 4.573838241739281,
      "learning_rate": 1.0669584679588082e-06,
      "loss": 0.9888,
      "step": 9215
    },
    {
      "epoch": 0.6649110782439306,
      "grad_norm": 3.9515516444638235,
      "learning_rate": 1.0665451088957194e-06,
      "loss": 0.8933,
      "step": 9216
    },
    {
      "epoch": 0.6649832257133581,
      "grad_norm": 4.277553549720627,
      "learning_rate": 1.0661318008051465e-06,
      "loss": 0.9785,
      "step": 9217
    },
    {
      "epoch": 0.6650553731827856,
      "grad_norm": 3.5106625947166825,
      "learning_rate": 1.0657185437096589e-06,
      "loss": 0.8539,
      "step": 9218
    },
    {
      "epoch": 0.6651275206522131,
      "grad_norm": 5.604881250517041,
      "learning_rate": 1.065305337631823e-06,
      "loss": 1.0056,
      "step": 9219
    },
    {
      "epoch": 0.6651996681216407,
      "grad_norm": 11.33424740265406,
      "learning_rate": 1.0648921825942035e-06,
      "loss": 0.9657,
      "step": 9220
    },
    {
      "epoch": 0.6652718155910682,
      "grad_norm": 5.255427325411385,
      "learning_rate": 1.0644790786193584e-06,
      "loss": 0.998,
      "step": 9221
    },
    {
      "epoch": 0.6653439630604957,
      "grad_norm": 5.587982614314041,
      "learning_rate": 1.0640660257298493e-06,
      "loss": 1.0196,
      "step": 9222
    },
    {
      "epoch": 0.6654161105299231,
      "grad_norm": 3.2463819465364594,
      "learning_rate": 1.063653023948229e-06,
      "loss": 0.9321,
      "step": 9223
    },
    {
      "epoch": 0.6654882579993506,
      "grad_norm": 4.231305012422799,
      "learning_rate": 1.063240073297051e-06,
      "loss": 1.0683,
      "step": 9224
    },
    {
      "epoch": 0.6655604054687781,
      "grad_norm": 2.7190170156964317,
      "learning_rate": 1.0628271737988647e-06,
      "loss": 0.925,
      "step": 9225
    },
    {
      "epoch": 0.6656325529382057,
      "grad_norm": 4.107010425778242,
      "learning_rate": 1.0624143254762171e-06,
      "loss": 0.974,
      "step": 9226
    },
    {
      "epoch": 0.6657047004076332,
      "grad_norm": 3.221115576322941,
      "learning_rate": 1.0620015283516536e-06,
      "loss": 0.9838,
      "step": 9227
    },
    {
      "epoch": 0.6657768478770607,
      "grad_norm": 3.3489620194877086,
      "learning_rate": 1.0615887824477119e-06,
      "loss": 1.0918,
      "step": 9228
    },
    {
      "epoch": 0.6658489953464882,
      "grad_norm": 3.846103474947397,
      "learning_rate": 1.0611760877869352e-06,
      "loss": 0.9979,
      "step": 9229
    },
    {
      "epoch": 0.6659211428159157,
      "grad_norm": 3.5979902379580473,
      "learning_rate": 1.060763444391855e-06,
      "loss": 0.9902,
      "step": 9230
    },
    {
      "epoch": 0.6659932902853433,
      "grad_norm": 4.069817628265884,
      "learning_rate": 1.0603508522850078e-06,
      "loss": 1.0355,
      "step": 9231
    },
    {
      "epoch": 0.6660654377547708,
      "grad_norm": 3.68365640273468,
      "learning_rate": 1.0599383114889214e-06,
      "loss": 0.9454,
      "step": 9232
    },
    {
      "epoch": 0.6661375852241983,
      "grad_norm": 5.080181186778964,
      "learning_rate": 1.0595258220261236e-06,
      "loss": 0.8826,
      "step": 9233
    },
    {
      "epoch": 0.6662097326936258,
      "grad_norm": 2.5297613593776767,
      "learning_rate": 1.0591133839191393e-06,
      "loss": 0.9356,
      "step": 9234
    },
    {
      "epoch": 0.6662818801630532,
      "grad_norm": 3.5356903515730886,
      "learning_rate": 1.0587009971904901e-06,
      "loss": 1.0421,
      "step": 9235
    },
    {
      "epoch": 0.6663540276324807,
      "grad_norm": 2.769881512831062,
      "learning_rate": 1.0582886618626956e-06,
      "loss": 0.9632,
      "step": 9236
    },
    {
      "epoch": 0.6664261751019083,
      "grad_norm": 4.348976997137472,
      "learning_rate": 1.0578763779582702e-06,
      "loss": 1.0088,
      "step": 9237
    },
    {
      "epoch": 0.6664983225713358,
      "grad_norm": 3.317197096564148,
      "learning_rate": 1.0574641454997286e-06,
      "loss": 0.9769,
      "step": 9238
    },
    {
      "epoch": 0.6665704700407633,
      "grad_norm": 5.852251203317277,
      "learning_rate": 1.0570519645095806e-06,
      "loss": 0.9371,
      "step": 9239
    },
    {
      "epoch": 0.6666426175101908,
      "grad_norm": 3.135819131530291,
      "learning_rate": 1.0566398350103342e-06,
      "loss": 0.9217,
      "step": 9240
    },
    {
      "epoch": 0.6667147649796183,
      "grad_norm": 2.985489881690976,
      "learning_rate": 1.0562277570244942e-06,
      "loss": 1.0097,
      "step": 9241
    },
    {
      "epoch": 0.6667869124490459,
      "grad_norm": 14.057823115513665,
      "learning_rate": 1.0558157305745634e-06,
      "loss": 0.8817,
      "step": 9242
    },
    {
      "epoch": 0.6668590599184734,
      "grad_norm": 0.7980852100129449,
      "learning_rate": 1.055403755683039e-06,
      "loss": 0.8507,
      "step": 9243
    },
    {
      "epoch": 0.6669312073879009,
      "grad_norm": 4.195495715127629,
      "learning_rate": 1.05499183237242e-06,
      "loss": 1.0112,
      "step": 9244
    },
    {
      "epoch": 0.6670033548573284,
      "grad_norm": 3.56764652542074,
      "learning_rate": 1.054579960665197e-06,
      "loss": 0.918,
      "step": 9245
    },
    {
      "epoch": 0.6670755023267559,
      "grad_norm": 3.57937067329008,
      "learning_rate": 1.0541681405838645e-06,
      "loss": 0.9504,
      "step": 9246
    },
    {
      "epoch": 0.6671476497961834,
      "grad_norm": 3.6511476567547607,
      "learning_rate": 1.0537563721509075e-06,
      "loss": 0.8518,
      "step": 9247
    },
    {
      "epoch": 0.6672197972656109,
      "grad_norm": 4.317223657048644,
      "learning_rate": 1.053344655388812e-06,
      "loss": 0.9642,
      "step": 9248
    },
    {
      "epoch": 0.6672919447350384,
      "grad_norm": 4.700814221244163,
      "learning_rate": 1.0529329903200609e-06,
      "loss": 0.8373,
      "step": 9249
    },
    {
      "epoch": 0.6673640922044659,
      "grad_norm": 3.2264346635446626,
      "learning_rate": 1.052521376967133e-06,
      "loss": 1.0398,
      "step": 9250
    },
    {
      "epoch": 0.6674362396738934,
      "grad_norm": 2.702760440299907,
      "learning_rate": 1.052109815352506e-06,
      "loss": 0.9768,
      "step": 9251
    },
    {
      "epoch": 0.667508387143321,
      "grad_norm": 2.908730545536256,
      "learning_rate": 1.0516983054986516e-06,
      "loss": 0.9377,
      "step": 9252
    },
    {
      "epoch": 0.6675805346127485,
      "grad_norm": 6.190460729754358,
      "learning_rate": 1.051286847428044e-06,
      "loss": 1.0102,
      "step": 9253
    },
    {
      "epoch": 0.667652682082176,
      "grad_norm": 2.9775392226447153,
      "learning_rate": 1.0508754411631482e-06,
      "loss": 1.0187,
      "step": 9254
    },
    {
      "epoch": 0.6677248295516035,
      "grad_norm": 3.461718606744383,
      "learning_rate": 1.0504640867264317e-06,
      "loss": 0.8616,
      "step": 9255
    },
    {
      "epoch": 0.667796977021031,
      "grad_norm": 5.7144986385457655,
      "learning_rate": 1.0500527841403558e-06,
      "loss": 0.9181,
      "step": 9256
    },
    {
      "epoch": 0.6678691244904585,
      "grad_norm": 4.712467280509547,
      "learning_rate": 1.0496415334273811e-06,
      "loss": 0.9513,
      "step": 9257
    },
    {
      "epoch": 0.667941271959886,
      "grad_norm": 3.107193531443721,
      "learning_rate": 1.0492303346099643e-06,
      "loss": 0.9376,
      "step": 9258
    },
    {
      "epoch": 0.6680134194293135,
      "grad_norm": 3.709756570937343,
      "learning_rate": 1.04881918771056e-06,
      "loss": 0.9062,
      "step": 9259
    },
    {
      "epoch": 0.668085566898741,
      "grad_norm": 4.9772939099612845,
      "learning_rate": 1.0484080927516178e-06,
      "loss": 1.0032,
      "step": 9260
    },
    {
      "epoch": 0.6681577143681685,
      "grad_norm": 2.3092374727132983,
      "learning_rate": 1.0479970497555868e-06,
      "loss": 0.974,
      "step": 9261
    },
    {
      "epoch": 0.668229861837596,
      "grad_norm": 3.8656792059438407,
      "learning_rate": 1.0475860587449127e-06,
      "loss": 1.0381,
      "step": 9262
    },
    {
      "epoch": 0.6683020093070235,
      "grad_norm": 4.609109334240236,
      "learning_rate": 1.0471751197420387e-06,
      "loss": 0.8521,
      "step": 9263
    },
    {
      "epoch": 0.6683741567764511,
      "grad_norm": 4.521603543583479,
      "learning_rate": 1.0467642327694038e-06,
      "loss": 0.9427,
      "step": 9264
    },
    {
      "epoch": 0.6684463042458786,
      "grad_norm": 4.989120472548204,
      "learning_rate": 1.0463533978494454e-06,
      "loss": 0.8725,
      "step": 9265
    },
    {
      "epoch": 0.6685184517153061,
      "grad_norm": 4.082622985295541,
      "learning_rate": 1.0459426150045985e-06,
      "loss": 0.8683,
      "step": 9266
    },
    {
      "epoch": 0.6685905991847336,
      "grad_norm": 3.177249958275754,
      "learning_rate": 1.0455318842572922e-06,
      "loss": 0.8811,
      "step": 9267
    },
    {
      "epoch": 0.6686627466541611,
      "grad_norm": 4.373981030193231,
      "learning_rate": 1.0451212056299578e-06,
      "loss": 1.0308,
      "step": 9268
    },
    {
      "epoch": 0.6687348941235887,
      "grad_norm": 3.9555409403208306,
      "learning_rate": 1.044710579145019e-06,
      "loss": 0.9758,
      "step": 9269
    },
    {
      "epoch": 0.6688070415930161,
      "grad_norm": 6.278721846240757,
      "learning_rate": 1.0443000048248986e-06,
      "loss": 0.8993,
      "step": 9270
    },
    {
      "epoch": 0.6688791890624436,
      "grad_norm": 3.602011960795029,
      "learning_rate": 1.0438894826920175e-06,
      "loss": 0.9252,
      "step": 9271
    },
    {
      "epoch": 0.6689513365318711,
      "grad_norm": 0.7758613137902863,
      "learning_rate": 1.043479012768792e-06,
      "loss": 0.8592,
      "step": 9272
    },
    {
      "epoch": 0.6690234840012986,
      "grad_norm": 4.081860932077824,
      "learning_rate": 1.0430685950776368e-06,
      "loss": 0.8457,
      "step": 9273
    },
    {
      "epoch": 0.6690956314707261,
      "grad_norm": 4.295812191003128,
      "learning_rate": 1.0426582296409634e-06,
      "loss": 0.935,
      "step": 9274
    },
    {
      "epoch": 0.6691677789401537,
      "grad_norm": 4.335930241759128,
      "learning_rate": 1.042247916481181e-06,
      "loss": 0.9011,
      "step": 9275
    },
    {
      "epoch": 0.6692399264095812,
      "grad_norm": 6.0896094707180835,
      "learning_rate": 1.0418376556206929e-06,
      "loss": 0.9273,
      "step": 9276
    },
    {
      "epoch": 0.6693120738790087,
      "grad_norm": 4.015320287639468,
      "learning_rate": 1.041427447081904e-06,
      "loss": 0.9654,
      "step": 9277
    },
    {
      "epoch": 0.6693842213484362,
      "grad_norm": 4.51187368680124,
      "learning_rate": 1.0410172908872138e-06,
      "loss": 0.9173,
      "step": 9278
    },
    {
      "epoch": 0.6694563688178637,
      "grad_norm": 3.6830656900878704,
      "learning_rate": 1.0406071870590189e-06,
      "loss": 0.9004,
      "step": 9279
    },
    {
      "epoch": 0.6695285162872913,
      "grad_norm": 3.5040945897682283,
      "learning_rate": 1.0401971356197142e-06,
      "loss": 0.9036,
      "step": 9280
    },
    {
      "epoch": 0.6696006637567188,
      "grad_norm": 3.2043819287386843,
      "learning_rate": 1.0397871365916918e-06,
      "loss": 0.8548,
      "step": 9281
    },
    {
      "epoch": 0.6696728112261462,
      "grad_norm": 5.646691270044284,
      "learning_rate": 1.0393771899973374e-06,
      "loss": 0.9018,
      "step": 9282
    },
    {
      "epoch": 0.6697449586955737,
      "grad_norm": 3.2268097346353715,
      "learning_rate": 1.0389672958590402e-06,
      "loss": 1.045,
      "step": 9283
    },
    {
      "epoch": 0.6698171061650012,
      "grad_norm": 5.507456839691479,
      "learning_rate": 1.0385574541991808e-06,
      "loss": 0.7191,
      "step": 9284
    },
    {
      "epoch": 0.6698892536344287,
      "grad_norm": 3.2967062088845216,
      "learning_rate": 1.0381476650401393e-06,
      "loss": 0.9398,
      "step": 9285
    },
    {
      "epoch": 0.6699614011038563,
      "grad_norm": 2.7456140221251037,
      "learning_rate": 1.0377379284042935e-06,
      "loss": 0.938,
      "step": 9286
    },
    {
      "epoch": 0.6700335485732838,
      "grad_norm": 3.444890072069438,
      "learning_rate": 1.0373282443140169e-06,
      "loss": 0.8466,
      "step": 9287
    },
    {
      "epoch": 0.6701056960427113,
      "grad_norm": 4.141815129669524,
      "learning_rate": 1.0369186127916816e-06,
      "loss": 1.0255,
      "step": 9288
    },
    {
      "epoch": 0.6701778435121388,
      "grad_norm": 3.612445796573107,
      "learning_rate": 1.0365090338596556e-06,
      "loss": 1.0234,
      "step": 9289
    },
    {
      "epoch": 0.6702499909815663,
      "grad_norm": 4.043232460451421,
      "learning_rate": 1.0360995075403049e-06,
      "loss": 0.8969,
      "step": 9290
    },
    {
      "epoch": 0.6703221384509939,
      "grad_norm": 5.0890063331041375,
      "learning_rate": 1.0356900338559914e-06,
      "loss": 0.9291,
      "step": 9291
    },
    {
      "epoch": 0.6703942859204214,
      "grad_norm": 3.5174708966947033,
      "learning_rate": 1.0352806128290757e-06,
      "loss": 1.0753,
      "step": 9292
    },
    {
      "epoch": 0.6704664333898489,
      "grad_norm": 0.755097944588605,
      "learning_rate": 1.0348712444819138e-06,
      "loss": 0.8418,
      "step": 9293
    },
    {
      "epoch": 0.6705385808592763,
      "grad_norm": 5.447309568384513,
      "learning_rate": 1.0344619288368611e-06,
      "loss": 0.8755,
      "step": 9294
    },
    {
      "epoch": 0.6706107283287038,
      "grad_norm": 3.3852995084395148,
      "learning_rate": 1.0340526659162676e-06,
      "loss": 0.8777,
      "step": 9295
    },
    {
      "epoch": 0.6706828757981314,
      "grad_norm": 5.327131698959724,
      "learning_rate": 1.0336434557424825e-06,
      "loss": 1.0025,
      "step": 9296
    },
    {
      "epoch": 0.6707550232675589,
      "grad_norm": 4.359804631326941,
      "learning_rate": 1.033234298337851e-06,
      "loss": 0.8761,
      "step": 9297
    },
    {
      "epoch": 0.6708271707369864,
      "grad_norm": 6.791424269143817,
      "learning_rate": 1.0328251937247162e-06,
      "loss": 0.9703,
      "step": 9298
    },
    {
      "epoch": 0.6708993182064139,
      "grad_norm": 6.88384385451754,
      "learning_rate": 1.0324161419254166e-06,
      "loss": 1.0139,
      "step": 9299
    },
    {
      "epoch": 0.6709714656758414,
      "grad_norm": 7.497515712007988,
      "learning_rate": 1.0320071429622895e-06,
      "loss": 1.0466,
      "step": 9300
    },
    {
      "epoch": 0.671043613145269,
      "grad_norm": 4.103093557244576,
      "learning_rate": 1.031598196857669e-06,
      "loss": 0.9646,
      "step": 9301
    },
    {
      "epoch": 0.6711157606146965,
      "grad_norm": 3.4053300087608096,
      "learning_rate": 1.0311893036338858e-06,
      "loss": 0.8508,
      "step": 9302
    },
    {
      "epoch": 0.671187908084124,
      "grad_norm": 5.164102574364646,
      "learning_rate": 1.0307804633132687e-06,
      "loss": 1.0106,
      "step": 9303
    },
    {
      "epoch": 0.6712600555535515,
      "grad_norm": 3.0276425318789344,
      "learning_rate": 1.0303716759181424e-06,
      "loss": 0.9531,
      "step": 9304
    },
    {
      "epoch": 0.671332203022979,
      "grad_norm": 2.507141118481801,
      "learning_rate": 1.0299629414708305e-06,
      "loss": 0.8939,
      "step": 9305
    },
    {
      "epoch": 0.6714043504924064,
      "grad_norm": 6.067576062857758,
      "learning_rate": 1.0295542599936493e-06,
      "loss": 0.9776,
      "step": 9306
    },
    {
      "epoch": 0.671476497961834,
      "grad_norm": 3.916700863519816,
      "learning_rate": 1.0291456315089195e-06,
      "loss": 0.9032,
      "step": 9307
    },
    {
      "epoch": 0.6715486454312615,
      "grad_norm": 2.794067422811073,
      "learning_rate": 1.0287370560389518e-06,
      "loss": 0.9615,
      "step": 9308
    },
    {
      "epoch": 0.671620792900689,
      "grad_norm": 2.909453564493737,
      "learning_rate": 1.028328533606058e-06,
      "loss": 0.9387,
      "step": 9309
    },
    {
      "epoch": 0.6716929403701165,
      "grad_norm": 3.760453214318082,
      "learning_rate": 1.0279200642325462e-06,
      "loss": 1.0288,
      "step": 9310
    },
    {
      "epoch": 0.671765087839544,
      "grad_norm": 4.208483602025274,
      "learning_rate": 1.027511647940721e-06,
      "loss": 0.9006,
      "step": 9311
    },
    {
      "epoch": 0.6718372353089715,
      "grad_norm": 5.789457197028964,
      "learning_rate": 1.0271032847528847e-06,
      "loss": 0.9455,
      "step": 9312
    },
    {
      "epoch": 0.6719093827783991,
      "grad_norm": 2.9556160567679313,
      "learning_rate": 1.026694974691337e-06,
      "loss": 0.8982,
      "step": 9313
    },
    {
      "epoch": 0.6719815302478266,
      "grad_norm": 4.353792725100685,
      "learning_rate": 1.0262867177783733e-06,
      "loss": 0.955,
      "step": 9314
    },
    {
      "epoch": 0.6720536777172541,
      "grad_norm": 3.1999840199548353,
      "learning_rate": 1.0258785140362873e-06,
      "loss": 0.9853,
      "step": 9315
    },
    {
      "epoch": 0.6721258251866816,
      "grad_norm": 0.7908791129914352,
      "learning_rate": 1.0254703634873693e-06,
      "loss": 0.844,
      "step": 9316
    },
    {
      "epoch": 0.672197972656109,
      "grad_norm": 3.3784110174922164,
      "learning_rate": 1.0250622661539074e-06,
      "loss": 0.9751,
      "step": 9317
    },
    {
      "epoch": 0.6722701201255366,
      "grad_norm": 4.475432834258054,
      "learning_rate": 1.024654222058186e-06,
      "loss": 0.8822,
      "step": 9318
    },
    {
      "epoch": 0.6723422675949641,
      "grad_norm": 3.780546643124166,
      "learning_rate": 1.0242462312224871e-06,
      "loss": 0.8292,
      "step": 9319
    },
    {
      "epoch": 0.6724144150643916,
      "grad_norm": 4.719637351571271,
      "learning_rate": 1.02383829366909e-06,
      "loss": 0.9893,
      "step": 9320
    },
    {
      "epoch": 0.6724865625338191,
      "grad_norm": 3.1478540528333667,
      "learning_rate": 1.0234304094202684e-06,
      "loss": 0.8807,
      "step": 9321
    },
    {
      "epoch": 0.6725587100032466,
      "grad_norm": 4.773761744498494,
      "learning_rate": 1.0230225784982986e-06,
      "loss": 0.871,
      "step": 9322
    },
    {
      "epoch": 0.6726308574726741,
      "grad_norm": 3.533231061378984,
      "learning_rate": 1.0226148009254486e-06,
      "loss": 0.9066,
      "step": 9323
    },
    {
      "epoch": 0.6727030049421017,
      "grad_norm": 5.8672066801242755,
      "learning_rate": 1.0222070767239857e-06,
      "loss": 0.9831,
      "step": 9324
    },
    {
      "epoch": 0.6727751524115292,
      "grad_norm": 3.1694224229144727,
      "learning_rate": 1.021799405916175e-06,
      "loss": 0.9989,
      "step": 9325
    },
    {
      "epoch": 0.6728472998809567,
      "grad_norm": 0.6949356214976693,
      "learning_rate": 1.0213917885242778e-06,
      "loss": 0.8139,
      "step": 9326
    },
    {
      "epoch": 0.6729194473503842,
      "grad_norm": 2.7469454186692888,
      "learning_rate": 1.0209842245705519e-06,
      "loss": 0.9992,
      "step": 9327
    },
    {
      "epoch": 0.6729915948198117,
      "grad_norm": 3.2896712004941833,
      "learning_rate": 1.0205767140772535e-06,
      "loss": 0.997,
      "step": 9328
    },
    {
      "epoch": 0.6730637422892392,
      "grad_norm": 2.877994967659692,
      "learning_rate": 1.0201692570666358e-06,
      "loss": 0.9467,
      "step": 9329
    },
    {
      "epoch": 0.6731358897586667,
      "grad_norm": 3.069208077544858,
      "learning_rate": 1.019761853560947e-06,
      "loss": 0.9174,
      "step": 9330
    },
    {
      "epoch": 0.6732080372280942,
      "grad_norm": 3.0211977684042046,
      "learning_rate": 1.0193545035824346e-06,
      "loss": 0.8555,
      "step": 9331
    },
    {
      "epoch": 0.6732801846975217,
      "grad_norm": 3.9185285539130668,
      "learning_rate": 1.0189472071533428e-06,
      "loss": 1.0311,
      "step": 9332
    },
    {
      "epoch": 0.6733523321669492,
      "grad_norm": 0.7254426591377001,
      "learning_rate": 1.0185399642959118e-06,
      "loss": 0.805,
      "step": 9333
    },
    {
      "epoch": 0.6734244796363767,
      "grad_norm": 3.455785080611418,
      "learning_rate": 1.0181327750323804e-06,
      "loss": 0.9662,
      "step": 9334
    },
    {
      "epoch": 0.6734966271058043,
      "grad_norm": 12.11279122678629,
      "learning_rate": 1.017725639384984e-06,
      "loss": 0.9769,
      "step": 9335
    },
    {
      "epoch": 0.6735687745752318,
      "grad_norm": 6.134535237734625,
      "learning_rate": 1.0173185573759529e-06,
      "loss": 0.9279,
      "step": 9336
    },
    {
      "epoch": 0.6736409220446593,
      "grad_norm": 3.613566685305247,
      "learning_rate": 1.016911529027519e-06,
      "loss": 0.9106,
      "step": 9337
    },
    {
      "epoch": 0.6737130695140868,
      "grad_norm": 3.9205906197274816,
      "learning_rate": 1.0165045543619066e-06,
      "loss": 0.9289,
      "step": 9338
    },
    {
      "epoch": 0.6737852169835143,
      "grad_norm": 15.281326387366594,
      "learning_rate": 1.0160976334013396e-06,
      "loss": 0.9514,
      "step": 9339
    },
    {
      "epoch": 0.6738573644529419,
      "grad_norm": 5.317765263306559,
      "learning_rate": 1.0156907661680385e-06,
      "loss": 0.9743,
      "step": 9340
    },
    {
      "epoch": 0.6739295119223693,
      "grad_norm": 4.710102570342236,
      "learning_rate": 1.015283952684221e-06,
      "loss": 0.9533,
      "step": 9341
    },
    {
      "epoch": 0.6740016593917968,
      "grad_norm": 3.7309126504955112,
      "learning_rate": 1.0148771929721017e-06,
      "loss": 0.9522,
      "step": 9342
    },
    {
      "epoch": 0.6740738068612243,
      "grad_norm": 7.908252809179565,
      "learning_rate": 1.0144704870538916e-06,
      "loss": 0.9057,
      "step": 9343
    },
    {
      "epoch": 0.6741459543306518,
      "grad_norm": 5.594069487755572,
      "learning_rate": 1.0140638349518012e-06,
      "loss": 0.9392,
      "step": 9344
    },
    {
      "epoch": 0.6742181018000794,
      "grad_norm": 3.8193409314241196,
      "learning_rate": 1.013657236688033e-06,
      "loss": 0.9696,
      "step": 9345
    },
    {
      "epoch": 0.6742902492695069,
      "grad_norm": 2.939661083279142,
      "learning_rate": 1.0132506922847936e-06,
      "loss": 0.9278,
      "step": 9346
    },
    {
      "epoch": 0.6743623967389344,
      "grad_norm": 6.248930877796498,
      "learning_rate": 1.0128442017642802e-06,
      "loss": 0.8841,
      "step": 9347
    },
    {
      "epoch": 0.6744345442083619,
      "grad_norm": 6.327899542665821,
      "learning_rate": 1.0124377651486906e-06,
      "loss": 0.9592,
      "step": 9348
    },
    {
      "epoch": 0.6745066916777894,
      "grad_norm": 2.7760403306873154,
      "learning_rate": 1.012031382460219e-06,
      "loss": 1.0494,
      "step": 9349
    },
    {
      "epoch": 0.674578839147217,
      "grad_norm": 4.176130203497312,
      "learning_rate": 1.011625053721056e-06,
      "loss": 0.8551,
      "step": 9350
    },
    {
      "epoch": 0.6746509866166445,
      "grad_norm": 4.79498432685908,
      "learning_rate": 1.01121877895339e-06,
      "loss": 0.95,
      "step": 9351
    },
    {
      "epoch": 0.674723134086072,
      "grad_norm": 2.3854358677875607,
      "learning_rate": 1.0108125581794072e-06,
      "loss": 0.9108,
      "step": 9352
    },
    {
      "epoch": 0.6747952815554994,
      "grad_norm": 3.666524682041368,
      "learning_rate": 1.0104063914212876e-06,
      "loss": 0.9659,
      "step": 9353
    },
    {
      "epoch": 0.6748674290249269,
      "grad_norm": 3.322309544187215,
      "learning_rate": 1.0100002787012118e-06,
      "loss": 1.0001,
      "step": 9354
    },
    {
      "epoch": 0.6749395764943544,
      "grad_norm": 3.8676231061875863,
      "learning_rate": 1.009594220041356e-06,
      "loss": 0.9527,
      "step": 9355
    },
    {
      "epoch": 0.675011723963782,
      "grad_norm": 3.5789795234139175,
      "learning_rate": 1.0091882154638934e-06,
      "loss": 0.9936,
      "step": 9356
    },
    {
      "epoch": 0.6750838714332095,
      "grad_norm": 5.274102243982426,
      "learning_rate": 1.0087822649909945e-06,
      "loss": 0.8759,
      "step": 9357
    },
    {
      "epoch": 0.675156018902637,
      "grad_norm": 7.730393496586092,
      "learning_rate": 1.0083763686448267e-06,
      "loss": 0.9941,
      "step": 9358
    },
    {
      "epoch": 0.6752281663720645,
      "grad_norm": 11.36941827337705,
      "learning_rate": 1.0079705264475556e-06,
      "loss": 1.0109,
      "step": 9359
    },
    {
      "epoch": 0.675300313841492,
      "grad_norm": 5.51718852330244,
      "learning_rate": 1.0075647384213397e-06,
      "loss": 0.9037,
      "step": 9360
    },
    {
      "epoch": 0.6753724613109195,
      "grad_norm": 3.3666760970131513,
      "learning_rate": 1.0071590045883413e-06,
      "loss": 1.016,
      "step": 9361
    },
    {
      "epoch": 0.6754446087803471,
      "grad_norm": 4.89777044837651,
      "learning_rate": 1.0067533249707137e-06,
      "loss": 1.0804,
      "step": 9362
    },
    {
      "epoch": 0.6755167562497746,
      "grad_norm": 11.55962736169341,
      "learning_rate": 1.0063476995906097e-06,
      "loss": 1.0219,
      "step": 9363
    },
    {
      "epoch": 0.675588903719202,
      "grad_norm": 8.067158617303757,
      "learning_rate": 1.0059421284701798e-06,
      "loss": 0.9517,
      "step": 9364
    },
    {
      "epoch": 0.6756610511886295,
      "grad_norm": 4.079053517305932,
      "learning_rate": 1.0055366116315705e-06,
      "loss": 1.037,
      "step": 9365
    },
    {
      "epoch": 0.675733198658057,
      "grad_norm": 4.62466140744865,
      "learning_rate": 1.005131149096925e-06,
      "loss": 0.9092,
      "step": 9366
    },
    {
      "epoch": 0.6758053461274846,
      "grad_norm": 3.250091551444888,
      "learning_rate": 1.004725740888385e-06,
      "loss": 0.9584,
      "step": 9367
    },
    {
      "epoch": 0.6758774935969121,
      "grad_norm": 3.7040610116337547,
      "learning_rate": 1.0043203870280885e-06,
      "loss": 0.9931,
      "step": 9368
    },
    {
      "epoch": 0.6759496410663396,
      "grad_norm": 2.9376829374448477,
      "learning_rate": 1.0039150875381692e-06,
      "loss": 0.9831,
      "step": 9369
    },
    {
      "epoch": 0.6760217885357671,
      "grad_norm": 0.7218396083198211,
      "learning_rate": 1.0035098424407592e-06,
      "loss": 0.8517,
      "step": 9370
    },
    {
      "epoch": 0.6760939360051946,
      "grad_norm": 0.8145735763654909,
      "learning_rate": 1.0031046517579882e-06,
      "loss": 0.8927,
      "step": 9371
    },
    {
      "epoch": 0.6761660834746221,
      "grad_norm": 3.67968374199483,
      "learning_rate": 1.0026995155119816e-06,
      "loss": 0.9761,
      "step": 9372
    },
    {
      "epoch": 0.6762382309440497,
      "grad_norm": 6.031247470044806,
      "learning_rate": 1.0022944337248626e-06,
      "loss": 1.0002,
      "step": 9373
    },
    {
      "epoch": 0.6763103784134772,
      "grad_norm": 6.526826519565531,
      "learning_rate": 1.001889406418752e-06,
      "loss": 1.0333,
      "step": 9374
    },
    {
      "epoch": 0.6763825258829047,
      "grad_norm": 3.9312959265035192,
      "learning_rate": 1.0014844336157646e-06,
      "loss": 1.06,
      "step": 9375
    },
    {
      "epoch": 0.6764546733523321,
      "grad_norm": 4.198707954034951,
      "learning_rate": 1.0010795153380171e-06,
      "loss": 0.8857,
      "step": 9376
    },
    {
      "epoch": 0.6765268208217596,
      "grad_norm": 3.2073772211280107,
      "learning_rate": 1.000674651607619e-06,
      "loss": 1.0407,
      "step": 9377
    },
    {
      "epoch": 0.6765989682911872,
      "grad_norm": 4.17572735144121,
      "learning_rate": 1.0002698424466784e-06,
      "loss": 0.986,
      "step": 9378
    },
    {
      "epoch": 0.6766711157606147,
      "grad_norm": 2.6139428822202695,
      "learning_rate": 9.99865087877301e-07,
      "loss": 0.9776,
      "step": 9379
    },
    {
      "epoch": 0.6767432632300422,
      "grad_norm": 3.7302634617107695,
      "learning_rate": 9.994603879215891e-07,
      "loss": 0.9724,
      "step": 9380
    },
    {
      "epoch": 0.6768154106994697,
      "grad_norm": 4.316945093275393,
      "learning_rate": 9.990557426016412e-07,
      "loss": 0.8508,
      "step": 9381
    },
    {
      "epoch": 0.6768875581688972,
      "grad_norm": 3.0337314477203163,
      "learning_rate": 9.98651151939554e-07,
      "loss": 0.9155,
      "step": 9382
    },
    {
      "epoch": 0.6769597056383247,
      "grad_norm": 4.517074194021169,
      "learning_rate": 9.982466159574213e-07,
      "loss": 0.9789,
      "step": 9383
    },
    {
      "epoch": 0.6770318531077523,
      "grad_norm": 5.669902700299131,
      "learning_rate": 9.97842134677331e-07,
      "loss": 0.9433,
      "step": 9384
    },
    {
      "epoch": 0.6771040005771798,
      "grad_norm": 8.846310517827872,
      "learning_rate": 9.974377081213732e-07,
      "loss": 1.0326,
      "step": 9385
    },
    {
      "epoch": 0.6771761480466073,
      "grad_norm": 6.997506106129688,
      "learning_rate": 9.970333363116302e-07,
      "loss": 0.9119,
      "step": 9386
    },
    {
      "epoch": 0.6772482955160348,
      "grad_norm": 3.3661579584966437,
      "learning_rate": 9.966290192701839e-07,
      "loss": 0.967,
      "step": 9387
    },
    {
      "epoch": 0.6773204429854622,
      "grad_norm": 9.121647885335356,
      "learning_rate": 9.962247570191122e-07,
      "loss": 0.9996,
      "step": 9388
    },
    {
      "epoch": 0.6773925904548898,
      "grad_norm": 2.9402657332964757,
      "learning_rate": 9.95820549580491e-07,
      "loss": 0.9699,
      "step": 9389
    },
    {
      "epoch": 0.6774647379243173,
      "grad_norm": 3.809749799507977,
      "learning_rate": 9.95416396976392e-07,
      "loss": 0.9651,
      "step": 9390
    },
    {
      "epoch": 0.6775368853937448,
      "grad_norm": 4.779787126533522,
      "learning_rate": 9.950122992288857e-07,
      "loss": 0.9797,
      "step": 9391
    },
    {
      "epoch": 0.6776090328631723,
      "grad_norm": 0.6741722886574506,
      "learning_rate": 9.946082563600365e-07,
      "loss": 0.8505,
      "step": 9392
    },
    {
      "epoch": 0.6776811803325998,
      "grad_norm": 4.221713007509326,
      "learning_rate": 9.942042683919088e-07,
      "loss": 0.9315,
      "step": 9393
    },
    {
      "epoch": 0.6777533278020273,
      "grad_norm": 3.463253994039982,
      "learning_rate": 9.938003353465622e-07,
      "loss": 1.0603,
      "step": 9394
    },
    {
      "epoch": 0.6778254752714549,
      "grad_norm": 3.47023720691797,
      "learning_rate": 9.933964572460549e-07,
      "loss": 0.8918,
      "step": 9395
    },
    {
      "epoch": 0.6778976227408824,
      "grad_norm": 3.90677401270413,
      "learning_rate": 9.929926341124415e-07,
      "loss": 0.8897,
      "step": 9396
    },
    {
      "epoch": 0.6779697702103099,
      "grad_norm": 4.730321376278358,
      "learning_rate": 9.92588865967771e-07,
      "loss": 0.9473,
      "step": 9397
    },
    {
      "epoch": 0.6780419176797374,
      "grad_norm": 3.378513767622553,
      "learning_rate": 9.921851528340944e-07,
      "loss": 0.9348,
      "step": 9398
    },
    {
      "epoch": 0.678114065149165,
      "grad_norm": 4.0768346389346934,
      "learning_rate": 9.917814947334547e-07,
      "loss": 0.9524,
      "step": 9399
    },
    {
      "epoch": 0.6781862126185924,
      "grad_norm": 3.2619323295062133,
      "learning_rate": 9.913778916878965e-07,
      "loss": 0.9197,
      "step": 9400
    },
    {
      "epoch": 0.6782583600880199,
      "grad_norm": 4.706911102914235,
      "learning_rate": 9.90974343719457e-07,
      "loss": 1.0121,
      "step": 9401
    },
    {
      "epoch": 0.6783305075574474,
      "grad_norm": 2.5656270741536797,
      "learning_rate": 9.905708508501738e-07,
      "loss": 0.9502,
      "step": 9402
    },
    {
      "epoch": 0.6784026550268749,
      "grad_norm": 4.895239265057246,
      "learning_rate": 9.901674131020792e-07,
      "loss": 0.9559,
      "step": 9403
    },
    {
      "epoch": 0.6784748024963024,
      "grad_norm": 4.412212477007509,
      "learning_rate": 9.89764030497204e-07,
      "loss": 1.0039,
      "step": 9404
    },
    {
      "epoch": 0.67854694996573,
      "grad_norm": 2.548685565337928,
      "learning_rate": 9.893607030575754e-07,
      "loss": 1.0062,
      "step": 9405
    },
    {
      "epoch": 0.6786190974351575,
      "grad_norm": 2.207940540528996,
      "learning_rate": 9.889574308052176e-07,
      "loss": 0.9972,
      "step": 9406
    },
    {
      "epoch": 0.678691244904585,
      "grad_norm": 3.405094195616043,
      "learning_rate": 9.885542137621526e-07,
      "loss": 0.8848,
      "step": 9407
    },
    {
      "epoch": 0.6787633923740125,
      "grad_norm": 3.9384743832128857,
      "learning_rate": 9.881510519503967e-07,
      "loss": 0.9829,
      "step": 9408
    },
    {
      "epoch": 0.67883553984344,
      "grad_norm": 5.565785841248264,
      "learning_rate": 9.877479453919662e-07,
      "loss": 0.9948,
      "step": 9409
    },
    {
      "epoch": 0.6789076873128675,
      "grad_norm": 5.702549066595885,
      "learning_rate": 9.873448941088732e-07,
      "loss": 0.9519,
      "step": 9410
    },
    {
      "epoch": 0.6789798347822951,
      "grad_norm": 3.8394801296550063,
      "learning_rate": 9.86941898123127e-07,
      "loss": 0.9754,
      "step": 9411
    },
    {
      "epoch": 0.6790519822517225,
      "grad_norm": 5.767085027494867,
      "learning_rate": 9.865389574567333e-07,
      "loss": 0.9443,
      "step": 9412
    },
    {
      "epoch": 0.67912412972115,
      "grad_norm": 7.297478791768332,
      "learning_rate": 9.861360721316962e-07,
      "loss": 0.9803,
      "step": 9413
    },
    {
      "epoch": 0.6791962771905775,
      "grad_norm": 3.8644616586839087,
      "learning_rate": 9.857332421700135e-07,
      "loss": 1.0356,
      "step": 9414
    },
    {
      "epoch": 0.679268424660005,
      "grad_norm": 0.7002977112787834,
      "learning_rate": 9.853304675936854e-07,
      "loss": 0.7585,
      "step": 9415
    },
    {
      "epoch": 0.6793405721294326,
      "grad_norm": 4.821576563264336,
      "learning_rate": 9.849277484247033e-07,
      "loss": 0.9364,
      "step": 9416
    },
    {
      "epoch": 0.6794127195988601,
      "grad_norm": 2.8913066988084655,
      "learning_rate": 9.845250846850595e-07,
      "loss": 0.8835,
      "step": 9417
    },
    {
      "epoch": 0.6794848670682876,
      "grad_norm": 3.4186522754042934,
      "learning_rate": 9.841224763967416e-07,
      "loss": 1.0386,
      "step": 9418
    },
    {
      "epoch": 0.6795570145377151,
      "grad_norm": 4.6134987181490885,
      "learning_rate": 9.837199235817343e-07,
      "loss": 0.9613,
      "step": 9419
    },
    {
      "epoch": 0.6796291620071426,
      "grad_norm": 5.649695882798935,
      "learning_rate": 9.833174262620208e-07,
      "loss": 0.919,
      "step": 9420
    },
    {
      "epoch": 0.6797013094765701,
      "grad_norm": 0.7726903061260788,
      "learning_rate": 9.829149844595775e-07,
      "loss": 0.8589,
      "step": 9421
    },
    {
      "epoch": 0.6797734569459977,
      "grad_norm": 6.49396469451546,
      "learning_rate": 9.825125981963834e-07,
      "loss": 0.9841,
      "step": 9422
    },
    {
      "epoch": 0.6798456044154251,
      "grad_norm": 10.980258913861261,
      "learning_rate": 9.821102674944089e-07,
      "loss": 0.822,
      "step": 9423
    },
    {
      "epoch": 0.6799177518848526,
      "grad_norm": 10.294680464289137,
      "learning_rate": 9.817079923756244e-07,
      "loss": 0.9778,
      "step": 9424
    },
    {
      "epoch": 0.6799898993542801,
      "grad_norm": 8.699411253903955,
      "learning_rate": 9.81305772861997e-07,
      "loss": 1.0169,
      "step": 9425
    },
    {
      "epoch": 0.6800620468237076,
      "grad_norm": 3.4212192969709365,
      "learning_rate": 9.809036089754903e-07,
      "loss": 0.9177,
      "step": 9426
    },
    {
      "epoch": 0.6801341942931352,
      "grad_norm": 3.966740858094756,
      "learning_rate": 9.805015007380648e-07,
      "loss": 0.9917,
      "step": 9427
    },
    {
      "epoch": 0.6802063417625627,
      "grad_norm": 2.8875744368301337,
      "learning_rate": 9.800994481716784e-07,
      "loss": 1.0559,
      "step": 9428
    },
    {
      "epoch": 0.6802784892319902,
      "grad_norm": 3.1768757903469376,
      "learning_rate": 9.796974512982857e-07,
      "loss": 1.0016,
      "step": 9429
    },
    {
      "epoch": 0.6803506367014177,
      "grad_norm": 2.901078740447485,
      "learning_rate": 9.79295510139839e-07,
      "loss": 0.9594,
      "step": 9430
    },
    {
      "epoch": 0.6804227841708452,
      "grad_norm": 0.7137562134534081,
      "learning_rate": 9.788936247182848e-07,
      "loss": 0.8261,
      "step": 9431
    },
    {
      "epoch": 0.6804949316402727,
      "grad_norm": 3.877002690764707,
      "learning_rate": 9.7849179505557e-07,
      "loss": 0.942,
      "step": 9432
    },
    {
      "epoch": 0.6805670791097003,
      "grad_norm": 2.8762862603837545,
      "learning_rate": 9.780900211736368e-07,
      "loss": 0.9882,
      "step": 9433
    },
    {
      "epoch": 0.6806392265791278,
      "grad_norm": 3.6553368365460988,
      "learning_rate": 9.776883030944243e-07,
      "loss": 0.8674,
      "step": 9434
    },
    {
      "epoch": 0.6807113740485552,
      "grad_norm": 3.635577980696212,
      "learning_rate": 9.772866408398702e-07,
      "loss": 1.0009,
      "step": 9435
    },
    {
      "epoch": 0.6807835215179827,
      "grad_norm": 8.629565371357653,
      "learning_rate": 9.768850344319051e-07,
      "loss": 0.9406,
      "step": 9436
    },
    {
      "epoch": 0.6808556689874102,
      "grad_norm": 4.021199791108341,
      "learning_rate": 9.76483483892462e-07,
      "loss": 0.8969,
      "step": 9437
    },
    {
      "epoch": 0.6809278164568378,
      "grad_norm": 3.351469309829497,
      "learning_rate": 9.760819892434657e-07,
      "loss": 0.8689,
      "step": 9438
    },
    {
      "epoch": 0.6809999639262653,
      "grad_norm": 3.1713284928377985,
      "learning_rate": 9.756805505068431e-07,
      "loss": 0.921,
      "step": 9439
    },
    {
      "epoch": 0.6810721113956928,
      "grad_norm": 3.3485154748507497,
      "learning_rate": 9.752791677045128e-07,
      "loss": 0.9977,
      "step": 9440
    },
    {
      "epoch": 0.6811442588651203,
      "grad_norm": 3.9970104251255103,
      "learning_rate": 9.74877840858394e-07,
      "loss": 0.9956,
      "step": 9441
    },
    {
      "epoch": 0.6812164063345478,
      "grad_norm": 9.747601336304406,
      "learning_rate": 9.744765699904015e-07,
      "loss": 0.9683,
      "step": 9442
    },
    {
      "epoch": 0.6812885538039753,
      "grad_norm": 4.3650060675055595,
      "learning_rate": 9.74075355122447e-07,
      "loss": 0.9638,
      "step": 9443
    },
    {
      "epoch": 0.6813607012734029,
      "grad_norm": 3.7751829797722016,
      "learning_rate": 9.7367419627644e-07,
      "loss": 0.8635,
      "step": 9444
    },
    {
      "epoch": 0.6814328487428304,
      "grad_norm": 2.8339187540860697,
      "learning_rate": 9.732730934742855e-07,
      "loss": 0.9749,
      "step": 9445
    },
    {
      "epoch": 0.6815049962122579,
      "grad_norm": 3.362821859958747,
      "learning_rate": 9.728720467378865e-07,
      "loss": 0.9846,
      "step": 9446
    },
    {
      "epoch": 0.6815771436816853,
      "grad_norm": 2.9532094469056003,
      "learning_rate": 9.724710560891424e-07,
      "loss": 0.9416,
      "step": 9447
    },
    {
      "epoch": 0.6816492911511128,
      "grad_norm": 5.085846368351162,
      "learning_rate": 9.720701215499502e-07,
      "loss": 1.0465,
      "step": 9448
    },
    {
      "epoch": 0.6817214386205404,
      "grad_norm": 3.4956484036793514,
      "learning_rate": 9.716692431422036e-07,
      "loss": 0.9587,
      "step": 9449
    },
    {
      "epoch": 0.6817935860899679,
      "grad_norm": 5.167009813952405,
      "learning_rate": 9.712684208877927e-07,
      "loss": 0.94,
      "step": 9450
    },
    {
      "epoch": 0.6818657335593954,
      "grad_norm": 3.450659791079731,
      "learning_rate": 9.708676548086055e-07,
      "loss": 0.9723,
      "step": 9451
    },
    {
      "epoch": 0.6819378810288229,
      "grad_norm": 7.256285212100314,
      "learning_rate": 9.704669449265264e-07,
      "loss": 1.0239,
      "step": 9452
    },
    {
      "epoch": 0.6820100284982504,
      "grad_norm": 9.165245870818666,
      "learning_rate": 9.700662912634347e-07,
      "loss": 0.9426,
      "step": 9453
    },
    {
      "epoch": 0.682082175967678,
      "grad_norm": 0.636185943848955,
      "learning_rate": 9.696656938412117e-07,
      "loss": 0.72,
      "step": 9454
    },
    {
      "epoch": 0.6821543234371055,
      "grad_norm": 5.3397889207705385,
      "learning_rate": 9.692651526817302e-07,
      "loss": 0.9637,
      "step": 9455
    },
    {
      "epoch": 0.682226470906533,
      "grad_norm": 4.739987310722524,
      "learning_rate": 9.68864667806863e-07,
      "loss": 0.8926,
      "step": 9456
    },
    {
      "epoch": 0.6822986183759605,
      "grad_norm": 4.169989113187415,
      "learning_rate": 9.684642392384795e-07,
      "loss": 0.984,
      "step": 9457
    },
    {
      "epoch": 0.682370765845388,
      "grad_norm": 4.2739940329738975,
      "learning_rate": 9.680638669984448e-07,
      "loss": 0.975,
      "step": 9458
    },
    {
      "epoch": 0.6824429133148154,
      "grad_norm": 8.37265892900839,
      "learning_rate": 9.676635511086234e-07,
      "loss": 0.9769,
      "step": 9459
    },
    {
      "epoch": 0.682515060784243,
      "grad_norm": 4.154359315762527,
      "learning_rate": 9.672632915908724e-07,
      "loss": 0.9019,
      "step": 9460
    },
    {
      "epoch": 0.6825872082536705,
      "grad_norm": 5.4857440337441705,
      "learning_rate": 9.66863088467051e-07,
      "loss": 1.0744,
      "step": 9461
    },
    {
      "epoch": 0.682659355723098,
      "grad_norm": 3.268077166950105,
      "learning_rate": 9.664629417590115e-07,
      "loss": 0.8644,
      "step": 9462
    },
    {
      "epoch": 0.6827315031925255,
      "grad_norm": 4.6205140086336165,
      "learning_rate": 9.660628514886044e-07,
      "loss": 1.0276,
      "step": 9463
    },
    {
      "epoch": 0.682803650661953,
      "grad_norm": 2.9011122708087176,
      "learning_rate": 9.656628176776778e-07,
      "loss": 0.9661,
      "step": 9464
    },
    {
      "epoch": 0.6828757981313806,
      "grad_norm": 2.9312107042879276,
      "learning_rate": 9.652628403480757e-07,
      "loss": 1.0233,
      "step": 9465
    },
    {
      "epoch": 0.6829479456008081,
      "grad_norm": 3.5447370496755854,
      "learning_rate": 9.64862919521639e-07,
      "loss": 0.9758,
      "step": 9466
    },
    {
      "epoch": 0.6830200930702356,
      "grad_norm": 3.2616667052359785,
      "learning_rate": 9.644630552202076e-07,
      "loss": 0.9446,
      "step": 9467
    },
    {
      "epoch": 0.6830922405396631,
      "grad_norm": 4.376157770914744,
      "learning_rate": 9.640632474656136e-07,
      "loss": 0.937,
      "step": 9468
    },
    {
      "epoch": 0.6831643880090906,
      "grad_norm": 4.232472798349628,
      "learning_rate": 9.636634962796923e-07,
      "loss": 0.9587,
      "step": 9469
    },
    {
      "epoch": 0.683236535478518,
      "grad_norm": 6.362166408344779,
      "learning_rate": 9.632638016842703e-07,
      "loss": 0.9661,
      "step": 9470
    },
    {
      "epoch": 0.6833086829479456,
      "grad_norm": 0.6126470019599113,
      "learning_rate": 9.628641637011743e-07,
      "loss": 0.7898,
      "step": 9471
    },
    {
      "epoch": 0.6833808304173731,
      "grad_norm": 0.8474661266680587,
      "learning_rate": 9.624645823522267e-07,
      "loss": 0.8977,
      "step": 9472
    },
    {
      "epoch": 0.6834529778868006,
      "grad_norm": 3.491491056288499,
      "learning_rate": 9.620650576592477e-07,
      "loss": 0.9409,
      "step": 9473
    },
    {
      "epoch": 0.6835251253562281,
      "grad_norm": 4.55388025684435,
      "learning_rate": 9.616655896440545e-07,
      "loss": 0.95,
      "step": 9474
    },
    {
      "epoch": 0.6835972728256556,
      "grad_norm": 3.7808213424464445,
      "learning_rate": 9.612661783284577e-07,
      "loss": 0.882,
      "step": 9475
    },
    {
      "epoch": 0.6836694202950832,
      "grad_norm": 2.8686032658700733,
      "learning_rate": 9.608668237342715e-07,
      "loss": 0.8936,
      "step": 9476
    },
    {
      "epoch": 0.6837415677645107,
      "grad_norm": 5.255909454889615,
      "learning_rate": 9.604675258832997e-07,
      "loss": 0.9233,
      "step": 9477
    },
    {
      "epoch": 0.6838137152339382,
      "grad_norm": 3.11922791513542,
      "learning_rate": 9.600682847973497e-07,
      "loss": 0.9402,
      "step": 9478
    },
    {
      "epoch": 0.6838858627033657,
      "grad_norm": 5.2680745299330844,
      "learning_rate": 9.5966910049822e-07,
      "loss": 0.9158,
      "step": 9479
    },
    {
      "epoch": 0.6839580101727932,
      "grad_norm": 2.9316965767478793,
      "learning_rate": 9.592699730077096e-07,
      "loss": 0.9035,
      "step": 9480
    },
    {
      "epoch": 0.6840301576422207,
      "grad_norm": 2.8951191375937912,
      "learning_rate": 9.588709023476134e-07,
      "loss": 0.9022,
      "step": 9481
    },
    {
      "epoch": 0.6841023051116482,
      "grad_norm": 2.295073756609754,
      "learning_rate": 9.584718885397231e-07,
      "loss": 1.0005,
      "step": 9482
    },
    {
      "epoch": 0.6841744525810757,
      "grad_norm": 2.994311502635772,
      "learning_rate": 9.58072931605828e-07,
      "loss": 0.8798,
      "step": 9483
    },
    {
      "epoch": 0.6842466000505032,
      "grad_norm": 3.6864361278982227,
      "learning_rate": 9.576740315677125e-07,
      "loss": 1.0328,
      "step": 9484
    },
    {
      "epoch": 0.6843187475199307,
      "grad_norm": 4.58785674243821,
      "learning_rate": 9.572751884471594e-07,
      "loss": 1.0577,
      "step": 9485
    },
    {
      "epoch": 0.6843908949893582,
      "grad_norm": 4.2442460097506345,
      "learning_rate": 9.568764022659483e-07,
      "loss": 0.9127,
      "step": 9486
    },
    {
      "epoch": 0.6844630424587858,
      "grad_norm": 3.471182168113059,
      "learning_rate": 9.564776730458553e-07,
      "loss": 0.8982,
      "step": 9487
    },
    {
      "epoch": 0.6845351899282133,
      "grad_norm": 3.3325771586706963,
      "learning_rate": 9.560790008086538e-07,
      "loss": 0.8651,
      "step": 9488
    },
    {
      "epoch": 0.6846073373976408,
      "grad_norm": 4.925588992766163,
      "learning_rate": 9.55680385576114e-07,
      "loss": 0.9244,
      "step": 9489
    },
    {
      "epoch": 0.6846794848670683,
      "grad_norm": 0.6783581372497647,
      "learning_rate": 9.552818273700012e-07,
      "loss": 0.8276,
      "step": 9490
    },
    {
      "epoch": 0.6847516323364958,
      "grad_norm": 4.328726275712516,
      "learning_rate": 9.548833262120816e-07,
      "loss": 1.0209,
      "step": 9491
    },
    {
      "epoch": 0.6848237798059233,
      "grad_norm": 0.7145667868460333,
      "learning_rate": 9.544848821241133e-07,
      "loss": 0.8181,
      "step": 9492
    },
    {
      "epoch": 0.6848959272753509,
      "grad_norm": 3.0733005262376722,
      "learning_rate": 9.540864951278568e-07,
      "loss": 0.8764,
      "step": 9493
    },
    {
      "epoch": 0.6849680747447783,
      "grad_norm": 2.686306000663219,
      "learning_rate": 9.53688165245064e-07,
      "loss": 0.938,
      "step": 9494
    },
    {
      "epoch": 0.6850402222142058,
      "grad_norm": 4.435985037940946,
      "learning_rate": 9.532898924974875e-07,
      "loss": 0.9137,
      "step": 9495
    },
    {
      "epoch": 0.6851123696836333,
      "grad_norm": 2.777554060086259,
      "learning_rate": 9.528916769068751e-07,
      "loss": 0.9194,
      "step": 9496
    },
    {
      "epoch": 0.6851845171530608,
      "grad_norm": 3.5374334430162118,
      "learning_rate": 9.524935184949718e-07,
      "loss": 0.9826,
      "step": 9497
    },
    {
      "epoch": 0.6852566646224884,
      "grad_norm": 3.95036913438638,
      "learning_rate": 9.520954172835207e-07,
      "loss": 0.9568,
      "step": 9498
    },
    {
      "epoch": 0.6853288120919159,
      "grad_norm": 2.449792140784711,
      "learning_rate": 9.516973732942579e-07,
      "loss": 1.0439,
      "step": 9499
    },
    {
      "epoch": 0.6854009595613434,
      "grad_norm": 5.232500611809605,
      "learning_rate": 9.512993865489225e-07,
      "loss": 0.9019,
      "step": 9500
    },
    {
      "epoch": 0.6854731070307709,
      "grad_norm": 5.483573833320167,
      "learning_rate": 9.509014570692448e-07,
      "loss": 0.958,
      "step": 9501
    },
    {
      "epoch": 0.6855452545001984,
      "grad_norm": 4.295333752348865,
      "learning_rate": 9.505035848769547e-07,
      "loss": 1.0747,
      "step": 9502
    },
    {
      "epoch": 0.685617401969626,
      "grad_norm": 0.7277955288052836,
      "learning_rate": 9.501057699937788e-07,
      "loss": 0.8253,
      "step": 9503
    },
    {
      "epoch": 0.6856895494390535,
      "grad_norm": 4.053122157652482,
      "learning_rate": 9.497080124414403e-07,
      "loss": 0.9664,
      "step": 9504
    },
    {
      "epoch": 0.685761696908481,
      "grad_norm": 10.062947328469273,
      "learning_rate": 9.49310312241659e-07,
      "loss": 0.9389,
      "step": 9505
    },
    {
      "epoch": 0.6858338443779084,
      "grad_norm": 2.8717674658632655,
      "learning_rate": 9.489126694161529e-07,
      "loss": 0.9597,
      "step": 9506
    },
    {
      "epoch": 0.6859059918473359,
      "grad_norm": 4.357058548837512,
      "learning_rate": 9.485150839866342e-07,
      "loss": 0.9947,
      "step": 9507
    },
    {
      "epoch": 0.6859781393167634,
      "grad_norm": 2.6414723137280647,
      "learning_rate": 9.481175559748142e-07,
      "loss": 0.9522,
      "step": 9508
    },
    {
      "epoch": 0.686050286786191,
      "grad_norm": 5.846308347073604,
      "learning_rate": 9.477200854024007e-07,
      "loss": 0.866,
      "step": 9509
    },
    {
      "epoch": 0.6861224342556185,
      "grad_norm": 8.174729499657962,
      "learning_rate": 9.473226722910979e-07,
      "loss": 0.9765,
      "step": 9510
    },
    {
      "epoch": 0.686194581725046,
      "grad_norm": 4.501071378537281,
      "learning_rate": 9.46925316662607e-07,
      "loss": 0.9361,
      "step": 9511
    },
    {
      "epoch": 0.6862667291944735,
      "grad_norm": 3.0456028556684087,
      "learning_rate": 9.465280185386262e-07,
      "loss": 0.9572,
      "step": 9512
    },
    {
      "epoch": 0.686338876663901,
      "grad_norm": 4.559329041425218,
      "learning_rate": 9.461307779408514e-07,
      "loss": 1.0036,
      "step": 9513
    },
    {
      "epoch": 0.6864110241333286,
      "grad_norm": 2.8748467445990302,
      "learning_rate": 9.457335948909717e-07,
      "loss": 1.0177,
      "step": 9514
    },
    {
      "epoch": 0.6864831716027561,
      "grad_norm": 4.2293477638194075,
      "learning_rate": 9.453364694106794e-07,
      "loss": 0.9223,
      "step": 9515
    },
    {
      "epoch": 0.6865553190721836,
      "grad_norm": 5.250021071618799,
      "learning_rate": 9.449394015216567e-07,
      "loss": 1.0597,
      "step": 9516
    },
    {
      "epoch": 0.686627466541611,
      "grad_norm": 5.679514550431617,
      "learning_rate": 9.44542391245589e-07,
      "loss": 0.933,
      "step": 9517
    },
    {
      "epoch": 0.6866996140110385,
      "grad_norm": 4.138141160595497,
      "learning_rate": 9.441454386041536e-07,
      "loss": 0.8842,
      "step": 9518
    },
    {
      "epoch": 0.686771761480466,
      "grad_norm": 9.857887216267608,
      "learning_rate": 9.437485436190269e-07,
      "loss": 1.0278,
      "step": 9519
    },
    {
      "epoch": 0.6868439089498936,
      "grad_norm": 3.7691189713418964,
      "learning_rate": 9.433517063118823e-07,
      "loss": 0.9342,
      "step": 9520
    },
    {
      "epoch": 0.6869160564193211,
      "grad_norm": 5.403054764987768,
      "learning_rate": 9.429549267043895e-07,
      "loss": 0.9696,
      "step": 9521
    },
    {
      "epoch": 0.6869882038887486,
      "grad_norm": 4.728687256475698,
      "learning_rate": 9.425582048182157e-07,
      "loss": 0.9601,
      "step": 9522
    },
    {
      "epoch": 0.6870603513581761,
      "grad_norm": 3.6737148139672997,
      "learning_rate": 9.421615406750232e-07,
      "loss": 0.9824,
      "step": 9523
    },
    {
      "epoch": 0.6871324988276036,
      "grad_norm": 4.309143529350464,
      "learning_rate": 9.417649342964728e-07,
      "loss": 0.9019,
      "step": 9524
    },
    {
      "epoch": 0.6872046462970312,
      "grad_norm": 3.525807330159539,
      "learning_rate": 9.41368385704222e-07,
      "loss": 0.8713,
      "step": 9525
    },
    {
      "epoch": 0.6872767937664587,
      "grad_norm": 0.6853753334460647,
      "learning_rate": 9.409718949199248e-07,
      "loss": 0.7602,
      "step": 9526
    },
    {
      "epoch": 0.6873489412358862,
      "grad_norm": 5.445921416422802,
      "learning_rate": 9.405754619652319e-07,
      "loss": 0.929,
      "step": 9527
    },
    {
      "epoch": 0.6874210887053137,
      "grad_norm": 2.825751810550658,
      "learning_rate": 9.40179086861792e-07,
      "loss": 1.0496,
      "step": 9528
    },
    {
      "epoch": 0.6874932361747411,
      "grad_norm": 3.87059804627655,
      "learning_rate": 9.397827696312472e-07,
      "loss": 0.9129,
      "step": 9529
    },
    {
      "epoch": 0.6875653836441686,
      "grad_norm": 3.3714143991749745,
      "learning_rate": 9.393865102952423e-07,
      "loss": 0.9312,
      "step": 9530
    },
    {
      "epoch": 0.6876375311135962,
      "grad_norm": 3.134259962136821,
      "learning_rate": 9.389903088754127e-07,
      "loss": 0.9087,
      "step": 9531
    },
    {
      "epoch": 0.6877096785830237,
      "grad_norm": 3.6880018248398376,
      "learning_rate": 9.38594165393395e-07,
      "loss": 0.9653,
      "step": 9532
    },
    {
      "epoch": 0.6877818260524512,
      "grad_norm": 4.108759032524725,
      "learning_rate": 9.381980798708207e-07,
      "loss": 1.0204,
      "step": 9533
    },
    {
      "epoch": 0.6878539735218787,
      "grad_norm": 2.7403020653279673,
      "learning_rate": 9.378020523293184e-07,
      "loss": 1.0163,
      "step": 9534
    },
    {
      "epoch": 0.6879261209913062,
      "grad_norm": 4.221462705684877,
      "learning_rate": 9.374060827905139e-07,
      "loss": 0.9593,
      "step": 9535
    },
    {
      "epoch": 0.6879982684607338,
      "grad_norm": 3.479472179285723,
      "learning_rate": 9.370101712760299e-07,
      "loss": 1.0096,
      "step": 9536
    },
    {
      "epoch": 0.6880704159301613,
      "grad_norm": 4.610186592678383,
      "learning_rate": 9.366143178074859e-07,
      "loss": 0.9828,
      "step": 9537
    },
    {
      "epoch": 0.6881425633995888,
      "grad_norm": 5.009441426218093,
      "learning_rate": 9.36218522406496e-07,
      "loss": 0.8455,
      "step": 9538
    },
    {
      "epoch": 0.6882147108690163,
      "grad_norm": 0.7719498401143811,
      "learning_rate": 9.35822785094676e-07,
      "loss": 0.8264,
      "step": 9539
    },
    {
      "epoch": 0.6882868583384438,
      "grad_norm": 4.247695522467084,
      "learning_rate": 9.354271058936339e-07,
      "loss": 1.0132,
      "step": 9540
    },
    {
      "epoch": 0.6883590058078712,
      "grad_norm": 5.446148888868772,
      "learning_rate": 9.350314848249763e-07,
      "loss": 0.8834,
      "step": 9541
    },
    {
      "epoch": 0.6884311532772988,
      "grad_norm": 3.7895978097217973,
      "learning_rate": 9.34635921910307e-07,
      "loss": 0.9522,
      "step": 9542
    },
    {
      "epoch": 0.6885033007467263,
      "grad_norm": 3.639949752335041,
      "learning_rate": 9.342404171712262e-07,
      "loss": 0.877,
      "step": 9543
    },
    {
      "epoch": 0.6885754482161538,
      "grad_norm": 4.501090023690615,
      "learning_rate": 9.33844970629331e-07,
      "loss": 0.8426,
      "step": 9544
    },
    {
      "epoch": 0.6886475956855813,
      "grad_norm": 6.299674119543897,
      "learning_rate": 9.334495823062159e-07,
      "loss": 0.9007,
      "step": 9545
    },
    {
      "epoch": 0.6887197431550088,
      "grad_norm": 4.501891692232694,
      "learning_rate": 9.330542522234699e-07,
      "loss": 1.0188,
      "step": 9546
    },
    {
      "epoch": 0.6887918906244364,
      "grad_norm": 5.193591666074731,
      "learning_rate": 9.326589804026816e-07,
      "loss": 1.0913,
      "step": 9547
    },
    {
      "epoch": 0.6888640380938639,
      "grad_norm": 3.4337876127368108,
      "learning_rate": 9.322637668654354e-07,
      "loss": 0.9095,
      "step": 9548
    },
    {
      "epoch": 0.6889361855632914,
      "grad_norm": 4.980008212469599,
      "learning_rate": 9.318686116333121e-07,
      "loss": 0.9097,
      "step": 9549
    },
    {
      "epoch": 0.6890083330327189,
      "grad_norm": 4.630552721750519,
      "learning_rate": 9.314735147278898e-07,
      "loss": 0.9129,
      "step": 9550
    },
    {
      "epoch": 0.6890804805021464,
      "grad_norm": 3.714446968204291,
      "learning_rate": 9.310784761707436e-07,
      "loss": 1.039,
      "step": 9551
    },
    {
      "epoch": 0.689152627971574,
      "grad_norm": 6.537507924062329,
      "learning_rate": 9.306834959834451e-07,
      "loss": 0.9266,
      "step": 9552
    },
    {
      "epoch": 0.6892247754410014,
      "grad_norm": 5.688560638461294,
      "learning_rate": 9.302885741875612e-07,
      "loss": 0.9304,
      "step": 9553
    },
    {
      "epoch": 0.6892969229104289,
      "grad_norm": 4.883464800179258,
      "learning_rate": 9.298937108046598e-07,
      "loss": 0.9838,
      "step": 9554
    },
    {
      "epoch": 0.6893690703798564,
      "grad_norm": 2.5256678878212657,
      "learning_rate": 9.294989058563008e-07,
      "loss": 1.0014,
      "step": 9555
    },
    {
      "epoch": 0.6894412178492839,
      "grad_norm": 10.524964219709817,
      "learning_rate": 9.291041593640434e-07,
      "loss": 0.8246,
      "step": 9556
    },
    {
      "epoch": 0.6895133653187114,
      "grad_norm": 4.968827732845675,
      "learning_rate": 9.287094713494437e-07,
      "loss": 0.9755,
      "step": 9557
    },
    {
      "epoch": 0.689585512788139,
      "grad_norm": 4.248769694252081,
      "learning_rate": 9.283148418340541e-07,
      "loss": 0.8996,
      "step": 9558
    },
    {
      "epoch": 0.6896576602575665,
      "grad_norm": 4.818318404327738,
      "learning_rate": 9.279202708394236e-07,
      "loss": 0.9568,
      "step": 9559
    },
    {
      "epoch": 0.689729807726994,
      "grad_norm": 10.126295972229308,
      "learning_rate": 9.275257583870984e-07,
      "loss": 0.9492,
      "step": 9560
    },
    {
      "epoch": 0.6898019551964215,
      "grad_norm": 5.434453527355921,
      "learning_rate": 9.271313044986222e-07,
      "loss": 0.8687,
      "step": 9561
    },
    {
      "epoch": 0.689874102665849,
      "grad_norm": 0.7906923740313025,
      "learning_rate": 9.26736909195533e-07,
      "loss": 0.8416,
      "step": 9562
    },
    {
      "epoch": 0.6899462501352766,
      "grad_norm": 3.764838932031896,
      "learning_rate": 9.26342572499368e-07,
      "loss": 0.9756,
      "step": 9563
    },
    {
      "epoch": 0.6900183976047041,
      "grad_norm": 5.445342973708668,
      "learning_rate": 9.259482944316608e-07,
      "loss": 0.9657,
      "step": 9564
    },
    {
      "epoch": 0.6900905450741315,
      "grad_norm": 0.7952583594340045,
      "learning_rate": 9.255540750139412e-07,
      "loss": 0.8897,
      "step": 9565
    },
    {
      "epoch": 0.690162692543559,
      "grad_norm": 0.857761567762974,
      "learning_rate": 9.251599142677362e-07,
      "loss": 0.895,
      "step": 9566
    },
    {
      "epoch": 0.6902348400129865,
      "grad_norm": 7.671266115969481,
      "learning_rate": 9.247658122145698e-07,
      "loss": 0.9756,
      "step": 9567
    },
    {
      "epoch": 0.690306987482414,
      "grad_norm": 5.652647947444545,
      "learning_rate": 9.243717688759607e-07,
      "loss": 0.9314,
      "step": 9568
    },
    {
      "epoch": 0.6903791349518416,
      "grad_norm": 6.223739692540012,
      "learning_rate": 9.239777842734291e-07,
      "loss": 0.9727,
      "step": 9569
    },
    {
      "epoch": 0.6904512824212691,
      "grad_norm": 2.6973829195565795,
      "learning_rate": 9.235838584284865e-07,
      "loss": 0.9647,
      "step": 9570
    },
    {
      "epoch": 0.6905234298906966,
      "grad_norm": 4.918157620852032,
      "learning_rate": 9.231899913626447e-07,
      "loss": 0.8828,
      "step": 9571
    },
    {
      "epoch": 0.6905955773601241,
      "grad_norm": 3.2618088030233268,
      "learning_rate": 9.227961830974111e-07,
      "loss": 0.9779,
      "step": 9572
    },
    {
      "epoch": 0.6906677248295516,
      "grad_norm": 5.51843639618951,
      "learning_rate": 9.224024336542904e-07,
      "loss": 0.9696,
      "step": 9573
    },
    {
      "epoch": 0.6907398722989792,
      "grad_norm": 3.2424861908253164,
      "learning_rate": 9.220087430547836e-07,
      "loss": 1.1058,
      "step": 9574
    },
    {
      "epoch": 0.6908120197684067,
      "grad_norm": 5.313918518969323,
      "learning_rate": 9.216151113203887e-07,
      "loss": 0.8555,
      "step": 9575
    },
    {
      "epoch": 0.6908841672378341,
      "grad_norm": 2.96013394929042,
      "learning_rate": 9.212215384726015e-07,
      "loss": 0.9881,
      "step": 9576
    },
    {
      "epoch": 0.6909563147072616,
      "grad_norm": 3.346689233691156,
      "learning_rate": 9.208280245329117e-07,
      "loss": 0.9755,
      "step": 9577
    },
    {
      "epoch": 0.6910284621766891,
      "grad_norm": 3.404430218272066,
      "learning_rate": 9.204345695228084e-07,
      "loss": 1.0433,
      "step": 9578
    },
    {
      "epoch": 0.6911006096461166,
      "grad_norm": 4.731267834809929,
      "learning_rate": 9.20041173463777e-07,
      "loss": 0.9581,
      "step": 9579
    },
    {
      "epoch": 0.6911727571155442,
      "grad_norm": 2.8240983655078242,
      "learning_rate": 9.196478363772992e-07,
      "loss": 1.0027,
      "step": 9580
    },
    {
      "epoch": 0.6912449045849717,
      "grad_norm": 5.282596574470126,
      "learning_rate": 9.192545582848538e-07,
      "loss": 0.9359,
      "step": 9581
    },
    {
      "epoch": 0.6913170520543992,
      "grad_norm": 4.092802913689788,
      "learning_rate": 9.18861339207916e-07,
      "loss": 1.0036,
      "step": 9582
    },
    {
      "epoch": 0.6913891995238267,
      "grad_norm": 2.6608133050738467,
      "learning_rate": 9.184681791679585e-07,
      "loss": 0.9332,
      "step": 9583
    },
    {
      "epoch": 0.6914613469932542,
      "grad_norm": 6.700386537248068,
      "learning_rate": 9.180750781864505e-07,
      "loss": 0.9923,
      "step": 9584
    },
    {
      "epoch": 0.6915334944626818,
      "grad_norm": 2.6341306742794077,
      "learning_rate": 9.176820362848569e-07,
      "loss": 0.9989,
      "step": 9585
    },
    {
      "epoch": 0.6916056419321093,
      "grad_norm": 0.782672159859921,
      "learning_rate": 9.172890534846405e-07,
      "loss": 0.8442,
      "step": 9586
    },
    {
      "epoch": 0.6916777894015368,
      "grad_norm": 10.002105872625377,
      "learning_rate": 9.168961298072608e-07,
      "loss": 0.8932,
      "step": 9587
    },
    {
      "epoch": 0.6917499368709642,
      "grad_norm": 3.1462027899102845,
      "learning_rate": 9.16503265274174e-07,
      "loss": 0.8384,
      "step": 9588
    },
    {
      "epoch": 0.6918220843403917,
      "grad_norm": 3.188825892521406,
      "learning_rate": 9.16110459906833e-07,
      "loss": 0.9226,
      "step": 9589
    },
    {
      "epoch": 0.6918942318098192,
      "grad_norm": 5.849525901950336,
      "learning_rate": 9.157177137266872e-07,
      "loss": 0.9501,
      "step": 9590
    },
    {
      "epoch": 0.6919663792792468,
      "grad_norm": 3.533215541195707,
      "learning_rate": 9.15325026755184e-07,
      "loss": 0.9008,
      "step": 9591
    },
    {
      "epoch": 0.6920385267486743,
      "grad_norm": 3.687691247150982,
      "learning_rate": 9.149323990137643e-07,
      "loss": 1.0238,
      "step": 9592
    },
    {
      "epoch": 0.6921106742181018,
      "grad_norm": 3.8600788169680635,
      "learning_rate": 9.14539830523871e-07,
      "loss": 0.9835,
      "step": 9593
    },
    {
      "epoch": 0.6921828216875293,
      "grad_norm": 4.102538016431966,
      "learning_rate": 9.141473213069384e-07,
      "loss": 1.0111,
      "step": 9594
    },
    {
      "epoch": 0.6922549691569568,
      "grad_norm": 2.8155705962180377,
      "learning_rate": 9.137548713844013e-07,
      "loss": 0.8365,
      "step": 9595
    },
    {
      "epoch": 0.6923271166263844,
      "grad_norm": 2.726892489600811,
      "learning_rate": 9.133624807776892e-07,
      "loss": 0.8899,
      "step": 9596
    },
    {
      "epoch": 0.6923992640958119,
      "grad_norm": 2.969467879384173,
      "learning_rate": 9.129701495082294e-07,
      "loss": 0.9206,
      "step": 9597
    },
    {
      "epoch": 0.6924714115652394,
      "grad_norm": 3.8697145162552875,
      "learning_rate": 9.125778775974458e-07,
      "loss": 0.9113,
      "step": 9598
    },
    {
      "epoch": 0.6925435590346669,
      "grad_norm": 2.954687119135436,
      "learning_rate": 9.121856650667594e-07,
      "loss": 0.924,
      "step": 9599
    },
    {
      "epoch": 0.6926157065040943,
      "grad_norm": 5.51273536093181,
      "learning_rate": 9.117935119375863e-07,
      "loss": 0.9549,
      "step": 9600
    },
    {
      "epoch": 0.6926878539735218,
      "grad_norm": 4.834024818328291,
      "learning_rate": 9.114014182313412e-07,
      "loss": 0.8912,
      "step": 9601
    },
    {
      "epoch": 0.6927600014429494,
      "grad_norm": 3.959538018800509,
      "learning_rate": 9.110093839694346e-07,
      "loss": 0.9209,
      "step": 9602
    },
    {
      "epoch": 0.6928321489123769,
      "grad_norm": 4.150925057503184,
      "learning_rate": 9.106174091732743e-07,
      "loss": 1.0119,
      "step": 9603
    },
    {
      "epoch": 0.6929042963818044,
      "grad_norm": 5.017569571993002,
      "learning_rate": 9.102254938642646e-07,
      "loss": 0.8773,
      "step": 9604
    },
    {
      "epoch": 0.6929764438512319,
      "grad_norm": 3.887598047522604,
      "learning_rate": 9.098336380638064e-07,
      "loss": 1.0058,
      "step": 9605
    },
    {
      "epoch": 0.6930485913206594,
      "grad_norm": 4.6343511949932275,
      "learning_rate": 9.094418417932983e-07,
      "loss": 0.976,
      "step": 9606
    },
    {
      "epoch": 0.693120738790087,
      "grad_norm": 0.6738142371474175,
      "learning_rate": 9.090501050741326e-07,
      "loss": 0.8151,
      "step": 9607
    },
    {
      "epoch": 0.6931928862595145,
      "grad_norm": 7.611434375862189,
      "learning_rate": 9.086584279277037e-07,
      "loss": 0.9671,
      "step": 9608
    },
    {
      "epoch": 0.693265033728942,
      "grad_norm": 4.585429226784299,
      "learning_rate": 9.082668103753972e-07,
      "loss": 0.9481,
      "step": 9609
    },
    {
      "epoch": 0.6933371811983695,
      "grad_norm": 4.450046050444222,
      "learning_rate": 9.078752524385989e-07,
      "loss": 0.9652,
      "step": 9610
    },
    {
      "epoch": 0.693409328667797,
      "grad_norm": 3.801672647670226,
      "learning_rate": 9.074837541386899e-07,
      "loss": 0.9759,
      "step": 9611
    },
    {
      "epoch": 0.6934814761372244,
      "grad_norm": 3.6780063814208455,
      "learning_rate": 9.070923154970489e-07,
      "loss": 1.0279,
      "step": 9612
    },
    {
      "epoch": 0.693553623606652,
      "grad_norm": 19.54339778474194,
      "learning_rate": 9.067009365350506e-07,
      "loss": 1.0306,
      "step": 9613
    },
    {
      "epoch": 0.6936257710760795,
      "grad_norm": 6.820080660345818,
      "learning_rate": 9.06309617274067e-07,
      "loss": 0.9634,
      "step": 9614
    },
    {
      "epoch": 0.693697918545507,
      "grad_norm": 4.433571809554743,
      "learning_rate": 9.059183577354672e-07,
      "loss": 0.9406,
      "step": 9615
    },
    {
      "epoch": 0.6937700660149345,
      "grad_norm": 3.709546280130651,
      "learning_rate": 9.055271579406152e-07,
      "loss": 1.0009,
      "step": 9616
    },
    {
      "epoch": 0.693842213484362,
      "grad_norm": 2.5478079555606166,
      "learning_rate": 9.051360179108733e-07,
      "loss": 1.0582,
      "step": 9617
    },
    {
      "epoch": 0.6939143609537896,
      "grad_norm": 3.4557888061313884,
      "learning_rate": 9.047449376676004e-07,
      "loss": 1.0064,
      "step": 9618
    },
    {
      "epoch": 0.6939865084232171,
      "grad_norm": 0.856205000704299,
      "learning_rate": 9.043539172321519e-07,
      "loss": 0.8531,
      "step": 9619
    },
    {
      "epoch": 0.6940586558926446,
      "grad_norm": 3.4783627767250427,
      "learning_rate": 9.039629566258802e-07,
      "loss": 0.9566,
      "step": 9620
    },
    {
      "epoch": 0.6941308033620721,
      "grad_norm": 3.3674700058675326,
      "learning_rate": 9.035720558701348e-07,
      "loss": 1.0174,
      "step": 9621
    },
    {
      "epoch": 0.6942029508314996,
      "grad_norm": 2.94951461419512,
      "learning_rate": 9.03181214986259e-07,
      "loss": 0.9416,
      "step": 9622
    },
    {
      "epoch": 0.694275098300927,
      "grad_norm": 4.485425022021194,
      "learning_rate": 9.027904339955983e-07,
      "loss": 0.9419,
      "step": 9623
    },
    {
      "epoch": 0.6943472457703546,
      "grad_norm": 6.162127946683583,
      "learning_rate": 9.023997129194896e-07,
      "loss": 0.8279,
      "step": 9624
    },
    {
      "epoch": 0.6944193932397821,
      "grad_norm": 3.440074857961865,
      "learning_rate": 9.020090517792694e-07,
      "loss": 1.0478,
      "step": 9625
    },
    {
      "epoch": 0.6944915407092096,
      "grad_norm": 4.246706415666072,
      "learning_rate": 9.016184505962701e-07,
      "loss": 1.0405,
      "step": 9626
    },
    {
      "epoch": 0.6945636881786371,
      "grad_norm": 7.4224118329699875,
      "learning_rate": 9.012279093918211e-07,
      "loss": 1.0002,
      "step": 9627
    },
    {
      "epoch": 0.6946358356480646,
      "grad_norm": 4.138492596547647,
      "learning_rate": 9.008374281872486e-07,
      "loss": 0.9463,
      "step": 9628
    },
    {
      "epoch": 0.6947079831174922,
      "grad_norm": 5.170537032194426,
      "learning_rate": 9.004470070038752e-07,
      "loss": 0.986,
      "step": 9629
    },
    {
      "epoch": 0.6947801305869197,
      "grad_norm": 5.865647684093058,
      "learning_rate": 9.000566458630209e-07,
      "loss": 0.8733,
      "step": 9630
    },
    {
      "epoch": 0.6948522780563472,
      "grad_norm": 3.396157247310836,
      "learning_rate": 8.99666344786e-07,
      "loss": 0.9354,
      "step": 9631
    },
    {
      "epoch": 0.6949244255257747,
      "grad_norm": 2.724953425298118,
      "learning_rate": 8.992761037941284e-07,
      "loss": 0.9809,
      "step": 9632
    },
    {
      "epoch": 0.6949965729952022,
      "grad_norm": 3.4220905671522606,
      "learning_rate": 8.988859229087129e-07,
      "loss": 0.9201,
      "step": 9633
    },
    {
      "epoch": 0.6950687204646298,
      "grad_norm": 3.057944822104248,
      "learning_rate": 8.984958021510611e-07,
      "loss": 0.9813,
      "step": 9634
    },
    {
      "epoch": 0.6951408679340572,
      "grad_norm": 2.6945984710068385,
      "learning_rate": 8.98105741542476e-07,
      "loss": 0.9,
      "step": 9635
    },
    {
      "epoch": 0.6952130154034847,
      "grad_norm": 2.698745708170628,
      "learning_rate": 8.977157411042573e-07,
      "loss": 0.9217,
      "step": 9636
    },
    {
      "epoch": 0.6952851628729122,
      "grad_norm": 7.252093308889499,
      "learning_rate": 8.973258008577014e-07,
      "loss": 1.0065,
      "step": 9637
    },
    {
      "epoch": 0.6953573103423397,
      "grad_norm": 4.03834583410159,
      "learning_rate": 8.969359208241023e-07,
      "loss": 1.0378,
      "step": 9638
    },
    {
      "epoch": 0.6954294578117672,
      "grad_norm": 5.783020408292246,
      "learning_rate": 8.965461010247484e-07,
      "loss": 0.9664,
      "step": 9639
    },
    {
      "epoch": 0.6955016052811948,
      "grad_norm": 3.3547170594205244,
      "learning_rate": 8.96156341480927e-07,
      "loss": 0.9565,
      "step": 9640
    },
    {
      "epoch": 0.6955737527506223,
      "grad_norm": 5.5940203894365315,
      "learning_rate": 8.957666422139216e-07,
      "loss": 0.9412,
      "step": 9641
    },
    {
      "epoch": 0.6956459002200498,
      "grad_norm": 3.9526696225564524,
      "learning_rate": 8.953770032450121e-07,
      "loss": 0.9405,
      "step": 9642
    },
    {
      "epoch": 0.6957180476894773,
      "grad_norm": 2.4696113918566884,
      "learning_rate": 8.949874245954754e-07,
      "loss": 1.0208,
      "step": 9643
    },
    {
      "epoch": 0.6957901951589048,
      "grad_norm": 10.302900435569292,
      "learning_rate": 8.945979062865847e-07,
      "loss": 0.9162,
      "step": 9644
    },
    {
      "epoch": 0.6958623426283324,
      "grad_norm": 6.900608433974007,
      "learning_rate": 8.942084483396111e-07,
      "loss": 0.9785,
      "step": 9645
    },
    {
      "epoch": 0.6959344900977599,
      "grad_norm": 4.216241839774474,
      "learning_rate": 8.93819050775819e-07,
      "loss": 1.0152,
      "step": 9646
    },
    {
      "epoch": 0.6960066375671873,
      "grad_norm": 3.7159523578295754,
      "learning_rate": 8.934297136164752e-07,
      "loss": 0.9391,
      "step": 9647
    },
    {
      "epoch": 0.6960787850366148,
      "grad_norm": 5.56941410647295,
      "learning_rate": 8.930404368828375e-07,
      "loss": 0.8993,
      "step": 9648
    },
    {
      "epoch": 0.6961509325060423,
      "grad_norm": 0.8332513093317173,
      "learning_rate": 8.926512205961638e-07,
      "loss": 0.8485,
      "step": 9649
    },
    {
      "epoch": 0.6962230799754698,
      "grad_norm": 3.026416976336976,
      "learning_rate": 8.922620647777077e-07,
      "loss": 0.9944,
      "step": 9650
    },
    {
      "epoch": 0.6962952274448974,
      "grad_norm": 4.171747212828128,
      "learning_rate": 8.918729694487197e-07,
      "loss": 0.9523,
      "step": 9651
    },
    {
      "epoch": 0.6963673749143249,
      "grad_norm": 6.82024090741348,
      "learning_rate": 8.914839346304463e-07,
      "loss": 0.9514,
      "step": 9652
    },
    {
      "epoch": 0.6964395223837524,
      "grad_norm": 5.451233806415482,
      "learning_rate": 8.91094960344132e-07,
      "loss": 0.9621,
      "step": 9653
    },
    {
      "epoch": 0.6965116698531799,
      "grad_norm": 0.7040107869587089,
      "learning_rate": 8.907060466110174e-07,
      "loss": 0.7811,
      "step": 9654
    },
    {
      "epoch": 0.6965838173226074,
      "grad_norm": 2.2398521675986864,
      "learning_rate": 8.903171934523385e-07,
      "loss": 0.8787,
      "step": 9655
    },
    {
      "epoch": 0.696655964792035,
      "grad_norm": 3.0767462881358316,
      "learning_rate": 8.899284008893298e-07,
      "loss": 0.9103,
      "step": 9656
    },
    {
      "epoch": 0.6967281122614625,
      "grad_norm": 3.872794107950445,
      "learning_rate": 8.895396689432217e-07,
      "loss": 0.9571,
      "step": 9657
    },
    {
      "epoch": 0.69680025973089,
      "grad_norm": 11.770360310370176,
      "learning_rate": 8.891509976352416e-07,
      "loss": 0.839,
      "step": 9658
    },
    {
      "epoch": 0.6968724072003174,
      "grad_norm": 4.061591648373781,
      "learning_rate": 8.887623869866132e-07,
      "loss": 0.9873,
      "step": 9659
    },
    {
      "epoch": 0.6969445546697449,
      "grad_norm": 7.30494312512283,
      "learning_rate": 8.883738370185579e-07,
      "loss": 0.985,
      "step": 9660
    },
    {
      "epoch": 0.6970167021391724,
      "grad_norm": 9.177165128853577,
      "learning_rate": 8.879853477522907e-07,
      "loss": 0.9289,
      "step": 9661
    },
    {
      "epoch": 0.6970888496086,
      "grad_norm": 4.415051050809039,
      "learning_rate": 8.875969192090289e-07,
      "loss": 0.9865,
      "step": 9662
    },
    {
      "epoch": 0.6971609970780275,
      "grad_norm": 3.7185776614478216,
      "learning_rate": 8.872085514099803e-07,
      "loss": 0.9575,
      "step": 9663
    },
    {
      "epoch": 0.697233144547455,
      "grad_norm": 7.5803838450490915,
      "learning_rate": 8.868202443763534e-07,
      "loss": 1.0339,
      "step": 9664
    },
    {
      "epoch": 0.6973052920168825,
      "grad_norm": 4.505139277228459,
      "learning_rate": 8.864319981293522e-07,
      "loss": 1.0379,
      "step": 9665
    },
    {
      "epoch": 0.69737743948631,
      "grad_norm": 4.709023056599214,
      "learning_rate": 8.860438126901772e-07,
      "loss": 1.0002,
      "step": 9666
    },
    {
      "epoch": 0.6974495869557376,
      "grad_norm": 0.8295285548606284,
      "learning_rate": 8.856556880800268e-07,
      "loss": 0.8612,
      "step": 9667
    },
    {
      "epoch": 0.6975217344251651,
      "grad_norm": 3.188643456103069,
      "learning_rate": 8.852676243200925e-07,
      "loss": 0.9731,
      "step": 9668
    },
    {
      "epoch": 0.6975938818945926,
      "grad_norm": 4.695722971411661,
      "learning_rate": 8.848796214315681e-07,
      "loss": 1.127,
      "step": 9669
    },
    {
      "epoch": 0.69766602936402,
      "grad_norm": 5.504780339181719,
      "learning_rate": 8.844916794356381e-07,
      "loss": 0.9571,
      "step": 9670
    },
    {
      "epoch": 0.6977381768334475,
      "grad_norm": 5.165073867077621,
      "learning_rate": 8.841037983534893e-07,
      "loss": 0.8171,
      "step": 9671
    },
    {
      "epoch": 0.697810324302875,
      "grad_norm": 4.584658182729201,
      "learning_rate": 8.837159782063006e-07,
      "loss": 0.9747,
      "step": 9672
    },
    {
      "epoch": 0.6978824717723026,
      "grad_norm": 0.7256438898327002,
      "learning_rate": 8.8332821901525e-07,
      "loss": 0.8443,
      "step": 9673
    },
    {
      "epoch": 0.6979546192417301,
      "grad_norm": 3.446533402006994,
      "learning_rate": 8.829405208015117e-07,
      "loss": 0.9508,
      "step": 9674
    },
    {
      "epoch": 0.6980267667111576,
      "grad_norm": 5.693450832777421,
      "learning_rate": 8.825528835862562e-07,
      "loss": 1.0101,
      "step": 9675
    },
    {
      "epoch": 0.6980989141805851,
      "grad_norm": 4.949269235017768,
      "learning_rate": 8.82165307390651e-07,
      "loss": 0.8738,
      "step": 9676
    },
    {
      "epoch": 0.6981710616500126,
      "grad_norm": 2.2398596186586355,
      "learning_rate": 8.817777922358612e-07,
      "loss": 0.9614,
      "step": 9677
    },
    {
      "epoch": 0.6982432091194402,
      "grad_norm": 4.80083985133833,
      "learning_rate": 8.813903381430456e-07,
      "loss": 0.9659,
      "step": 9678
    },
    {
      "epoch": 0.6983153565888677,
      "grad_norm": 10.626560231261704,
      "learning_rate": 8.810029451333629e-07,
      "loss": 0.8664,
      "step": 9679
    },
    {
      "epoch": 0.6983875040582952,
      "grad_norm": 4.794761167113353,
      "learning_rate": 8.806156132279669e-07,
      "loss": 0.9684,
      "step": 9680
    },
    {
      "epoch": 0.6984596515277227,
      "grad_norm": 5.511415425909051,
      "learning_rate": 8.802283424480086e-07,
      "loss": 0.9068,
      "step": 9681
    },
    {
      "epoch": 0.6985317989971501,
      "grad_norm": 7.397872371499795,
      "learning_rate": 8.79841132814636e-07,
      "loss": 0.971,
      "step": 9682
    },
    {
      "epoch": 0.6986039464665776,
      "grad_norm": 4.8746570564345015,
      "learning_rate": 8.79453984348991e-07,
      "loss": 1.0768,
      "step": 9683
    },
    {
      "epoch": 0.6986760939360052,
      "grad_norm": 0.7491493169057726,
      "learning_rate": 8.79066897072217e-07,
      "loss": 0.7892,
      "step": 9684
    },
    {
      "epoch": 0.6987482414054327,
      "grad_norm": 3.288113056067632,
      "learning_rate": 8.786798710054491e-07,
      "loss": 1.0039,
      "step": 9685
    },
    {
      "epoch": 0.6988203888748602,
      "grad_norm": 4.521203419996742,
      "learning_rate": 8.782929061698237e-07,
      "loss": 1.0079,
      "step": 9686
    },
    {
      "epoch": 0.6988925363442877,
      "grad_norm": 4.0354665542746035,
      "learning_rate": 8.779060025864695e-07,
      "loss": 0.9865,
      "step": 9687
    },
    {
      "epoch": 0.6989646838137152,
      "grad_norm": 4.297799971465234,
      "learning_rate": 8.775191602765147e-07,
      "loss": 1.0242,
      "step": 9688
    },
    {
      "epoch": 0.6990368312831428,
      "grad_norm": 5.294076365889625,
      "learning_rate": 8.771323792610832e-07,
      "loss": 0.97,
      "step": 9689
    },
    {
      "epoch": 0.6991089787525703,
      "grad_norm": 3.5451412584921878,
      "learning_rate": 8.767456595612958e-07,
      "loss": 0.9728,
      "step": 9690
    },
    {
      "epoch": 0.6991811262219978,
      "grad_norm": 4.013185940398465,
      "learning_rate": 8.763590011982706e-07,
      "loss": 1.0311,
      "step": 9691
    },
    {
      "epoch": 0.6992532736914253,
      "grad_norm": 4.150263784826524,
      "learning_rate": 8.759724041931194e-07,
      "loss": 0.987,
      "step": 9692
    },
    {
      "epoch": 0.6993254211608528,
      "grad_norm": 4.119910134381678,
      "learning_rate": 8.755858685669555e-07,
      "loss": 0.9241,
      "step": 9693
    },
    {
      "epoch": 0.6993975686302802,
      "grad_norm": 3.924594377101954,
      "learning_rate": 8.751993943408842e-07,
      "loss": 0.9348,
      "step": 9694
    },
    {
      "epoch": 0.6994697160997078,
      "grad_norm": 8.076380882649177,
      "learning_rate": 8.748129815360099e-07,
      "loss": 1.056,
      "step": 9695
    },
    {
      "epoch": 0.6995418635691353,
      "grad_norm": 3.0543206426327765,
      "learning_rate": 8.744266301734336e-07,
      "loss": 0.9857,
      "step": 9696
    },
    {
      "epoch": 0.6996140110385628,
      "grad_norm": 3.24150854148369,
      "learning_rate": 8.740403402742522e-07,
      "loss": 0.9092,
      "step": 9697
    },
    {
      "epoch": 0.6996861585079903,
      "grad_norm": 4.167512477222202,
      "learning_rate": 8.736541118595597e-07,
      "loss": 0.7692,
      "step": 9698
    },
    {
      "epoch": 0.6997583059774178,
      "grad_norm": 3.6603661215427334,
      "learning_rate": 8.732679449504475e-07,
      "loss": 0.9281,
      "step": 9699
    },
    {
      "epoch": 0.6998304534468454,
      "grad_norm": 4.316565546056647,
      "learning_rate": 8.72881839568e-07,
      "loss": 0.9059,
      "step": 9700
    },
    {
      "epoch": 0.6999026009162729,
      "grad_norm": 3.638355906881453,
      "learning_rate": 8.724957957333044e-07,
      "loss": 0.9105,
      "step": 9701
    },
    {
      "epoch": 0.6999747483857004,
      "grad_norm": 3.3790341038852696,
      "learning_rate": 8.721098134674389e-07,
      "loss": 0.9838,
      "step": 9702
    },
    {
      "epoch": 0.7000468958551279,
      "grad_norm": 3.5266530320661746,
      "learning_rate": 8.717238927914809e-07,
      "loss": 1.0088,
      "step": 9703
    },
    {
      "epoch": 0.7001190433245554,
      "grad_norm": 10.479513982903553,
      "learning_rate": 8.713380337265046e-07,
      "loss": 0.9882,
      "step": 9704
    },
    {
      "epoch": 0.700191190793983,
      "grad_norm": 5.735768827006998,
      "learning_rate": 8.709522362935802e-07,
      "loss": 0.935,
      "step": 9705
    },
    {
      "epoch": 0.7002633382634104,
      "grad_norm": 3.588759963321707,
      "learning_rate": 8.70566500513775e-07,
      "loss": 0.9446,
      "step": 9706
    },
    {
      "epoch": 0.7003354857328379,
      "grad_norm": 2.3399434012716602,
      "learning_rate": 8.701808264081509e-07,
      "loss": 0.7857,
      "step": 9707
    },
    {
      "epoch": 0.7004076332022654,
      "grad_norm": 4.092656113170143,
      "learning_rate": 8.697952139977708e-07,
      "loss": 0.9511,
      "step": 9708
    },
    {
      "epoch": 0.7004797806716929,
      "grad_norm": 5.476818127727013,
      "learning_rate": 8.694096633036889e-07,
      "loss": 1.0158,
      "step": 9709
    },
    {
      "epoch": 0.7005519281411204,
      "grad_norm": 4.631467884616642,
      "learning_rate": 8.690241743469615e-07,
      "loss": 0.918,
      "step": 9710
    },
    {
      "epoch": 0.700624075610548,
      "grad_norm": 3.0312264038425996,
      "learning_rate": 8.686387471486363e-07,
      "loss": 1.0346,
      "step": 9711
    },
    {
      "epoch": 0.7006962230799755,
      "grad_norm": 7.542857235953921,
      "learning_rate": 8.68253381729761e-07,
      "loss": 0.9607,
      "step": 9712
    },
    {
      "epoch": 0.700768370549403,
      "grad_norm": 4.026057721591668,
      "learning_rate": 8.678680781113789e-07,
      "loss": 1.027,
      "step": 9713
    },
    {
      "epoch": 0.7008405180188305,
      "grad_norm": 8.646117963280453,
      "learning_rate": 8.674828363145301e-07,
      "loss": 0.9253,
      "step": 9714
    },
    {
      "epoch": 0.700912665488258,
      "grad_norm": 4.276596101538962,
      "learning_rate": 8.670976563602519e-07,
      "loss": 0.9485,
      "step": 9715
    },
    {
      "epoch": 0.7009848129576856,
      "grad_norm": 4.135837075542228,
      "learning_rate": 8.667125382695761e-07,
      "loss": 0.9626,
      "step": 9716
    },
    {
      "epoch": 0.7010569604271131,
      "grad_norm": 4.149753167958363,
      "learning_rate": 8.663274820635333e-07,
      "loss": 0.9948,
      "step": 9717
    },
    {
      "epoch": 0.7011291078965405,
      "grad_norm": 4.134452393514561,
      "learning_rate": 8.6594248776315e-07,
      "loss": 0.9896,
      "step": 9718
    },
    {
      "epoch": 0.701201255365968,
      "grad_norm": 3.633733685919086,
      "learning_rate": 8.655575553894494e-07,
      "loss": 1.028,
      "step": 9719
    },
    {
      "epoch": 0.7012734028353955,
      "grad_norm": 2.842759850802726,
      "learning_rate": 8.651726849634509e-07,
      "loss": 0.989,
      "step": 9720
    },
    {
      "epoch": 0.701345550304823,
      "grad_norm": 3.136635595148166,
      "learning_rate": 8.64787876506172e-07,
      "loss": 0.9584,
      "step": 9721
    },
    {
      "epoch": 0.7014176977742506,
      "grad_norm": 5.135950624305846,
      "learning_rate": 8.644031300386232e-07,
      "loss": 0.9897,
      "step": 9722
    },
    {
      "epoch": 0.7014898452436781,
      "grad_norm": 3.8723130293174237,
      "learning_rate": 8.640184455818171e-07,
      "loss": 0.9253,
      "step": 9723
    },
    {
      "epoch": 0.7015619927131056,
      "grad_norm": 2.725493650606409,
      "learning_rate": 8.636338231567571e-07,
      "loss": 1.0297,
      "step": 9724
    },
    {
      "epoch": 0.7016341401825331,
      "grad_norm": 5.921836349643,
      "learning_rate": 8.632492627844486e-07,
      "loss": 0.9399,
      "step": 9725
    },
    {
      "epoch": 0.7017062876519606,
      "grad_norm": 0.7608316978515005,
      "learning_rate": 8.62864764485889e-07,
      "loss": 0.8804,
      "step": 9726
    },
    {
      "epoch": 0.7017784351213882,
      "grad_norm": 7.251921563525449,
      "learning_rate": 8.624803282820754e-07,
      "loss": 0.9658,
      "step": 9727
    },
    {
      "epoch": 0.7018505825908157,
      "grad_norm": 4.295550443928346,
      "learning_rate": 8.620959541939998e-07,
      "loss": 1.0036,
      "step": 9728
    },
    {
      "epoch": 0.7019227300602431,
      "grad_norm": 0.7321051561305066,
      "learning_rate": 8.617116422426518e-07,
      "loss": 0.8376,
      "step": 9729
    },
    {
      "epoch": 0.7019948775296706,
      "grad_norm": 5.868619335210909,
      "learning_rate": 8.613273924490181e-07,
      "loss": 0.9318,
      "step": 9730
    },
    {
      "epoch": 0.7020670249990981,
      "grad_norm": 4.1895036600973095,
      "learning_rate": 8.609432048340788e-07,
      "loss": 1.0106,
      "step": 9731
    },
    {
      "epoch": 0.7021391724685256,
      "grad_norm": 3.1045000208791547,
      "learning_rate": 8.605590794188158e-07,
      "loss": 0.9654,
      "step": 9732
    },
    {
      "epoch": 0.7022113199379532,
      "grad_norm": 5.288936100721453,
      "learning_rate": 8.601750162242028e-07,
      "loss": 0.8692,
      "step": 9733
    },
    {
      "epoch": 0.7022834674073807,
      "grad_norm": 3.0832617639706776,
      "learning_rate": 8.597910152712125e-07,
      "loss": 0.9794,
      "step": 9734
    },
    {
      "epoch": 0.7023556148768082,
      "grad_norm": 3.288807911205311,
      "learning_rate": 8.594070765808141e-07,
      "loss": 0.9311,
      "step": 9735
    },
    {
      "epoch": 0.7024277623462357,
      "grad_norm": 5.433971795012755,
      "learning_rate": 8.590232001739728e-07,
      "loss": 1.0161,
      "step": 9736
    },
    {
      "epoch": 0.7024999098156632,
      "grad_norm": 7.812281491084872,
      "learning_rate": 8.586393860716508e-07,
      "loss": 0.8796,
      "step": 9737
    },
    {
      "epoch": 0.7025720572850908,
      "grad_norm": 5.246594005374016,
      "learning_rate": 8.582556342948075e-07,
      "loss": 0.8809,
      "step": 9738
    },
    {
      "epoch": 0.7026442047545183,
      "grad_norm": 5.352748856298371,
      "learning_rate": 8.578719448643968e-07,
      "loss": 0.9897,
      "step": 9739
    },
    {
      "epoch": 0.7027163522239458,
      "grad_norm": 3.9331953951253618,
      "learning_rate": 8.574883178013712e-07,
      "loss": 0.8582,
      "step": 9740
    },
    {
      "epoch": 0.7027884996933732,
      "grad_norm": 5.245187824620941,
      "learning_rate": 8.571047531266789e-07,
      "loss": 0.9853,
      "step": 9741
    },
    {
      "epoch": 0.7028606471628007,
      "grad_norm": 7.868493677256367,
      "learning_rate": 8.567212508612653e-07,
      "loss": 0.9689,
      "step": 9742
    },
    {
      "epoch": 0.7029327946322282,
      "grad_norm": 4.9081726194709585,
      "learning_rate": 8.56337811026072e-07,
      "loss": 0.9483,
      "step": 9743
    },
    {
      "epoch": 0.7030049421016558,
      "grad_norm": 0.8757974532956914,
      "learning_rate": 8.55954433642037e-07,
      "loss": 0.8446,
      "step": 9744
    },
    {
      "epoch": 0.7030770895710833,
      "grad_norm": 3.3220542019642867,
      "learning_rate": 8.55571118730096e-07,
      "loss": 0.8844,
      "step": 9745
    },
    {
      "epoch": 0.7031492370405108,
      "grad_norm": 4.827446849812906,
      "learning_rate": 8.551878663111784e-07,
      "loss": 1.1598,
      "step": 9746
    },
    {
      "epoch": 0.7032213845099383,
      "grad_norm": 2.822825910134559,
      "learning_rate": 8.548046764062146e-07,
      "loss": 0.9704,
      "step": 9747
    },
    {
      "epoch": 0.7032935319793658,
      "grad_norm": 2.8970618260784655,
      "learning_rate": 8.544215490361277e-07,
      "loss": 1.0384,
      "step": 9748
    },
    {
      "epoch": 0.7033656794487934,
      "grad_norm": 4.6435981243039395,
      "learning_rate": 8.540384842218391e-07,
      "loss": 0.9398,
      "step": 9749
    },
    {
      "epoch": 0.7034378269182209,
      "grad_norm": 7.195548419211885,
      "learning_rate": 8.536554819842665e-07,
      "loss": 1.045,
      "step": 9750
    },
    {
      "epoch": 0.7035099743876484,
      "grad_norm": 4.810027639266008,
      "learning_rate": 8.532725423443246e-07,
      "loss": 0.9783,
      "step": 9751
    },
    {
      "epoch": 0.7035821218570759,
      "grad_norm": 3.4359360344897967,
      "learning_rate": 8.52889665322924e-07,
      "loss": 0.9471,
      "step": 9752
    },
    {
      "epoch": 0.7036542693265033,
      "grad_norm": 3.6054224608519623,
      "learning_rate": 8.525068509409726e-07,
      "loss": 1.0239,
      "step": 9753
    },
    {
      "epoch": 0.7037264167959308,
      "grad_norm": 7.147687962326497,
      "learning_rate": 8.521240992193746e-07,
      "loss": 0.9048,
      "step": 9754
    },
    {
      "epoch": 0.7037985642653584,
      "grad_norm": 3.5908196485330497,
      "learning_rate": 8.517414101790296e-07,
      "loss": 1.024,
      "step": 9755
    },
    {
      "epoch": 0.7038707117347859,
      "grad_norm": 3.472217098656159,
      "learning_rate": 8.513587838408356e-07,
      "loss": 0.9724,
      "step": 9756
    },
    {
      "epoch": 0.7039428592042134,
      "grad_norm": 2.9018158259244244,
      "learning_rate": 8.50976220225686e-07,
      "loss": 0.8865,
      "step": 9757
    },
    {
      "epoch": 0.7040150066736409,
      "grad_norm": 2.927676393063277,
      "learning_rate": 8.505937193544719e-07,
      "loss": 0.8412,
      "step": 9758
    },
    {
      "epoch": 0.7040871541430684,
      "grad_norm": 5.300613389417978,
      "learning_rate": 8.502112812480795e-07,
      "loss": 1.0045,
      "step": 9759
    },
    {
      "epoch": 0.704159301612496,
      "grad_norm": 7.8903803721362,
      "learning_rate": 8.498289059273935e-07,
      "loss": 1.0076,
      "step": 9760
    },
    {
      "epoch": 0.7042314490819235,
      "grad_norm": 3.250222271874549,
      "learning_rate": 8.494465934132917e-07,
      "loss": 0.9929,
      "step": 9761
    },
    {
      "epoch": 0.704303596551351,
      "grad_norm": 3.28010321059316,
      "learning_rate": 8.490643437266536e-07,
      "loss": 0.8251,
      "step": 9762
    },
    {
      "epoch": 0.7043757440207785,
      "grad_norm": 4.57914686028059,
      "learning_rate": 8.486821568883505e-07,
      "loss": 0.9578,
      "step": 9763
    },
    {
      "epoch": 0.704447891490206,
      "grad_norm": 4.317984590178676,
      "learning_rate": 8.483000329192527e-07,
      "loss": 0.9402,
      "step": 9764
    },
    {
      "epoch": 0.7045200389596334,
      "grad_norm": 4.9939265557733235,
      "learning_rate": 8.479179718402268e-07,
      "loss": 1.0519,
      "step": 9765
    },
    {
      "epoch": 0.704592186429061,
      "grad_norm": 3.6437783398842964,
      "learning_rate": 8.475359736721352e-07,
      "loss": 0.9785,
      "step": 9766
    },
    {
      "epoch": 0.7046643338984885,
      "grad_norm": 16.00012588451455,
      "learning_rate": 8.471540384358382e-07,
      "loss": 0.9429,
      "step": 9767
    },
    {
      "epoch": 0.704736481367916,
      "grad_norm": 3.971585559878578,
      "learning_rate": 8.467721661521912e-07,
      "loss": 0.9177,
      "step": 9768
    },
    {
      "epoch": 0.7048086288373435,
      "grad_norm": 11.075907540433798,
      "learning_rate": 8.463903568420481e-07,
      "loss": 0.9575,
      "step": 9769
    },
    {
      "epoch": 0.704880776306771,
      "grad_norm": 3.4253895447562157,
      "learning_rate": 8.460086105262561e-07,
      "loss": 0.9729,
      "step": 9770
    },
    {
      "epoch": 0.7049529237761986,
      "grad_norm": 7.673935001800917,
      "learning_rate": 8.456269272256622e-07,
      "loss": 0.9463,
      "step": 9771
    },
    {
      "epoch": 0.7050250712456261,
      "grad_norm": 5.150575601947036,
      "learning_rate": 8.452453069611083e-07,
      "loss": 0.8761,
      "step": 9772
    },
    {
      "epoch": 0.7050972187150536,
      "grad_norm": 4.642528002463971,
      "learning_rate": 8.448637497534334e-07,
      "loss": 0.8417,
      "step": 9773
    },
    {
      "epoch": 0.7051693661844811,
      "grad_norm": 5.5809994084915475,
      "learning_rate": 8.444822556234729e-07,
      "loss": 0.9136,
      "step": 9774
    },
    {
      "epoch": 0.7052415136539086,
      "grad_norm": 3.872266482065753,
      "learning_rate": 8.441008245920591e-07,
      "loss": 0.8687,
      "step": 9775
    },
    {
      "epoch": 0.705313661123336,
      "grad_norm": 7.178970716226219,
      "learning_rate": 8.437194566800201e-07,
      "loss": 0.9911,
      "step": 9776
    },
    {
      "epoch": 0.7053858085927636,
      "grad_norm": 3.208885706791073,
      "learning_rate": 8.433381519081819e-07,
      "loss": 0.8875,
      "step": 9777
    },
    {
      "epoch": 0.7054579560621911,
      "grad_norm": 2.4756660644084008,
      "learning_rate": 8.429569102973648e-07,
      "loss": 0.9747,
      "step": 9778
    },
    {
      "epoch": 0.7055301035316186,
      "grad_norm": 2.8024239504429658,
      "learning_rate": 8.425757318683874e-07,
      "loss": 0.9434,
      "step": 9779
    },
    {
      "epoch": 0.7056022510010461,
      "grad_norm": 3.6677369520823615,
      "learning_rate": 8.421946166420649e-07,
      "loss": 0.7333,
      "step": 9780
    },
    {
      "epoch": 0.7056743984704736,
      "grad_norm": 3.88205839083514,
      "learning_rate": 8.418135646392081e-07,
      "loss": 0.8776,
      "step": 9781
    },
    {
      "epoch": 0.7057465459399012,
      "grad_norm": 5.649432209483603,
      "learning_rate": 8.414325758806253e-07,
      "loss": 0.9693,
      "step": 9782
    },
    {
      "epoch": 0.7058186934093287,
      "grad_norm": 4.537251464090201,
      "learning_rate": 8.410516503871206e-07,
      "loss": 0.8896,
      "step": 9783
    },
    {
      "epoch": 0.7058908408787562,
      "grad_norm": 3.9027400001757955,
      "learning_rate": 8.406707881794955e-07,
      "loss": 0.9086,
      "step": 9784
    },
    {
      "epoch": 0.7059629883481837,
      "grad_norm": 4.62005082370827,
      "learning_rate": 8.402899892785458e-07,
      "loss": 0.9317,
      "step": 9785
    },
    {
      "epoch": 0.7060351358176112,
      "grad_norm": 3.6901907721106117,
      "learning_rate": 8.399092537050678e-07,
      "loss": 0.97,
      "step": 9786
    },
    {
      "epoch": 0.7061072832870388,
      "grad_norm": 4.783710918521053,
      "learning_rate": 8.395285814798505e-07,
      "loss": 1.0122,
      "step": 9787
    },
    {
      "epoch": 0.7061794307564662,
      "grad_norm": 5.981703836887641,
      "learning_rate": 8.391479726236815e-07,
      "loss": 0.9493,
      "step": 9788
    },
    {
      "epoch": 0.7062515782258937,
      "grad_norm": 5.855587643160037,
      "learning_rate": 8.387674271573441e-07,
      "loss": 0.999,
      "step": 9789
    },
    {
      "epoch": 0.7063237256953212,
      "grad_norm": 5.430066168225632,
      "learning_rate": 8.38386945101619e-07,
      "loss": 0.9387,
      "step": 9790
    },
    {
      "epoch": 0.7063958731647487,
      "grad_norm": 4.158079734107478,
      "learning_rate": 8.380065264772827e-07,
      "loss": 0.9971,
      "step": 9791
    },
    {
      "epoch": 0.7064680206341762,
      "grad_norm": 3.5351423568874103,
      "learning_rate": 8.376261713051089e-07,
      "loss": 1.0163,
      "step": 9792
    },
    {
      "epoch": 0.7065401681036038,
      "grad_norm": 7.442963204492397,
      "learning_rate": 8.372458796058666e-07,
      "loss": 0.936,
      "step": 9793
    },
    {
      "epoch": 0.7066123155730313,
      "grad_norm": 7.302604565993261,
      "learning_rate": 8.368656514003221e-07,
      "loss": 1.0336,
      "step": 9794
    },
    {
      "epoch": 0.7066844630424588,
      "grad_norm": 0.7674892073130279,
      "learning_rate": 8.364854867092388e-07,
      "loss": 0.8032,
      "step": 9795
    },
    {
      "epoch": 0.7067566105118863,
      "grad_norm": 3.8476035797077106,
      "learning_rate": 8.361053855533757e-07,
      "loss": 0.949,
      "step": 9796
    },
    {
      "epoch": 0.7068287579813138,
      "grad_norm": 3.0379562914317746,
      "learning_rate": 8.357253479534891e-07,
      "loss": 0.8725,
      "step": 9797
    },
    {
      "epoch": 0.7069009054507414,
      "grad_norm": 5.8981762152393875,
      "learning_rate": 8.353453739303311e-07,
      "loss": 0.9525,
      "step": 9798
    },
    {
      "epoch": 0.7069730529201689,
      "grad_norm": 0.7500826074883272,
      "learning_rate": 8.349654635046517e-07,
      "loss": 0.9005,
      "step": 9799
    },
    {
      "epoch": 0.7070452003895963,
      "grad_norm": 4.846506626437494,
      "learning_rate": 8.345856166971938e-07,
      "loss": 1.0067,
      "step": 9800
    },
    {
      "epoch": 0.7071173478590238,
      "grad_norm": 5.536837067543564,
      "learning_rate": 8.34205833528703e-07,
      "loss": 1.022,
      "step": 9801
    },
    {
      "epoch": 0.7071894953284513,
      "grad_norm": 4.920484558372739,
      "learning_rate": 8.338261140199149e-07,
      "loss": 0.9616,
      "step": 9802
    },
    {
      "epoch": 0.7072616427978788,
      "grad_norm": 7.5251288013270905,
      "learning_rate": 8.334464581915659e-07,
      "loss": 0.9124,
      "step": 9803
    },
    {
      "epoch": 0.7073337902673064,
      "grad_norm": 4.6644053884244965,
      "learning_rate": 8.330668660643874e-07,
      "loss": 0.9027,
      "step": 9804
    },
    {
      "epoch": 0.7074059377367339,
      "grad_norm": 3.6119287217332166,
      "learning_rate": 8.326873376591075e-07,
      "loss": 1.0107,
      "step": 9805
    },
    {
      "epoch": 0.7074780852061614,
      "grad_norm": 3.8645665390321504,
      "learning_rate": 8.323078729964508e-07,
      "loss": 0.8762,
      "step": 9806
    },
    {
      "epoch": 0.7075502326755889,
      "grad_norm": 3.0160677095700184,
      "learning_rate": 8.319284720971385e-07,
      "loss": 0.9524,
      "step": 9807
    },
    {
      "epoch": 0.7076223801450164,
      "grad_norm": 2.876935721847821,
      "learning_rate": 8.315491349818891e-07,
      "loss": 0.8873,
      "step": 9808
    },
    {
      "epoch": 0.707694527614444,
      "grad_norm": 5.157270850648929,
      "learning_rate": 8.311698616714151e-07,
      "loss": 1.0453,
      "step": 9809
    },
    {
      "epoch": 0.7077666750838715,
      "grad_norm": 5.1421107779951445,
      "learning_rate": 8.307906521864283e-07,
      "loss": 0.8991,
      "step": 9810
    },
    {
      "epoch": 0.707838822553299,
      "grad_norm": 3.4111206508783347,
      "learning_rate": 8.304115065476355e-07,
      "loss": 0.8836,
      "step": 9811
    },
    {
      "epoch": 0.7079109700227264,
      "grad_norm": 3.554470717977704,
      "learning_rate": 8.300324247757409e-07,
      "loss": 0.9217,
      "step": 9812
    },
    {
      "epoch": 0.7079831174921539,
      "grad_norm": 5.0402955431797505,
      "learning_rate": 8.296534068914443e-07,
      "loss": 0.9495,
      "step": 9813
    },
    {
      "epoch": 0.7080552649615814,
      "grad_norm": 3.145133203739183,
      "learning_rate": 8.292744529154434e-07,
      "loss": 0.9524,
      "step": 9814
    },
    {
      "epoch": 0.708127412431009,
      "grad_norm": 6.86144126410065,
      "learning_rate": 8.288955628684293e-07,
      "loss": 0.8986,
      "step": 9815
    },
    {
      "epoch": 0.7081995599004365,
      "grad_norm": 3.1960017200126365,
      "learning_rate": 8.285167367710948e-07,
      "loss": 0.938,
      "step": 9816
    },
    {
      "epoch": 0.708271707369864,
      "grad_norm": 4.601541982193751,
      "learning_rate": 8.281379746441239e-07,
      "loss": 0.9742,
      "step": 9817
    },
    {
      "epoch": 0.7083438548392915,
      "grad_norm": 4.868300505077645,
      "learning_rate": 8.277592765082e-07,
      "loss": 1.0527,
      "step": 9818
    },
    {
      "epoch": 0.708416002308719,
      "grad_norm": 8.71691612727916,
      "learning_rate": 8.273806423840026e-07,
      "loss": 1.0302,
      "step": 9819
    },
    {
      "epoch": 0.7084881497781466,
      "grad_norm": 4.281916211494655,
      "learning_rate": 8.270020722922073e-07,
      "loss": 0.9298,
      "step": 9820
    },
    {
      "epoch": 0.7085602972475741,
      "grad_norm": 3.9484631940989154,
      "learning_rate": 8.266235662534869e-07,
      "loss": 0.838,
      "step": 9821
    },
    {
      "epoch": 0.7086324447170016,
      "grad_norm": 5.004709314828158,
      "learning_rate": 8.262451242885096e-07,
      "loss": 0.8837,
      "step": 9822
    },
    {
      "epoch": 0.7087045921864291,
      "grad_norm": 5.500284707763324,
      "learning_rate": 8.258667464179416e-07,
      "loss": 1.0101,
      "step": 9823
    },
    {
      "epoch": 0.7087767396558565,
      "grad_norm": 3.376414108722676,
      "learning_rate": 8.254884326624427e-07,
      "loss": 0.9589,
      "step": 9824
    },
    {
      "epoch": 0.708848887125284,
      "grad_norm": 2.8812273003931437,
      "learning_rate": 8.251101830426741e-07,
      "loss": 1.0482,
      "step": 9825
    },
    {
      "epoch": 0.7089210345947116,
      "grad_norm": 4.9131446551289,
      "learning_rate": 8.247319975792883e-07,
      "loss": 1.0291,
      "step": 9826
    },
    {
      "epoch": 0.7089931820641391,
      "grad_norm": 3.5288836897558973,
      "learning_rate": 8.243538762929376e-07,
      "loss": 1.0698,
      "step": 9827
    },
    {
      "epoch": 0.7090653295335666,
      "grad_norm": 3.869216664013352,
      "learning_rate": 8.239758192042696e-07,
      "loss": 0.9835,
      "step": 9828
    },
    {
      "epoch": 0.7091374770029941,
      "grad_norm": 0.7292894850337583,
      "learning_rate": 8.235978263339285e-07,
      "loss": 0.8403,
      "step": 9829
    },
    {
      "epoch": 0.7092096244724216,
      "grad_norm": 3.7618551417706154,
      "learning_rate": 8.232198977025553e-07,
      "loss": 0.9167,
      "step": 9830
    },
    {
      "epoch": 0.7092817719418492,
      "grad_norm": 3.6785487784060042,
      "learning_rate": 8.228420333307877e-07,
      "loss": 0.9941,
      "step": 9831
    },
    {
      "epoch": 0.7093539194112767,
      "grad_norm": 0.7411469644272931,
      "learning_rate": 8.224642332392587e-07,
      "loss": 0.8567,
      "step": 9832
    },
    {
      "epoch": 0.7094260668807042,
      "grad_norm": 7.564273342573752,
      "learning_rate": 8.220864974485984e-07,
      "loss": 0.9645,
      "step": 9833
    },
    {
      "epoch": 0.7094982143501317,
      "grad_norm": 3.3127832831404302,
      "learning_rate": 8.217088259794343e-07,
      "loss": 0.9317,
      "step": 9834
    },
    {
      "epoch": 0.7095703618195591,
      "grad_norm": 3.695382567178946,
      "learning_rate": 8.213312188523891e-07,
      "loss": 1.0153,
      "step": 9835
    },
    {
      "epoch": 0.7096425092889866,
      "grad_norm": 12.26703378155663,
      "learning_rate": 8.209536760880829e-07,
      "loss": 0.9366,
      "step": 9836
    },
    {
      "epoch": 0.7097146567584142,
      "grad_norm": 5.2685828348703625,
      "learning_rate": 8.205761977071315e-07,
      "loss": 1.0098,
      "step": 9837
    },
    {
      "epoch": 0.7097868042278417,
      "grad_norm": 3.025993902690667,
      "learning_rate": 8.201987837301487e-07,
      "loss": 0.9876,
      "step": 9838
    },
    {
      "epoch": 0.7098589516972692,
      "grad_norm": 2.5851548038961654,
      "learning_rate": 8.198214341777416e-07,
      "loss": 0.918,
      "step": 9839
    },
    {
      "epoch": 0.7099310991666967,
      "grad_norm": 7.9228009467175555,
      "learning_rate": 8.194441490705184e-07,
      "loss": 1.0504,
      "step": 9840
    },
    {
      "epoch": 0.7100032466361242,
      "grad_norm": 5.853505523744204,
      "learning_rate": 8.190669284290794e-07,
      "loss": 0.956,
      "step": 9841
    },
    {
      "epoch": 0.7100753941055518,
      "grad_norm": 4.3526487244641165,
      "learning_rate": 8.186897722740236e-07,
      "loss": 1.0143,
      "step": 9842
    },
    {
      "epoch": 0.7101475415749793,
      "grad_norm": 4.333127872780266,
      "learning_rate": 8.183126806259463e-07,
      "loss": 1.0191,
      "step": 9843
    },
    {
      "epoch": 0.7102196890444068,
      "grad_norm": 0.6799981076200183,
      "learning_rate": 8.179356535054392e-07,
      "loss": 0.777,
      "step": 9844
    },
    {
      "epoch": 0.7102918365138343,
      "grad_norm": 4.7916836061040025,
      "learning_rate": 8.1755869093309e-07,
      "loss": 1.0621,
      "step": 9845
    },
    {
      "epoch": 0.7103639839832618,
      "grad_norm": 3.0827796713873887,
      "learning_rate": 8.171817929294836e-07,
      "loss": 0.8597,
      "step": 9846
    },
    {
      "epoch": 0.7104361314526892,
      "grad_norm": 4.021449988849468,
      "learning_rate": 8.168049595152014e-07,
      "loss": 1.0441,
      "step": 9847
    },
    {
      "epoch": 0.7105082789221168,
      "grad_norm": 7.2808290286198964,
      "learning_rate": 8.164281907108197e-07,
      "loss": 0.9721,
      "step": 9848
    },
    {
      "epoch": 0.7105804263915443,
      "grad_norm": 3.6108524881983644,
      "learning_rate": 8.16051486536913e-07,
      "loss": 0.9729,
      "step": 9849
    },
    {
      "epoch": 0.7106525738609718,
      "grad_norm": 5.0766730947238825,
      "learning_rate": 8.156748470140518e-07,
      "loss": 1.0562,
      "step": 9850
    },
    {
      "epoch": 0.7107247213303993,
      "grad_norm": 2.8336406054959973,
      "learning_rate": 8.15298272162803e-07,
      "loss": 0.9841,
      "step": 9851
    },
    {
      "epoch": 0.7107968687998268,
      "grad_norm": 6.06039486545992,
      "learning_rate": 8.149217620037301e-07,
      "loss": 1.0152,
      "step": 9852
    },
    {
      "epoch": 0.7108690162692544,
      "grad_norm": 6.55075504522934,
      "learning_rate": 8.145453165573935e-07,
      "loss": 0.87,
      "step": 9853
    },
    {
      "epoch": 0.7109411637386819,
      "grad_norm": 4.332045999679016,
      "learning_rate": 8.14168935844347e-07,
      "loss": 0.9582,
      "step": 9854
    },
    {
      "epoch": 0.7110133112081094,
      "grad_norm": 0.7263228318533517,
      "learning_rate": 8.137926198851468e-07,
      "loss": 0.8254,
      "step": 9855
    },
    {
      "epoch": 0.7110854586775369,
      "grad_norm": 0.6792952403158006,
      "learning_rate": 8.134163687003396e-07,
      "loss": 0.7861,
      "step": 9856
    },
    {
      "epoch": 0.7111576061469644,
      "grad_norm": 3.5330085089262817,
      "learning_rate": 8.13040182310472e-07,
      "loss": 0.9579,
      "step": 9857
    },
    {
      "epoch": 0.711229753616392,
      "grad_norm": 2.4247996729270036,
      "learning_rate": 8.126640607360862e-07,
      "loss": 1.0435,
      "step": 9858
    },
    {
      "epoch": 0.7113019010858194,
      "grad_norm": 5.787875281539695,
      "learning_rate": 8.122880039977205e-07,
      "loss": 0.9407,
      "step": 9859
    },
    {
      "epoch": 0.7113740485552469,
      "grad_norm": 5.346195308395747,
      "learning_rate": 8.1191201211591e-07,
      "loss": 0.946,
      "step": 9860
    },
    {
      "epoch": 0.7114461960246744,
      "grad_norm": 5.9996642972492635,
      "learning_rate": 8.115360851111865e-07,
      "loss": 0.8428,
      "step": 9861
    },
    {
      "epoch": 0.7115183434941019,
      "grad_norm": 3.9472513703509313,
      "learning_rate": 8.111602230040789e-07,
      "loss": 0.9274,
      "step": 9862
    },
    {
      "epoch": 0.7115904909635294,
      "grad_norm": 5.685858185611453,
      "learning_rate": 8.107844258151087e-07,
      "loss": 0.9863,
      "step": 9863
    },
    {
      "epoch": 0.711662638432957,
      "grad_norm": 0.7537864629425187,
      "learning_rate": 8.104086935648003e-07,
      "loss": 0.8188,
      "step": 9864
    },
    {
      "epoch": 0.7117347859023845,
      "grad_norm": 7.091564228230962,
      "learning_rate": 8.100330262736687e-07,
      "loss": 0.9164,
      "step": 9865
    },
    {
      "epoch": 0.711806933371812,
      "grad_norm": 2.793564526163808,
      "learning_rate": 8.096574239622285e-07,
      "loss": 0.9616,
      "step": 9866
    },
    {
      "epoch": 0.7118790808412395,
      "grad_norm": 3.7439503509457333,
      "learning_rate": 8.092818866509896e-07,
      "loss": 0.8629,
      "step": 9867
    },
    {
      "epoch": 0.711951228310667,
      "grad_norm": 4.794632875392029,
      "learning_rate": 8.089064143604594e-07,
      "loss": 1.0098,
      "step": 9868
    },
    {
      "epoch": 0.7120233757800946,
      "grad_norm": 7.2052716347862065,
      "learning_rate": 8.085310071111401e-07,
      "loss": 0.9172,
      "step": 9869
    },
    {
      "epoch": 0.7120955232495221,
      "grad_norm": 4.824983531548259,
      "learning_rate": 8.08155664923533e-07,
      "loss": 0.9407,
      "step": 9870
    },
    {
      "epoch": 0.7121676707189495,
      "grad_norm": 3.794256492627791,
      "learning_rate": 8.077803878181322e-07,
      "loss": 0.9842,
      "step": 9871
    },
    {
      "epoch": 0.712239818188377,
      "grad_norm": 3.1832275179973313,
      "learning_rate": 8.074051758154308e-07,
      "loss": 0.9525,
      "step": 9872
    },
    {
      "epoch": 0.7123119656578045,
      "grad_norm": 4.60365960046124,
      "learning_rate": 8.07030028935918e-07,
      "loss": 0.9385,
      "step": 9873
    },
    {
      "epoch": 0.712384113127232,
      "grad_norm": 4.598007201718021,
      "learning_rate": 8.066549472000792e-07,
      "loss": 1.0879,
      "step": 9874
    },
    {
      "epoch": 0.7124562605966596,
      "grad_norm": 3.1115060529102885,
      "learning_rate": 8.062799306283968e-07,
      "loss": 0.9117,
      "step": 9875
    },
    {
      "epoch": 0.7125284080660871,
      "grad_norm": 3.720970452584754,
      "learning_rate": 8.059049792413468e-07,
      "loss": 0.994,
      "step": 9876
    },
    {
      "epoch": 0.7126005555355146,
      "grad_norm": 3.3027825963917636,
      "learning_rate": 8.055300930594071e-07,
      "loss": 0.8949,
      "step": 9877
    },
    {
      "epoch": 0.7126727030049421,
      "grad_norm": 22.56193529093339,
      "learning_rate": 8.051552721030457e-07,
      "loss": 0.8235,
      "step": 9878
    },
    {
      "epoch": 0.7127448504743696,
      "grad_norm": 4.20342585398055,
      "learning_rate": 8.047805163927332e-07,
      "loss": 1.0148,
      "step": 9879
    },
    {
      "epoch": 0.7128169979437972,
      "grad_norm": 3.5249731942097746,
      "learning_rate": 8.044058259489315e-07,
      "loss": 0.9201,
      "step": 9880
    },
    {
      "epoch": 0.7128891454132247,
      "grad_norm": 6.011360539502573,
      "learning_rate": 8.040312007921015e-07,
      "loss": 1.0345,
      "step": 9881
    },
    {
      "epoch": 0.7129612928826521,
      "grad_norm": 2.5262503521374926,
      "learning_rate": 8.036566409427006e-07,
      "loss": 0.9011,
      "step": 9882
    },
    {
      "epoch": 0.7130334403520796,
      "grad_norm": 3.5826411502820563,
      "learning_rate": 8.032821464211817e-07,
      "loss": 0.9938,
      "step": 9883
    },
    {
      "epoch": 0.7131055878215071,
      "grad_norm": 7.652592746953965,
      "learning_rate": 8.029077172479945e-07,
      "loss": 0.8985,
      "step": 9884
    },
    {
      "epoch": 0.7131777352909346,
      "grad_norm": 5.169938476904775,
      "learning_rate": 8.025333534435857e-07,
      "loss": 0.8594,
      "step": 9885
    },
    {
      "epoch": 0.7132498827603622,
      "grad_norm": 0.7859299676431811,
      "learning_rate": 8.021590550283981e-07,
      "loss": 0.8503,
      "step": 9886
    },
    {
      "epoch": 0.7133220302297897,
      "grad_norm": 7.928844388291413,
      "learning_rate": 8.017848220228696e-07,
      "loss": 0.8465,
      "step": 9887
    },
    {
      "epoch": 0.7133941776992172,
      "grad_norm": 2.892802639940627,
      "learning_rate": 8.014106544474361e-07,
      "loss": 0.9656,
      "step": 9888
    },
    {
      "epoch": 0.7134663251686447,
      "grad_norm": 4.287624764364894,
      "learning_rate": 8.0103655232253e-07,
      "loss": 0.9495,
      "step": 9889
    },
    {
      "epoch": 0.7135384726380722,
      "grad_norm": 4.7783750978810575,
      "learning_rate": 8.006625156685793e-07,
      "loss": 0.9808,
      "step": 9890
    },
    {
      "epoch": 0.7136106201074998,
      "grad_norm": 3.755345286051372,
      "learning_rate": 8.00288544506009e-07,
      "loss": 0.8727,
      "step": 9891
    },
    {
      "epoch": 0.7136827675769273,
      "grad_norm": 2.736713651826048,
      "learning_rate": 7.999146388552407e-07,
      "loss": 0.9966,
      "step": 9892
    },
    {
      "epoch": 0.7137549150463548,
      "grad_norm": 4.938294358878049,
      "learning_rate": 7.995407987366898e-07,
      "loss": 1.0116,
      "step": 9893
    },
    {
      "epoch": 0.7138270625157822,
      "grad_norm": 0.7139719665697991,
      "learning_rate": 7.991670241707735e-07,
      "loss": 0.7755,
      "step": 9894
    },
    {
      "epoch": 0.7138992099852097,
      "grad_norm": 4.499650941662226,
      "learning_rate": 7.987933151779001e-07,
      "loss": 0.9592,
      "step": 9895
    },
    {
      "epoch": 0.7139713574546372,
      "grad_norm": 3.3886110855498495,
      "learning_rate": 7.984196717784768e-07,
      "loss": 1.0478,
      "step": 9896
    },
    {
      "epoch": 0.7140435049240648,
      "grad_norm": 6.58599860727508,
      "learning_rate": 7.980460939929072e-07,
      "loss": 0.9104,
      "step": 9897
    },
    {
      "epoch": 0.7141156523934923,
      "grad_norm": 5.036170969684078,
      "learning_rate": 7.97672581841591e-07,
      "loss": 0.8992,
      "step": 9898
    },
    {
      "epoch": 0.7141877998629198,
      "grad_norm": 3.554063276879298,
      "learning_rate": 7.972991353449241e-07,
      "loss": 0.9942,
      "step": 9899
    },
    {
      "epoch": 0.7142599473323473,
      "grad_norm": 5.468500622104297,
      "learning_rate": 7.969257545232993e-07,
      "loss": 0.9987,
      "step": 9900
    },
    {
      "epoch": 0.7143320948017748,
      "grad_norm": 11.888598556380211,
      "learning_rate": 7.965524393971062e-07,
      "loss": 1.0149,
      "step": 9901
    },
    {
      "epoch": 0.7144042422712024,
      "grad_norm": 3.7719072670808758,
      "learning_rate": 7.961791899867285e-07,
      "loss": 1.0289,
      "step": 9902
    },
    {
      "epoch": 0.7144763897406299,
      "grad_norm": 7.025755502930833,
      "learning_rate": 7.95806006312549e-07,
      "loss": 0.9947,
      "step": 9903
    },
    {
      "epoch": 0.7145485372100574,
      "grad_norm": 5.345699735395802,
      "learning_rate": 7.954328883949455e-07,
      "loss": 0.9469,
      "step": 9904
    },
    {
      "epoch": 0.7146206846794849,
      "grad_norm": 3.672722771355989,
      "learning_rate": 7.950598362542929e-07,
      "loss": 0.9315,
      "step": 9905
    },
    {
      "epoch": 0.7146928321489123,
      "grad_norm": 3.1655682783703174,
      "learning_rate": 7.946868499109623e-07,
      "loss": 0.9108,
      "step": 9906
    },
    {
      "epoch": 0.7147649796183398,
      "grad_norm": 6.219629886814414,
      "learning_rate": 7.943139293853207e-07,
      "loss": 1.073,
      "step": 9907
    },
    {
      "epoch": 0.7148371270877674,
      "grad_norm": 3.198882778165274,
      "learning_rate": 7.939410746977322e-07,
      "loss": 0.9178,
      "step": 9908
    },
    {
      "epoch": 0.7149092745571949,
      "grad_norm": 2.345273247681968,
      "learning_rate": 7.935682858685575e-07,
      "loss": 0.9405,
      "step": 9909
    },
    {
      "epoch": 0.7149814220266224,
      "grad_norm": 3.240054538880016,
      "learning_rate": 7.931955629181521e-07,
      "loss": 0.8956,
      "step": 9910
    },
    {
      "epoch": 0.7150535694960499,
      "grad_norm": 3.2273183311434694,
      "learning_rate": 7.928229058668694e-07,
      "loss": 0.9413,
      "step": 9911
    },
    {
      "epoch": 0.7151257169654774,
      "grad_norm": 3.6790304386987707,
      "learning_rate": 7.924503147350592e-07,
      "loss": 0.986,
      "step": 9912
    },
    {
      "epoch": 0.715197864434905,
      "grad_norm": 4.946566984572364,
      "learning_rate": 7.92077789543067e-07,
      "loss": 0.9476,
      "step": 9913
    },
    {
      "epoch": 0.7152700119043325,
      "grad_norm": 4.755156478794087,
      "learning_rate": 7.917053303112359e-07,
      "loss": 0.7812,
      "step": 9914
    },
    {
      "epoch": 0.71534215937376,
      "grad_norm": 6.301217146267383,
      "learning_rate": 7.913329370599025e-07,
      "loss": 0.9521,
      "step": 9915
    },
    {
      "epoch": 0.7154143068431875,
      "grad_norm": 4.465181433055385,
      "learning_rate": 7.909606098094041e-07,
      "loss": 0.9794,
      "step": 9916
    },
    {
      "epoch": 0.715486454312615,
      "grad_norm": 2.893542822389612,
      "learning_rate": 7.9058834858007e-07,
      "loss": 1.0303,
      "step": 9917
    },
    {
      "epoch": 0.7155586017820424,
      "grad_norm": 4.711783413829924,
      "learning_rate": 7.902161533922305e-07,
      "loss": 0.9309,
      "step": 9918
    },
    {
      "epoch": 0.71563074925147,
      "grad_norm": 0.832483338942499,
      "learning_rate": 7.898440242662074e-07,
      "loss": 0.8542,
      "step": 9919
    },
    {
      "epoch": 0.7157028967208975,
      "grad_norm": 4.2439383868060405,
      "learning_rate": 7.894719612223226e-07,
      "loss": 0.9597,
      "step": 9920
    },
    {
      "epoch": 0.715775044190325,
      "grad_norm": 7.636334359331692,
      "learning_rate": 7.890999642808926e-07,
      "loss": 0.9785,
      "step": 9921
    },
    {
      "epoch": 0.7158471916597525,
      "grad_norm": 4.252756683299508,
      "learning_rate": 7.887280334622311e-07,
      "loss": 0.8792,
      "step": 9922
    },
    {
      "epoch": 0.71591933912918,
      "grad_norm": 10.462189671370261,
      "learning_rate": 7.883561687866476e-07,
      "loss": 0.9432,
      "step": 9923
    },
    {
      "epoch": 0.7159914865986076,
      "grad_norm": 2.873246860869865,
      "learning_rate": 7.879843702744491e-07,
      "loss": 1.0653,
      "step": 9924
    },
    {
      "epoch": 0.7160636340680351,
      "grad_norm": 3.8205568824410627,
      "learning_rate": 7.876126379459367e-07,
      "loss": 0.8836,
      "step": 9925
    },
    {
      "epoch": 0.7161357815374626,
      "grad_norm": 4.623901442919498,
      "learning_rate": 7.872409718214099e-07,
      "loss": 0.8852,
      "step": 9926
    },
    {
      "epoch": 0.7162079290068901,
      "grad_norm": 4.515915382842676,
      "learning_rate": 7.868693719211643e-07,
      "loss": 0.9872,
      "step": 9927
    },
    {
      "epoch": 0.7162800764763176,
      "grad_norm": 0.7588112532072301,
      "learning_rate": 7.864978382654915e-07,
      "loss": 0.8208,
      "step": 9928
    },
    {
      "epoch": 0.716352223945745,
      "grad_norm": 2.7773955135108968,
      "learning_rate": 7.861263708746795e-07,
      "loss": 1.0088,
      "step": 9929
    },
    {
      "epoch": 0.7164243714151726,
      "grad_norm": 4.300353927240889,
      "learning_rate": 7.857549697690127e-07,
      "loss": 1.0279,
      "step": 9930
    },
    {
      "epoch": 0.7164965188846001,
      "grad_norm": 9.429558855015186,
      "learning_rate": 7.853836349687728e-07,
      "loss": 0.9415,
      "step": 9931
    },
    {
      "epoch": 0.7165686663540276,
      "grad_norm": 4.504868522867809,
      "learning_rate": 7.850123664942348e-07,
      "loss": 0.9427,
      "step": 9932
    },
    {
      "epoch": 0.7166408138234551,
      "grad_norm": 2.99278615834192,
      "learning_rate": 7.846411643656751e-07,
      "loss": 0.9274,
      "step": 9933
    },
    {
      "epoch": 0.7167129612928826,
      "grad_norm": 5.252696480142366,
      "learning_rate": 7.842700286033616e-07,
      "loss": 0.9947,
      "step": 9934
    },
    {
      "epoch": 0.7167851087623102,
      "grad_norm": 2.8562950681640817,
      "learning_rate": 7.838989592275616e-07,
      "loss": 0.9402,
      "step": 9935
    },
    {
      "epoch": 0.7168572562317377,
      "grad_norm": 3.3574802466838967,
      "learning_rate": 7.835279562585373e-07,
      "loss": 0.9479,
      "step": 9936
    },
    {
      "epoch": 0.7169294037011652,
      "grad_norm": 5.049489575264328,
      "learning_rate": 7.831570197165481e-07,
      "loss": 0.9484,
      "step": 9937
    },
    {
      "epoch": 0.7170015511705927,
      "grad_norm": 4.829372800300999,
      "learning_rate": 7.827861496218502e-07,
      "loss": 0.9389,
      "step": 9938
    },
    {
      "epoch": 0.7170736986400202,
      "grad_norm": 19.99039839102841,
      "learning_rate": 7.824153459946934e-07,
      "loss": 0.9465,
      "step": 9939
    },
    {
      "epoch": 0.7171458461094478,
      "grad_norm": 5.560191672060749,
      "learning_rate": 7.820446088553285e-07,
      "loss": 0.9867,
      "step": 9940
    },
    {
      "epoch": 0.7172179935788752,
      "grad_norm": 7.984048437319434,
      "learning_rate": 7.81673938223998e-07,
      "loss": 0.9692,
      "step": 9941
    },
    {
      "epoch": 0.7172901410483027,
      "grad_norm": 3.631260825582575,
      "learning_rate": 7.813033341209439e-07,
      "loss": 1.0163,
      "step": 9942
    },
    {
      "epoch": 0.7173622885177302,
      "grad_norm": 3.0674128149897717,
      "learning_rate": 7.809327965664032e-07,
      "loss": 0.8044,
      "step": 9943
    },
    {
      "epoch": 0.7174344359871577,
      "grad_norm": 3.004275136850883,
      "learning_rate": 7.805623255806095e-07,
      "loss": 0.8821,
      "step": 9944
    },
    {
      "epoch": 0.7175065834565852,
      "grad_norm": 5.688175035597595,
      "learning_rate": 7.801919211837931e-07,
      "loss": 0.9973,
      "step": 9945
    },
    {
      "epoch": 0.7175787309260128,
      "grad_norm": 2.70840071569886,
      "learning_rate": 7.79821583396181e-07,
      "loss": 0.8686,
      "step": 9946
    },
    {
      "epoch": 0.7176508783954403,
      "grad_norm": 4.054819681883264,
      "learning_rate": 7.794513122379938e-07,
      "loss": 1.0353,
      "step": 9947
    },
    {
      "epoch": 0.7177230258648678,
      "grad_norm": 4.827406351375325,
      "learning_rate": 7.790811077294537e-07,
      "loss": 0.9995,
      "step": 9948
    },
    {
      "epoch": 0.7177951733342953,
      "grad_norm": 6.576038455544404,
      "learning_rate": 7.787109698907737e-07,
      "loss": 0.998,
      "step": 9949
    },
    {
      "epoch": 0.7178673208037228,
      "grad_norm": 3.388621357917763,
      "learning_rate": 7.783408987421667e-07,
      "loss": 1.0021,
      "step": 9950
    },
    {
      "epoch": 0.7179394682731504,
      "grad_norm": 5.409134464656826,
      "learning_rate": 7.779708943038408e-07,
      "loss": 0.9315,
      "step": 9951
    },
    {
      "epoch": 0.7180116157425779,
      "grad_norm": 4.120938237587141,
      "learning_rate": 7.776009565960007e-07,
      "loss": 0.9401,
      "step": 9952
    },
    {
      "epoch": 0.7180837632120053,
      "grad_norm": 6.069202297025269,
      "learning_rate": 7.772310856388475e-07,
      "loss": 0.8968,
      "step": 9953
    },
    {
      "epoch": 0.7181559106814328,
      "grad_norm": 4.2991251654389355,
      "learning_rate": 7.768612814525772e-07,
      "loss": 0.968,
      "step": 9954
    },
    {
      "epoch": 0.7182280581508603,
      "grad_norm": 3.7829313717253177,
      "learning_rate": 7.764915440573856e-07,
      "loss": 0.889,
      "step": 9955
    },
    {
      "epoch": 0.7183002056202878,
      "grad_norm": 4.180925360651382,
      "learning_rate": 7.7612187347346e-07,
      "loss": 0.9148,
      "step": 9956
    },
    {
      "epoch": 0.7183723530897154,
      "grad_norm": 4.645715663543052,
      "learning_rate": 7.757522697209899e-07,
      "loss": 0.9732,
      "step": 9957
    },
    {
      "epoch": 0.7184445005591429,
      "grad_norm": 6.175136969278938,
      "learning_rate": 7.753827328201554e-07,
      "loss": 0.9672,
      "step": 9958
    },
    {
      "epoch": 0.7185166480285704,
      "grad_norm": 4.064939500046475,
      "learning_rate": 7.750132627911367e-07,
      "loss": 0.8374,
      "step": 9959
    },
    {
      "epoch": 0.7185887954979979,
      "grad_norm": 2.9045221258060465,
      "learning_rate": 7.746438596541089e-07,
      "loss": 0.9062,
      "step": 9960
    },
    {
      "epoch": 0.7186609429674254,
      "grad_norm": 3.2362304378078885,
      "learning_rate": 7.742745234292439e-07,
      "loss": 0.9428,
      "step": 9961
    },
    {
      "epoch": 0.718733090436853,
      "grad_norm": 8.032680517662092,
      "learning_rate": 7.739052541367102e-07,
      "loss": 1.0653,
      "step": 9962
    },
    {
      "epoch": 0.7188052379062805,
      "grad_norm": 4.680284974640717,
      "learning_rate": 7.735360517966711e-07,
      "loss": 0.9676,
      "step": 9963
    },
    {
      "epoch": 0.718877385375708,
      "grad_norm": 3.5045702931248397,
      "learning_rate": 7.73166916429288e-07,
      "loss": 1.0524,
      "step": 9964
    },
    {
      "epoch": 0.7189495328451354,
      "grad_norm": 4.981576355087553,
      "learning_rate": 7.727978480547182e-07,
      "loss": 1.0244,
      "step": 9965
    },
    {
      "epoch": 0.7190216803145629,
      "grad_norm": 4.393900779426891,
      "learning_rate": 7.724288466931146e-07,
      "loss": 0.9606,
      "step": 9966
    },
    {
      "epoch": 0.7190938277839904,
      "grad_norm": 6.190153535531815,
      "learning_rate": 7.720599123646276e-07,
      "loss": 0.8902,
      "step": 9967
    },
    {
      "epoch": 0.719165975253418,
      "grad_norm": 5.239044748259703,
      "learning_rate": 7.716910450894039e-07,
      "loss": 0.9724,
      "step": 9968
    },
    {
      "epoch": 0.7192381227228455,
      "grad_norm": 4.156636614493296,
      "learning_rate": 7.713222448875835e-07,
      "loss": 0.9393,
      "step": 9969
    },
    {
      "epoch": 0.719310270192273,
      "grad_norm": 5.4440116039884,
      "learning_rate": 7.709535117793085e-07,
      "loss": 1.0221,
      "step": 9970
    },
    {
      "epoch": 0.7193824176617005,
      "grad_norm": 4.597761621296773,
      "learning_rate": 7.705848457847111e-07,
      "loss": 0.951,
      "step": 9971
    },
    {
      "epoch": 0.719454565131128,
      "grad_norm": 3.4418841061628362,
      "learning_rate": 7.702162469239252e-07,
      "loss": 1.0481,
      "step": 9972
    },
    {
      "epoch": 0.7195267126005556,
      "grad_norm": 4.60803182396231,
      "learning_rate": 7.698477152170773e-07,
      "loss": 0.8853,
      "step": 9973
    },
    {
      "epoch": 0.7195988600699831,
      "grad_norm": 3.5536879926351195,
      "learning_rate": 7.694792506842916e-07,
      "loss": 1.0351,
      "step": 9974
    },
    {
      "epoch": 0.7196710075394106,
      "grad_norm": 3.4149473252709184,
      "learning_rate": 7.691108533456887e-07,
      "loss": 1.0482,
      "step": 9975
    },
    {
      "epoch": 0.7197431550088381,
      "grad_norm": 3.9614057696728153,
      "learning_rate": 7.687425232213855e-07,
      "loss": 0.9807,
      "step": 9976
    },
    {
      "epoch": 0.7198153024782655,
      "grad_norm": 2.668120395265932,
      "learning_rate": 7.68374260331496e-07,
      "loss": 0.9979,
      "step": 9977
    },
    {
      "epoch": 0.719887449947693,
      "grad_norm": 9.09208784610923,
      "learning_rate": 7.680060646961273e-07,
      "loss": 1.0265,
      "step": 9978
    },
    {
      "epoch": 0.7199595974171206,
      "grad_norm": 8.752392686950353,
      "learning_rate": 7.676379363353882e-07,
      "loss": 0.9296,
      "step": 9979
    },
    {
      "epoch": 0.7200317448865481,
      "grad_norm": 5.890489976382715,
      "learning_rate": 7.672698752693787e-07,
      "loss": 0.9992,
      "step": 9980
    },
    {
      "epoch": 0.7201038923559756,
      "grad_norm": 3.567923850722325,
      "learning_rate": 7.669018815181978e-07,
      "loss": 1.0113,
      "step": 9981
    },
    {
      "epoch": 0.7201760398254031,
      "grad_norm": 0.7549651225667758,
      "learning_rate": 7.665339551019403e-07,
      "loss": 0.8015,
      "step": 9982
    },
    {
      "epoch": 0.7202481872948306,
      "grad_norm": 4.828421080563043,
      "learning_rate": 7.661660960406975e-07,
      "loss": 0.9903,
      "step": 9983
    },
    {
      "epoch": 0.7203203347642582,
      "grad_norm": 0.621041061757674,
      "learning_rate": 7.657983043545568e-07,
      "loss": 0.7895,
      "step": 9984
    },
    {
      "epoch": 0.7203924822336857,
      "grad_norm": 6.990717045975651,
      "learning_rate": 7.654305800636023e-07,
      "loss": 0.9766,
      "step": 9985
    },
    {
      "epoch": 0.7204646297031132,
      "grad_norm": 5.346859389816854,
      "learning_rate": 7.650629231879131e-07,
      "loss": 0.9727,
      "step": 9986
    },
    {
      "epoch": 0.7205367771725407,
      "grad_norm": 3.576167608203353,
      "learning_rate": 7.646953337475659e-07,
      "loss": 0.9032,
      "step": 9987
    },
    {
      "epoch": 0.7206089246419681,
      "grad_norm": 2.8629601521012358,
      "learning_rate": 7.643278117626338e-07,
      "loss": 0.9034,
      "step": 9988
    },
    {
      "epoch": 0.7206810721113956,
      "grad_norm": 10.321032514152988,
      "learning_rate": 7.639603572531854e-07,
      "loss": 0.8939,
      "step": 9989
    },
    {
      "epoch": 0.7207532195808232,
      "grad_norm": 3.2407997194605027,
      "learning_rate": 7.635929702392863e-07,
      "loss": 0.8565,
      "step": 9990
    },
    {
      "epoch": 0.7208253670502507,
      "grad_norm": 3.7577104772728114,
      "learning_rate": 7.632256507409982e-07,
      "loss": 0.9436,
      "step": 9991
    },
    {
      "epoch": 0.7208975145196782,
      "grad_norm": 4.38122703992457,
      "learning_rate": 7.628583987783794e-07,
      "loss": 0.8743,
      "step": 9992
    },
    {
      "epoch": 0.7209696619891057,
      "grad_norm": 5.223661932201738,
      "learning_rate": 7.624912143714826e-07,
      "loss": 1.0662,
      "step": 9993
    },
    {
      "epoch": 0.7210418094585332,
      "grad_norm": 3.9399786519667974,
      "learning_rate": 7.621240975403607e-07,
      "loss": 1.0269,
      "step": 9994
    },
    {
      "epoch": 0.7211139569279608,
      "grad_norm": 3.487793618151014,
      "learning_rate": 7.61757048305058e-07,
      "loss": 0.9161,
      "step": 9995
    },
    {
      "epoch": 0.7211861043973883,
      "grad_norm": 4.66072471713531,
      "learning_rate": 7.613900666856203e-07,
      "loss": 1.0326,
      "step": 9996
    },
    {
      "epoch": 0.7212582518668158,
      "grad_norm": 4.664014857574406,
      "learning_rate": 7.610231527020854e-07,
      "loss": 0.899,
      "step": 9997
    },
    {
      "epoch": 0.7213303993362433,
      "grad_norm": 4.2217353713003005,
      "learning_rate": 7.606563063744895e-07,
      "loss": 0.9578,
      "step": 9998
    },
    {
      "epoch": 0.7214025468056708,
      "grad_norm": 3.013715070018132,
      "learning_rate": 7.602895277228647e-07,
      "loss": 1.0496,
      "step": 9999
    },
    {
      "epoch": 0.7214746942750982,
      "grad_norm": 0.7785736782021475,
      "learning_rate": 7.599228167672398e-07,
      "loss": 0.8525,
      "step": 10000
    },
    {
      "epoch": 0.7215468417445258,
      "grad_norm": 4.174161020372719,
      "learning_rate": 7.595561735276398e-07,
      "loss": 0.8583,
      "step": 10001
    },
    {
      "epoch": 0.7216189892139533,
      "grad_norm": 3.190105196375259,
      "learning_rate": 7.591895980240846e-07,
      "loss": 0.9861,
      "step": 10002
    },
    {
      "epoch": 0.7216911366833808,
      "grad_norm": 5.051270074046914,
      "learning_rate": 7.58823090276592e-07,
      "loss": 0.87,
      "step": 10003
    },
    {
      "epoch": 0.7217632841528083,
      "grad_norm": 12.41476125413092,
      "learning_rate": 7.584566503051759e-07,
      "loss": 0.9125,
      "step": 10004
    },
    {
      "epoch": 0.7218354316222358,
      "grad_norm": 0.754416693204509,
      "learning_rate": 7.580902781298459e-07,
      "loss": 0.8462,
      "step": 10005
    },
    {
      "epoch": 0.7219075790916634,
      "grad_norm": 5.422272681347167,
      "learning_rate": 7.577239737706087e-07,
      "loss": 0.9776,
      "step": 10006
    },
    {
      "epoch": 0.7219797265610909,
      "grad_norm": 0.722612905491179,
      "learning_rate": 7.573577372474671e-07,
      "loss": 0.8952,
      "step": 10007
    },
    {
      "epoch": 0.7220518740305184,
      "grad_norm": 3.3696465596122307,
      "learning_rate": 7.569915685804178e-07,
      "loss": 1.0416,
      "step": 10008
    },
    {
      "epoch": 0.7221240214999459,
      "grad_norm": 4.358824336230609,
      "learning_rate": 7.56625467789459e-07,
      "loss": 1.0442,
      "step": 10009
    },
    {
      "epoch": 0.7221961689693734,
      "grad_norm": 5.171758091460949,
      "learning_rate": 7.562594348945801e-07,
      "loss": 0.8376,
      "step": 10010
    },
    {
      "epoch": 0.722268316438801,
      "grad_norm": 10.575158113059631,
      "learning_rate": 7.558934699157691e-07,
      "loss": 0.9619,
      "step": 10011
    },
    {
      "epoch": 0.7223404639082284,
      "grad_norm": 3.699568058266941,
      "learning_rate": 7.555275728730101e-07,
      "loss": 1.0216,
      "step": 10012
    },
    {
      "epoch": 0.7224126113776559,
      "grad_norm": 6.5207938131124035,
      "learning_rate": 7.551617437862837e-07,
      "loss": 0.9135,
      "step": 10013
    },
    {
      "epoch": 0.7224847588470834,
      "grad_norm": 8.531969871471377,
      "learning_rate": 7.54795982675566e-07,
      "loss": 0.8902,
      "step": 10014
    },
    {
      "epoch": 0.7225569063165109,
      "grad_norm": 3.7943866880485366,
      "learning_rate": 7.544302895608302e-07,
      "loss": 0.866,
      "step": 10015
    },
    {
      "epoch": 0.7226290537859384,
      "grad_norm": 2.730973487666648,
      "learning_rate": 7.54064664462046e-07,
      "loss": 1.04,
      "step": 10016
    },
    {
      "epoch": 0.722701201255366,
      "grad_norm": 3.9007799835783667,
      "learning_rate": 7.53699107399177e-07,
      "loss": 1.0169,
      "step": 10017
    },
    {
      "epoch": 0.7227733487247935,
      "grad_norm": 2.6319649348024727,
      "learning_rate": 7.533336183921872e-07,
      "loss": 0.9709,
      "step": 10018
    },
    {
      "epoch": 0.722845496194221,
      "grad_norm": 7.053930163789207,
      "learning_rate": 7.529681974610329e-07,
      "loss": 1.0218,
      "step": 10019
    },
    {
      "epoch": 0.7229176436636485,
      "grad_norm": 3.0126242457683645,
      "learning_rate": 7.526028446256689e-07,
      "loss": 1.0127,
      "step": 10020
    },
    {
      "epoch": 0.722989791133076,
      "grad_norm": 3.536133351888249,
      "learning_rate": 7.52237559906046e-07,
      "loss": 1.0062,
      "step": 10021
    },
    {
      "epoch": 0.7230619386025036,
      "grad_norm": 5.318444729392194,
      "learning_rate": 7.518723433221109e-07,
      "loss": 1.0203,
      "step": 10022
    },
    {
      "epoch": 0.7231340860719311,
      "grad_norm": 2.951331184172151,
      "learning_rate": 7.515071948938067e-07,
      "loss": 0.9343,
      "step": 10023
    },
    {
      "epoch": 0.7232062335413585,
      "grad_norm": 2.613800772671385,
      "learning_rate": 7.511421146410735e-07,
      "loss": 0.9843,
      "step": 10024
    },
    {
      "epoch": 0.723278381010786,
      "grad_norm": 6.860140746872074,
      "learning_rate": 7.507771025838456e-07,
      "loss": 0.8805,
      "step": 10025
    },
    {
      "epoch": 0.7233505284802135,
      "grad_norm": 5.115476179434169,
      "learning_rate": 7.504121587420555e-07,
      "loss": 1.0151,
      "step": 10026
    },
    {
      "epoch": 0.723422675949641,
      "grad_norm": 4.400160890585378,
      "learning_rate": 7.500472831356319e-07,
      "loss": 0.9107,
      "step": 10027
    },
    {
      "epoch": 0.7234948234190686,
      "grad_norm": 4.349230266300205,
      "learning_rate": 7.496824757844988e-07,
      "loss": 0.8971,
      "step": 10028
    },
    {
      "epoch": 0.7235669708884961,
      "grad_norm": 3.133208518552115,
      "learning_rate": 7.493177367085772e-07,
      "loss": 0.9358,
      "step": 10029
    },
    {
      "epoch": 0.7236391183579236,
      "grad_norm": 4.552249838477557,
      "learning_rate": 7.489530659277842e-07,
      "loss": 0.9629,
      "step": 10030
    },
    {
      "epoch": 0.7237112658273511,
      "grad_norm": 2.8530184438639425,
      "learning_rate": 7.485884634620336e-07,
      "loss": 0.8873,
      "step": 10031
    },
    {
      "epoch": 0.7237834132967786,
      "grad_norm": 6.651188748455281,
      "learning_rate": 7.482239293312334e-07,
      "loss": 0.9722,
      "step": 10032
    },
    {
      "epoch": 0.7238555607662062,
      "grad_norm": 4.716043466615199,
      "learning_rate": 7.478594635552915e-07,
      "loss": 0.959,
      "step": 10033
    },
    {
      "epoch": 0.7239277082356337,
      "grad_norm": 4.948677448878841,
      "learning_rate": 7.474950661541086e-07,
      "loss": 1.0072,
      "step": 10034
    },
    {
      "epoch": 0.7239998557050611,
      "grad_norm": 3.566676318669554,
      "learning_rate": 7.471307371475834e-07,
      "loss": 0.9966,
      "step": 10035
    },
    {
      "epoch": 0.7240720031744886,
      "grad_norm": 4.165377124872036,
      "learning_rate": 7.467664765556108e-07,
      "loss": 0.9239,
      "step": 10036
    },
    {
      "epoch": 0.7241441506439161,
      "grad_norm": 4.896570656088144,
      "learning_rate": 7.464022843980815e-07,
      "loss": 0.8895,
      "step": 10037
    },
    {
      "epoch": 0.7242162981133436,
      "grad_norm": 4.733743656981202,
      "learning_rate": 7.460381606948827e-07,
      "loss": 0.883,
      "step": 10038
    },
    {
      "epoch": 0.7242884455827712,
      "grad_norm": 4.425720699263087,
      "learning_rate": 7.45674105465898e-07,
      "loss": 0.9327,
      "step": 10039
    },
    {
      "epoch": 0.7243605930521987,
      "grad_norm": 7.603264268257351,
      "learning_rate": 7.453101187310078e-07,
      "loss": 0.9753,
      "step": 10040
    },
    {
      "epoch": 0.7244327405216262,
      "grad_norm": 2.0926599654258204,
      "learning_rate": 7.449462005100866e-07,
      "loss": 0.894,
      "step": 10041
    },
    {
      "epoch": 0.7245048879910537,
      "grad_norm": 3.5120155624879037,
      "learning_rate": 7.445823508230073e-07,
      "loss": 0.9644,
      "step": 10042
    },
    {
      "epoch": 0.7245770354604812,
      "grad_norm": 3.8517767980047792,
      "learning_rate": 7.442185696896384e-07,
      "loss": 0.8583,
      "step": 10043
    },
    {
      "epoch": 0.7246491829299088,
      "grad_norm": 4.326200532495377,
      "learning_rate": 7.438548571298449e-07,
      "loss": 0.9746,
      "step": 10044
    },
    {
      "epoch": 0.7247213303993363,
      "grad_norm": 26.291182121088028,
      "learning_rate": 7.434912131634872e-07,
      "loss": 0.8419,
      "step": 10045
    },
    {
      "epoch": 0.7247934778687638,
      "grad_norm": 4.308221878784325,
      "learning_rate": 7.431276378104239e-07,
      "loss": 0.9592,
      "step": 10046
    },
    {
      "epoch": 0.7248656253381912,
      "grad_norm": 4.116677182038709,
      "learning_rate": 7.427641310905059e-07,
      "loss": 0.9179,
      "step": 10047
    },
    {
      "epoch": 0.7249377728076187,
      "grad_norm": 2.973017145573832,
      "learning_rate": 7.424006930235862e-07,
      "loss": 0.8632,
      "step": 10048
    },
    {
      "epoch": 0.7250099202770462,
      "grad_norm": 4.530435002137626,
      "learning_rate": 7.420373236295083e-07,
      "loss": 1.059,
      "step": 10049
    },
    {
      "epoch": 0.7250820677464738,
      "grad_norm": 4.816168241660836,
      "learning_rate": 7.416740229281153e-07,
      "loss": 0.961,
      "step": 10050
    },
    {
      "epoch": 0.7251542152159013,
      "grad_norm": 3.529142172652227,
      "learning_rate": 7.41310790939246e-07,
      "loss": 1.0041,
      "step": 10051
    },
    {
      "epoch": 0.7252263626853288,
      "grad_norm": 4.817988844944513,
      "learning_rate": 7.409476276827347e-07,
      "loss": 0.9903,
      "step": 10052
    },
    {
      "epoch": 0.7252985101547563,
      "grad_norm": 3.7839025856205417,
      "learning_rate": 7.405845331784126e-07,
      "loss": 1.0141,
      "step": 10053
    },
    {
      "epoch": 0.7253706576241838,
      "grad_norm": 4.48355722447387,
      "learning_rate": 7.402215074461072e-07,
      "loss": 0.8878,
      "step": 10054
    },
    {
      "epoch": 0.7254428050936114,
      "grad_norm": 4.851680244331576,
      "learning_rate": 7.39858550505642e-07,
      "loss": 0.9623,
      "step": 10055
    },
    {
      "epoch": 0.7255149525630389,
      "grad_norm": 3.72365163682791,
      "learning_rate": 7.394956623768361e-07,
      "loss": 0.9472,
      "step": 10056
    },
    {
      "epoch": 0.7255871000324664,
      "grad_norm": 4.8836274710506915,
      "learning_rate": 7.391328430795057e-07,
      "loss": 0.9849,
      "step": 10057
    },
    {
      "epoch": 0.7256592475018939,
      "grad_norm": 5.135819899672312,
      "learning_rate": 7.387700926334633e-07,
      "loss": 0.9458,
      "step": 10058
    },
    {
      "epoch": 0.7257313949713213,
      "grad_norm": 5.398396515311138,
      "learning_rate": 7.38407411058517e-07,
      "loss": 0.9425,
      "step": 10059
    },
    {
      "epoch": 0.7258035424407489,
      "grad_norm": 4.580217840019854,
      "learning_rate": 7.380447983744719e-07,
      "loss": 0.9159,
      "step": 10060
    },
    {
      "epoch": 0.7258756899101764,
      "grad_norm": 4.138657588313,
      "learning_rate": 7.376822546011287e-07,
      "loss": 0.9722,
      "step": 10061
    },
    {
      "epoch": 0.7259478373796039,
      "grad_norm": 3.1190536381792726,
      "learning_rate": 7.373197797582848e-07,
      "loss": 0.9876,
      "step": 10062
    },
    {
      "epoch": 0.7260199848490314,
      "grad_norm": 3.4599919459078325,
      "learning_rate": 7.36957373865734e-07,
      "loss": 0.9839,
      "step": 10063
    },
    {
      "epoch": 0.7260921323184589,
      "grad_norm": 6.005157796928872,
      "learning_rate": 7.365950369432647e-07,
      "loss": 1.0297,
      "step": 10064
    },
    {
      "epoch": 0.7261642797878864,
      "grad_norm": 4.033674826759369,
      "learning_rate": 7.362327690106635e-07,
      "loss": 0.8418,
      "step": 10065
    },
    {
      "epoch": 0.726236427257314,
      "grad_norm": 11.068010412544231,
      "learning_rate": 7.358705700877124e-07,
      "loss": 0.9782,
      "step": 10066
    },
    {
      "epoch": 0.7263085747267415,
      "grad_norm": 5.099528773360059,
      "learning_rate": 7.3550844019419e-07,
      "loss": 1.0235,
      "step": 10067
    },
    {
      "epoch": 0.726380722196169,
      "grad_norm": 3.8891146382399886,
      "learning_rate": 7.351463793498704e-07,
      "loss": 1.0059,
      "step": 10068
    },
    {
      "epoch": 0.7264528696655965,
      "grad_norm": 6.093798201932757,
      "learning_rate": 7.34784387574525e-07,
      "loss": 0.9176,
      "step": 10069
    },
    {
      "epoch": 0.726525017135024,
      "grad_norm": 5.164774373860463,
      "learning_rate": 7.344224648879211e-07,
      "loss": 0.9381,
      "step": 10070
    },
    {
      "epoch": 0.7265971646044515,
      "grad_norm": 2.8053848972180817,
      "learning_rate": 7.3406061130982e-07,
      "loss": 0.9542,
      "step": 10071
    },
    {
      "epoch": 0.726669312073879,
      "grad_norm": 3.7569879908142583,
      "learning_rate": 7.33698826859984e-07,
      "loss": 0.9991,
      "step": 10072
    },
    {
      "epoch": 0.7267414595433065,
      "grad_norm": 6.698363705000117,
      "learning_rate": 7.333371115581665e-07,
      "loss": 0.9866,
      "step": 10073
    },
    {
      "epoch": 0.726813607012734,
      "grad_norm": 8.29898280149747,
      "learning_rate": 7.329754654241205e-07,
      "loss": 1.0027,
      "step": 10074
    },
    {
      "epoch": 0.7268857544821615,
      "grad_norm": 3.025770445513081,
      "learning_rate": 7.326138884775939e-07,
      "loss": 0.9637,
      "step": 10075
    },
    {
      "epoch": 0.726957901951589,
      "grad_norm": 3.54757159278465,
      "learning_rate": 7.322523807383312e-07,
      "loss": 0.9063,
      "step": 10076
    },
    {
      "epoch": 0.7270300494210166,
      "grad_norm": 3.7565354935345106,
      "learning_rate": 7.318909422260729e-07,
      "loss": 0.9362,
      "step": 10077
    },
    {
      "epoch": 0.7271021968904441,
      "grad_norm": 6.481647110694791,
      "learning_rate": 7.315295729605563e-07,
      "loss": 0.9085,
      "step": 10078
    },
    {
      "epoch": 0.7271743443598716,
      "grad_norm": 12.386247551941175,
      "learning_rate": 7.311682729615137e-07,
      "loss": 1.0118,
      "step": 10079
    },
    {
      "epoch": 0.7272464918292991,
      "grad_norm": 4.456258320633449,
      "learning_rate": 7.308070422486746e-07,
      "loss": 0.9366,
      "step": 10080
    },
    {
      "epoch": 0.7273186392987266,
      "grad_norm": 6.495761369653628,
      "learning_rate": 7.304458808417644e-07,
      "loss": 0.871,
      "step": 10081
    },
    {
      "epoch": 0.727390786768154,
      "grad_norm": 0.7073418372389386,
      "learning_rate": 7.30084788760505e-07,
      "loss": 0.8217,
      "step": 10082
    },
    {
      "epoch": 0.7274629342375816,
      "grad_norm": 4.1677337551636695,
      "learning_rate": 7.297237660246142e-07,
      "loss": 0.948,
      "step": 10083
    },
    {
      "epoch": 0.7275350817070091,
      "grad_norm": 4.346634380556113,
      "learning_rate": 7.29362812653806e-07,
      "loss": 1.1108,
      "step": 10084
    },
    {
      "epoch": 0.7276072291764366,
      "grad_norm": 4.957197662999121,
      "learning_rate": 7.290019286677917e-07,
      "loss": 0.9279,
      "step": 10085
    },
    {
      "epoch": 0.7276793766458641,
      "grad_norm": 10.585010625935709,
      "learning_rate": 7.286411140862759e-07,
      "loss": 0.8952,
      "step": 10086
    },
    {
      "epoch": 0.7277515241152916,
      "grad_norm": 3.957322135393626,
      "learning_rate": 7.282803689289634e-07,
      "loss": 0.9915,
      "step": 10087
    },
    {
      "epoch": 0.7278236715847192,
      "grad_norm": 4.245098204358579,
      "learning_rate": 7.279196932155516e-07,
      "loss": 1.0609,
      "step": 10088
    },
    {
      "epoch": 0.7278958190541467,
      "grad_norm": 3.4205829846258538,
      "learning_rate": 7.275590869657364e-07,
      "loss": 0.9759,
      "step": 10089
    },
    {
      "epoch": 0.7279679665235742,
      "grad_norm": 5.164536908818055,
      "learning_rate": 7.271985501992088e-07,
      "loss": 0.9815,
      "step": 10090
    },
    {
      "epoch": 0.7280401139930017,
      "grad_norm": 3.2565009315146165,
      "learning_rate": 7.268380829356569e-07,
      "loss": 0.9777,
      "step": 10091
    },
    {
      "epoch": 0.7281122614624292,
      "grad_norm": 4.534239315681849,
      "learning_rate": 7.26477685194764e-07,
      "loss": 0.8992,
      "step": 10092
    },
    {
      "epoch": 0.7281844089318568,
      "grad_norm": 3.5617771419182596,
      "learning_rate": 7.261173569962105e-07,
      "loss": 0.9405,
      "step": 10093
    },
    {
      "epoch": 0.7282565564012842,
      "grad_norm": 0.8417564081435837,
      "learning_rate": 7.257570983596731e-07,
      "loss": 0.8734,
      "step": 10094
    },
    {
      "epoch": 0.7283287038707117,
      "grad_norm": 4.641690218652846,
      "learning_rate": 7.253969093048227e-07,
      "loss": 0.9483,
      "step": 10095
    },
    {
      "epoch": 0.7284008513401392,
      "grad_norm": 4.604367189945719,
      "learning_rate": 7.250367898513286e-07,
      "loss": 0.9836,
      "step": 10096
    },
    {
      "epoch": 0.7284729988095667,
      "grad_norm": 4.993419131130555,
      "learning_rate": 7.246767400188556e-07,
      "loss": 0.9552,
      "step": 10097
    },
    {
      "epoch": 0.7285451462789942,
      "grad_norm": 4.982264535319491,
      "learning_rate": 7.243167598270648e-07,
      "loss": 1.0268,
      "step": 10098
    },
    {
      "epoch": 0.7286172937484218,
      "grad_norm": 4.038301909090003,
      "learning_rate": 7.239568492956134e-07,
      "loss": 0.9504,
      "step": 10099
    },
    {
      "epoch": 0.7286894412178493,
      "grad_norm": 5.963624361178379,
      "learning_rate": 7.235970084441554e-07,
      "loss": 1.0791,
      "step": 10100
    },
    {
      "epoch": 0.7287615886872768,
      "grad_norm": 4.0639084941894925,
      "learning_rate": 7.232372372923381e-07,
      "loss": 0.949,
      "step": 10101
    },
    {
      "epoch": 0.7288337361567043,
      "grad_norm": 5.018873642173949,
      "learning_rate": 7.228775358598105e-07,
      "loss": 1.0405,
      "step": 10102
    },
    {
      "epoch": 0.7289058836261318,
      "grad_norm": 3.81137255898531,
      "learning_rate": 7.225179041662119e-07,
      "loss": 1.0064,
      "step": 10103
    },
    {
      "epoch": 0.7289780310955594,
      "grad_norm": 6.565385747124454,
      "learning_rate": 7.221583422311819e-07,
      "loss": 0.9426,
      "step": 10104
    },
    {
      "epoch": 0.7290501785649869,
      "grad_norm": 3.8290999436125754,
      "learning_rate": 7.217988500743541e-07,
      "loss": 0.9252,
      "step": 10105
    },
    {
      "epoch": 0.7291223260344143,
      "grad_norm": 4.742557517483644,
      "learning_rate": 7.214394277153595e-07,
      "loss": 0.9431,
      "step": 10106
    },
    {
      "epoch": 0.7291944735038418,
      "grad_norm": 3.1968405504505353,
      "learning_rate": 7.210800751738246e-07,
      "loss": 0.8808,
      "step": 10107
    },
    {
      "epoch": 0.7292666209732693,
      "grad_norm": 4.104659126802274,
      "learning_rate": 7.207207924693726e-07,
      "loss": 0.9103,
      "step": 10108
    },
    {
      "epoch": 0.7293387684426968,
      "grad_norm": 4.127378529379183,
      "learning_rate": 7.20361579621623e-07,
      "loss": 0.9949,
      "step": 10109
    },
    {
      "epoch": 0.7294109159121244,
      "grad_norm": 3.74024699634783,
      "learning_rate": 7.200024366501889e-07,
      "loss": 0.8375,
      "step": 10110
    },
    {
      "epoch": 0.7294830633815519,
      "grad_norm": 5.272469529217162,
      "learning_rate": 7.196433635746848e-07,
      "loss": 0.963,
      "step": 10111
    },
    {
      "epoch": 0.7295552108509794,
      "grad_norm": 6.280039200873029,
      "learning_rate": 7.192843604147163e-07,
      "loss": 0.9271,
      "step": 10112
    },
    {
      "epoch": 0.7296273583204069,
      "grad_norm": 3.1143060431060645,
      "learning_rate": 7.189254271898879e-07,
      "loss": 0.9365,
      "step": 10113
    },
    {
      "epoch": 0.7296995057898344,
      "grad_norm": 3.9479249066725655,
      "learning_rate": 7.185665639197996e-07,
      "loss": 0.9918,
      "step": 10114
    },
    {
      "epoch": 0.729771653259262,
      "grad_norm": 3.304105811182023,
      "learning_rate": 7.182077706240473e-07,
      "loss": 0.9566,
      "step": 10115
    },
    {
      "epoch": 0.7298438007286895,
      "grad_norm": 3.8482246714885835,
      "learning_rate": 7.178490473222236e-07,
      "loss": 0.9075,
      "step": 10116
    },
    {
      "epoch": 0.729915948198117,
      "grad_norm": 7.4560221353456075,
      "learning_rate": 7.174903940339178e-07,
      "loss": 0.9488,
      "step": 10117
    },
    {
      "epoch": 0.7299880956675444,
      "grad_norm": 5.92292007444402,
      "learning_rate": 7.171318107787135e-07,
      "loss": 0.8813,
      "step": 10118
    },
    {
      "epoch": 0.7300602431369719,
      "grad_norm": 3.3053507398593527,
      "learning_rate": 7.167732975761915e-07,
      "loss": 0.9686,
      "step": 10119
    },
    {
      "epoch": 0.7301323906063995,
      "grad_norm": 3.6194103159010687,
      "learning_rate": 7.164148544459296e-07,
      "loss": 0.8906,
      "step": 10120
    },
    {
      "epoch": 0.730204538075827,
      "grad_norm": 4.64280264267007,
      "learning_rate": 7.16056481407501e-07,
      "loss": 1.0462,
      "step": 10121
    },
    {
      "epoch": 0.7302766855452545,
      "grad_norm": 0.7902399852061381,
      "learning_rate": 7.156981784804755e-07,
      "loss": 0.8069,
      "step": 10122
    },
    {
      "epoch": 0.730348833014682,
      "grad_norm": 2.8882418319492875,
      "learning_rate": 7.153399456844167e-07,
      "loss": 0.9321,
      "step": 10123
    },
    {
      "epoch": 0.7304209804841095,
      "grad_norm": 3.898429671835828,
      "learning_rate": 7.149817830388893e-07,
      "loss": 0.9264,
      "step": 10124
    },
    {
      "epoch": 0.730493127953537,
      "grad_norm": 3.513591941445912,
      "learning_rate": 7.146236905634484e-07,
      "loss": 0.972,
      "step": 10125
    },
    {
      "epoch": 0.7305652754229646,
      "grad_norm": 4.215424533882025,
      "learning_rate": 7.142656682776509e-07,
      "loss": 0.9837,
      "step": 10126
    },
    {
      "epoch": 0.7306374228923921,
      "grad_norm": 3.0457447009534846,
      "learning_rate": 7.139077162010447e-07,
      "loss": 0.9145,
      "step": 10127
    },
    {
      "epoch": 0.7307095703618196,
      "grad_norm": 2.873346599657392,
      "learning_rate": 7.135498343531774e-07,
      "loss": 0.9539,
      "step": 10128
    },
    {
      "epoch": 0.7307817178312471,
      "grad_norm": 5.413448414511056,
      "learning_rate": 7.131920227535912e-07,
      "loss": 0.9582,
      "step": 10129
    },
    {
      "epoch": 0.7308538653006745,
      "grad_norm": 3.2154131138338578,
      "learning_rate": 7.128342814218251e-07,
      "loss": 0.7905,
      "step": 10130
    },
    {
      "epoch": 0.730926012770102,
      "grad_norm": 7.373266759966239,
      "learning_rate": 7.12476610377414e-07,
      "loss": 0.8914,
      "step": 10131
    },
    {
      "epoch": 0.7309981602395296,
      "grad_norm": 3.939266958835569,
      "learning_rate": 7.12119009639889e-07,
      "loss": 1.0588,
      "step": 10132
    },
    {
      "epoch": 0.7310703077089571,
      "grad_norm": 3.805175499296233,
      "learning_rate": 7.117614792287778e-07,
      "loss": 0.988,
      "step": 10133
    },
    {
      "epoch": 0.7311424551783846,
      "grad_norm": 2.6592617239400025,
      "learning_rate": 7.114040191636028e-07,
      "loss": 0.968,
      "step": 10134
    },
    {
      "epoch": 0.7312146026478121,
      "grad_norm": 25.49905932318264,
      "learning_rate": 7.110466294638842e-07,
      "loss": 0.9728,
      "step": 10135
    },
    {
      "epoch": 0.7312867501172396,
      "grad_norm": 5.596230068993538,
      "learning_rate": 7.106893101491373e-07,
      "loss": 0.9077,
      "step": 10136
    },
    {
      "epoch": 0.7313588975866672,
      "grad_norm": 3.5229089852280717,
      "learning_rate": 7.103320612388746e-07,
      "loss": 1.0852,
      "step": 10137
    },
    {
      "epoch": 0.7314310450560947,
      "grad_norm": 3.4046424081940443,
      "learning_rate": 7.099748827526037e-07,
      "loss": 1.0045,
      "step": 10138
    },
    {
      "epoch": 0.7315031925255222,
      "grad_norm": 4.268296296121483,
      "learning_rate": 7.096177747098298e-07,
      "loss": 0.9944,
      "step": 10139
    },
    {
      "epoch": 0.7315753399949497,
      "grad_norm": 10.405909443916013,
      "learning_rate": 7.092607371300507e-07,
      "loss": 0.9808,
      "step": 10140
    },
    {
      "epoch": 0.7316474874643771,
      "grad_norm": 3.850866041993212,
      "learning_rate": 7.089037700327662e-07,
      "loss": 0.99,
      "step": 10141
    },
    {
      "epoch": 0.7317196349338047,
      "grad_norm": 2.76206232116352,
      "learning_rate": 7.085468734374666e-07,
      "loss": 0.8471,
      "step": 10142
    },
    {
      "epoch": 0.7317917824032322,
      "grad_norm": 0.7622594234976289,
      "learning_rate": 7.081900473636416e-07,
      "loss": 0.816,
      "step": 10143
    },
    {
      "epoch": 0.7318639298726597,
      "grad_norm": 5.780655783968165,
      "learning_rate": 7.078332918307757e-07,
      "loss": 1.0465,
      "step": 10144
    },
    {
      "epoch": 0.7319360773420872,
      "grad_norm": 3.24334284408236,
      "learning_rate": 7.074766068583504e-07,
      "loss": 0.982,
      "step": 10145
    },
    {
      "epoch": 0.7320082248115147,
      "grad_norm": 3.483968257684998,
      "learning_rate": 7.071199924658435e-07,
      "loss": 0.9781,
      "step": 10146
    },
    {
      "epoch": 0.7320803722809422,
      "grad_norm": 4.179514316286516,
      "learning_rate": 7.067634486727265e-07,
      "loss": 0.9731,
      "step": 10147
    },
    {
      "epoch": 0.7321525197503698,
      "grad_norm": 4.974235146161331,
      "learning_rate": 7.064069754984716e-07,
      "loss": 0.9708,
      "step": 10148
    },
    {
      "epoch": 0.7322246672197973,
      "grad_norm": 31.102659488315364,
      "learning_rate": 7.060505729625414e-07,
      "loss": 0.9541,
      "step": 10149
    },
    {
      "epoch": 0.7322968146892248,
      "grad_norm": 2.8522455102933844,
      "learning_rate": 7.056942410844012e-07,
      "loss": 0.9069,
      "step": 10150
    },
    {
      "epoch": 0.7323689621586523,
      "grad_norm": 4.047802910782714,
      "learning_rate": 7.053379798835059e-07,
      "loss": 0.8638,
      "step": 10151
    },
    {
      "epoch": 0.7324411096280798,
      "grad_norm": 2.8123993749631673,
      "learning_rate": 7.04981789379311e-07,
      "loss": 0.9451,
      "step": 10152
    },
    {
      "epoch": 0.7325132570975073,
      "grad_norm": 2.9056722415358434,
      "learning_rate": 7.046256695912669e-07,
      "loss": 0.9496,
      "step": 10153
    },
    {
      "epoch": 0.7325854045669348,
      "grad_norm": 6.253996219976665,
      "learning_rate": 7.042696205388195e-07,
      "loss": 0.972,
      "step": 10154
    },
    {
      "epoch": 0.7326575520363623,
      "grad_norm": 0.6796131751101914,
      "learning_rate": 7.039136422414114e-07,
      "loss": 0.8003,
      "step": 10155
    },
    {
      "epoch": 0.7327296995057898,
      "grad_norm": 4.665607445716411,
      "learning_rate": 7.035577347184821e-07,
      "loss": 0.9135,
      "step": 10156
    },
    {
      "epoch": 0.7328018469752173,
      "grad_norm": 4.235080786861124,
      "learning_rate": 7.032018979894649e-07,
      "loss": 0.8296,
      "step": 10157
    },
    {
      "epoch": 0.7328739944446448,
      "grad_norm": 8.166134979171412,
      "learning_rate": 7.028461320737915e-07,
      "loss": 0.9963,
      "step": 10158
    },
    {
      "epoch": 0.7329461419140724,
      "grad_norm": 5.604643374828481,
      "learning_rate": 7.024904369908891e-07,
      "loss": 0.9406,
      "step": 10159
    },
    {
      "epoch": 0.7330182893834999,
      "grad_norm": 3.4346812395454043,
      "learning_rate": 7.021348127601805e-07,
      "loss": 0.9032,
      "step": 10160
    },
    {
      "epoch": 0.7330904368529274,
      "grad_norm": 3.3605194271589176,
      "learning_rate": 7.017792594010858e-07,
      "loss": 0.939,
      "step": 10161
    },
    {
      "epoch": 0.7331625843223549,
      "grad_norm": 3.361308406437296,
      "learning_rate": 7.014237769330189e-07,
      "loss": 0.9685,
      "step": 10162
    },
    {
      "epoch": 0.7332347317917824,
      "grad_norm": 13.332646415817782,
      "learning_rate": 7.010683653753935e-07,
      "loss": 0.9773,
      "step": 10163
    },
    {
      "epoch": 0.73330687926121,
      "grad_norm": 0.8153781166444005,
      "learning_rate": 7.007130247476149e-07,
      "loss": 0.8051,
      "step": 10164
    },
    {
      "epoch": 0.7333790267306374,
      "grad_norm": 4.845514775851289,
      "learning_rate": 7.003577550690896e-07,
      "loss": 0.9643,
      "step": 10165
    },
    {
      "epoch": 0.7334511742000649,
      "grad_norm": 2.753997065352107,
      "learning_rate": 7.000025563592154e-07,
      "loss": 0.9434,
      "step": 10166
    },
    {
      "epoch": 0.7335233216694924,
      "grad_norm": 0.7613053985382245,
      "learning_rate": 6.996474286373893e-07,
      "loss": 0.8151,
      "step": 10167
    },
    {
      "epoch": 0.7335954691389199,
      "grad_norm": 4.765074250962675,
      "learning_rate": 6.992923719230033e-07,
      "loss": 0.8721,
      "step": 10168
    },
    {
      "epoch": 0.7336676166083475,
      "grad_norm": 3.7766490088305638,
      "learning_rate": 6.989373862354457e-07,
      "loss": 0.941,
      "step": 10169
    },
    {
      "epoch": 0.733739764077775,
      "grad_norm": 5.162203778873346,
      "learning_rate": 6.98582471594102e-07,
      "loss": 0.8653,
      "step": 10170
    },
    {
      "epoch": 0.7338119115472025,
      "grad_norm": 5.660560609262373,
      "learning_rate": 6.982276280183503e-07,
      "loss": 0.9147,
      "step": 10171
    },
    {
      "epoch": 0.73388405901663,
      "grad_norm": 4.2491101848117445,
      "learning_rate": 6.978728555275704e-07,
      "loss": 0.9702,
      "step": 10172
    },
    {
      "epoch": 0.7339562064860575,
      "grad_norm": 4.934252673052761,
      "learning_rate": 6.975181541411328e-07,
      "loss": 1.0002,
      "step": 10173
    },
    {
      "epoch": 0.734028353955485,
      "grad_norm": 3.3345634257052708,
      "learning_rate": 6.97163523878407e-07,
      "loss": 1.0177,
      "step": 10174
    },
    {
      "epoch": 0.7341005014249126,
      "grad_norm": 0.8871770002355839,
      "learning_rate": 6.968089647587583e-07,
      "loss": 0.9566,
      "step": 10175
    },
    {
      "epoch": 0.7341726488943401,
      "grad_norm": 5.092466789515601,
      "learning_rate": 6.964544768015479e-07,
      "loss": 1.0556,
      "step": 10176
    },
    {
      "epoch": 0.7342447963637675,
      "grad_norm": 2.926544375142002,
      "learning_rate": 6.961000600261329e-07,
      "loss": 0.9635,
      "step": 10177
    },
    {
      "epoch": 0.734316943833195,
      "grad_norm": 2.940747028863339,
      "learning_rate": 6.957457144518675e-07,
      "loss": 0.9858,
      "step": 10178
    },
    {
      "epoch": 0.7343890913026225,
      "grad_norm": 3.651464737973792,
      "learning_rate": 6.953914400980989e-07,
      "loss": 0.9627,
      "step": 10179
    },
    {
      "epoch": 0.73446123877205,
      "grad_norm": 3.3545643983312154,
      "learning_rate": 6.950372369841757e-07,
      "loss": 0.9532,
      "step": 10180
    },
    {
      "epoch": 0.7345333862414776,
      "grad_norm": 2.6441304792045246,
      "learning_rate": 6.946831051294376e-07,
      "loss": 1.0279,
      "step": 10181
    },
    {
      "epoch": 0.7346055337109051,
      "grad_norm": 3.826041265637032,
      "learning_rate": 6.943290445532228e-07,
      "loss": 0.9518,
      "step": 10182
    },
    {
      "epoch": 0.7346776811803326,
      "grad_norm": 4.145252794165763,
      "learning_rate": 6.939750552748658e-07,
      "loss": 0.9022,
      "step": 10183
    },
    {
      "epoch": 0.7347498286497601,
      "grad_norm": 3.8667702189643403,
      "learning_rate": 6.936211373136959e-07,
      "loss": 0.9417,
      "step": 10184
    },
    {
      "epoch": 0.7348219761191876,
      "grad_norm": 4.443253990851939,
      "learning_rate": 6.932672906890407e-07,
      "loss": 1.0008,
      "step": 10185
    },
    {
      "epoch": 0.7348941235886152,
      "grad_norm": 2.5763571486755747,
      "learning_rate": 6.929135154202197e-07,
      "loss": 0.9011,
      "step": 10186
    },
    {
      "epoch": 0.7349662710580427,
      "grad_norm": 5.538517891564553,
      "learning_rate": 6.925598115265545e-07,
      "loss": 0.8585,
      "step": 10187
    },
    {
      "epoch": 0.7350384185274701,
      "grad_norm": 6.113985083285108,
      "learning_rate": 6.922061790273566e-07,
      "loss": 0.9385,
      "step": 10188
    },
    {
      "epoch": 0.7351105659968976,
      "grad_norm": 4.941860735954559,
      "learning_rate": 6.918526179419395e-07,
      "loss": 0.9247,
      "step": 10189
    },
    {
      "epoch": 0.7351827134663251,
      "grad_norm": 6.882795872922595,
      "learning_rate": 6.914991282896074e-07,
      "loss": 0.9034,
      "step": 10190
    },
    {
      "epoch": 0.7352548609357527,
      "grad_norm": 4.353595799648694,
      "learning_rate": 6.911457100896641e-07,
      "loss": 1.0484,
      "step": 10191
    },
    {
      "epoch": 0.7353270084051802,
      "grad_norm": 3.2626084992940125,
      "learning_rate": 6.907923633614084e-07,
      "loss": 0.9961,
      "step": 10192
    },
    {
      "epoch": 0.7353991558746077,
      "grad_norm": 5.327495638462038,
      "learning_rate": 6.904390881241351e-07,
      "loss": 1.0073,
      "step": 10193
    },
    {
      "epoch": 0.7354713033440352,
      "grad_norm": 12.54318407242773,
      "learning_rate": 6.900858843971362e-07,
      "loss": 1.0132,
      "step": 10194
    },
    {
      "epoch": 0.7355434508134627,
      "grad_norm": 3.3073677755094044,
      "learning_rate": 6.897327521996972e-07,
      "loss": 1.0215,
      "step": 10195
    },
    {
      "epoch": 0.7356155982828902,
      "grad_norm": 3.731922190068685,
      "learning_rate": 6.893796915511023e-07,
      "loss": 0.9718,
      "step": 10196
    },
    {
      "epoch": 0.7356877457523178,
      "grad_norm": 2.3971802836620215,
      "learning_rate": 6.890267024706306e-07,
      "loss": 0.8986,
      "step": 10197
    },
    {
      "epoch": 0.7357598932217453,
      "grad_norm": 2.8180568477184162,
      "learning_rate": 6.886737849775577e-07,
      "loss": 0.9964,
      "step": 10198
    },
    {
      "epoch": 0.7358320406911728,
      "grad_norm": 4.200510112120475,
      "learning_rate": 6.883209390911548e-07,
      "loss": 0.9863,
      "step": 10199
    },
    {
      "epoch": 0.7359041881606002,
      "grad_norm": 4.920336286015466,
      "learning_rate": 6.879681648306908e-07,
      "loss": 0.9791,
      "step": 10200
    },
    {
      "epoch": 0.7359763356300277,
      "grad_norm": 4.054152378058403,
      "learning_rate": 6.876154622154267e-07,
      "loss": 0.9079,
      "step": 10201
    },
    {
      "epoch": 0.7360484830994553,
      "grad_norm": 5.252449917234494,
      "learning_rate": 6.872628312646254e-07,
      "loss": 1.0452,
      "step": 10202
    },
    {
      "epoch": 0.7361206305688828,
      "grad_norm": 3.25297190427799,
      "learning_rate": 6.869102719975399e-07,
      "loss": 1.0181,
      "step": 10203
    },
    {
      "epoch": 0.7361927780383103,
      "grad_norm": 4.407763498514625,
      "learning_rate": 6.865577844334249e-07,
      "loss": 0.9543,
      "step": 10204
    },
    {
      "epoch": 0.7362649255077378,
      "grad_norm": 5.489757016334027,
      "learning_rate": 6.862053685915266e-07,
      "loss": 0.9379,
      "step": 10205
    },
    {
      "epoch": 0.7363370729771653,
      "grad_norm": 0.7534438380350981,
      "learning_rate": 6.858530244910894e-07,
      "loss": 0.8067,
      "step": 10206
    },
    {
      "epoch": 0.7364092204465928,
      "grad_norm": 4.477352159232964,
      "learning_rate": 6.855007521513538e-07,
      "loss": 0.9038,
      "step": 10207
    },
    {
      "epoch": 0.7364813679160204,
      "grad_norm": 4.6762548221174765,
      "learning_rate": 6.85148551591556e-07,
      "loss": 1.1062,
      "step": 10208
    },
    {
      "epoch": 0.7365535153854479,
      "grad_norm": 3.5087590289297017,
      "learning_rate": 6.847964228309291e-07,
      "loss": 0.8863,
      "step": 10209
    },
    {
      "epoch": 0.7366256628548754,
      "grad_norm": 6.606105708312162,
      "learning_rate": 6.844443658886994e-07,
      "loss": 0.939,
      "step": 10210
    },
    {
      "epoch": 0.7366978103243029,
      "grad_norm": 4.460601847458786,
      "learning_rate": 6.840923807840944e-07,
      "loss": 0.9142,
      "step": 10211
    },
    {
      "epoch": 0.7367699577937303,
      "grad_norm": 7.158153563948482,
      "learning_rate": 6.837404675363323e-07,
      "loss": 1.0175,
      "step": 10212
    },
    {
      "epoch": 0.7368421052631579,
      "grad_norm": 4.252509105746631,
      "learning_rate": 6.833886261646306e-07,
      "loss": 0.9915,
      "step": 10213
    },
    {
      "epoch": 0.7369142527325854,
      "grad_norm": 5.107350359501701,
      "learning_rate": 6.830368566882021e-07,
      "loss": 0.9894,
      "step": 10214
    },
    {
      "epoch": 0.7369864002020129,
      "grad_norm": 4.879774054627811,
      "learning_rate": 6.826851591262559e-07,
      "loss": 0.9405,
      "step": 10215
    },
    {
      "epoch": 0.7370585476714404,
      "grad_norm": 3.8583019170175534,
      "learning_rate": 6.823335334979963e-07,
      "loss": 0.8735,
      "step": 10216
    },
    {
      "epoch": 0.7371306951408679,
      "grad_norm": 7.2643942169904125,
      "learning_rate": 6.819819798226254e-07,
      "loss": 0.9895,
      "step": 10217
    },
    {
      "epoch": 0.7372028426102955,
      "grad_norm": 7.121402836305049,
      "learning_rate": 6.816304981193381e-07,
      "loss": 1.0617,
      "step": 10218
    },
    {
      "epoch": 0.737274990079723,
      "grad_norm": 3.2574682259638483,
      "learning_rate": 6.812790884073303e-07,
      "loss": 0.922,
      "step": 10219
    },
    {
      "epoch": 0.7373471375491505,
      "grad_norm": 4.936845373868154,
      "learning_rate": 6.809277507057889e-07,
      "loss": 0.9506,
      "step": 10220
    },
    {
      "epoch": 0.737419285018578,
      "grad_norm": 2.814192283979927,
      "learning_rate": 6.805764850339002e-07,
      "loss": 0.9048,
      "step": 10221
    },
    {
      "epoch": 0.7374914324880055,
      "grad_norm": 6.118493185605806,
      "learning_rate": 6.802252914108453e-07,
      "loss": 0.9776,
      "step": 10222
    },
    {
      "epoch": 0.737563579957433,
      "grad_norm": 3.109211519630614,
      "learning_rate": 6.798741698558015e-07,
      "loss": 0.904,
      "step": 10223
    },
    {
      "epoch": 0.7376357274268605,
      "grad_norm": 4.453680920617886,
      "learning_rate": 6.795231203879428e-07,
      "loss": 0.9926,
      "step": 10224
    },
    {
      "epoch": 0.737707874896288,
      "grad_norm": 2.8066966901367887,
      "learning_rate": 6.791721430264372e-07,
      "loss": 0.9584,
      "step": 10225
    },
    {
      "epoch": 0.7377800223657155,
      "grad_norm": 5.2057781614739875,
      "learning_rate": 6.788212377904528e-07,
      "loss": 0.9133,
      "step": 10226
    },
    {
      "epoch": 0.737852169835143,
      "grad_norm": 3.610526293487806,
      "learning_rate": 6.784704046991488e-07,
      "loss": 0.9696,
      "step": 10227
    },
    {
      "epoch": 0.7379243173045705,
      "grad_norm": 3.5888446002690575,
      "learning_rate": 6.781196437716839e-07,
      "loss": 0.9573,
      "step": 10228
    },
    {
      "epoch": 0.737996464773998,
      "grad_norm": 8.37808808286929,
      "learning_rate": 6.777689550272119e-07,
      "loss": 0.7548,
      "step": 10229
    },
    {
      "epoch": 0.7380686122434256,
      "grad_norm": 3.2005916346557886,
      "learning_rate": 6.774183384848826e-07,
      "loss": 0.9178,
      "step": 10230
    },
    {
      "epoch": 0.7381407597128531,
      "grad_norm": 4.0359330292845685,
      "learning_rate": 6.770677941638416e-07,
      "loss": 1.0229,
      "step": 10231
    },
    {
      "epoch": 0.7382129071822806,
      "grad_norm": 3.262329044466937,
      "learning_rate": 6.767173220832312e-07,
      "loss": 0.9241,
      "step": 10232
    },
    {
      "epoch": 0.7382850546517081,
      "grad_norm": 3.8268804278686446,
      "learning_rate": 6.763669222621899e-07,
      "loss": 0.9835,
      "step": 10233
    },
    {
      "epoch": 0.7383572021211356,
      "grad_norm": 4.627091836842786,
      "learning_rate": 6.760165947198501e-07,
      "loss": 0.8458,
      "step": 10234
    },
    {
      "epoch": 0.7384293495905631,
      "grad_norm": 2.4974203151633927,
      "learning_rate": 6.756663394753433e-07,
      "loss": 0.9875,
      "step": 10235
    },
    {
      "epoch": 0.7385014970599906,
      "grad_norm": 0.6201495545117858,
      "learning_rate": 6.753161565477949e-07,
      "loss": 0.79,
      "step": 10236
    },
    {
      "epoch": 0.7385736445294181,
      "grad_norm": 2.8601530090163623,
      "learning_rate": 6.749660459563276e-07,
      "loss": 0.9368,
      "step": 10237
    },
    {
      "epoch": 0.7386457919988456,
      "grad_norm": 2.7389199985270287,
      "learning_rate": 6.746160077200591e-07,
      "loss": 1.0064,
      "step": 10238
    },
    {
      "epoch": 0.7387179394682731,
      "grad_norm": 4.002754216886266,
      "learning_rate": 6.742660418581047e-07,
      "loss": 0.9221,
      "step": 10239
    },
    {
      "epoch": 0.7387900869377007,
      "grad_norm": 2.984222887566507,
      "learning_rate": 6.73916148389573e-07,
      "loss": 0.932,
      "step": 10240
    },
    {
      "epoch": 0.7388622344071282,
      "grad_norm": 0.7227545181682106,
      "learning_rate": 6.735663273335725e-07,
      "loss": 0.8157,
      "step": 10241
    },
    {
      "epoch": 0.7389343818765557,
      "grad_norm": 3.9135711422418504,
      "learning_rate": 6.732165787092035e-07,
      "loss": 0.952,
      "step": 10242
    },
    {
      "epoch": 0.7390065293459832,
      "grad_norm": 2.8982632731180886,
      "learning_rate": 6.728669025355669e-07,
      "loss": 0.8536,
      "step": 10243
    },
    {
      "epoch": 0.7390786768154107,
      "grad_norm": 5.9552507585054455,
      "learning_rate": 6.725172988317549e-07,
      "loss": 1.0069,
      "step": 10244
    },
    {
      "epoch": 0.7391508242848382,
      "grad_norm": 4.5413369341970595,
      "learning_rate": 6.721677676168593e-07,
      "loss": 1.0091,
      "step": 10245
    },
    {
      "epoch": 0.7392229717542658,
      "grad_norm": 3.0807466912221253,
      "learning_rate": 6.718183089099662e-07,
      "loss": 0.9728,
      "step": 10246
    },
    {
      "epoch": 0.7392951192236932,
      "grad_norm": 7.746714972489322,
      "learning_rate": 6.714689227301589e-07,
      "loss": 0.798,
      "step": 10247
    },
    {
      "epoch": 0.7393672666931207,
      "grad_norm": 4.318032737604297,
      "learning_rate": 6.711196090965159e-07,
      "loss": 0.8794,
      "step": 10248
    },
    {
      "epoch": 0.7394394141625482,
      "grad_norm": 4.335227894496066,
      "learning_rate": 6.707703680281114e-07,
      "loss": 0.9258,
      "step": 10249
    },
    {
      "epoch": 0.7395115616319757,
      "grad_norm": 3.1863163264941083,
      "learning_rate": 6.704211995440161e-07,
      "loss": 0.9933,
      "step": 10250
    },
    {
      "epoch": 0.7395837091014033,
      "grad_norm": 3.476251412521344,
      "learning_rate": 6.700721036632975e-07,
      "loss": 0.9062,
      "step": 10251
    },
    {
      "epoch": 0.7396558565708308,
      "grad_norm": 2.7655615502961712,
      "learning_rate": 6.697230804050178e-07,
      "loss": 0.8695,
      "step": 10252
    },
    {
      "epoch": 0.7397280040402583,
      "grad_norm": 3.092329219213245,
      "learning_rate": 6.693741297882364e-07,
      "loss": 0.9854,
      "step": 10253
    },
    {
      "epoch": 0.7398001515096858,
      "grad_norm": 4.3292012443055,
      "learning_rate": 6.690252518320078e-07,
      "loss": 0.9852,
      "step": 10254
    },
    {
      "epoch": 0.7398722989791133,
      "grad_norm": 5.499546032290238,
      "learning_rate": 6.686764465553831e-07,
      "loss": 0.9516,
      "step": 10255
    },
    {
      "epoch": 0.7399444464485408,
      "grad_norm": 4.038627084545854,
      "learning_rate": 6.683277139774101e-07,
      "loss": 0.9995,
      "step": 10256
    },
    {
      "epoch": 0.7400165939179684,
      "grad_norm": 3.2690458495716395,
      "learning_rate": 6.679790541171302e-07,
      "loss": 0.9724,
      "step": 10257
    },
    {
      "epoch": 0.7400887413873959,
      "grad_norm": 7.282303238687955,
      "learning_rate": 6.67630466993583e-07,
      "loss": 0.9656,
      "step": 10258
    },
    {
      "epoch": 0.7401608888568233,
      "grad_norm": 5.017640086348642,
      "learning_rate": 6.67281952625804e-07,
      "loss": 0.8522,
      "step": 10259
    },
    {
      "epoch": 0.7402330363262508,
      "grad_norm": 5.655713134915271,
      "learning_rate": 6.66933511032824e-07,
      "loss": 0.9083,
      "step": 10260
    },
    {
      "epoch": 0.7403051837956783,
      "grad_norm": 5.292759920061663,
      "learning_rate": 6.665851422336699e-07,
      "loss": 0.9773,
      "step": 10261
    },
    {
      "epoch": 0.7403773312651059,
      "grad_norm": 3.0452087541252686,
      "learning_rate": 6.662368462473653e-07,
      "loss": 0.9305,
      "step": 10262
    },
    {
      "epoch": 0.7404494787345334,
      "grad_norm": 4.338818941571457,
      "learning_rate": 6.658886230929298e-07,
      "loss": 0.8917,
      "step": 10263
    },
    {
      "epoch": 0.7405216262039609,
      "grad_norm": 3.1668831767971732,
      "learning_rate": 6.655404727893763e-07,
      "loss": 0.9759,
      "step": 10264
    },
    {
      "epoch": 0.7405937736733884,
      "grad_norm": 3.581930211097141,
      "learning_rate": 6.65192395355719e-07,
      "loss": 0.9103,
      "step": 10265
    },
    {
      "epoch": 0.7406659211428159,
      "grad_norm": 3.7857931288248983,
      "learning_rate": 6.648443908109631e-07,
      "loss": 0.9443,
      "step": 10266
    },
    {
      "epoch": 0.7407380686122434,
      "grad_norm": 7.100080580992852,
      "learning_rate": 6.644964591741122e-07,
      "loss": 0.961,
      "step": 10267
    },
    {
      "epoch": 0.740810216081671,
      "grad_norm": 2.9355337176913223,
      "learning_rate": 6.641486004641661e-07,
      "loss": 0.9109,
      "step": 10268
    },
    {
      "epoch": 0.7408823635510985,
      "grad_norm": 2.8868799642060585,
      "learning_rate": 6.638008147001197e-07,
      "loss": 0.9218,
      "step": 10269
    },
    {
      "epoch": 0.740954511020526,
      "grad_norm": 5.2957072813497925,
      "learning_rate": 6.634531019009644e-07,
      "loss": 0.9284,
      "step": 10270
    },
    {
      "epoch": 0.7410266584899534,
      "grad_norm": 3.599917352575408,
      "learning_rate": 6.631054620856879e-07,
      "loss": 0.9719,
      "step": 10271
    },
    {
      "epoch": 0.7410988059593809,
      "grad_norm": 0.6671768982735081,
      "learning_rate": 6.627578952732726e-07,
      "loss": 0.7936,
      "step": 10272
    },
    {
      "epoch": 0.7411709534288085,
      "grad_norm": 3.2407989837821924,
      "learning_rate": 6.624104014826986e-07,
      "loss": 0.9983,
      "step": 10273
    },
    {
      "epoch": 0.741243100898236,
      "grad_norm": 5.901781117190353,
      "learning_rate": 6.620629807329405e-07,
      "loss": 0.9971,
      "step": 10274
    },
    {
      "epoch": 0.7413152483676635,
      "grad_norm": 0.7143738666702318,
      "learning_rate": 6.617156330429705e-07,
      "loss": 0.8362,
      "step": 10275
    },
    {
      "epoch": 0.741387395837091,
      "grad_norm": 0.6950831945971262,
      "learning_rate": 6.613683584317556e-07,
      "loss": 0.8543,
      "step": 10276
    },
    {
      "epoch": 0.7414595433065185,
      "grad_norm": 0.7770051841882013,
      "learning_rate": 6.610211569182591e-07,
      "loss": 0.8324,
      "step": 10277
    },
    {
      "epoch": 0.741531690775946,
      "grad_norm": 3.57248359628359,
      "learning_rate": 6.606740285214412e-07,
      "loss": 0.9608,
      "step": 10278
    },
    {
      "epoch": 0.7416038382453736,
      "grad_norm": 4.0215287208427934,
      "learning_rate": 6.60326973260255e-07,
      "loss": 0.9482,
      "step": 10279
    },
    {
      "epoch": 0.7416759857148011,
      "grad_norm": 2.871666509886959,
      "learning_rate": 6.599799911536552e-07,
      "loss": 0.9844,
      "step": 10280
    },
    {
      "epoch": 0.7417481331842286,
      "grad_norm": 0.6589644469191542,
      "learning_rate": 6.596330822205867e-07,
      "loss": 0.8336,
      "step": 10281
    },
    {
      "epoch": 0.7418202806536561,
      "grad_norm": 0.6717820214098847,
      "learning_rate": 6.592862464799935e-07,
      "loss": 0.8311,
      "step": 10282
    },
    {
      "epoch": 0.7418924281230835,
      "grad_norm": 3.840211930188516,
      "learning_rate": 6.589394839508156e-07,
      "loss": 1.0092,
      "step": 10283
    },
    {
      "epoch": 0.741964575592511,
      "grad_norm": 4.027652288360133,
      "learning_rate": 6.585927946519878e-07,
      "loss": 0.9404,
      "step": 10284
    },
    {
      "epoch": 0.7420367230619386,
      "grad_norm": 25.55757851080453,
      "learning_rate": 6.582461786024418e-07,
      "loss": 0.9043,
      "step": 10285
    },
    {
      "epoch": 0.7421088705313661,
      "grad_norm": 5.092605555867865,
      "learning_rate": 6.578996358211051e-07,
      "loss": 0.9945,
      "step": 10286
    },
    {
      "epoch": 0.7421810180007936,
      "grad_norm": 3.425716943256898,
      "learning_rate": 6.575531663269016e-07,
      "loss": 0.9061,
      "step": 10287
    },
    {
      "epoch": 0.7422531654702211,
      "grad_norm": 6.816239476828022,
      "learning_rate": 6.572067701387493e-07,
      "loss": 0.9764,
      "step": 10288
    },
    {
      "epoch": 0.7423253129396487,
      "grad_norm": 8.406587615199834,
      "learning_rate": 6.568604472755646e-07,
      "loss": 0.9557,
      "step": 10289
    },
    {
      "epoch": 0.7423974604090762,
      "grad_norm": 3.515754527778805,
      "learning_rate": 6.565141977562587e-07,
      "loss": 0.9436,
      "step": 10290
    },
    {
      "epoch": 0.7424696078785037,
      "grad_norm": 0.7350535808547888,
      "learning_rate": 6.561680215997391e-07,
      "loss": 0.8601,
      "step": 10291
    },
    {
      "epoch": 0.7425417553479312,
      "grad_norm": 3.182285756273707,
      "learning_rate": 6.55821918824909e-07,
      "loss": 0.89,
      "step": 10292
    },
    {
      "epoch": 0.7426139028173587,
      "grad_norm": 3.955281027781468,
      "learning_rate": 6.554758894506689e-07,
      "loss": 0.9603,
      "step": 10293
    },
    {
      "epoch": 0.7426860502867861,
      "grad_norm": 8.612020705662848,
      "learning_rate": 6.551299334959115e-07,
      "loss": 0.9882,
      "step": 10294
    },
    {
      "epoch": 0.7427581977562137,
      "grad_norm": 2.760857130693266,
      "learning_rate": 6.547840509795315e-07,
      "loss": 0.9111,
      "step": 10295
    },
    {
      "epoch": 0.7428303452256412,
      "grad_norm": 4.474099112247172,
      "learning_rate": 6.54438241920414e-07,
      "loss": 1.0347,
      "step": 10296
    },
    {
      "epoch": 0.7429024926950687,
      "grad_norm": 3.05474884674053,
      "learning_rate": 6.540925063374427e-07,
      "loss": 0.9592,
      "step": 10297
    },
    {
      "epoch": 0.7429746401644962,
      "grad_norm": 3.853898583788683,
      "learning_rate": 6.537468442494976e-07,
      "loss": 0.9019,
      "step": 10298
    },
    {
      "epoch": 0.7430467876339237,
      "grad_norm": 7.417496138347653,
      "learning_rate": 6.534012556754536e-07,
      "loss": 1.0124,
      "step": 10299
    },
    {
      "epoch": 0.7431189351033513,
      "grad_norm": 3.1499751195984635,
      "learning_rate": 6.530557406341819e-07,
      "loss": 0.9976,
      "step": 10300
    },
    {
      "epoch": 0.7431910825727788,
      "grad_norm": 4.038144625291485,
      "learning_rate": 6.527102991445501e-07,
      "loss": 0.9648,
      "step": 10301
    },
    {
      "epoch": 0.7432632300422063,
      "grad_norm": 4.352855113714585,
      "learning_rate": 6.52364931225422e-07,
      "loss": 0.9343,
      "step": 10302
    },
    {
      "epoch": 0.7433353775116338,
      "grad_norm": 4.998242832410523,
      "learning_rate": 6.52019636895655e-07,
      "loss": 0.9312,
      "step": 10303
    },
    {
      "epoch": 0.7434075249810613,
      "grad_norm": 4.5340075290720065,
      "learning_rate": 6.51674416174107e-07,
      "loss": 0.962,
      "step": 10304
    },
    {
      "epoch": 0.7434796724504888,
      "grad_norm": 5.131259909839795,
      "learning_rate": 6.513292690796268e-07,
      "loss": 0.9404,
      "step": 10305
    },
    {
      "epoch": 0.7435518199199163,
      "grad_norm": 5.2330103675335735,
      "learning_rate": 6.509841956310629e-07,
      "loss": 0.8427,
      "step": 10306
    },
    {
      "epoch": 0.7436239673893438,
      "grad_norm": 3.785617292452066,
      "learning_rate": 6.506391958472581e-07,
      "loss": 0.8324,
      "step": 10307
    },
    {
      "epoch": 0.7436961148587713,
      "grad_norm": 17.51595342159764,
      "learning_rate": 6.502942697470517e-07,
      "loss": 0.8315,
      "step": 10308
    },
    {
      "epoch": 0.7437682623281988,
      "grad_norm": 3.3238312868620374,
      "learning_rate": 6.499494173492788e-07,
      "loss": 0.9324,
      "step": 10309
    },
    {
      "epoch": 0.7438404097976263,
      "grad_norm": 3.246068190050543,
      "learning_rate": 6.496046386727709e-07,
      "loss": 0.8182,
      "step": 10310
    },
    {
      "epoch": 0.7439125572670539,
      "grad_norm": 7.85470144810528,
      "learning_rate": 6.492599337363542e-07,
      "loss": 1.0319,
      "step": 10311
    },
    {
      "epoch": 0.7439847047364814,
      "grad_norm": 2.532401305795784,
      "learning_rate": 6.489153025588525e-07,
      "loss": 0.9541,
      "step": 10312
    },
    {
      "epoch": 0.7440568522059089,
      "grad_norm": 4.015743743684525,
      "learning_rate": 6.485707451590843e-07,
      "loss": 0.8939,
      "step": 10313
    },
    {
      "epoch": 0.7441289996753364,
      "grad_norm": 3.920061398823293,
      "learning_rate": 6.482262615558649e-07,
      "loss": 0.8942,
      "step": 10314
    },
    {
      "epoch": 0.7442011471447639,
      "grad_norm": 2.9638475596549427,
      "learning_rate": 6.478818517680054e-07,
      "loss": 0.9804,
      "step": 10315
    },
    {
      "epoch": 0.7442732946141914,
      "grad_norm": 4.643580051356859,
      "learning_rate": 6.475375158143124e-07,
      "loss": 0.9372,
      "step": 10316
    },
    {
      "epoch": 0.744345442083619,
      "grad_norm": 9.310857640804322,
      "learning_rate": 6.471932537135898e-07,
      "loss": 0.8315,
      "step": 10317
    },
    {
      "epoch": 0.7444175895530464,
      "grad_norm": 5.568840442269823,
      "learning_rate": 6.468490654846343e-07,
      "loss": 0.8767,
      "step": 10318
    },
    {
      "epoch": 0.7444897370224739,
      "grad_norm": 3.5282398999191984,
      "learning_rate": 6.465049511462437e-07,
      "loss": 1.0544,
      "step": 10319
    },
    {
      "epoch": 0.7445618844919014,
      "grad_norm": 0.8155644653773724,
      "learning_rate": 6.461609107172063e-07,
      "loss": 0.8307,
      "step": 10320
    },
    {
      "epoch": 0.7446340319613289,
      "grad_norm": 4.885388773477253,
      "learning_rate": 6.4581694421631e-07,
      "loss": 0.9917,
      "step": 10321
    },
    {
      "epoch": 0.7447061794307565,
      "grad_norm": 6.37982115109208,
      "learning_rate": 6.454730516623373e-07,
      "loss": 0.9306,
      "step": 10322
    },
    {
      "epoch": 0.744778326900184,
      "grad_norm": 4.8510885452496675,
      "learning_rate": 6.451292330740669e-07,
      "loss": 0.8372,
      "step": 10323
    },
    {
      "epoch": 0.7448504743696115,
      "grad_norm": 4.830371321785028,
      "learning_rate": 6.447854884702737e-07,
      "loss": 0.9155,
      "step": 10324
    },
    {
      "epoch": 0.744922621839039,
      "grad_norm": 2.6842334323630057,
      "learning_rate": 6.444418178697279e-07,
      "loss": 1.0133,
      "step": 10325
    },
    {
      "epoch": 0.7449947693084665,
      "grad_norm": 5.208024933903274,
      "learning_rate": 6.440982212911971e-07,
      "loss": 1.0149,
      "step": 10326
    },
    {
      "epoch": 0.745066916777894,
      "grad_norm": 2.2017963271713428,
      "learning_rate": 6.437546987534426e-07,
      "loss": 1.0973,
      "step": 10327
    },
    {
      "epoch": 0.7451390642473216,
      "grad_norm": 0.7486177263299848,
      "learning_rate": 6.434112502752231e-07,
      "loss": 0.8657,
      "step": 10328
    },
    {
      "epoch": 0.7452112117167491,
      "grad_norm": 8.074475288468102,
      "learning_rate": 6.430678758752934e-07,
      "loss": 0.8995,
      "step": 10329
    },
    {
      "epoch": 0.7452833591861765,
      "grad_norm": 5.225023348085356,
      "learning_rate": 6.427245755724038e-07,
      "loss": 1.1147,
      "step": 10330
    },
    {
      "epoch": 0.745355506655604,
      "grad_norm": 2.7232449085380956,
      "learning_rate": 6.423813493853006e-07,
      "loss": 1.0264,
      "step": 10331
    },
    {
      "epoch": 0.7454276541250315,
      "grad_norm": 3.8781540862659503,
      "learning_rate": 6.42038197332727e-07,
      "loss": 0.9794,
      "step": 10332
    },
    {
      "epoch": 0.745499801594459,
      "grad_norm": 4.655849362745128,
      "learning_rate": 6.416951194334188e-07,
      "loss": 0.9425,
      "step": 10333
    },
    {
      "epoch": 0.7455719490638866,
      "grad_norm": 4.542772677545456,
      "learning_rate": 6.413521157061135e-07,
      "loss": 0.9175,
      "step": 10334
    },
    {
      "epoch": 0.7456440965333141,
      "grad_norm": 2.5326407108963367,
      "learning_rate": 6.410091861695388e-07,
      "loss": 0.8837,
      "step": 10335
    },
    {
      "epoch": 0.7457162440027416,
      "grad_norm": 5.900134599087847,
      "learning_rate": 6.406663308424216e-07,
      "loss": 1.0079,
      "step": 10336
    },
    {
      "epoch": 0.7457883914721691,
      "grad_norm": 6.184711791395534,
      "learning_rate": 6.40323549743484e-07,
      "loss": 0.9658,
      "step": 10337
    },
    {
      "epoch": 0.7458605389415967,
      "grad_norm": 3.9351006115237945,
      "learning_rate": 6.399808428914441e-07,
      "loss": 0.923,
      "step": 10338
    },
    {
      "epoch": 0.7459326864110242,
      "grad_norm": 3.9953910978043266,
      "learning_rate": 6.396382103050157e-07,
      "loss": 0.9424,
      "step": 10339
    },
    {
      "epoch": 0.7460048338804517,
      "grad_norm": 5.160854249075457,
      "learning_rate": 6.392956520029089e-07,
      "loss": 0.9473,
      "step": 10340
    },
    {
      "epoch": 0.7460769813498791,
      "grad_norm": 4.58928728342162,
      "learning_rate": 6.389531680038299e-07,
      "loss": 0.9869,
      "step": 10341
    },
    {
      "epoch": 0.7461491288193066,
      "grad_norm": 4.422462451316097,
      "learning_rate": 6.386107583264785e-07,
      "loss": 0.8955,
      "step": 10342
    },
    {
      "epoch": 0.7462212762887341,
      "grad_norm": 2.7655620675550767,
      "learning_rate": 6.382684229895555e-07,
      "loss": 0.9479,
      "step": 10343
    },
    {
      "epoch": 0.7462934237581617,
      "grad_norm": 3.617624342968852,
      "learning_rate": 6.379261620117522e-07,
      "loss": 0.9202,
      "step": 10344
    },
    {
      "epoch": 0.7463655712275892,
      "grad_norm": 3.240286616846412,
      "learning_rate": 6.375839754117589e-07,
      "loss": 0.953,
      "step": 10345
    },
    {
      "epoch": 0.7464377186970167,
      "grad_norm": 3.0877931834028716,
      "learning_rate": 6.372418632082612e-07,
      "loss": 0.9169,
      "step": 10346
    },
    {
      "epoch": 0.7465098661664442,
      "grad_norm": 3.4526241163612634,
      "learning_rate": 6.368998254199407e-07,
      "loss": 0.9494,
      "step": 10347
    },
    {
      "epoch": 0.7465820136358717,
      "grad_norm": 6.016644914047073,
      "learning_rate": 6.365578620654746e-07,
      "loss": 0.9016,
      "step": 10348
    },
    {
      "epoch": 0.7466541611052993,
      "grad_norm": 3.104415849417291,
      "learning_rate": 6.36215973163537e-07,
      "loss": 0.914,
      "step": 10349
    },
    {
      "epoch": 0.7467263085747268,
      "grad_norm": 3.7642490516504834,
      "learning_rate": 6.35874158732796e-07,
      "loss": 0.8933,
      "step": 10350
    },
    {
      "epoch": 0.7467984560441543,
      "grad_norm": 8.29152194971923,
      "learning_rate": 6.355324187919173e-07,
      "loss": 0.9671,
      "step": 10351
    },
    {
      "epoch": 0.7468706035135818,
      "grad_norm": 4.551419742963259,
      "learning_rate": 6.351907533595624e-07,
      "loss": 0.8244,
      "step": 10352
    },
    {
      "epoch": 0.7469427509830092,
      "grad_norm": 3.536074827783358,
      "learning_rate": 6.348491624543877e-07,
      "loss": 0.9825,
      "step": 10353
    },
    {
      "epoch": 0.7470148984524367,
      "grad_norm": 3.3450792469165593,
      "learning_rate": 6.34507646095047e-07,
      "loss": 0.9025,
      "step": 10354
    },
    {
      "epoch": 0.7470870459218643,
      "grad_norm": 6.481946375487634,
      "learning_rate": 6.341662043001885e-07,
      "loss": 0.959,
      "step": 10355
    },
    {
      "epoch": 0.7471591933912918,
      "grad_norm": 5.1447362721343115,
      "learning_rate": 6.338248370884583e-07,
      "loss": 0.8941,
      "step": 10356
    },
    {
      "epoch": 0.7472313408607193,
      "grad_norm": 0.7062273038745062,
      "learning_rate": 6.334835444784949e-07,
      "loss": 0.7739,
      "step": 10357
    },
    {
      "epoch": 0.7473034883301468,
      "grad_norm": 3.6576504103943837,
      "learning_rate": 6.33142326488938e-07,
      "loss": 0.9394,
      "step": 10358
    },
    {
      "epoch": 0.7473756357995743,
      "grad_norm": 5.238919690714569,
      "learning_rate": 6.32801183138418e-07,
      "loss": 0.9452,
      "step": 10359
    },
    {
      "epoch": 0.7474477832690019,
      "grad_norm": 6.898954516830283,
      "learning_rate": 6.324601144455639e-07,
      "loss": 1.0018,
      "step": 10360
    },
    {
      "epoch": 0.7475199307384294,
      "grad_norm": 3.010599326605732,
      "learning_rate": 6.321191204290009e-07,
      "loss": 1.0076,
      "step": 10361
    },
    {
      "epoch": 0.7475920782078569,
      "grad_norm": 5.3365806985030915,
      "learning_rate": 6.317782011073487e-07,
      "loss": 0.9853,
      "step": 10362
    },
    {
      "epoch": 0.7476642256772844,
      "grad_norm": 4.320678695417896,
      "learning_rate": 6.314373564992242e-07,
      "loss": 1.0076,
      "step": 10363
    },
    {
      "epoch": 0.7477363731467119,
      "grad_norm": 2.727739839577039,
      "learning_rate": 6.310965866232392e-07,
      "loss": 0.9329,
      "step": 10364
    },
    {
      "epoch": 0.7478085206161393,
      "grad_norm": 3.653046076356871,
      "learning_rate": 6.307558914980029e-07,
      "loss": 0.8682,
      "step": 10365
    },
    {
      "epoch": 0.7478806680855669,
      "grad_norm": 3.3936829730698332,
      "learning_rate": 6.304152711421178e-07,
      "loss": 0.8688,
      "step": 10366
    },
    {
      "epoch": 0.7479528155549944,
      "grad_norm": 8.885624074061814,
      "learning_rate": 6.300747255741847e-07,
      "loss": 0.8845,
      "step": 10367
    },
    {
      "epoch": 0.7480249630244219,
      "grad_norm": 3.881785450733008,
      "learning_rate": 6.297342548127997e-07,
      "loss": 0.9328,
      "step": 10368
    },
    {
      "epoch": 0.7480971104938494,
      "grad_norm": 2.9923221567795157,
      "learning_rate": 6.293938588765544e-07,
      "loss": 0.8814,
      "step": 10369
    },
    {
      "epoch": 0.7481692579632769,
      "grad_norm": 4.137268548582606,
      "learning_rate": 6.290535377840367e-07,
      "loss": 0.8739,
      "step": 10370
    },
    {
      "epoch": 0.7482414054327045,
      "grad_norm": 12.566576244187457,
      "learning_rate": 6.28713291553831e-07,
      "loss": 1.0174,
      "step": 10371
    },
    {
      "epoch": 0.748313552902132,
      "grad_norm": 3.248649610059606,
      "learning_rate": 6.283731202045146e-07,
      "loss": 0.9655,
      "step": 10372
    },
    {
      "epoch": 0.7483857003715595,
      "grad_norm": 5.015235957093371,
      "learning_rate": 6.28033023754666e-07,
      "loss": 1.0291,
      "step": 10373
    },
    {
      "epoch": 0.748457847840987,
      "grad_norm": 5.284810147227271,
      "learning_rate": 6.276930022228543e-07,
      "loss": 0.8672,
      "step": 10374
    },
    {
      "epoch": 0.7485299953104145,
      "grad_norm": 3.486099705839983,
      "learning_rate": 6.273530556276478e-07,
      "loss": 0.9908,
      "step": 10375
    },
    {
      "epoch": 0.748602142779842,
      "grad_norm": 5.579671351554446,
      "learning_rate": 6.270131839876093e-07,
      "loss": 0.9646,
      "step": 10376
    },
    {
      "epoch": 0.7486742902492695,
      "grad_norm": 7.055212531507899,
      "learning_rate": 6.266733873212986e-07,
      "loss": 0.8842,
      "step": 10377
    },
    {
      "epoch": 0.748746437718697,
      "grad_norm": 3.1878099477828967,
      "learning_rate": 6.263336656472702e-07,
      "loss": 0.9875,
      "step": 10378
    },
    {
      "epoch": 0.7488185851881245,
      "grad_norm": 6.905480803725055,
      "learning_rate": 6.259940189840751e-07,
      "loss": 0.9128,
      "step": 10379
    },
    {
      "epoch": 0.748890732657552,
      "grad_norm": 2.934891211915912,
      "learning_rate": 6.256544473502608e-07,
      "loss": 0.8778,
      "step": 10380
    },
    {
      "epoch": 0.7489628801269795,
      "grad_norm": 2.249017500942013,
      "learning_rate": 6.253149507643689e-07,
      "loss": 0.8961,
      "step": 10381
    },
    {
      "epoch": 0.749035027596407,
      "grad_norm": 6.274406082801346,
      "learning_rate": 6.249755292449388e-07,
      "loss": 0.9598,
      "step": 10382
    },
    {
      "epoch": 0.7491071750658346,
      "grad_norm": 6.545504988851665,
      "learning_rate": 6.246361828105046e-07,
      "loss": 1.0011,
      "step": 10383
    },
    {
      "epoch": 0.7491793225352621,
      "grad_norm": 3.2915639901544886,
      "learning_rate": 6.242969114795971e-07,
      "loss": 0.9229,
      "step": 10384
    },
    {
      "epoch": 0.7492514700046896,
      "grad_norm": 3.565007448335761,
      "learning_rate": 6.239577152707427e-07,
      "loss": 0.9673,
      "step": 10385
    },
    {
      "epoch": 0.7493236174741171,
      "grad_norm": 3.476690465736554,
      "learning_rate": 6.236185942024635e-07,
      "loss": 0.9981,
      "step": 10386
    },
    {
      "epoch": 0.7493957649435447,
      "grad_norm": 4.652308491942384,
      "learning_rate": 6.232795482932778e-07,
      "loss": 0.981,
      "step": 10387
    },
    {
      "epoch": 0.7494679124129722,
      "grad_norm": 2.5028288095709446,
      "learning_rate": 6.229405775616999e-07,
      "loss": 0.996,
      "step": 10388
    },
    {
      "epoch": 0.7495400598823996,
      "grad_norm": 4.3406343332276816,
      "learning_rate": 6.226016820262388e-07,
      "loss": 1.0201,
      "step": 10389
    },
    {
      "epoch": 0.7496122073518271,
      "grad_norm": 4.959760679501059,
      "learning_rate": 6.222628617054011e-07,
      "loss": 0.9946,
      "step": 10390
    },
    {
      "epoch": 0.7496843548212546,
      "grad_norm": 5.838429032758439,
      "learning_rate": 6.21924116617688e-07,
      "loss": 0.8664,
      "step": 10391
    },
    {
      "epoch": 0.7497565022906821,
      "grad_norm": 4.443184878098357,
      "learning_rate": 6.215854467815975e-07,
      "loss": 1.0019,
      "step": 10392
    },
    {
      "epoch": 0.7498286497601097,
      "grad_norm": 3.163686320230326,
      "learning_rate": 6.212468522156238e-07,
      "loss": 1.0098,
      "step": 10393
    },
    {
      "epoch": 0.7499007972295372,
      "grad_norm": 3.7422988652042193,
      "learning_rate": 6.209083329382543e-07,
      "loss": 0.9166,
      "step": 10394
    },
    {
      "epoch": 0.7499729446989647,
      "grad_norm": 6.858487638805686,
      "learning_rate": 6.205698889679767e-07,
      "loss": 0.9627,
      "step": 10395
    },
    {
      "epoch": 0.7500450921683922,
      "grad_norm": 4.324553520662702,
      "learning_rate": 6.202315203232698e-07,
      "loss": 1.0918,
      "step": 10396
    },
    {
      "epoch": 0.7501172396378197,
      "grad_norm": 3.329532093017735,
      "learning_rate": 6.198932270226129e-07,
      "loss": 0.9795,
      "step": 10397
    },
    {
      "epoch": 0.7501893871072473,
      "grad_norm": 2.9542071262130336,
      "learning_rate": 6.195550090844775e-07,
      "loss": 1.0064,
      "step": 10398
    },
    {
      "epoch": 0.7502615345766748,
      "grad_norm": 3.1080923837606904,
      "learning_rate": 6.192168665273328e-07,
      "loss": 0.8375,
      "step": 10399
    },
    {
      "epoch": 0.7503336820461022,
      "grad_norm": 7.70044319561774,
      "learning_rate": 6.188787993696434e-07,
      "loss": 1.0028,
      "step": 10400
    },
    {
      "epoch": 0.7504058295155297,
      "grad_norm": 3.4510569183599054,
      "learning_rate": 6.185408076298699e-07,
      "loss": 0.9496,
      "step": 10401
    },
    {
      "epoch": 0.7504779769849572,
      "grad_norm": 3.8566264508215933,
      "learning_rate": 6.182028913264692e-07,
      "loss": 0.8711,
      "step": 10402
    },
    {
      "epoch": 0.7505501244543847,
      "grad_norm": 4.405114643107589,
      "learning_rate": 6.17865050477894e-07,
      "loss": 0.9516,
      "step": 10403
    },
    {
      "epoch": 0.7506222719238123,
      "grad_norm": 0.8116141037954965,
      "learning_rate": 6.17527285102591e-07,
      "loss": 0.8818,
      "step": 10404
    },
    {
      "epoch": 0.7506944193932398,
      "grad_norm": 9.201805136680385,
      "learning_rate": 6.171895952190054e-07,
      "loss": 0.9813,
      "step": 10405
    },
    {
      "epoch": 0.7507665668626673,
      "grad_norm": 3.275245913525789,
      "learning_rate": 6.16851980845577e-07,
      "loss": 0.9246,
      "step": 10406
    },
    {
      "epoch": 0.7508387143320948,
      "grad_norm": 3.1575801048477414,
      "learning_rate": 6.165144420007418e-07,
      "loss": 0.9626,
      "step": 10407
    },
    {
      "epoch": 0.7509108618015223,
      "grad_norm": 21.533553798873616,
      "learning_rate": 6.161769787029312e-07,
      "loss": 1.0285,
      "step": 10408
    },
    {
      "epoch": 0.7509830092709499,
      "grad_norm": 4.241674458934369,
      "learning_rate": 6.158395909705732e-07,
      "loss": 0.8517,
      "step": 10409
    },
    {
      "epoch": 0.7510551567403774,
      "grad_norm": 2.8159026861065475,
      "learning_rate": 6.155022788220918e-07,
      "loss": 0.9956,
      "step": 10410
    },
    {
      "epoch": 0.7511273042098049,
      "grad_norm": 5.478582817658359,
      "learning_rate": 6.151650422759042e-07,
      "loss": 1.0077,
      "step": 10411
    },
    {
      "epoch": 0.7511994516792323,
      "grad_norm": 3.478880377467023,
      "learning_rate": 6.148278813504288e-07,
      "loss": 0.9757,
      "step": 10412
    },
    {
      "epoch": 0.7512715991486598,
      "grad_norm": 4.509797979596477,
      "learning_rate": 6.144907960640742e-07,
      "loss": 0.8761,
      "step": 10413
    },
    {
      "epoch": 0.7513437466180873,
      "grad_norm": 6.491497935043485,
      "learning_rate": 6.141537864352484e-07,
      "loss": 0.9119,
      "step": 10414
    },
    {
      "epoch": 0.7514158940875149,
      "grad_norm": 3.4900872317454863,
      "learning_rate": 6.138168524823539e-07,
      "loss": 0.967,
      "step": 10415
    },
    {
      "epoch": 0.7514880415569424,
      "grad_norm": 4.763858255053687,
      "learning_rate": 6.134799942237898e-07,
      "loss": 1.0027,
      "step": 10416
    },
    {
      "epoch": 0.7515601890263699,
      "grad_norm": 8.161899758390545,
      "learning_rate": 6.131432116779511e-07,
      "loss": 0.9937,
      "step": 10417
    },
    {
      "epoch": 0.7516323364957974,
      "grad_norm": 7.391707294729482,
      "learning_rate": 6.128065048632263e-07,
      "loss": 0.9601,
      "step": 10418
    },
    {
      "epoch": 0.7517044839652249,
      "grad_norm": 0.7197157549726003,
      "learning_rate": 6.124698737980043e-07,
      "loss": 0.8763,
      "step": 10419
    },
    {
      "epoch": 0.7517766314346525,
      "grad_norm": 3.056763706837026,
      "learning_rate": 6.121333185006654e-07,
      "loss": 0.9486,
      "step": 10420
    },
    {
      "epoch": 0.75184877890408,
      "grad_norm": 5.867629602184983,
      "learning_rate": 6.117968389895885e-07,
      "loss": 0.9647,
      "step": 10421
    },
    {
      "epoch": 0.7519209263735075,
      "grad_norm": 5.391992014168731,
      "learning_rate": 6.11460435283147e-07,
      "loss": 0.8336,
      "step": 10422
    },
    {
      "epoch": 0.751993073842935,
      "grad_norm": 5.409208337253745,
      "learning_rate": 6.111241073997109e-07,
      "loss": 0.9482,
      "step": 10423
    },
    {
      "epoch": 0.7520652213123624,
      "grad_norm": 0.8080636834811515,
      "learning_rate": 6.10787855357646e-07,
      "loss": 0.8409,
      "step": 10424
    },
    {
      "epoch": 0.7521373687817899,
      "grad_norm": 4.673888726166356,
      "learning_rate": 6.104516791753143e-07,
      "loss": 0.9158,
      "step": 10425
    },
    {
      "epoch": 0.7522095162512175,
      "grad_norm": 3.837124332094754,
      "learning_rate": 6.101155788710713e-07,
      "loss": 1.0,
      "step": 10426
    },
    {
      "epoch": 0.752281663720645,
      "grad_norm": 4.08048247383055,
      "learning_rate": 6.097795544632722e-07,
      "loss": 0.941,
      "step": 10427
    },
    {
      "epoch": 0.7523538111900725,
      "grad_norm": 8.389569264063766,
      "learning_rate": 6.094436059702651e-07,
      "loss": 1.062,
      "step": 10428
    },
    {
      "epoch": 0.7524259586595,
      "grad_norm": 11.01411157045921,
      "learning_rate": 6.091077334103947e-07,
      "loss": 0.9515,
      "step": 10429
    },
    {
      "epoch": 0.7524981061289275,
      "grad_norm": 6.248581992460682,
      "learning_rate": 6.087719368020021e-07,
      "loss": 0.9774,
      "step": 10430
    },
    {
      "epoch": 0.752570253598355,
      "grad_norm": 8.822897978983923,
      "learning_rate": 6.084362161634237e-07,
      "loss": 0.8658,
      "step": 10431
    },
    {
      "epoch": 0.7526424010677826,
      "grad_norm": 4.80095049666071,
      "learning_rate": 6.081005715129932e-07,
      "loss": 0.9279,
      "step": 10432
    },
    {
      "epoch": 0.7527145485372101,
      "grad_norm": 3.1150999799575994,
      "learning_rate": 6.077650028690362e-07,
      "loss": 0.9315,
      "step": 10433
    },
    {
      "epoch": 0.7527866960066376,
      "grad_norm": 3.6456165067592434,
      "learning_rate": 6.074295102498801e-07,
      "loss": 1.0246,
      "step": 10434
    },
    {
      "epoch": 0.7528588434760651,
      "grad_norm": 4.7204689785358545,
      "learning_rate": 6.070940936738418e-07,
      "loss": 0.9484,
      "step": 10435
    },
    {
      "epoch": 0.7529309909454925,
      "grad_norm": 7.592311895638528,
      "learning_rate": 6.067587531592402e-07,
      "loss": 1.0056,
      "step": 10436
    },
    {
      "epoch": 0.7530031384149201,
      "grad_norm": 4.639896014462438,
      "learning_rate": 6.064234887243847e-07,
      "loss": 1.0312,
      "step": 10437
    },
    {
      "epoch": 0.7530752858843476,
      "grad_norm": 5.651635242233427,
      "learning_rate": 6.060883003875837e-07,
      "loss": 1.0122,
      "step": 10438
    },
    {
      "epoch": 0.7531474333537751,
      "grad_norm": 6.311413662035636,
      "learning_rate": 6.057531881671405e-07,
      "loss": 0.9921,
      "step": 10439
    },
    {
      "epoch": 0.7532195808232026,
      "grad_norm": 3.7872681449200134,
      "learning_rate": 6.054181520813542e-07,
      "loss": 0.9063,
      "step": 10440
    },
    {
      "epoch": 0.7532917282926301,
      "grad_norm": 3.9691300097581577,
      "learning_rate": 6.050831921485209e-07,
      "loss": 0.8528,
      "step": 10441
    },
    {
      "epoch": 0.7533638757620577,
      "grad_norm": 1.016688215829906,
      "learning_rate": 6.047483083869301e-07,
      "loss": 0.8373,
      "step": 10442
    },
    {
      "epoch": 0.7534360232314852,
      "grad_norm": 4.278818595619444,
      "learning_rate": 6.044135008148689e-07,
      "loss": 1.0039,
      "step": 10443
    },
    {
      "epoch": 0.7535081707009127,
      "grad_norm": 4.079950267638242,
      "learning_rate": 6.040787694506202e-07,
      "loss": 0.9141,
      "step": 10444
    },
    {
      "epoch": 0.7535803181703402,
      "grad_norm": 3.916444582763918,
      "learning_rate": 6.037441143124624e-07,
      "loss": 0.9621,
      "step": 10445
    },
    {
      "epoch": 0.7536524656397677,
      "grad_norm": 6.496774313230434,
      "learning_rate": 6.034095354186699e-07,
      "loss": 0.9028,
      "step": 10446
    },
    {
      "epoch": 0.7537246131091951,
      "grad_norm": 4.671529393505905,
      "learning_rate": 6.030750327875132e-07,
      "loss": 0.937,
      "step": 10447
    },
    {
      "epoch": 0.7537967605786227,
      "grad_norm": 4.489059500289961,
      "learning_rate": 6.027406064372564e-07,
      "loss": 0.8969,
      "step": 10448
    },
    {
      "epoch": 0.7538689080480502,
      "grad_norm": 3.77581547903738,
      "learning_rate": 6.024062563861638e-07,
      "loss": 0.993,
      "step": 10449
    },
    {
      "epoch": 0.7539410555174777,
      "grad_norm": 3.4058129266423576,
      "learning_rate": 6.020719826524907e-07,
      "loss": 0.9523,
      "step": 10450
    },
    {
      "epoch": 0.7540132029869052,
      "grad_norm": 3.798710383377789,
      "learning_rate": 6.017377852544929e-07,
      "loss": 0.9988,
      "step": 10451
    },
    {
      "epoch": 0.7540853504563327,
      "grad_norm": 3.38922244829343,
      "learning_rate": 6.014036642104178e-07,
      "loss": 0.871,
      "step": 10452
    },
    {
      "epoch": 0.7541574979257603,
      "grad_norm": 8.797587445478438,
      "learning_rate": 6.010696195385112e-07,
      "loss": 0.8476,
      "step": 10453
    },
    {
      "epoch": 0.7542296453951878,
      "grad_norm": 9.144167899792187,
      "learning_rate": 6.007356512570141e-07,
      "loss": 0.9365,
      "step": 10454
    },
    {
      "epoch": 0.7543017928646153,
      "grad_norm": 3.6622002592889076,
      "learning_rate": 6.00401759384163e-07,
      "loss": 0.892,
      "step": 10455
    },
    {
      "epoch": 0.7543739403340428,
      "grad_norm": 4.115149092910823,
      "learning_rate": 6.000679439381915e-07,
      "loss": 0.9772,
      "step": 10456
    },
    {
      "epoch": 0.7544460878034703,
      "grad_norm": 0.8382124205077512,
      "learning_rate": 5.99734204937326e-07,
      "loss": 0.8808,
      "step": 10457
    },
    {
      "epoch": 0.7545182352728979,
      "grad_norm": 2.3930022941925873,
      "learning_rate": 5.99400542399793e-07,
      "loss": 0.9373,
      "step": 10458
    },
    {
      "epoch": 0.7545903827423253,
      "grad_norm": 4.505214636767025,
      "learning_rate": 5.990669563438109e-07,
      "loss": 0.9275,
      "step": 10459
    },
    {
      "epoch": 0.7546625302117528,
      "grad_norm": 3.980150923239612,
      "learning_rate": 5.987334467875964e-07,
      "loss": 0.9188,
      "step": 10460
    },
    {
      "epoch": 0.7547346776811803,
      "grad_norm": 3.356918785478702,
      "learning_rate": 5.98400013749361e-07,
      "loss": 0.9582,
      "step": 10461
    },
    {
      "epoch": 0.7548068251506078,
      "grad_norm": 3.770601774802084,
      "learning_rate": 5.980666572473123e-07,
      "loss": 0.964,
      "step": 10462
    },
    {
      "epoch": 0.7548789726200353,
      "grad_norm": 5.285143799354791,
      "learning_rate": 5.977333772996535e-07,
      "loss": 0.9141,
      "step": 10463
    },
    {
      "epoch": 0.7549511200894629,
      "grad_norm": 3.8275390176509796,
      "learning_rate": 5.974001739245848e-07,
      "loss": 0.9676,
      "step": 10464
    },
    {
      "epoch": 0.7550232675588904,
      "grad_norm": 2.67549156235826,
      "learning_rate": 5.970670471402999e-07,
      "loss": 0.921,
      "step": 10465
    },
    {
      "epoch": 0.7550954150283179,
      "grad_norm": 8.64961820432361,
      "learning_rate": 5.967339969649898e-07,
      "loss": 1.0183,
      "step": 10466
    },
    {
      "epoch": 0.7551675624977454,
      "grad_norm": 0.6792157388483929,
      "learning_rate": 5.964010234168416e-07,
      "loss": 0.8441,
      "step": 10467
    },
    {
      "epoch": 0.7552397099671729,
      "grad_norm": 3.443199703606164,
      "learning_rate": 5.960681265140373e-07,
      "loss": 0.9846,
      "step": 10468
    },
    {
      "epoch": 0.7553118574366005,
      "grad_norm": 2.5945353410614485,
      "learning_rate": 5.957353062747554e-07,
      "loss": 0.9067,
      "step": 10469
    },
    {
      "epoch": 0.755384004906028,
      "grad_norm": 3.1059952199669465,
      "learning_rate": 5.954025627171703e-07,
      "loss": 0.8439,
      "step": 10470
    },
    {
      "epoch": 0.7554561523754554,
      "grad_norm": 3.794838692876952,
      "learning_rate": 5.95069895859452e-07,
      "loss": 0.9232,
      "step": 10471
    },
    {
      "epoch": 0.7555282998448829,
      "grad_norm": 8.796220609402503,
      "learning_rate": 5.947373057197647e-07,
      "loss": 0.904,
      "step": 10472
    },
    {
      "epoch": 0.7556004473143104,
      "grad_norm": 3.3231559515791047,
      "learning_rate": 5.944047923162723e-07,
      "loss": 0.9275,
      "step": 10473
    },
    {
      "epoch": 0.7556725947837379,
      "grad_norm": 4.287050647376798,
      "learning_rate": 5.940723556671297e-07,
      "loss": 0.9682,
      "step": 10474
    },
    {
      "epoch": 0.7557447422531655,
      "grad_norm": 0.8537974296118299,
      "learning_rate": 5.937399957904923e-07,
      "loss": 0.8617,
      "step": 10475
    },
    {
      "epoch": 0.755816889722593,
      "grad_norm": 3.424737438523886,
      "learning_rate": 5.934077127045072e-07,
      "loss": 1.0117,
      "step": 10476
    },
    {
      "epoch": 0.7558890371920205,
      "grad_norm": 4.251961648044972,
      "learning_rate": 5.9307550642732e-07,
      "loss": 1.0656,
      "step": 10477
    },
    {
      "epoch": 0.755961184661448,
      "grad_norm": 3.180402745588286,
      "learning_rate": 5.927433769770711e-07,
      "loss": 0.8192,
      "step": 10478
    },
    {
      "epoch": 0.7560333321308755,
      "grad_norm": 2.715590032740883,
      "learning_rate": 5.924113243718969e-07,
      "loss": 0.9915,
      "step": 10479
    },
    {
      "epoch": 0.756105479600303,
      "grad_norm": 3.990612697358312,
      "learning_rate": 5.920793486299302e-07,
      "loss": 1.0008,
      "step": 10480
    },
    {
      "epoch": 0.7561776270697306,
      "grad_norm": 3.4954653382937986,
      "learning_rate": 5.917474497692978e-07,
      "loss": 0.8742,
      "step": 10481
    },
    {
      "epoch": 0.7562497745391581,
      "grad_norm": 3.3688609485367857,
      "learning_rate": 5.91415627808124e-07,
      "loss": 0.8578,
      "step": 10482
    },
    {
      "epoch": 0.7563219220085855,
      "grad_norm": 4.240949530893286,
      "learning_rate": 5.910838827645281e-07,
      "loss": 1.0932,
      "step": 10483
    },
    {
      "epoch": 0.756394069478013,
      "grad_norm": 2.960162139266542,
      "learning_rate": 5.90752214656626e-07,
      "loss": 0.9811,
      "step": 10484
    },
    {
      "epoch": 0.7564662169474405,
      "grad_norm": 3.4482691527991958,
      "learning_rate": 5.904206235025284e-07,
      "loss": 1.0924,
      "step": 10485
    },
    {
      "epoch": 0.7565383644168681,
      "grad_norm": 3.998951178376472,
      "learning_rate": 5.900891093203433e-07,
      "loss": 0.8483,
      "step": 10486
    },
    {
      "epoch": 0.7566105118862956,
      "grad_norm": 6.284998747735213,
      "learning_rate": 5.897576721281712e-07,
      "loss": 0.9119,
      "step": 10487
    },
    {
      "epoch": 0.7566826593557231,
      "grad_norm": 3.230648972121065,
      "learning_rate": 5.894263119441134e-07,
      "loss": 0.9082,
      "step": 10488
    },
    {
      "epoch": 0.7567548068251506,
      "grad_norm": 4.970546757425085,
      "learning_rate": 5.890950287862624e-07,
      "loss": 0.9368,
      "step": 10489
    },
    {
      "epoch": 0.7568269542945781,
      "grad_norm": 2.548657407926239,
      "learning_rate": 5.887638226727087e-07,
      "loss": 0.9491,
      "step": 10490
    },
    {
      "epoch": 0.7568991017640057,
      "grad_norm": 2.9889662013716203,
      "learning_rate": 5.884326936215387e-07,
      "loss": 0.9498,
      "step": 10491
    },
    {
      "epoch": 0.7569712492334332,
      "grad_norm": 3.7452349588310994,
      "learning_rate": 5.881016416508338e-07,
      "loss": 0.991,
      "step": 10492
    },
    {
      "epoch": 0.7570433967028607,
      "grad_norm": 5.0571656556804765,
      "learning_rate": 5.877706667786713e-07,
      "loss": 0.9643,
      "step": 10493
    },
    {
      "epoch": 0.7571155441722881,
      "grad_norm": 3.2233412505890566,
      "learning_rate": 5.87439769023125e-07,
      "loss": 0.9848,
      "step": 10494
    },
    {
      "epoch": 0.7571876916417156,
      "grad_norm": 0.7281441649178201,
      "learning_rate": 5.871089484022644e-07,
      "loss": 0.8299,
      "step": 10495
    },
    {
      "epoch": 0.7572598391111431,
      "grad_norm": 3.57733745320306,
      "learning_rate": 5.867782049341526e-07,
      "loss": 0.8903,
      "step": 10496
    },
    {
      "epoch": 0.7573319865805707,
      "grad_norm": 6.023343452877177,
      "learning_rate": 5.864475386368526e-07,
      "loss": 0.8261,
      "step": 10497
    },
    {
      "epoch": 0.7574041340499982,
      "grad_norm": 3.6610026973661287,
      "learning_rate": 5.86116949528419e-07,
      "loss": 0.8614,
      "step": 10498
    },
    {
      "epoch": 0.7574762815194257,
      "grad_norm": 4.478649990829429,
      "learning_rate": 5.857864376269051e-07,
      "loss": 0.9935,
      "step": 10499
    },
    {
      "epoch": 0.7575484289888532,
      "grad_norm": 2.637825288645548,
      "learning_rate": 5.854560029503584e-07,
      "loss": 1.044,
      "step": 10500
    },
    {
      "epoch": 0.7576205764582807,
      "grad_norm": 4.3881703095677,
      "learning_rate": 5.85125645516823e-07,
      "loss": 0.9547,
      "step": 10501
    },
    {
      "epoch": 0.7576927239277083,
      "grad_norm": 3.398358363841547,
      "learning_rate": 5.847953653443383e-07,
      "loss": 0.9249,
      "step": 10502
    },
    {
      "epoch": 0.7577648713971358,
      "grad_norm": 2.3043139138750126,
      "learning_rate": 5.844651624509405e-07,
      "loss": 0.9421,
      "step": 10503
    },
    {
      "epoch": 0.7578370188665633,
      "grad_norm": 2.947843481083221,
      "learning_rate": 5.841350368546594e-07,
      "loss": 0.9883,
      "step": 10504
    },
    {
      "epoch": 0.7579091663359908,
      "grad_norm": 4.489159135478753,
      "learning_rate": 5.838049885735223e-07,
      "loss": 0.9861,
      "step": 10505
    },
    {
      "epoch": 0.7579813138054182,
      "grad_norm": 9.14004530413395,
      "learning_rate": 5.834750176255525e-07,
      "loss": 0.9181,
      "step": 10506
    },
    {
      "epoch": 0.7580534612748457,
      "grad_norm": 5.240873851565914,
      "learning_rate": 5.831451240287682e-07,
      "loss": 0.9687,
      "step": 10507
    },
    {
      "epoch": 0.7581256087442733,
      "grad_norm": 33.844348736624404,
      "learning_rate": 5.828153078011837e-07,
      "loss": 0.9796,
      "step": 10508
    },
    {
      "epoch": 0.7581977562137008,
      "grad_norm": 4.578912348304675,
      "learning_rate": 5.824855689608088e-07,
      "loss": 0.9429,
      "step": 10509
    },
    {
      "epoch": 0.7582699036831283,
      "grad_norm": 4.669711118989531,
      "learning_rate": 5.821559075256504e-07,
      "loss": 0.8841,
      "step": 10510
    },
    {
      "epoch": 0.7583420511525558,
      "grad_norm": 5.379174673880663,
      "learning_rate": 5.818263235137078e-07,
      "loss": 0.9716,
      "step": 10511
    },
    {
      "epoch": 0.7584141986219833,
      "grad_norm": 6.053442097217519,
      "learning_rate": 5.81496816942981e-07,
      "loss": 1.0048,
      "step": 10512
    },
    {
      "epoch": 0.7584863460914109,
      "grad_norm": 3.7263857881565743,
      "learning_rate": 5.811673878314613e-07,
      "loss": 0.9368,
      "step": 10513
    },
    {
      "epoch": 0.7585584935608384,
      "grad_norm": 3.2226925009798224,
      "learning_rate": 5.808380361971384e-07,
      "loss": 0.8824,
      "step": 10514
    },
    {
      "epoch": 0.7586306410302659,
      "grad_norm": 7.288187937385021,
      "learning_rate": 5.805087620579965e-07,
      "loss": 0.9692,
      "step": 10515
    },
    {
      "epoch": 0.7587027884996934,
      "grad_norm": 3.2614012055382466,
      "learning_rate": 5.801795654320165e-07,
      "loss": 0.8887,
      "step": 10516
    },
    {
      "epoch": 0.7587749359691209,
      "grad_norm": 3.7637476376488403,
      "learning_rate": 5.798504463371742e-07,
      "loss": 0.9314,
      "step": 10517
    },
    {
      "epoch": 0.7588470834385483,
      "grad_norm": 3.375289904541129,
      "learning_rate": 5.79521404791442e-07,
      "loss": 0.8524,
      "step": 10518
    },
    {
      "epoch": 0.7589192309079759,
      "grad_norm": 6.394981037556181,
      "learning_rate": 5.791924408127881e-07,
      "loss": 0.9026,
      "step": 10519
    },
    {
      "epoch": 0.7589913783774034,
      "grad_norm": 3.450712716387384,
      "learning_rate": 5.788635544191749e-07,
      "loss": 1.0202,
      "step": 10520
    },
    {
      "epoch": 0.7590635258468309,
      "grad_norm": 3.1245235842899604,
      "learning_rate": 5.785347456285619e-07,
      "loss": 1.0339,
      "step": 10521
    },
    {
      "epoch": 0.7591356733162584,
      "grad_norm": 18.37626051958324,
      "learning_rate": 5.782060144589045e-07,
      "loss": 0.9358,
      "step": 10522
    },
    {
      "epoch": 0.7592078207856859,
      "grad_norm": 2.872884220625187,
      "learning_rate": 5.778773609281531e-07,
      "loss": 1.1153,
      "step": 10523
    },
    {
      "epoch": 0.7592799682551135,
      "grad_norm": 2.9537641852801406,
      "learning_rate": 5.775487850542546e-07,
      "loss": 0.8622,
      "step": 10524
    },
    {
      "epoch": 0.759352115724541,
      "grad_norm": 9.355969454242086,
      "learning_rate": 5.772202868551519e-07,
      "loss": 0.9358,
      "step": 10525
    },
    {
      "epoch": 0.7594242631939685,
      "grad_norm": 4.575912409617689,
      "learning_rate": 5.768918663487811e-07,
      "loss": 0.9975,
      "step": 10526
    },
    {
      "epoch": 0.759496410663396,
      "grad_norm": 3.5902095438123727,
      "learning_rate": 5.765635235530786e-07,
      "loss": 0.9807,
      "step": 10527
    },
    {
      "epoch": 0.7595685581328235,
      "grad_norm": 0.800695907883319,
      "learning_rate": 5.762352584859722e-07,
      "loss": 0.8172,
      "step": 10528
    },
    {
      "epoch": 0.759640705602251,
      "grad_norm": 3.203375383220829,
      "learning_rate": 5.759070711653875e-07,
      "loss": 1.0097,
      "step": 10529
    },
    {
      "epoch": 0.7597128530716785,
      "grad_norm": 4.445517161506854,
      "learning_rate": 5.755789616092459e-07,
      "loss": 0.8718,
      "step": 10530
    },
    {
      "epoch": 0.759785000541106,
      "grad_norm": 4.10843801513353,
      "learning_rate": 5.752509298354642e-07,
      "loss": 1.0301,
      "step": 10531
    },
    {
      "epoch": 0.7598571480105335,
      "grad_norm": 5.146604081269118,
      "learning_rate": 5.749229758619549e-07,
      "loss": 1.0263,
      "step": 10532
    },
    {
      "epoch": 0.759929295479961,
      "grad_norm": 5.876625647759626,
      "learning_rate": 5.745950997066266e-07,
      "loss": 0.9735,
      "step": 10533
    },
    {
      "epoch": 0.7600014429493885,
      "grad_norm": 0.8148992034973753,
      "learning_rate": 5.742673013873838e-07,
      "loss": 0.7994,
      "step": 10534
    },
    {
      "epoch": 0.7600735904188161,
      "grad_norm": 5.784233024746696,
      "learning_rate": 5.739395809221252e-07,
      "loss": 0.9166,
      "step": 10535
    },
    {
      "epoch": 0.7601457378882436,
      "grad_norm": 19.53572058210573,
      "learning_rate": 5.73611938328747e-07,
      "loss": 0.9018,
      "step": 10536
    },
    {
      "epoch": 0.7602178853576711,
      "grad_norm": 5.3121752471396535,
      "learning_rate": 5.732843736251405e-07,
      "loss": 0.8946,
      "step": 10537
    },
    {
      "epoch": 0.7602900328270986,
      "grad_norm": 4.3610893351557225,
      "learning_rate": 5.729568868291928e-07,
      "loss": 1.0046,
      "step": 10538
    },
    {
      "epoch": 0.7603621802965261,
      "grad_norm": 3.468225525392331,
      "learning_rate": 5.726294779587868e-07,
      "loss": 0.9049,
      "step": 10539
    },
    {
      "epoch": 0.7604343277659537,
      "grad_norm": 4.112709462127497,
      "learning_rate": 5.72302147031801e-07,
      "loss": 0.9255,
      "step": 10540
    },
    {
      "epoch": 0.7605064752353812,
      "grad_norm": 3.4776971572723516,
      "learning_rate": 5.719748940661097e-07,
      "loss": 0.9562,
      "step": 10541
    },
    {
      "epoch": 0.7605786227048086,
      "grad_norm": 5.6677136669427135,
      "learning_rate": 5.716477190795839e-07,
      "loss": 0.9158,
      "step": 10542
    },
    {
      "epoch": 0.7606507701742361,
      "grad_norm": 5.124491224388807,
      "learning_rate": 5.713206220900879e-07,
      "loss": 1.0,
      "step": 10543
    },
    {
      "epoch": 0.7607229176436636,
      "grad_norm": 5.164626097979755,
      "learning_rate": 5.70993603115484e-07,
      "loss": 0.9392,
      "step": 10544
    },
    {
      "epoch": 0.7607950651130911,
      "grad_norm": 3.777937391748852,
      "learning_rate": 5.706666621736293e-07,
      "loss": 0.9537,
      "step": 10545
    },
    {
      "epoch": 0.7608672125825187,
      "grad_norm": 4.256642928170548,
      "learning_rate": 5.703397992823771e-07,
      "loss": 0.9862,
      "step": 10546
    },
    {
      "epoch": 0.7609393600519462,
      "grad_norm": 0.7704996899078755,
      "learning_rate": 5.700130144595761e-07,
      "loss": 0.8688,
      "step": 10547
    },
    {
      "epoch": 0.7610115075213737,
      "grad_norm": 4.164997961952837,
      "learning_rate": 5.696863077230704e-07,
      "loss": 0.9707,
      "step": 10548
    },
    {
      "epoch": 0.7610836549908012,
      "grad_norm": 3.96426486595298,
      "learning_rate": 5.693596790907016e-07,
      "loss": 1.0566,
      "step": 10549
    },
    {
      "epoch": 0.7611558024602287,
      "grad_norm": 4.087260686652406,
      "learning_rate": 5.690331285803032e-07,
      "loss": 0.9679,
      "step": 10550
    },
    {
      "epoch": 0.7612279499296563,
      "grad_norm": 9.293718279724786,
      "learning_rate": 5.687066562097099e-07,
      "loss": 1.0009,
      "step": 10551
    },
    {
      "epoch": 0.7613000973990838,
      "grad_norm": 3.639920931994858,
      "learning_rate": 5.683802619967466e-07,
      "loss": 0.9799,
      "step": 10552
    },
    {
      "epoch": 0.7613722448685112,
      "grad_norm": 9.287463230969022,
      "learning_rate": 5.680539459592377e-07,
      "loss": 0.9581,
      "step": 10553
    },
    {
      "epoch": 0.7614443923379387,
      "grad_norm": 4.275944227377683,
      "learning_rate": 5.67727708115002e-07,
      "loss": 0.8898,
      "step": 10554
    },
    {
      "epoch": 0.7615165398073662,
      "grad_norm": 4.1537885906436935,
      "learning_rate": 5.674015484818541e-07,
      "loss": 0.9161,
      "step": 10555
    },
    {
      "epoch": 0.7615886872767937,
      "grad_norm": 5.199348430893525,
      "learning_rate": 5.670754670776043e-07,
      "loss": 1.0124,
      "step": 10556
    },
    {
      "epoch": 0.7616608347462213,
      "grad_norm": 7.120671346067109,
      "learning_rate": 5.667494639200594e-07,
      "loss": 0.8902,
      "step": 10557
    },
    {
      "epoch": 0.7617329822156488,
      "grad_norm": 3.216385055890975,
      "learning_rate": 5.664235390270198e-07,
      "loss": 0.9027,
      "step": 10558
    },
    {
      "epoch": 0.7618051296850763,
      "grad_norm": 3.960049441250505,
      "learning_rate": 5.660976924162839e-07,
      "loss": 0.9405,
      "step": 10559
    },
    {
      "epoch": 0.7618772771545038,
      "grad_norm": 3.085048731043417,
      "learning_rate": 5.657719241056449e-07,
      "loss": 1.0647,
      "step": 10560
    },
    {
      "epoch": 0.7619494246239313,
      "grad_norm": 23.05264625833049,
      "learning_rate": 5.654462341128918e-07,
      "loss": 1.0429,
      "step": 10561
    },
    {
      "epoch": 0.7620215720933589,
      "grad_norm": 4.267485160600709,
      "learning_rate": 5.651206224558091e-07,
      "loss": 0.9805,
      "step": 10562
    },
    {
      "epoch": 0.7620937195627864,
      "grad_norm": 3.2928747704468226,
      "learning_rate": 5.647950891521778e-07,
      "loss": 0.942,
      "step": 10563
    },
    {
      "epoch": 0.7621658670322139,
      "grad_norm": 3.93848079999927,
      "learning_rate": 5.644696342197744e-07,
      "loss": 0.8336,
      "step": 10564
    },
    {
      "epoch": 0.7622380145016413,
      "grad_norm": 3.021496842401153,
      "learning_rate": 5.641442576763687e-07,
      "loss": 0.8206,
      "step": 10565
    },
    {
      "epoch": 0.7623101619710688,
      "grad_norm": 8.591015633880666,
      "learning_rate": 5.638189595397309e-07,
      "loss": 0.975,
      "step": 10566
    },
    {
      "epoch": 0.7623823094404963,
      "grad_norm": 4.254431713987707,
      "learning_rate": 5.634937398276225e-07,
      "loss": 0.9467,
      "step": 10567
    },
    {
      "epoch": 0.7624544569099239,
      "grad_norm": 5.139303762282469,
      "learning_rate": 5.631685985578034e-07,
      "loss": 0.8607,
      "step": 10568
    },
    {
      "epoch": 0.7625266043793514,
      "grad_norm": 3.552361769218374,
      "learning_rate": 5.628435357480281e-07,
      "loss": 1.0427,
      "step": 10569
    },
    {
      "epoch": 0.7625987518487789,
      "grad_norm": 4.208868137533594,
      "learning_rate": 5.625185514160471e-07,
      "loss": 0.9719,
      "step": 10570
    },
    {
      "epoch": 0.7626708993182064,
      "grad_norm": 2.9617939618743256,
      "learning_rate": 5.621936455796066e-07,
      "loss": 0.8544,
      "step": 10571
    },
    {
      "epoch": 0.7627430467876339,
      "grad_norm": 4.364538709948024,
      "learning_rate": 5.618688182564486e-07,
      "loss": 0.9692,
      "step": 10572
    },
    {
      "epoch": 0.7628151942570615,
      "grad_norm": 3.5630657349164143,
      "learning_rate": 5.615440694643114e-07,
      "loss": 0.9247,
      "step": 10573
    },
    {
      "epoch": 0.762887341726489,
      "grad_norm": 2.959750700066303,
      "learning_rate": 5.612193992209268e-07,
      "loss": 0.9042,
      "step": 10574
    },
    {
      "epoch": 0.7629594891959165,
      "grad_norm": 4.6511005370830985,
      "learning_rate": 5.608948075440247e-07,
      "loss": 1.014,
      "step": 10575
    },
    {
      "epoch": 0.763031636665344,
      "grad_norm": 4.300151449685741,
      "learning_rate": 5.605702944513298e-07,
      "loss": 0.9654,
      "step": 10576
    },
    {
      "epoch": 0.7631037841347714,
      "grad_norm": 4.269353892559401,
      "learning_rate": 5.602458599605624e-07,
      "loss": 1.0595,
      "step": 10577
    },
    {
      "epoch": 0.7631759316041989,
      "grad_norm": 3.2513370697905257,
      "learning_rate": 5.59921504089439e-07,
      "loss": 1.0633,
      "step": 10578
    },
    {
      "epoch": 0.7632480790736265,
      "grad_norm": 5.582007331343087,
      "learning_rate": 5.595972268556716e-07,
      "loss": 1.0171,
      "step": 10579
    },
    {
      "epoch": 0.763320226543054,
      "grad_norm": 5.527950358718594,
      "learning_rate": 5.592730282769664e-07,
      "loss": 0.9752,
      "step": 10580
    },
    {
      "epoch": 0.7633923740124815,
      "grad_norm": 4.333531277365975,
      "learning_rate": 5.58948908371029e-07,
      "loss": 0.8736,
      "step": 10581
    },
    {
      "epoch": 0.763464521481909,
      "grad_norm": 4.671752928092685,
      "learning_rate": 5.586248671555565e-07,
      "loss": 1.0352,
      "step": 10582
    },
    {
      "epoch": 0.7635366689513365,
      "grad_norm": 3.7500991808173283,
      "learning_rate": 5.583009046482441e-07,
      "loss": 0.8829,
      "step": 10583
    },
    {
      "epoch": 0.7636088164207641,
      "grad_norm": 5.0378013274608024,
      "learning_rate": 5.579770208667824e-07,
      "loss": 0.9114,
      "step": 10584
    },
    {
      "epoch": 0.7636809638901916,
      "grad_norm": 5.273644925029376,
      "learning_rate": 5.576532158288574e-07,
      "loss": 0.863,
      "step": 10585
    },
    {
      "epoch": 0.7637531113596191,
      "grad_norm": 4.476288738834112,
      "learning_rate": 5.573294895521508e-07,
      "loss": 1.0539,
      "step": 10586
    },
    {
      "epoch": 0.7638252588290466,
      "grad_norm": 2.6413312340018287,
      "learning_rate": 5.570058420543404e-07,
      "loss": 0.944,
      "step": 10587
    },
    {
      "epoch": 0.7638974062984741,
      "grad_norm": 0.714519281176558,
      "learning_rate": 5.566822733530996e-07,
      "loss": 0.7979,
      "step": 10588
    },
    {
      "epoch": 0.7639695537679015,
      "grad_norm": 9.193770364392597,
      "learning_rate": 5.563587834660956e-07,
      "loss": 0.9419,
      "step": 10589
    },
    {
      "epoch": 0.7640417012373291,
      "grad_norm": 3.9735166505047363,
      "learning_rate": 5.560353724109959e-07,
      "loss": 1.0176,
      "step": 10590
    },
    {
      "epoch": 0.7641138487067566,
      "grad_norm": 3.2054738434489147,
      "learning_rate": 5.55712040205458e-07,
      "loss": 0.9787,
      "step": 10591
    },
    {
      "epoch": 0.7641859961761841,
      "grad_norm": 3.792226826857203,
      "learning_rate": 5.553887868671394e-07,
      "loss": 0.8451,
      "step": 10592
    },
    {
      "epoch": 0.7642581436456116,
      "grad_norm": 3.5801854760057528,
      "learning_rate": 5.550656124136912e-07,
      "loss": 0.9624,
      "step": 10593
    },
    {
      "epoch": 0.7643302911150391,
      "grad_norm": 3.2938208150761046,
      "learning_rate": 5.547425168627611e-07,
      "loss": 0.9633,
      "step": 10594
    },
    {
      "epoch": 0.7644024385844667,
      "grad_norm": 2.642013997089234,
      "learning_rate": 5.544195002319918e-07,
      "loss": 0.9586,
      "step": 10595
    },
    {
      "epoch": 0.7644745860538942,
      "grad_norm": 3.441305931958678,
      "learning_rate": 5.540965625390228e-07,
      "loss": 0.9963,
      "step": 10596
    },
    {
      "epoch": 0.7645467335233217,
      "grad_norm": 7.305306050432009,
      "learning_rate": 5.537737038014876e-07,
      "loss": 0.8991,
      "step": 10597
    },
    {
      "epoch": 0.7646188809927492,
      "grad_norm": 3.255777652037503,
      "learning_rate": 5.534509240370164e-07,
      "loss": 0.998,
      "step": 10598
    },
    {
      "epoch": 0.7646910284621767,
      "grad_norm": 2.8024572999748583,
      "learning_rate": 5.531282232632355e-07,
      "loss": 0.9077,
      "step": 10599
    },
    {
      "epoch": 0.7647631759316041,
      "grad_norm": 6.978041130139318,
      "learning_rate": 5.528056014977662e-07,
      "loss": 0.9951,
      "step": 10600
    },
    {
      "epoch": 0.7648353234010317,
      "grad_norm": 2.7753128803271383,
      "learning_rate": 5.524830587582263e-07,
      "loss": 0.953,
      "step": 10601
    },
    {
      "epoch": 0.7649074708704592,
      "grad_norm": 2.6195990995362743,
      "learning_rate": 5.521605950622268e-07,
      "loss": 1.0242,
      "step": 10602
    },
    {
      "epoch": 0.7649796183398867,
      "grad_norm": 4.441429225355592,
      "learning_rate": 5.518382104273787e-07,
      "loss": 0.8244,
      "step": 10603
    },
    {
      "epoch": 0.7650517658093142,
      "grad_norm": 4.723115026310265,
      "learning_rate": 5.515159048712836e-07,
      "loss": 1.0103,
      "step": 10604
    },
    {
      "epoch": 0.7651239132787417,
      "grad_norm": 5.5295543765653985,
      "learning_rate": 5.511936784115442e-07,
      "loss": 0.9542,
      "step": 10605
    },
    {
      "epoch": 0.7651960607481693,
      "grad_norm": 4.190330573540918,
      "learning_rate": 5.508715310657542e-07,
      "loss": 0.869,
      "step": 10606
    },
    {
      "epoch": 0.7652682082175968,
      "grad_norm": 5.6384200332896235,
      "learning_rate": 5.505494628515053e-07,
      "loss": 0.9908,
      "step": 10607
    },
    {
      "epoch": 0.7653403556870243,
      "grad_norm": 3.6599494932681687,
      "learning_rate": 5.502274737863846e-07,
      "loss": 0.9509,
      "step": 10608
    },
    {
      "epoch": 0.7654125031564518,
      "grad_norm": 5.920163191530743,
      "learning_rate": 5.499055638879746e-07,
      "loss": 0.8862,
      "step": 10609
    },
    {
      "epoch": 0.7654846506258793,
      "grad_norm": 3.073619197561509,
      "learning_rate": 5.495837331738538e-07,
      "loss": 0.8972,
      "step": 10610
    },
    {
      "epoch": 0.7655567980953069,
      "grad_norm": 2.4104394483574976,
      "learning_rate": 5.49261981661596e-07,
      "loss": 0.9662,
      "step": 10611
    },
    {
      "epoch": 0.7656289455647343,
      "grad_norm": 4.6690334494465535,
      "learning_rate": 5.489403093687714e-07,
      "loss": 1.0129,
      "step": 10612
    },
    {
      "epoch": 0.7657010930341618,
      "grad_norm": 4.340001966062883,
      "learning_rate": 5.486187163129445e-07,
      "loss": 1.0004,
      "step": 10613
    },
    {
      "epoch": 0.7657732405035893,
      "grad_norm": 4.186079097344974,
      "learning_rate": 5.482972025116765e-07,
      "loss": 0.9914,
      "step": 10614
    },
    {
      "epoch": 0.7658453879730168,
      "grad_norm": 6.804838299312126,
      "learning_rate": 5.479757679825245e-07,
      "loss": 0.8715,
      "step": 10615
    },
    {
      "epoch": 0.7659175354424443,
      "grad_norm": 3.387497122787999,
      "learning_rate": 5.476544127430405e-07,
      "loss": 0.9796,
      "step": 10616
    },
    {
      "epoch": 0.7659896829118719,
      "grad_norm": 4.475715170885381,
      "learning_rate": 5.473331368107725e-07,
      "loss": 0.9654,
      "step": 10617
    },
    {
      "epoch": 0.7660618303812994,
      "grad_norm": 2.403175073704679,
      "learning_rate": 5.47011940203265e-07,
      "loss": 0.9406,
      "step": 10618
    },
    {
      "epoch": 0.7661339778507269,
      "grad_norm": 2.8900825610132066,
      "learning_rate": 5.466908229380554e-07,
      "loss": 1.015,
      "step": 10619
    },
    {
      "epoch": 0.7662061253201544,
      "grad_norm": 7.833428754292974,
      "learning_rate": 5.463697850326811e-07,
      "loss": 0.9276,
      "step": 10620
    },
    {
      "epoch": 0.7662782727895819,
      "grad_norm": 5.499082141929372,
      "learning_rate": 5.460488265046713e-07,
      "loss": 0.9731,
      "step": 10621
    },
    {
      "epoch": 0.7663504202590095,
      "grad_norm": 5.528590710712996,
      "learning_rate": 5.457279473715526e-07,
      "loss": 0.9299,
      "step": 10622
    },
    {
      "epoch": 0.766422567728437,
      "grad_norm": 2.9490509187552414,
      "learning_rate": 5.454071476508473e-07,
      "loss": 0.9754,
      "step": 10623
    },
    {
      "epoch": 0.7664947151978644,
      "grad_norm": 3.2516875654051165,
      "learning_rate": 5.450864273600728e-07,
      "loss": 0.963,
      "step": 10624
    },
    {
      "epoch": 0.7665668626672919,
      "grad_norm": 2.7971989960786363,
      "learning_rate": 5.447657865167428e-07,
      "loss": 1.0121,
      "step": 10625
    },
    {
      "epoch": 0.7666390101367194,
      "grad_norm": 4.3580298313837265,
      "learning_rate": 5.444452251383658e-07,
      "loss": 0.9996,
      "step": 10626
    },
    {
      "epoch": 0.7667111576061469,
      "grad_norm": 2.771783813903394,
      "learning_rate": 5.441247432424476e-07,
      "loss": 0.9821,
      "step": 10627
    },
    {
      "epoch": 0.7667833050755745,
      "grad_norm": 4.294132647614941,
      "learning_rate": 5.438043408464866e-07,
      "loss": 0.9786,
      "step": 10628
    },
    {
      "epoch": 0.766855452545002,
      "grad_norm": 4.4249039375524735,
      "learning_rate": 5.434840179679809e-07,
      "loss": 0.8577,
      "step": 10629
    },
    {
      "epoch": 0.7669276000144295,
      "grad_norm": 5.109245275314662,
      "learning_rate": 5.431637746244204e-07,
      "loss": 0.995,
      "step": 10630
    },
    {
      "epoch": 0.766999747483857,
      "grad_norm": 6.002565788979359,
      "learning_rate": 5.428436108332935e-07,
      "loss": 0.8456,
      "step": 10631
    },
    {
      "epoch": 0.7670718949532845,
      "grad_norm": 5.556312119573304,
      "learning_rate": 5.425235266120826e-07,
      "loss": 0.9191,
      "step": 10632
    },
    {
      "epoch": 0.7671440424227121,
      "grad_norm": 6.08114243164274,
      "learning_rate": 5.422035219782666e-07,
      "loss": 0.7526,
      "step": 10633
    },
    {
      "epoch": 0.7672161898921396,
      "grad_norm": 3.405384338690959,
      "learning_rate": 5.418835969493196e-07,
      "loss": 0.8401,
      "step": 10634
    },
    {
      "epoch": 0.7672883373615671,
      "grad_norm": 5.387039186104641,
      "learning_rate": 5.415637515427123e-07,
      "loss": 0.9452,
      "step": 10635
    },
    {
      "epoch": 0.7673604848309945,
      "grad_norm": 4.639251197376931,
      "learning_rate": 5.412439857759088e-07,
      "loss": 1.0112,
      "step": 10636
    },
    {
      "epoch": 0.767432632300422,
      "grad_norm": 4.07242721127257,
      "learning_rate": 5.409242996663714e-07,
      "loss": 0.83,
      "step": 10637
    },
    {
      "epoch": 0.7675047797698495,
      "grad_norm": 4.546798233485338,
      "learning_rate": 5.406046932315564e-07,
      "loss": 0.8893,
      "step": 10638
    },
    {
      "epoch": 0.7675769272392771,
      "grad_norm": 4.7629433012572395,
      "learning_rate": 5.402851664889168e-07,
      "loss": 0.939,
      "step": 10639
    },
    {
      "epoch": 0.7676490747087046,
      "grad_norm": 14.90871167599302,
      "learning_rate": 5.39965719455901e-07,
      "loss": 0.854,
      "step": 10640
    },
    {
      "epoch": 0.7677212221781321,
      "grad_norm": 5.017412194062001,
      "learning_rate": 5.396463521499515e-07,
      "loss": 1.0124,
      "step": 10641
    },
    {
      "epoch": 0.7677933696475596,
      "grad_norm": 3.941976273510277,
      "learning_rate": 5.393270645885095e-07,
      "loss": 0.9019,
      "step": 10642
    },
    {
      "epoch": 0.7678655171169871,
      "grad_norm": 4.337953500923203,
      "learning_rate": 5.390078567890084e-07,
      "loss": 0.8598,
      "step": 10643
    },
    {
      "epoch": 0.7679376645864147,
      "grad_norm": 6.581794784328042,
      "learning_rate": 5.386887287688808e-07,
      "loss": 0.9931,
      "step": 10644
    },
    {
      "epoch": 0.7680098120558422,
      "grad_norm": 4.263111086359768,
      "learning_rate": 5.383696805455515e-07,
      "loss": 0.8329,
      "step": 10645
    },
    {
      "epoch": 0.7680819595252697,
      "grad_norm": 9.303754758166063,
      "learning_rate": 5.380507121364435e-07,
      "loss": 1.0284,
      "step": 10646
    },
    {
      "epoch": 0.7681541069946971,
      "grad_norm": 3.44061041472875,
      "learning_rate": 5.377318235589739e-07,
      "loss": 0.8537,
      "step": 10647
    },
    {
      "epoch": 0.7682262544641246,
      "grad_norm": 7.166350453815687,
      "learning_rate": 5.374130148305563e-07,
      "loss": 0.9513,
      "step": 10648
    },
    {
      "epoch": 0.7682984019335521,
      "grad_norm": 4.950922045903597,
      "learning_rate": 5.370942859685996e-07,
      "loss": 0.9864,
      "step": 10649
    },
    {
      "epoch": 0.7683705494029797,
      "grad_norm": 3.7682160766346082,
      "learning_rate": 5.367756369905085e-07,
      "loss": 0.9224,
      "step": 10650
    },
    {
      "epoch": 0.7684426968724072,
      "grad_norm": 5.082299778827506,
      "learning_rate": 5.364570679136839e-07,
      "loss": 1.059,
      "step": 10651
    },
    {
      "epoch": 0.7685148443418347,
      "grad_norm": 3.304818658052188,
      "learning_rate": 5.361385787555204e-07,
      "loss": 0.8931,
      "step": 10652
    },
    {
      "epoch": 0.7685869918112622,
      "grad_norm": 4.04903374907489,
      "learning_rate": 5.358201695334099e-07,
      "loss": 0.8946,
      "step": 10653
    },
    {
      "epoch": 0.7686591392806897,
      "grad_norm": 2.71598745721272,
      "learning_rate": 5.355018402647397e-07,
      "loss": 0.9739,
      "step": 10654
    },
    {
      "epoch": 0.7687312867501173,
      "grad_norm": 2.67816928751236,
      "learning_rate": 5.351835909668929e-07,
      "loss": 0.8887,
      "step": 10655
    },
    {
      "epoch": 0.7688034342195448,
      "grad_norm": 4.648622961111205,
      "learning_rate": 5.348654216572475e-07,
      "loss": 0.8295,
      "step": 10656
    },
    {
      "epoch": 0.7688755816889723,
      "grad_norm": 5.261482760228493,
      "learning_rate": 5.345473323531782e-07,
      "loss": 1.0006,
      "step": 10657
    },
    {
      "epoch": 0.7689477291583998,
      "grad_norm": 2.6923797131431653,
      "learning_rate": 5.342293230720529e-07,
      "loss": 0.9345,
      "step": 10658
    },
    {
      "epoch": 0.7690198766278272,
      "grad_norm": 3.0850513586285686,
      "learning_rate": 5.339113938312396e-07,
      "loss": 0.901,
      "step": 10659
    },
    {
      "epoch": 0.7690920240972547,
      "grad_norm": 5.579948917825065,
      "learning_rate": 5.33593544648097e-07,
      "loss": 0.9851,
      "step": 10660
    },
    {
      "epoch": 0.7691641715666823,
      "grad_norm": 3.782373182713539,
      "learning_rate": 5.332757755399826e-07,
      "loss": 0.9539,
      "step": 10661
    },
    {
      "epoch": 0.7692363190361098,
      "grad_norm": 4.462054805827687,
      "learning_rate": 5.329580865242484e-07,
      "loss": 0.9849,
      "step": 10662
    },
    {
      "epoch": 0.7693084665055373,
      "grad_norm": 4.074153442444579,
      "learning_rate": 5.326404776182422e-07,
      "loss": 0.9709,
      "step": 10663
    },
    {
      "epoch": 0.7693806139749648,
      "grad_norm": 3.893883501094091,
      "learning_rate": 5.323229488393084e-07,
      "loss": 0.8077,
      "step": 10664
    },
    {
      "epoch": 0.7694527614443923,
      "grad_norm": 7.722506084149534,
      "learning_rate": 5.320055002047839e-07,
      "loss": 0.8828,
      "step": 10665
    },
    {
      "epoch": 0.7695249089138199,
      "grad_norm": 6.275740147440249,
      "learning_rate": 5.316881317320058e-07,
      "loss": 0.9174,
      "step": 10666
    },
    {
      "epoch": 0.7695970563832474,
      "grad_norm": 5.959967258030607,
      "learning_rate": 5.313708434383022e-07,
      "loss": 0.9533,
      "step": 10667
    },
    {
      "epoch": 0.7696692038526749,
      "grad_norm": 5.667058407575578,
      "learning_rate": 5.310536353410013e-07,
      "loss": 0.9023,
      "step": 10668
    },
    {
      "epoch": 0.7697413513221024,
      "grad_norm": 8.51271990297369,
      "learning_rate": 5.30736507457423e-07,
      "loss": 0.9662,
      "step": 10669
    },
    {
      "epoch": 0.7698134987915299,
      "grad_norm": 3.8995378122581346,
      "learning_rate": 5.30419459804885e-07,
      "loss": 1.012,
      "step": 10670
    },
    {
      "epoch": 0.7698856462609573,
      "grad_norm": 4.363612149633791,
      "learning_rate": 5.301024924007e-07,
      "loss": 1.0345,
      "step": 10671
    },
    {
      "epoch": 0.7699577937303849,
      "grad_norm": 0.7734187730776187,
      "learning_rate": 5.297856052621768e-07,
      "loss": 0.8822,
      "step": 10672
    },
    {
      "epoch": 0.7700299411998124,
      "grad_norm": 12.702163415904234,
      "learning_rate": 5.294687984066189e-07,
      "loss": 1.0324,
      "step": 10673
    },
    {
      "epoch": 0.7701020886692399,
      "grad_norm": 3.1345780646283976,
      "learning_rate": 5.291520718513267e-07,
      "loss": 0.9298,
      "step": 10674
    },
    {
      "epoch": 0.7701742361386674,
      "grad_norm": 6.033727265594308,
      "learning_rate": 5.288354256135945e-07,
      "loss": 0.9896,
      "step": 10675
    },
    {
      "epoch": 0.7702463836080949,
      "grad_norm": 6.524104978482771,
      "learning_rate": 5.285188597107137e-07,
      "loss": 1.0021,
      "step": 10676
    },
    {
      "epoch": 0.7703185310775225,
      "grad_norm": 7.471261468105951,
      "learning_rate": 5.282023741599706e-07,
      "loss": 0.8834,
      "step": 10677
    },
    {
      "epoch": 0.77039067854695,
      "grad_norm": 2.826941985661434,
      "learning_rate": 5.278859689786477e-07,
      "loss": 0.8279,
      "step": 10678
    },
    {
      "epoch": 0.7704628260163775,
      "grad_norm": 12.396742091272975,
      "learning_rate": 5.275696441840227e-07,
      "loss": 1.0031,
      "step": 10679
    },
    {
      "epoch": 0.770534973485805,
      "grad_norm": 6.415666609609191,
      "learning_rate": 5.272533997933676e-07,
      "loss": 0.8596,
      "step": 10680
    },
    {
      "epoch": 0.7706071209552325,
      "grad_norm": 5.400209436946925,
      "learning_rate": 5.269372358239539e-07,
      "loss": 1.0035,
      "step": 10681
    },
    {
      "epoch": 0.7706792684246601,
      "grad_norm": 4.698817717197926,
      "learning_rate": 5.26621152293043e-07,
      "loss": 0.9537,
      "step": 10682
    },
    {
      "epoch": 0.7707514158940875,
      "grad_norm": 6.322856301569943,
      "learning_rate": 5.263051492178983e-07,
      "loss": 0.8843,
      "step": 10683
    },
    {
      "epoch": 0.770823563363515,
      "grad_norm": 2.138740322463471,
      "learning_rate": 5.259892266157731e-07,
      "loss": 0.8199,
      "step": 10684
    },
    {
      "epoch": 0.7708957108329425,
      "grad_norm": 3.1018741172467985,
      "learning_rate": 5.256733845039198e-07,
      "loss": 0.8443,
      "step": 10685
    },
    {
      "epoch": 0.77096785830237,
      "grad_norm": 4.562458612959761,
      "learning_rate": 5.253576228995851e-07,
      "loss": 0.9801,
      "step": 10686
    },
    {
      "epoch": 0.7710400057717975,
      "grad_norm": 3.7122295499478946,
      "learning_rate": 5.250419418200118e-07,
      "loss": 0.9271,
      "step": 10687
    },
    {
      "epoch": 0.7711121532412251,
      "grad_norm": 3.5365056450010117,
      "learning_rate": 5.247263412824386e-07,
      "loss": 0.972,
      "step": 10688
    },
    {
      "epoch": 0.7711843007106526,
      "grad_norm": 5.830364380011848,
      "learning_rate": 5.244108213040972e-07,
      "loss": 0.961,
      "step": 10689
    },
    {
      "epoch": 0.7712564481800801,
      "grad_norm": 3.215346527646232,
      "learning_rate": 5.240953819022195e-07,
      "loss": 0.97,
      "step": 10690
    },
    {
      "epoch": 0.7713285956495076,
      "grad_norm": 11.913764084377494,
      "learning_rate": 5.23780023094029e-07,
      "loss": 0.928,
      "step": 10691
    },
    {
      "epoch": 0.7714007431189351,
      "grad_norm": 15.4879186984679,
      "learning_rate": 5.234647448967469e-07,
      "loss": 0.8589,
      "step": 10692
    },
    {
      "epoch": 0.7714728905883627,
      "grad_norm": 8.25626574622363,
      "learning_rate": 5.231495473275887e-07,
      "loss": 1.0011,
      "step": 10693
    },
    {
      "epoch": 0.7715450380577902,
      "grad_norm": 4.450616283818021,
      "learning_rate": 5.228344304037668e-07,
      "loss": 0.9965,
      "step": 10694
    },
    {
      "epoch": 0.7716171855272176,
      "grad_norm": 4.607129186734589,
      "learning_rate": 5.225193941424883e-07,
      "loss": 0.9785,
      "step": 10695
    },
    {
      "epoch": 0.7716893329966451,
      "grad_norm": 4.3021503039944085,
      "learning_rate": 5.22204438560957e-07,
      "loss": 0.9839,
      "step": 10696
    },
    {
      "epoch": 0.7717614804660726,
      "grad_norm": 3.3483958545681483,
      "learning_rate": 5.218895636763693e-07,
      "loss": 0.8093,
      "step": 10697
    },
    {
      "epoch": 0.7718336279355001,
      "grad_norm": 5.444013180595547,
      "learning_rate": 5.21574769505922e-07,
      "loss": 0.9545,
      "step": 10698
    },
    {
      "epoch": 0.7719057754049277,
      "grad_norm": 3.847940161440007,
      "learning_rate": 5.212600560668032e-07,
      "loss": 0.8303,
      "step": 10699
    },
    {
      "epoch": 0.7719779228743552,
      "grad_norm": 6.391353201940347,
      "learning_rate": 5.209454233761985e-07,
      "loss": 0.8686,
      "step": 10700
    },
    {
      "epoch": 0.7720500703437827,
      "grad_norm": 4.006932688136992,
      "learning_rate": 5.206308714512888e-07,
      "loss": 0.9025,
      "step": 10701
    },
    {
      "epoch": 0.7721222178132102,
      "grad_norm": 3.7021517863015716,
      "learning_rate": 5.203164003092511e-07,
      "loss": 0.9401,
      "step": 10702
    },
    {
      "epoch": 0.7721943652826377,
      "grad_norm": 4.872255775590341,
      "learning_rate": 5.200020099672575e-07,
      "loss": 0.9611,
      "step": 10703
    },
    {
      "epoch": 0.7722665127520653,
      "grad_norm": 3.610716995512067,
      "learning_rate": 5.196877004424743e-07,
      "loss": 0.9865,
      "step": 10704
    },
    {
      "epoch": 0.7723386602214928,
      "grad_norm": 3.5652173035723775,
      "learning_rate": 5.193734717520668e-07,
      "loss": 0.9418,
      "step": 10705
    },
    {
      "epoch": 0.7724108076909202,
      "grad_norm": 3.1817104643386136,
      "learning_rate": 5.190593239131926e-07,
      "loss": 0.9329,
      "step": 10706
    },
    {
      "epoch": 0.7724829551603477,
      "grad_norm": 2.8327608839194083,
      "learning_rate": 5.187452569430064e-07,
      "loss": 0.8846,
      "step": 10707
    },
    {
      "epoch": 0.7725551026297752,
      "grad_norm": 5.3132987824341305,
      "learning_rate": 5.184312708586583e-07,
      "loss": 1.0506,
      "step": 10708
    },
    {
      "epoch": 0.7726272500992027,
      "grad_norm": 3.7735783963327227,
      "learning_rate": 5.18117365677294e-07,
      "loss": 0.9146,
      "step": 10709
    },
    {
      "epoch": 0.7726993975686303,
      "grad_norm": 3.699361313003301,
      "learning_rate": 5.178035414160544e-07,
      "loss": 0.9748,
      "step": 10710
    },
    {
      "epoch": 0.7727715450380578,
      "grad_norm": 5.159444847997347,
      "learning_rate": 5.174897980920767e-07,
      "loss": 1.008,
      "step": 10711
    },
    {
      "epoch": 0.7728436925074853,
      "grad_norm": 3.0161728435438424,
      "learning_rate": 5.171761357224935e-07,
      "loss": 1.016,
      "step": 10712
    },
    {
      "epoch": 0.7729158399769128,
      "grad_norm": 5.084491201806726,
      "learning_rate": 5.168625543244318e-07,
      "loss": 0.9881,
      "step": 10713
    },
    {
      "epoch": 0.7729879874463403,
      "grad_norm": 4.6365346639031175,
      "learning_rate": 5.165490539150157e-07,
      "loss": 0.9768,
      "step": 10714
    },
    {
      "epoch": 0.7730601349157679,
      "grad_norm": 9.436295388249398,
      "learning_rate": 5.162356345113641e-07,
      "loss": 0.9697,
      "step": 10715
    },
    {
      "epoch": 0.7731322823851954,
      "grad_norm": 4.2969127306582084,
      "learning_rate": 5.159222961305918e-07,
      "loss": 0.9509,
      "step": 10716
    },
    {
      "epoch": 0.7732044298546229,
      "grad_norm": 6.492824408439529,
      "learning_rate": 5.156090387898091e-07,
      "loss": 0.9005,
      "step": 10717
    },
    {
      "epoch": 0.7732765773240503,
      "grad_norm": 4.812892947317535,
      "learning_rate": 5.152958625061223e-07,
      "loss": 1.0289,
      "step": 10718
    },
    {
      "epoch": 0.7733487247934778,
      "grad_norm": 3.4452464817495025,
      "learning_rate": 5.149827672966309e-07,
      "loss": 1.0159,
      "step": 10719
    },
    {
      "epoch": 0.7734208722629053,
      "grad_norm": 0.8065571961197097,
      "learning_rate": 5.146697531784348e-07,
      "loss": 0.8719,
      "step": 10720
    },
    {
      "epoch": 0.7734930197323329,
      "grad_norm": 3.8813288289049126,
      "learning_rate": 5.143568201686235e-07,
      "loss": 0.86,
      "step": 10721
    },
    {
      "epoch": 0.7735651672017604,
      "grad_norm": 5.342537011681941,
      "learning_rate": 5.140439682842879e-07,
      "loss": 1.0299,
      "step": 10722
    },
    {
      "epoch": 0.7736373146711879,
      "grad_norm": 3.899583055754554,
      "learning_rate": 5.137311975425098e-07,
      "loss": 0.9574,
      "step": 10723
    },
    {
      "epoch": 0.7737094621406154,
      "grad_norm": 4.045460339581994,
      "learning_rate": 5.13418507960369e-07,
      "loss": 0.9761,
      "step": 10724
    },
    {
      "epoch": 0.7737816096100429,
      "grad_norm": 3.975962415582792,
      "learning_rate": 5.131058995549402e-07,
      "loss": 0.976,
      "step": 10725
    },
    {
      "epoch": 0.7738537570794705,
      "grad_norm": 5.181740194400018,
      "learning_rate": 5.12793372343294e-07,
      "loss": 0.9071,
      "step": 10726
    },
    {
      "epoch": 0.773925904548898,
      "grad_norm": 4.768021978199632,
      "learning_rate": 5.124809263424969e-07,
      "loss": 1.0438,
      "step": 10727
    },
    {
      "epoch": 0.7739980520183255,
      "grad_norm": 4.392523847069053,
      "learning_rate": 5.121685615696094e-07,
      "loss": 0.905,
      "step": 10728
    },
    {
      "epoch": 0.774070199487753,
      "grad_norm": 0.8474655288392565,
      "learning_rate": 5.118562780416889e-07,
      "loss": 0.8631,
      "step": 10729
    },
    {
      "epoch": 0.7741423469571804,
      "grad_norm": 3.531835051826881,
      "learning_rate": 5.11544075775788e-07,
      "loss": 0.9518,
      "step": 10730
    },
    {
      "epoch": 0.774214494426608,
      "grad_norm": 3.8040714587650837,
      "learning_rate": 5.11231954788955e-07,
      "loss": 0.978,
      "step": 10731
    },
    {
      "epoch": 0.7742866418960355,
      "grad_norm": 0.7692715356643215,
      "learning_rate": 5.10919915098234e-07,
      "loss": 0.8715,
      "step": 10732
    },
    {
      "epoch": 0.774358789365463,
      "grad_norm": 4.8442797955600065,
      "learning_rate": 5.106079567206638e-07,
      "loss": 1.0149,
      "step": 10733
    },
    {
      "epoch": 0.7744309368348905,
      "grad_norm": 3.5159788504649034,
      "learning_rate": 5.102960796732796e-07,
      "loss": 1.0155,
      "step": 10734
    },
    {
      "epoch": 0.774503084304318,
      "grad_norm": 21.411554144407788,
      "learning_rate": 5.099842839731124e-07,
      "loss": 0.9384,
      "step": 10735
    },
    {
      "epoch": 0.7745752317737455,
      "grad_norm": 2.9522929962229147,
      "learning_rate": 5.096725696371871e-07,
      "loss": 1.0648,
      "step": 10736
    },
    {
      "epoch": 0.7746473792431731,
      "grad_norm": 7.3965230557478145,
      "learning_rate": 5.093609366825256e-07,
      "loss": 0.8595,
      "step": 10737
    },
    {
      "epoch": 0.7747195267126006,
      "grad_norm": 5.20415406461014,
      "learning_rate": 5.090493851261452e-07,
      "loss": 1.0197,
      "step": 10738
    },
    {
      "epoch": 0.7747916741820281,
      "grad_norm": 4.024201374316376,
      "learning_rate": 5.087379149850588e-07,
      "loss": 0.8983,
      "step": 10739
    },
    {
      "epoch": 0.7748638216514556,
      "grad_norm": 5.3882021559277575,
      "learning_rate": 5.084265262762741e-07,
      "loss": 0.9737,
      "step": 10740
    },
    {
      "epoch": 0.7749359691208831,
      "grad_norm": 4.827586122334297,
      "learning_rate": 5.081152190167954e-07,
      "loss": 0.9156,
      "step": 10741
    },
    {
      "epoch": 0.7750081165903105,
      "grad_norm": 3.588726081409688,
      "learning_rate": 5.078039932236224e-07,
      "loss": 0.9206,
      "step": 10742
    },
    {
      "epoch": 0.7750802640597381,
      "grad_norm": 14.271113230983858,
      "learning_rate": 5.074928489137482e-07,
      "loss": 0.9883,
      "step": 10743
    },
    {
      "epoch": 0.7751524115291656,
      "grad_norm": 5.223218455330438,
      "learning_rate": 5.071817861041657e-07,
      "loss": 1.076,
      "step": 10744
    },
    {
      "epoch": 0.7752245589985931,
      "grad_norm": 3.7114087899743446,
      "learning_rate": 5.068708048118589e-07,
      "loss": 0.9867,
      "step": 10745
    },
    {
      "epoch": 0.7752967064680206,
      "grad_norm": 7.480351459589608,
      "learning_rate": 5.065599050538101e-07,
      "loss": 0.9168,
      "step": 10746
    },
    {
      "epoch": 0.7753688539374481,
      "grad_norm": 0.720842797615969,
      "learning_rate": 5.062490868469964e-07,
      "loss": 0.8181,
      "step": 10747
    },
    {
      "epoch": 0.7754410014068757,
      "grad_norm": 3.2433950358601984,
      "learning_rate": 5.059383502083903e-07,
      "loss": 0.9128,
      "step": 10748
    },
    {
      "epoch": 0.7755131488763032,
      "grad_norm": 2.930201940510314,
      "learning_rate": 5.056276951549601e-07,
      "loss": 1.0044,
      "step": 10749
    },
    {
      "epoch": 0.7755852963457307,
      "grad_norm": 3.9055734277361216,
      "learning_rate": 5.053171217036702e-07,
      "loss": 0.8632,
      "step": 10750
    },
    {
      "epoch": 0.7756574438151582,
      "grad_norm": 9.347805202140659,
      "learning_rate": 5.050066298714784e-07,
      "loss": 0.9113,
      "step": 10751
    },
    {
      "epoch": 0.7757295912845857,
      "grad_norm": 4.01199711295392,
      "learning_rate": 5.046962196753404e-07,
      "loss": 0.8816,
      "step": 10752
    },
    {
      "epoch": 0.7758017387540131,
      "grad_norm": 5.205126678068576,
      "learning_rate": 5.043858911322063e-07,
      "loss": 1.0639,
      "step": 10753
    },
    {
      "epoch": 0.7758738862234407,
      "grad_norm": 5.029147638865928,
      "learning_rate": 5.040756442590222e-07,
      "loss": 0.8842,
      "step": 10754
    },
    {
      "epoch": 0.7759460336928682,
      "grad_norm": 2.886543485408085,
      "learning_rate": 5.037654790727295e-07,
      "loss": 0.9159,
      "step": 10755
    },
    {
      "epoch": 0.7760181811622957,
      "grad_norm": 3.185164250281221,
      "learning_rate": 5.03455395590265e-07,
      "loss": 0.893,
      "step": 10756
    },
    {
      "epoch": 0.7760903286317232,
      "grad_norm": 6.850520995437905,
      "learning_rate": 5.031453938285619e-07,
      "loss": 0.9515,
      "step": 10757
    },
    {
      "epoch": 0.7761624761011507,
      "grad_norm": 3.5147246839812376,
      "learning_rate": 5.028354738045466e-07,
      "loss": 0.8643,
      "step": 10758
    },
    {
      "epoch": 0.7762346235705783,
      "grad_norm": 3.4978284911416657,
      "learning_rate": 5.02525635535145e-07,
      "loss": 0.9586,
      "step": 10759
    },
    {
      "epoch": 0.7763067710400058,
      "grad_norm": 3.474582476884363,
      "learning_rate": 5.022158790372744e-07,
      "loss": 0.9717,
      "step": 10760
    },
    {
      "epoch": 0.7763789185094333,
      "grad_norm": 4.284299335326342,
      "learning_rate": 5.019062043278499e-07,
      "loss": 0.8355,
      "step": 10761
    },
    {
      "epoch": 0.7764510659788608,
      "grad_norm": 0.8652453281923452,
      "learning_rate": 5.015966114237818e-07,
      "loss": 0.9336,
      "step": 10762
    },
    {
      "epoch": 0.7765232134482883,
      "grad_norm": 11.596338522942139,
      "learning_rate": 5.012871003419759e-07,
      "loss": 0.9407,
      "step": 10763
    },
    {
      "epoch": 0.7765953609177159,
      "grad_norm": 6.7270785498186125,
      "learning_rate": 5.009776710993334e-07,
      "loss": 0.9678,
      "step": 10764
    },
    {
      "epoch": 0.7766675083871433,
      "grad_norm": 2.893100317905152,
      "learning_rate": 5.006683237127511e-07,
      "loss": 0.9516,
      "step": 10765
    },
    {
      "epoch": 0.7767396558565708,
      "grad_norm": 3.395278760564216,
      "learning_rate": 5.003590581991219e-07,
      "loss": 1.0522,
      "step": 10766
    },
    {
      "epoch": 0.7768118033259983,
      "grad_norm": 0.7578646455108077,
      "learning_rate": 5.000498745753326e-07,
      "loss": 0.8344,
      "step": 10767
    },
    {
      "epoch": 0.7768839507954258,
      "grad_norm": 4.664797317664785,
      "learning_rate": 4.997407728582668e-07,
      "loss": 0.8983,
      "step": 10768
    },
    {
      "epoch": 0.7769560982648533,
      "grad_norm": 2.9423411632569216,
      "learning_rate": 4.994317530648038e-07,
      "loss": 0.9908,
      "step": 10769
    },
    {
      "epoch": 0.7770282457342809,
      "grad_norm": 3.7088169825476935,
      "learning_rate": 4.991228152118176e-07,
      "loss": 0.8903,
      "step": 10770
    },
    {
      "epoch": 0.7771003932037084,
      "grad_norm": 4.461110661180557,
      "learning_rate": 4.988139593161787e-07,
      "loss": 1.0368,
      "step": 10771
    },
    {
      "epoch": 0.7771725406731359,
      "grad_norm": 4.480248017598393,
      "learning_rate": 4.985051853947526e-07,
      "loss": 1.0433,
      "step": 10772
    },
    {
      "epoch": 0.7772446881425634,
      "grad_norm": 3.6504495370028867,
      "learning_rate": 4.98196493464399e-07,
      "loss": 1.0736,
      "step": 10773
    },
    {
      "epoch": 0.7773168356119909,
      "grad_norm": 4.239329237513076,
      "learning_rate": 4.978878835419766e-07,
      "loss": 0.8613,
      "step": 10774
    },
    {
      "epoch": 0.7773889830814185,
      "grad_norm": 0.7889930770096483,
      "learning_rate": 4.975793556443355e-07,
      "loss": 0.8196,
      "step": 10775
    },
    {
      "epoch": 0.777461130550846,
      "grad_norm": 3.281626216573907,
      "learning_rate": 4.972709097883241e-07,
      "loss": 1.0492,
      "step": 10776
    },
    {
      "epoch": 0.7775332780202734,
      "grad_norm": 2.751447903327952,
      "learning_rate": 4.969625459907856e-07,
      "loss": 0.8764,
      "step": 10777
    },
    {
      "epoch": 0.7776054254897009,
      "grad_norm": 3.0342574783337386,
      "learning_rate": 4.966542642685581e-07,
      "loss": 1.0071,
      "step": 10778
    },
    {
      "epoch": 0.7776775729591284,
      "grad_norm": 3.4673893639833477,
      "learning_rate": 4.963460646384763e-07,
      "loss": 1.0207,
      "step": 10779
    },
    {
      "epoch": 0.777749720428556,
      "grad_norm": 3.176070872258349,
      "learning_rate": 4.960379471173692e-07,
      "loss": 0.9839,
      "step": 10780
    },
    {
      "epoch": 0.7778218678979835,
      "grad_norm": 4.820934286464521,
      "learning_rate": 4.957299117220632e-07,
      "loss": 1.0282,
      "step": 10781
    },
    {
      "epoch": 0.777894015367411,
      "grad_norm": 4.3449158784824435,
      "learning_rate": 4.954219584693768e-07,
      "loss": 0.9965,
      "step": 10782
    },
    {
      "epoch": 0.7779661628368385,
      "grad_norm": 3.693441873211026,
      "learning_rate": 4.951140873761288e-07,
      "loss": 0.9253,
      "step": 10783
    },
    {
      "epoch": 0.778038310306266,
      "grad_norm": 9.711928546447847,
      "learning_rate": 4.94806298459129e-07,
      "loss": 0.9312,
      "step": 10784
    },
    {
      "epoch": 0.7781104577756935,
      "grad_norm": 3.161325886616114,
      "learning_rate": 4.944985917351852e-07,
      "loss": 0.8949,
      "step": 10785
    },
    {
      "epoch": 0.7781826052451211,
      "grad_norm": 4.488285074655681,
      "learning_rate": 4.941909672211e-07,
      "loss": 0.9433,
      "step": 10786
    },
    {
      "epoch": 0.7782547527145486,
      "grad_norm": 7.405808834022384,
      "learning_rate": 4.938834249336719e-07,
      "loss": 1.0088,
      "step": 10787
    },
    {
      "epoch": 0.7783269001839761,
      "grad_norm": 5.186050396340061,
      "learning_rate": 4.935759648896945e-07,
      "loss": 0.8442,
      "step": 10788
    },
    {
      "epoch": 0.7783990476534035,
      "grad_norm": 7.724545422854739,
      "learning_rate": 4.932685871059577e-07,
      "loss": 0.9695,
      "step": 10789
    },
    {
      "epoch": 0.778471195122831,
      "grad_norm": 9.969883008325978,
      "learning_rate": 4.92961291599245e-07,
      "loss": 0.9268,
      "step": 10790
    },
    {
      "epoch": 0.7785433425922585,
      "grad_norm": 5.061357534090101,
      "learning_rate": 4.92654078386337e-07,
      "loss": 1.0292,
      "step": 10791
    },
    {
      "epoch": 0.7786154900616861,
      "grad_norm": 4.791694154522323,
      "learning_rate": 4.923469474840101e-07,
      "loss": 1.0388,
      "step": 10792
    },
    {
      "epoch": 0.7786876375311136,
      "grad_norm": 3.82519693303993,
      "learning_rate": 4.920398989090353e-07,
      "loss": 0.9236,
      "step": 10793
    },
    {
      "epoch": 0.7787597850005411,
      "grad_norm": 5.93262353716224,
      "learning_rate": 4.917329326781792e-07,
      "loss": 0.9121,
      "step": 10794
    },
    {
      "epoch": 0.7788319324699686,
      "grad_norm": 3.7726089484053196,
      "learning_rate": 4.914260488082042e-07,
      "loss": 0.9721,
      "step": 10795
    },
    {
      "epoch": 0.7789040799393961,
      "grad_norm": 4.805922534040498,
      "learning_rate": 4.911192473158687e-07,
      "loss": 0.9144,
      "step": 10796
    },
    {
      "epoch": 0.7789762274088237,
      "grad_norm": 14.249253102856226,
      "learning_rate": 4.908125282179245e-07,
      "loss": 0.9915,
      "step": 10797
    },
    {
      "epoch": 0.7790483748782512,
      "grad_norm": 0.8141811657775803,
      "learning_rate": 4.905058915311223e-07,
      "loss": 0.8854,
      "step": 10798
    },
    {
      "epoch": 0.7791205223476787,
      "grad_norm": 4.550993532133022,
      "learning_rate": 4.901993372722046e-07,
      "loss": 1.0681,
      "step": 10799
    },
    {
      "epoch": 0.7791926698171062,
      "grad_norm": 5.821218862943822,
      "learning_rate": 4.898928654579124e-07,
      "loss": 0.9016,
      "step": 10800
    },
    {
      "epoch": 0.7792648172865336,
      "grad_norm": 8.572357072948654,
      "learning_rate": 4.895864761049804e-07,
      "loss": 1.047,
      "step": 10801
    },
    {
      "epoch": 0.7793369647559611,
      "grad_norm": 8.008358879988059,
      "learning_rate": 4.892801692301396e-07,
      "loss": 0.9641,
      "step": 10802
    },
    {
      "epoch": 0.7794091122253887,
      "grad_norm": 7.242073461463936,
      "learning_rate": 4.889739448501162e-07,
      "loss": 0.9959,
      "step": 10803
    },
    {
      "epoch": 0.7794812596948162,
      "grad_norm": 4.6657916111462345,
      "learning_rate": 4.886678029816323e-07,
      "loss": 0.9366,
      "step": 10804
    },
    {
      "epoch": 0.7795534071642437,
      "grad_norm": 4.276134469825326,
      "learning_rate": 4.883617436414051e-07,
      "loss": 1.0742,
      "step": 10805
    },
    {
      "epoch": 0.7796255546336712,
      "grad_norm": 3.1669988123388926,
      "learning_rate": 4.880557668461469e-07,
      "loss": 0.8154,
      "step": 10806
    },
    {
      "epoch": 0.7796977021030987,
      "grad_norm": 3.2827913569587017,
      "learning_rate": 4.877498726125661e-07,
      "loss": 0.9461,
      "step": 10807
    },
    {
      "epoch": 0.7797698495725263,
      "grad_norm": 4.9698748005193165,
      "learning_rate": 4.874440609573667e-07,
      "loss": 1.0137,
      "step": 10808
    },
    {
      "epoch": 0.7798419970419538,
      "grad_norm": 0.6835520922365608,
      "learning_rate": 4.871383318972477e-07,
      "loss": 0.7985,
      "step": 10809
    },
    {
      "epoch": 0.7799141445113813,
      "grad_norm": 3.934919086676239,
      "learning_rate": 4.868326854489042e-07,
      "loss": 0.9704,
      "step": 10810
    },
    {
      "epoch": 0.7799862919808088,
      "grad_norm": 4.8032941084845655,
      "learning_rate": 4.865271216290267e-07,
      "loss": 0.9153,
      "step": 10811
    },
    {
      "epoch": 0.7800584394502362,
      "grad_norm": 7.758990825432802,
      "learning_rate": 4.862216404542992e-07,
      "loss": 0.8762,
      "step": 10812
    },
    {
      "epoch": 0.7801305869196637,
      "grad_norm": 3.8247718294192583,
      "learning_rate": 4.859162419414051e-07,
      "loss": 0.9268,
      "step": 10813
    },
    {
      "epoch": 0.7802027343890913,
      "grad_norm": 7.751883001147306,
      "learning_rate": 4.856109261070198e-07,
      "loss": 1.0007,
      "step": 10814
    },
    {
      "epoch": 0.7802748818585188,
      "grad_norm": 7.471645416472344,
      "learning_rate": 4.853056929678156e-07,
      "loss": 1.0044,
      "step": 10815
    },
    {
      "epoch": 0.7803470293279463,
      "grad_norm": 6.171634563156858,
      "learning_rate": 4.850005425404601e-07,
      "loss": 0.9664,
      "step": 10816
    },
    {
      "epoch": 0.7804191767973738,
      "grad_norm": 8.187160339298767,
      "learning_rate": 4.84695474841617e-07,
      "loss": 0.901,
      "step": 10817
    },
    {
      "epoch": 0.7804913242668013,
      "grad_norm": 5.595407080683189,
      "learning_rate": 4.843904898879445e-07,
      "loss": 0.9422,
      "step": 10818
    },
    {
      "epoch": 0.7805634717362289,
      "grad_norm": 3.4284302790288184,
      "learning_rate": 4.840855876960965e-07,
      "loss": 0.9547,
      "step": 10819
    },
    {
      "epoch": 0.7806356192056564,
      "grad_norm": 2.949903234414178,
      "learning_rate": 4.837807682827235e-07,
      "loss": 1.0231,
      "step": 10820
    },
    {
      "epoch": 0.7807077666750839,
      "grad_norm": 5.886592061198967,
      "learning_rate": 4.834760316644689e-07,
      "loss": 0.9345,
      "step": 10821
    },
    {
      "epoch": 0.7807799141445114,
      "grad_norm": 2.893913584095827,
      "learning_rate": 4.831713778579754e-07,
      "loss": 0.9627,
      "step": 10822
    },
    {
      "epoch": 0.7808520616139389,
      "grad_norm": 5.49830965596482,
      "learning_rate": 4.828668068798771e-07,
      "loss": 1.0301,
      "step": 10823
    },
    {
      "epoch": 0.7809242090833663,
      "grad_norm": 5.75413696075145,
      "learning_rate": 4.825623187468062e-07,
      "loss": 0.8289,
      "step": 10824
    },
    {
      "epoch": 0.7809963565527939,
      "grad_norm": 4.285577299562726,
      "learning_rate": 4.822579134753899e-07,
      "loss": 0.9714,
      "step": 10825
    },
    {
      "epoch": 0.7810685040222214,
      "grad_norm": 3.9327760242321075,
      "learning_rate": 4.819535910822503e-07,
      "loss": 0.986,
      "step": 10826
    },
    {
      "epoch": 0.7811406514916489,
      "grad_norm": 5.636797930187113,
      "learning_rate": 4.816493515840055e-07,
      "loss": 1.0315,
      "step": 10827
    },
    {
      "epoch": 0.7812127989610764,
      "grad_norm": 2.776505012324951,
      "learning_rate": 4.813451949972696e-07,
      "loss": 0.946,
      "step": 10828
    },
    {
      "epoch": 0.781284946430504,
      "grad_norm": 3.927831374478218,
      "learning_rate": 4.8104112133865e-07,
      "loss": 0.9646,
      "step": 10829
    },
    {
      "epoch": 0.7813570938999315,
      "grad_norm": 2.4209084827973317,
      "learning_rate": 4.807371306247518e-07,
      "loss": 0.9154,
      "step": 10830
    },
    {
      "epoch": 0.781429241369359,
      "grad_norm": 4.721036849498866,
      "learning_rate": 4.804332228721748e-07,
      "loss": 1.028,
      "step": 10831
    },
    {
      "epoch": 0.7815013888387865,
      "grad_norm": 2.8176341349668297,
      "learning_rate": 4.801293980975141e-07,
      "loss": 1.0574,
      "step": 10832
    },
    {
      "epoch": 0.781573536308214,
      "grad_norm": 3.6244449848299647,
      "learning_rate": 4.798256563173608e-07,
      "loss": 1.0149,
      "step": 10833
    },
    {
      "epoch": 0.7816456837776415,
      "grad_norm": 6.888916357388988,
      "learning_rate": 4.795219975483009e-07,
      "loss": 1.0534,
      "step": 10834
    },
    {
      "epoch": 0.7817178312470691,
      "grad_norm": 4.926748230757531,
      "learning_rate": 4.792184218069166e-07,
      "loss": 0.9342,
      "step": 10835
    },
    {
      "epoch": 0.7817899787164965,
      "grad_norm": 0.727393382803664,
      "learning_rate": 4.789149291097833e-07,
      "loss": 0.8382,
      "step": 10836
    },
    {
      "epoch": 0.781862126185924,
      "grad_norm": 3.2823995847080236,
      "learning_rate": 4.786115194734763e-07,
      "loss": 0.9868,
      "step": 10837
    },
    {
      "epoch": 0.7819342736553515,
      "grad_norm": 3.224188053326054,
      "learning_rate": 4.783081929145614e-07,
      "loss": 1.0292,
      "step": 10838
    },
    {
      "epoch": 0.782006421124779,
      "grad_norm": 0.7843514108340328,
      "learning_rate": 4.78004949449603e-07,
      "loss": 0.8643,
      "step": 10839
    },
    {
      "epoch": 0.7820785685942065,
      "grad_norm": 4.215433583250934,
      "learning_rate": 4.777017890951603e-07,
      "loss": 0.9156,
      "step": 10840
    },
    {
      "epoch": 0.7821507160636341,
      "grad_norm": 4.154532859423463,
      "learning_rate": 4.773987118677873e-07,
      "loss": 0.955,
      "step": 10841
    },
    {
      "epoch": 0.7822228635330616,
      "grad_norm": 5.074968783277895,
      "learning_rate": 4.770957177840342e-07,
      "loss": 0.9095,
      "step": 10842
    },
    {
      "epoch": 0.7822950110024891,
      "grad_norm": 4.212783323998629,
      "learning_rate": 4.7679280686044655e-07,
      "loss": 1.0141,
      "step": 10843
    },
    {
      "epoch": 0.7823671584719166,
      "grad_norm": 3.909278611078162,
      "learning_rate": 4.764899791135655e-07,
      "loss": 0.9102,
      "step": 10844
    },
    {
      "epoch": 0.7824393059413441,
      "grad_norm": 2.6216752932197025,
      "learning_rate": 4.7618723455992605e-07,
      "loss": 1.0471,
      "step": 10845
    },
    {
      "epoch": 0.7825114534107717,
      "grad_norm": 3.2212402192020213,
      "learning_rate": 4.75884573216061e-07,
      "loss": 0.9199,
      "step": 10846
    },
    {
      "epoch": 0.7825836008801992,
      "grad_norm": 3.795465027032042,
      "learning_rate": 4.755819950984972e-07,
      "loss": 1.0507,
      "step": 10847
    },
    {
      "epoch": 0.7826557483496266,
      "grad_norm": 4.326813316583292,
      "learning_rate": 4.7527950022375793e-07,
      "loss": 0.929,
      "step": 10848
    },
    {
      "epoch": 0.7827278958190541,
      "grad_norm": 5.013755378577601,
      "learning_rate": 4.7497708860835993e-07,
      "loss": 1.0546,
      "step": 10849
    },
    {
      "epoch": 0.7828000432884816,
      "grad_norm": 3.93150260368454,
      "learning_rate": 4.7467476026881814e-07,
      "loss": 0.8319,
      "step": 10850
    },
    {
      "epoch": 0.7828721907579091,
      "grad_norm": 9.056199823823645,
      "learning_rate": 4.743725152216418e-07,
      "loss": 0.9482,
      "step": 10851
    },
    {
      "epoch": 0.7829443382273367,
      "grad_norm": 3.499869889157,
      "learning_rate": 4.740703534833339e-07,
      "loss": 1.0573,
      "step": 10852
    },
    {
      "epoch": 0.7830164856967642,
      "grad_norm": 3.574919484972444,
      "learning_rate": 4.737682750703953e-07,
      "loss": 0.8991,
      "step": 10853
    },
    {
      "epoch": 0.7830886331661917,
      "grad_norm": 6.851548302525912,
      "learning_rate": 4.7346627999932075e-07,
      "loss": 0.985,
      "step": 10854
    },
    {
      "epoch": 0.7831607806356192,
      "grad_norm": 3.616092260660652,
      "learning_rate": 4.731643682866029e-07,
      "loss": 1.0036,
      "step": 10855
    },
    {
      "epoch": 0.7832329281050467,
      "grad_norm": 6.617764822716149,
      "learning_rate": 4.7286253994872604e-07,
      "loss": 0.9827,
      "step": 10856
    },
    {
      "epoch": 0.7833050755744743,
      "grad_norm": 0.8455817860164704,
      "learning_rate": 4.725607950021726e-07,
      "loss": 0.9314,
      "step": 10857
    },
    {
      "epoch": 0.7833772230439018,
      "grad_norm": 3.9457041933114705,
      "learning_rate": 4.722591334634203e-07,
      "loss": 0.8237,
      "step": 10858
    },
    {
      "epoch": 0.7834493705133292,
      "grad_norm": 3.3431539984768484,
      "learning_rate": 4.7195755534894075e-07,
      "loss": 0.9879,
      "step": 10859
    },
    {
      "epoch": 0.7835215179827567,
      "grad_norm": 4.201801304327682,
      "learning_rate": 4.716560606752018e-07,
      "loss": 0.9542,
      "step": 10860
    },
    {
      "epoch": 0.7835936654521842,
      "grad_norm": 3.0575911436430756,
      "learning_rate": 4.713546494586689e-07,
      "loss": 1.067,
      "step": 10861
    },
    {
      "epoch": 0.7836658129216117,
      "grad_norm": 3.9828691815981867,
      "learning_rate": 4.71053321715799e-07,
      "loss": 0.9821,
      "step": 10862
    },
    {
      "epoch": 0.7837379603910393,
      "grad_norm": 3.7364692721150488,
      "learning_rate": 4.7075207746304736e-07,
      "loss": 0.9789,
      "step": 10863
    },
    {
      "epoch": 0.7838101078604668,
      "grad_norm": 6.903453692325319,
      "learning_rate": 4.7045091671686354e-07,
      "loss": 1.0087,
      "step": 10864
    },
    {
      "epoch": 0.7838822553298943,
      "grad_norm": 3.548308364352675,
      "learning_rate": 4.701498394936938e-07,
      "loss": 1.018,
      "step": 10865
    },
    {
      "epoch": 0.7839544027993218,
      "grad_norm": 5.529059369623135,
      "learning_rate": 4.6984884580997677e-07,
      "loss": 0.9795,
      "step": 10866
    },
    {
      "epoch": 0.7840265502687493,
      "grad_norm": 9.245072160504193,
      "learning_rate": 4.6954793568215035e-07,
      "loss": 0.897,
      "step": 10867
    },
    {
      "epoch": 0.7840986977381769,
      "grad_norm": 2.7837576544613167,
      "learning_rate": 4.6924710912664613e-07,
      "loss": 0.9184,
      "step": 10868
    },
    {
      "epoch": 0.7841708452076044,
      "grad_norm": 3.934950593592152,
      "learning_rate": 4.6894636615989004e-07,
      "loss": 0.9049,
      "step": 10869
    },
    {
      "epoch": 0.7842429926770319,
      "grad_norm": 3.556754046018936,
      "learning_rate": 4.686457067983054e-07,
      "loss": 0.9887,
      "step": 10870
    },
    {
      "epoch": 0.7843151401464593,
      "grad_norm": 7.264759168000201,
      "learning_rate": 4.6834513105831017e-07,
      "loss": 0.9858,
      "step": 10871
    },
    {
      "epoch": 0.7843872876158868,
      "grad_norm": 2.6540927878676976,
      "learning_rate": 4.6804463895631686e-07,
      "loss": 0.9435,
      "step": 10872
    },
    {
      "epoch": 0.7844594350853143,
      "grad_norm": 7.456610738208873,
      "learning_rate": 4.67744230508734e-07,
      "loss": 0.8704,
      "step": 10873
    },
    {
      "epoch": 0.7845315825547419,
      "grad_norm": 4.435693722456328,
      "learning_rate": 4.674439057319672e-07,
      "loss": 0.8739,
      "step": 10874
    },
    {
      "epoch": 0.7846037300241694,
      "grad_norm": 2.908422171065694,
      "learning_rate": 4.671436646424158e-07,
      "loss": 0.9679,
      "step": 10875
    },
    {
      "epoch": 0.7846758774935969,
      "grad_norm": 3.284003274281736,
      "learning_rate": 4.6684350725647404e-07,
      "loss": 0.9406,
      "step": 10876
    },
    {
      "epoch": 0.7847480249630244,
      "grad_norm": 5.611137432040851,
      "learning_rate": 4.6654343359053274e-07,
      "loss": 0.9281,
      "step": 10877
    },
    {
      "epoch": 0.784820172432452,
      "grad_norm": 4.178740948602503,
      "learning_rate": 4.662434436609786e-07,
      "loss": 1.0561,
      "step": 10878
    },
    {
      "epoch": 0.7848923199018795,
      "grad_norm": 7.207357822038694,
      "learning_rate": 4.659435374841909e-07,
      "loss": 0.8663,
      "step": 10879
    },
    {
      "epoch": 0.784964467371307,
      "grad_norm": 7.213522863994749,
      "learning_rate": 4.6564371507654823e-07,
      "loss": 0.9088,
      "step": 10880
    },
    {
      "epoch": 0.7850366148407345,
      "grad_norm": 3.360815262888189,
      "learning_rate": 4.6534397645442246e-07,
      "loss": 0.9023,
      "step": 10881
    },
    {
      "epoch": 0.785108762310162,
      "grad_norm": 3.864124541627193,
      "learning_rate": 4.650443216341815e-07,
      "loss": 0.9762,
      "step": 10882
    },
    {
      "epoch": 0.7851809097795894,
      "grad_norm": 3.007413604839908,
      "learning_rate": 4.6474475063218753e-07,
      "loss": 0.9575,
      "step": 10883
    },
    {
      "epoch": 0.785253057249017,
      "grad_norm": 2.168853524235471,
      "learning_rate": 4.644452634647986e-07,
      "loss": 1.0151,
      "step": 10884
    },
    {
      "epoch": 0.7853252047184445,
      "grad_norm": 4.451820369815311,
      "learning_rate": 4.6414586014837074e-07,
      "loss": 0.9845,
      "step": 10885
    },
    {
      "epoch": 0.785397352187872,
      "grad_norm": 3.7021694318598204,
      "learning_rate": 4.638465406992511e-07,
      "loss": 0.9339,
      "step": 10886
    },
    {
      "epoch": 0.7854694996572995,
      "grad_norm": 6.5352362190639175,
      "learning_rate": 4.635473051337855e-07,
      "loss": 1.0298,
      "step": 10887
    },
    {
      "epoch": 0.785541647126727,
      "grad_norm": 4.079572281950904,
      "learning_rate": 4.6324815346831413e-07,
      "loss": 0.9146,
      "step": 10888
    },
    {
      "epoch": 0.7856137945961545,
      "grad_norm": 35.289364309922576,
      "learning_rate": 4.62949085719172e-07,
      "loss": 0.8672,
      "step": 10889
    },
    {
      "epoch": 0.7856859420655821,
      "grad_norm": 10.26960851292731,
      "learning_rate": 4.626501019026894e-07,
      "loss": 0.8867,
      "step": 10890
    },
    {
      "epoch": 0.7857580895350096,
      "grad_norm": 12.420636116483163,
      "learning_rate": 4.623512020351945e-07,
      "loss": 0.9458,
      "step": 10891
    },
    {
      "epoch": 0.7858302370044371,
      "grad_norm": 3.3097901686342874,
      "learning_rate": 4.620523861330088e-07,
      "loss": 0.9055,
      "step": 10892
    },
    {
      "epoch": 0.7859023844738646,
      "grad_norm": 6.154038395079708,
      "learning_rate": 4.617536542124483e-07,
      "loss": 0.9941,
      "step": 10893
    },
    {
      "epoch": 0.7859745319432921,
      "grad_norm": 1.0477707290401082,
      "learning_rate": 4.6145500628982635e-07,
      "loss": 0.9247,
      "step": 10894
    },
    {
      "epoch": 0.7860466794127196,
      "grad_norm": 4.387823004937215,
      "learning_rate": 4.611564423814518e-07,
      "loss": 1.0017,
      "step": 10895
    },
    {
      "epoch": 0.7861188268821471,
      "grad_norm": 3.043930273964295,
      "learning_rate": 4.608579625036266e-07,
      "loss": 0.9567,
      "step": 10896
    },
    {
      "epoch": 0.7861909743515746,
      "grad_norm": 4.408660445789574,
      "learning_rate": 4.6055956667264983e-07,
      "loss": 0.904,
      "step": 10897
    },
    {
      "epoch": 0.7862631218210021,
      "grad_norm": 5.492625754874027,
      "learning_rate": 4.602612549048168e-07,
      "loss": 0.9954,
      "step": 10898
    },
    {
      "epoch": 0.7863352692904296,
      "grad_norm": 5.656869085163602,
      "learning_rate": 4.599630272164172e-07,
      "loss": 1.0244,
      "step": 10899
    },
    {
      "epoch": 0.7864074167598571,
      "grad_norm": 3.3730006477090693,
      "learning_rate": 4.596648836237353e-07,
      "loss": 0.9171,
      "step": 10900
    },
    {
      "epoch": 0.7864795642292847,
      "grad_norm": 5.861391580067212,
      "learning_rate": 4.5936682414305194e-07,
      "loss": 0.9387,
      "step": 10901
    },
    {
      "epoch": 0.7865517116987122,
      "grad_norm": 4.399135175157345,
      "learning_rate": 4.590688487906436e-07,
      "loss": 0.8191,
      "step": 10902
    },
    {
      "epoch": 0.7866238591681397,
      "grad_norm": 6.572293195057329,
      "learning_rate": 4.5877095758278006e-07,
      "loss": 1.0187,
      "step": 10903
    },
    {
      "epoch": 0.7866960066375672,
      "grad_norm": 4.351227177245294,
      "learning_rate": 4.5847315053572954e-07,
      "loss": 0.8945,
      "step": 10904
    },
    {
      "epoch": 0.7867681541069947,
      "grad_norm": 7.244072957775765,
      "learning_rate": 4.5817542766575437e-07,
      "loss": 0.9922,
      "step": 10905
    },
    {
      "epoch": 0.7868403015764222,
      "grad_norm": 4.629132254833432,
      "learning_rate": 4.57877788989111e-07,
      "loss": 1.0634,
      "step": 10906
    },
    {
      "epoch": 0.7869124490458497,
      "grad_norm": 0.8299593851769154,
      "learning_rate": 4.575802345220525e-07,
      "loss": 0.8571,
      "step": 10907
    },
    {
      "epoch": 0.7869845965152772,
      "grad_norm": 4.714596366555268,
      "learning_rate": 4.5728276428082726e-07,
      "loss": 1.0317,
      "step": 10908
    },
    {
      "epoch": 0.7870567439847047,
      "grad_norm": 0.6984332270139517,
      "learning_rate": 4.5698537828168036e-07,
      "loss": 0.8409,
      "step": 10909
    },
    {
      "epoch": 0.7871288914541322,
      "grad_norm": 10.867777914631759,
      "learning_rate": 4.566880765408494e-07,
      "loss": 0.9886,
      "step": 10910
    },
    {
      "epoch": 0.7872010389235597,
      "grad_norm": 8.512299558545509,
      "learning_rate": 4.563908590745693e-07,
      "loss": 0.973,
      "step": 10911
    },
    {
      "epoch": 0.7872731863929873,
      "grad_norm": 5.79689869014701,
      "learning_rate": 4.560937258990709e-07,
      "loss": 0.9238,
      "step": 10912
    },
    {
      "epoch": 0.7873453338624148,
      "grad_norm": 5.250020526664206,
      "learning_rate": 4.557966770305781e-07,
      "loss": 0.8926,
      "step": 10913
    },
    {
      "epoch": 0.7874174813318423,
      "grad_norm": 3.3472868850251873,
      "learning_rate": 4.554997124853117e-07,
      "loss": 0.9937,
      "step": 10914
    },
    {
      "epoch": 0.7874896288012698,
      "grad_norm": 0.7072897591281913,
      "learning_rate": 4.5520283227948964e-07,
      "loss": 0.833,
      "step": 10915
    },
    {
      "epoch": 0.7875617762706973,
      "grad_norm": 3.7965576290407395,
      "learning_rate": 4.549060364293218e-07,
      "loss": 0.9495,
      "step": 10916
    },
    {
      "epoch": 0.7876339237401249,
      "grad_norm": 2.6774296715882695,
      "learning_rate": 4.5460932495101545e-07,
      "loss": 0.9214,
      "step": 10917
    },
    {
      "epoch": 0.7877060712095523,
      "grad_norm": 2.6968364443183357,
      "learning_rate": 4.543126978607732e-07,
      "loss": 1.016,
      "step": 10918
    },
    {
      "epoch": 0.7877782186789798,
      "grad_norm": 7.115177544893825,
      "learning_rate": 4.54016155174793e-07,
      "loss": 0.8755,
      "step": 10919
    },
    {
      "epoch": 0.7878503661484073,
      "grad_norm": 4.54610685696362,
      "learning_rate": 4.537196969092669e-07,
      "loss": 0.8814,
      "step": 10920
    },
    {
      "epoch": 0.7879225136178348,
      "grad_norm": 2.485571615074108,
      "learning_rate": 4.5342332308038346e-07,
      "loss": 0.9136,
      "step": 10921
    },
    {
      "epoch": 0.7879946610872623,
      "grad_norm": 5.384587276301536,
      "learning_rate": 4.5312703370432845e-07,
      "loss": 1.0243,
      "step": 10922
    },
    {
      "epoch": 0.7880668085566899,
      "grad_norm": 3.503953880232987,
      "learning_rate": 4.528308287972791e-07,
      "loss": 1.0175,
      "step": 10923
    },
    {
      "epoch": 0.7881389560261174,
      "grad_norm": 3.976211742227866,
      "learning_rate": 4.5253470837541074e-07,
      "loss": 0.9297,
      "step": 10924
    },
    {
      "epoch": 0.7882111034955449,
      "grad_norm": 8.393269667299379,
      "learning_rate": 4.5223867245489346e-07,
      "loss": 1.0403,
      "step": 10925
    },
    {
      "epoch": 0.7882832509649724,
      "grad_norm": 9.468527611868634,
      "learning_rate": 4.5194272105189314e-07,
      "loss": 0.9827,
      "step": 10926
    },
    {
      "epoch": 0.7883553984344,
      "grad_norm": 6.430601998205624,
      "learning_rate": 4.516468541825691e-07,
      "loss": 0.8717,
      "step": 10927
    },
    {
      "epoch": 0.7884275459038275,
      "grad_norm": 6.604579331013326,
      "learning_rate": 4.513510718630791e-07,
      "loss": 0.9727,
      "step": 10928
    },
    {
      "epoch": 0.788499693373255,
      "grad_norm": 4.637401964328709,
      "learning_rate": 4.5105537410957464e-07,
      "loss": 0.9546,
      "step": 10929
    },
    {
      "epoch": 0.7885718408426824,
      "grad_norm": 4.021035671907906,
      "learning_rate": 4.5075976093820167e-07,
      "loss": 0.97,
      "step": 10930
    },
    {
      "epoch": 0.7886439883121099,
      "grad_norm": 3.5722036882957595,
      "learning_rate": 4.504642323651031e-07,
      "loss": 1.0178,
      "step": 10931
    },
    {
      "epoch": 0.7887161357815374,
      "grad_norm": 4.21414566269153,
      "learning_rate": 4.501687884064174e-07,
      "loss": 0.9803,
      "step": 10932
    },
    {
      "epoch": 0.788788283250965,
      "grad_norm": 10.567893913874341,
      "learning_rate": 4.498734290782755e-07,
      "loss": 0.8414,
      "step": 10933
    },
    {
      "epoch": 0.7888604307203925,
      "grad_norm": 4.178318034629574,
      "learning_rate": 4.495781543968078e-07,
      "loss": 0.8837,
      "step": 10934
    },
    {
      "epoch": 0.78893257818982,
      "grad_norm": 5.182499877223056,
      "learning_rate": 4.492829643781377e-07,
      "loss": 0.9548,
      "step": 10935
    },
    {
      "epoch": 0.7890047256592475,
      "grad_norm": 4.39438063978024,
      "learning_rate": 4.489878590383849e-07,
      "loss": 0.8704,
      "step": 10936
    },
    {
      "epoch": 0.789076873128675,
      "grad_norm": 4.087148220778552,
      "learning_rate": 4.48692838393663e-07,
      "loss": 0.9284,
      "step": 10937
    },
    {
      "epoch": 0.7891490205981025,
      "grad_norm": 4.084988602397058,
      "learning_rate": 4.483979024600817e-07,
      "loss": 0.9665,
      "step": 10938
    },
    {
      "epoch": 0.7892211680675301,
      "grad_norm": 17.279733877605274,
      "learning_rate": 4.4810305125374825e-07,
      "loss": 1.0558,
      "step": 10939
    },
    {
      "epoch": 0.7892933155369576,
      "grad_norm": 2.8107807414697845,
      "learning_rate": 4.478082847907618e-07,
      "loss": 0.9048,
      "step": 10940
    },
    {
      "epoch": 0.7893654630063851,
      "grad_norm": 3.454015694614044,
      "learning_rate": 4.475136030872189e-07,
      "loss": 1.0394,
      "step": 10941
    },
    {
      "epoch": 0.7894376104758125,
      "grad_norm": 3.5354115768033316,
      "learning_rate": 4.4721900615921117e-07,
      "loss": 1.0386,
      "step": 10942
    },
    {
      "epoch": 0.78950975794524,
      "grad_norm": 3.3308985082761864,
      "learning_rate": 4.469244940228256e-07,
      "loss": 0.932,
      "step": 10943
    },
    {
      "epoch": 0.7895819054146676,
      "grad_norm": 6.713948650073969,
      "learning_rate": 4.466300666941432e-07,
      "loss": 0.8652,
      "step": 10944
    },
    {
      "epoch": 0.7896540528840951,
      "grad_norm": 5.5156792916281585,
      "learning_rate": 4.4633572418924295e-07,
      "loss": 0.9012,
      "step": 10945
    },
    {
      "epoch": 0.7897262003535226,
      "grad_norm": 5.463028361521125,
      "learning_rate": 4.4604146652419784e-07,
      "loss": 0.8032,
      "step": 10946
    },
    {
      "epoch": 0.7897983478229501,
      "grad_norm": 3.196023652047688,
      "learning_rate": 4.457472937150752e-07,
      "loss": 0.9305,
      "step": 10947
    },
    {
      "epoch": 0.7898704952923776,
      "grad_norm": 6.2139110045062464,
      "learning_rate": 4.454532057779392e-07,
      "loss": 0.9767,
      "step": 10948
    },
    {
      "epoch": 0.7899426427618051,
      "grad_norm": 3.102446845595994,
      "learning_rate": 4.451592027288493e-07,
      "loss": 0.9423,
      "step": 10949
    },
    {
      "epoch": 0.7900147902312327,
      "grad_norm": 2.228706157807238,
      "learning_rate": 4.448652845838592e-07,
      "loss": 0.9481,
      "step": 10950
    },
    {
      "epoch": 0.7900869377006602,
      "grad_norm": 3.5933152972484135,
      "learning_rate": 4.445714513590184e-07,
      "loss": 1.0157,
      "step": 10951
    },
    {
      "epoch": 0.7901590851700877,
      "grad_norm": 3.800516148698884,
      "learning_rate": 4.4427770307037307e-07,
      "loss": 0.9798,
      "step": 10952
    },
    {
      "epoch": 0.7902312326395152,
      "grad_norm": 3.401971632429348,
      "learning_rate": 4.439840397339638e-07,
      "loss": 0.9579,
      "step": 10953
    },
    {
      "epoch": 0.7903033801089426,
      "grad_norm": 0.6412583685955358,
      "learning_rate": 4.4369046136582564e-07,
      "loss": 0.816,
      "step": 10954
    },
    {
      "epoch": 0.7903755275783702,
      "grad_norm": 3.720361953272086,
      "learning_rate": 4.4339696798199e-07,
      "loss": 0.9515,
      "step": 10955
    },
    {
      "epoch": 0.7904476750477977,
      "grad_norm": 5.355276069541687,
      "learning_rate": 4.43103559598484e-07,
      "loss": 0.938,
      "step": 10956
    },
    {
      "epoch": 0.7905198225172252,
      "grad_norm": 3.349482248635353,
      "learning_rate": 4.428102362313282e-07,
      "loss": 0.8266,
      "step": 10957
    },
    {
      "epoch": 0.7905919699866527,
      "grad_norm": 6.772309346829939,
      "learning_rate": 4.4251699789654153e-07,
      "loss": 0.8907,
      "step": 10958
    },
    {
      "epoch": 0.7906641174560802,
      "grad_norm": 3.0654055766712673,
      "learning_rate": 4.422238446101365e-07,
      "loss": 0.8552,
      "step": 10959
    },
    {
      "epoch": 0.7907362649255077,
      "grad_norm": 4.174236186621358,
      "learning_rate": 4.419307763881199e-07,
      "loss": 0.9571,
      "step": 10960
    },
    {
      "epoch": 0.7908084123949353,
      "grad_norm": 3.9369018342538427,
      "learning_rate": 4.416377932464961e-07,
      "loss": 1.0322,
      "step": 10961
    },
    {
      "epoch": 0.7908805598643628,
      "grad_norm": 3.637913164617641,
      "learning_rate": 4.4134489520126284e-07,
      "loss": 0.9785,
      "step": 10962
    },
    {
      "epoch": 0.7909527073337903,
      "grad_norm": 3.2431809707196253,
      "learning_rate": 4.4105208226841604e-07,
      "loss": 0.9362,
      "step": 10963
    },
    {
      "epoch": 0.7910248548032178,
      "grad_norm": 4.137000458862781,
      "learning_rate": 4.407593544639434e-07,
      "loss": 0.9878,
      "step": 10964
    },
    {
      "epoch": 0.7910970022726452,
      "grad_norm": 5.529452791646641,
      "learning_rate": 4.404667118038303e-07,
      "loss": 1.0177,
      "step": 10965
    },
    {
      "epoch": 0.7911691497420728,
      "grad_norm": 3.2784221270866847,
      "learning_rate": 4.401741543040576e-07,
      "loss": 0.942,
      "step": 10966
    },
    {
      "epoch": 0.7912412972115003,
      "grad_norm": 2.622377629802499,
      "learning_rate": 4.398816819805993e-07,
      "loss": 0.819,
      "step": 10967
    },
    {
      "epoch": 0.7913134446809278,
      "grad_norm": 3.5005870054362873,
      "learning_rate": 4.395892948494264e-07,
      "loss": 1.056,
      "step": 10968
    },
    {
      "epoch": 0.7913855921503553,
      "grad_norm": 3.2220735150219735,
      "learning_rate": 4.3929699292650647e-07,
      "loss": 0.9023,
      "step": 10969
    },
    {
      "epoch": 0.7914577396197828,
      "grad_norm": 6.612425765141845,
      "learning_rate": 4.3900477622780063e-07,
      "loss": 0.9703,
      "step": 10970
    },
    {
      "epoch": 0.7915298870892103,
      "grad_norm": 2.9484498493237172,
      "learning_rate": 4.3871264476926463e-07,
      "loss": 0.9626,
      "step": 10971
    },
    {
      "epoch": 0.7916020345586379,
      "grad_norm": 2.6838653293769585,
      "learning_rate": 4.3842059856685163e-07,
      "loss": 0.9374,
      "step": 10972
    },
    {
      "epoch": 0.7916741820280654,
      "grad_norm": 4.823456219784192,
      "learning_rate": 4.381286376365094e-07,
      "loss": 0.8959,
      "step": 10973
    },
    {
      "epoch": 0.7917463294974929,
      "grad_norm": 4.786000204594022,
      "learning_rate": 4.3783676199418005e-07,
      "loss": 0.9174,
      "step": 10974
    },
    {
      "epoch": 0.7918184769669204,
      "grad_norm": 3.853449546839063,
      "learning_rate": 4.375449716558015e-07,
      "loss": 0.9664,
      "step": 10975
    },
    {
      "epoch": 0.791890624436348,
      "grad_norm": 3.1739454605714523,
      "learning_rate": 4.372532666373092e-07,
      "loss": 0.8762,
      "step": 10976
    },
    {
      "epoch": 0.7919627719057754,
      "grad_norm": 5.140267866999427,
      "learning_rate": 4.3696164695463044e-07,
      "loss": 0.9514,
      "step": 10977
    },
    {
      "epoch": 0.7920349193752029,
      "grad_norm": 3.8336019145473608,
      "learning_rate": 4.366701126236898e-07,
      "loss": 0.872,
      "step": 10978
    },
    {
      "epoch": 0.7921070668446304,
      "grad_norm": 4.312998867205465,
      "learning_rate": 4.363786636604072e-07,
      "loss": 0.8592,
      "step": 10979
    },
    {
      "epoch": 0.7921792143140579,
      "grad_norm": 3.163467916500502,
      "learning_rate": 4.360873000806979e-07,
      "loss": 0.9506,
      "step": 10980
    },
    {
      "epoch": 0.7922513617834854,
      "grad_norm": 5.859227537206904,
      "learning_rate": 4.357960219004708e-07,
      "loss": 0.9478,
      "step": 10981
    },
    {
      "epoch": 0.792323509252913,
      "grad_norm": 5.398545259981828,
      "learning_rate": 4.355048291356329e-07,
      "loss": 0.9279,
      "step": 10982
    },
    {
      "epoch": 0.7923956567223405,
      "grad_norm": 5.596592016004339,
      "learning_rate": 4.3521372180208524e-07,
      "loss": 0.9024,
      "step": 10983
    },
    {
      "epoch": 0.792467804191768,
      "grad_norm": 3.5934534365151904,
      "learning_rate": 4.3492269991572295e-07,
      "loss": 0.978,
      "step": 10984
    },
    {
      "epoch": 0.7925399516611955,
      "grad_norm": 4.8362451310674315,
      "learning_rate": 4.3463176349243834e-07,
      "loss": 1.0594,
      "step": 10985
    },
    {
      "epoch": 0.792612099130623,
      "grad_norm": 27.891437969594158,
      "learning_rate": 4.3434091254811786e-07,
      "loss": 0.8888,
      "step": 10986
    },
    {
      "epoch": 0.7926842466000505,
      "grad_norm": 3.648870732399976,
      "learning_rate": 4.340501470986455e-07,
      "loss": 0.8994,
      "step": 10987
    },
    {
      "epoch": 0.7927563940694781,
      "grad_norm": 3.866693761884093,
      "learning_rate": 4.3375946715989696e-07,
      "loss": 0.8149,
      "step": 10988
    },
    {
      "epoch": 0.7928285415389055,
      "grad_norm": 7.17609462562526,
      "learning_rate": 4.3346887274774577e-07,
      "loss": 0.937,
      "step": 10989
    },
    {
      "epoch": 0.792900689008333,
      "grad_norm": 5.949297297679924,
      "learning_rate": 4.3317836387806104e-07,
      "loss": 0.9386,
      "step": 10990
    },
    {
      "epoch": 0.7929728364777605,
      "grad_norm": 8.193732997664235,
      "learning_rate": 4.3288794056670517e-07,
      "loss": 0.876,
      "step": 10991
    },
    {
      "epoch": 0.793044983947188,
      "grad_norm": 0.711157837982423,
      "learning_rate": 4.3259760282953704e-07,
      "loss": 0.8142,
      "step": 10992
    },
    {
      "epoch": 0.7931171314166156,
      "grad_norm": 10.162387718715904,
      "learning_rate": 4.3230735068241263e-07,
      "loss": 0.9546,
      "step": 10993
    },
    {
      "epoch": 0.7931892788860431,
      "grad_norm": 3.3845197849177233,
      "learning_rate": 4.320171841411797e-07,
      "loss": 0.9108,
      "step": 10994
    },
    {
      "epoch": 0.7932614263554706,
      "grad_norm": 3.9051225179484432,
      "learning_rate": 4.31727103221684e-07,
      "loss": 0.9966,
      "step": 10995
    },
    {
      "epoch": 0.7933335738248981,
      "grad_norm": 8.226856019959541,
      "learning_rate": 4.3143710793976563e-07,
      "loss": 0.9635,
      "step": 10996
    },
    {
      "epoch": 0.7934057212943256,
      "grad_norm": 0.6181774650193514,
      "learning_rate": 4.3114719831126067e-07,
      "loss": 0.7713,
      "step": 10997
    },
    {
      "epoch": 0.7934778687637531,
      "grad_norm": 4.298575879625194,
      "learning_rate": 4.30857374351999e-07,
      "loss": 0.8574,
      "step": 10998
    },
    {
      "epoch": 0.7935500162331807,
      "grad_norm": 2.431827988357671,
      "learning_rate": 4.305676360778068e-07,
      "loss": 0.9186,
      "step": 10999
    },
    {
      "epoch": 0.7936221637026082,
      "grad_norm": 3.9737229317087874,
      "learning_rate": 4.3027798350450717e-07,
      "loss": 0.9994,
      "step": 11000
    },
    {
      "epoch": 0.7936943111720356,
      "grad_norm": 0.7805939781066153,
      "learning_rate": 4.2998841664791554e-07,
      "loss": 0.8718,
      "step": 11001
    },
    {
      "epoch": 0.7937664586414631,
      "grad_norm": 4.960905973078956,
      "learning_rate": 4.296989355238445e-07,
      "loss": 0.965,
      "step": 11002
    },
    {
      "epoch": 0.7938386061108906,
      "grad_norm": 3.6336252923489174,
      "learning_rate": 4.294095401481019e-07,
      "loss": 0.9736,
      "step": 11003
    },
    {
      "epoch": 0.7939107535803182,
      "grad_norm": 3.621260457074806,
      "learning_rate": 4.291202305364898e-07,
      "loss": 0.8845,
      "step": 11004
    },
    {
      "epoch": 0.7939829010497457,
      "grad_norm": 3.902010639013884,
      "learning_rate": 4.288310067048062e-07,
      "loss": 1.0074,
      "step": 11005
    },
    {
      "epoch": 0.7940550485191732,
      "grad_norm": 15.59660359113809,
      "learning_rate": 4.285418686688453e-07,
      "loss": 0.8684,
      "step": 11006
    },
    {
      "epoch": 0.7941271959886007,
      "grad_norm": 3.4619268721113508,
      "learning_rate": 4.282528164443966e-07,
      "loss": 1.0055,
      "step": 11007
    },
    {
      "epoch": 0.7941993434580282,
      "grad_norm": 12.314742295623454,
      "learning_rate": 4.279638500472422e-07,
      "loss": 0.9747,
      "step": 11008
    },
    {
      "epoch": 0.7942714909274557,
      "grad_norm": 5.025781538429768,
      "learning_rate": 4.2767496949316274e-07,
      "loss": 0.9516,
      "step": 11009
    },
    {
      "epoch": 0.7943436383968833,
      "grad_norm": 5.505351497505313,
      "learning_rate": 4.2738617479793326e-07,
      "loss": 0.9378,
      "step": 11010
    },
    {
      "epoch": 0.7944157858663108,
      "grad_norm": 3.9569120699830616,
      "learning_rate": 4.270974659773219e-07,
      "loss": 0.9788,
      "step": 11011
    },
    {
      "epoch": 0.7944879333357382,
      "grad_norm": 12.218563302801615,
      "learning_rate": 4.2680884304709575e-07,
      "loss": 0.9684,
      "step": 11012
    },
    {
      "epoch": 0.7945600808051657,
      "grad_norm": 4.169891914778201,
      "learning_rate": 4.2652030602301514e-07,
      "loss": 0.9865,
      "step": 11013
    },
    {
      "epoch": 0.7946322282745932,
      "grad_norm": 4.6368102761212,
      "learning_rate": 4.262318549208361e-07,
      "loss": 0.9241,
      "step": 11014
    },
    {
      "epoch": 0.7947043757440208,
      "grad_norm": 3.7845077982754267,
      "learning_rate": 4.259434897563092e-07,
      "loss": 0.9882,
      "step": 11015
    },
    {
      "epoch": 0.7947765232134483,
      "grad_norm": 5.2760090301297575,
      "learning_rate": 4.2565521054518074e-07,
      "loss": 0.9477,
      "step": 11016
    },
    {
      "epoch": 0.7948486706828758,
      "grad_norm": 9.732191205185854,
      "learning_rate": 4.2536701730319426e-07,
      "loss": 0.976,
      "step": 11017
    },
    {
      "epoch": 0.7949208181523033,
      "grad_norm": 9.69410332848076,
      "learning_rate": 4.2507891004608543e-07,
      "loss": 1.0609,
      "step": 11018
    },
    {
      "epoch": 0.7949929656217308,
      "grad_norm": 4.251126757088524,
      "learning_rate": 4.2479088878958725e-07,
      "loss": 0.888,
      "step": 11019
    },
    {
      "epoch": 0.7950651130911583,
      "grad_norm": 4.364859026787746,
      "learning_rate": 4.245029535494278e-07,
      "loss": 1.0162,
      "step": 11020
    },
    {
      "epoch": 0.7951372605605859,
      "grad_norm": 4.542389955001211,
      "learning_rate": 4.2421510434132913e-07,
      "loss": 1.0323,
      "step": 11021
    },
    {
      "epoch": 0.7952094080300134,
      "grad_norm": 4.206690300715375,
      "learning_rate": 4.2392734118101047e-07,
      "loss": 0.998,
      "step": 11022
    },
    {
      "epoch": 0.7952815554994409,
      "grad_norm": 5.50303982046514,
      "learning_rate": 4.236396640841846e-07,
      "loss": 0.8956,
      "step": 11023
    },
    {
      "epoch": 0.7953537029688683,
      "grad_norm": 4.629647265647923,
      "learning_rate": 4.233520730665623e-07,
      "loss": 0.947,
      "step": 11024
    },
    {
      "epoch": 0.7954258504382958,
      "grad_norm": 23.79149464168143,
      "learning_rate": 4.230645681438463e-07,
      "loss": 0.8881,
      "step": 11025
    },
    {
      "epoch": 0.7954979979077234,
      "grad_norm": 4.281336205721399,
      "learning_rate": 4.227771493317365e-07,
      "loss": 1.0123,
      "step": 11026
    },
    {
      "epoch": 0.7955701453771509,
      "grad_norm": 3.0910611268253025,
      "learning_rate": 4.224898166459283e-07,
      "loss": 0.8251,
      "step": 11027
    },
    {
      "epoch": 0.7956422928465784,
      "grad_norm": 0.662791483112883,
      "learning_rate": 4.2220257010211127e-07,
      "loss": 0.8263,
      "step": 11028
    },
    {
      "epoch": 0.7957144403160059,
      "grad_norm": 5.741729759604175,
      "learning_rate": 4.219154097159703e-07,
      "loss": 0.9176,
      "step": 11029
    },
    {
      "epoch": 0.7957865877854334,
      "grad_norm": 3.628563412012743,
      "learning_rate": 4.2162833550318754e-07,
      "loss": 0.9763,
      "step": 11030
    },
    {
      "epoch": 0.795858735254861,
      "grad_norm": 4.148838634457902,
      "learning_rate": 4.2134134747943897e-07,
      "loss": 0.931,
      "step": 11031
    },
    {
      "epoch": 0.7959308827242885,
      "grad_norm": 5.260009306692731,
      "learning_rate": 4.2105444566039485e-07,
      "loss": 0.8578,
      "step": 11032
    },
    {
      "epoch": 0.796003030193716,
      "grad_norm": 2.9900438084789887,
      "learning_rate": 4.2076763006172245e-07,
      "loss": 0.9784,
      "step": 11033
    },
    {
      "epoch": 0.7960751776631435,
      "grad_norm": 3.608576566560693,
      "learning_rate": 4.2048090069908414e-07,
      "loss": 0.8344,
      "step": 11034
    },
    {
      "epoch": 0.796147325132571,
      "grad_norm": 4.971907661949304,
      "learning_rate": 4.201942575881356e-07,
      "loss": 0.9221,
      "step": 11035
    },
    {
      "epoch": 0.7962194726019984,
      "grad_norm": 3.7271225226559377,
      "learning_rate": 4.199077007445311e-07,
      "loss": 1.0212,
      "step": 11036
    },
    {
      "epoch": 0.796291620071426,
      "grad_norm": 4.522711344726175,
      "learning_rate": 4.196212301839181e-07,
      "loss": 0.9221,
      "step": 11037
    },
    {
      "epoch": 0.7963637675408535,
      "grad_norm": 2.9271984869591035,
      "learning_rate": 4.19334845921939e-07,
      "loss": 0.9657,
      "step": 11038
    },
    {
      "epoch": 0.796435915010281,
      "grad_norm": 3.0121735898537785,
      "learning_rate": 4.1904854797423253e-07,
      "loss": 0.9495,
      "step": 11039
    },
    {
      "epoch": 0.7965080624797085,
      "grad_norm": 3.471825825422807,
      "learning_rate": 4.1876233635643165e-07,
      "loss": 1.0084,
      "step": 11040
    },
    {
      "epoch": 0.796580209949136,
      "grad_norm": 4.57186659774999,
      "learning_rate": 4.1847621108416733e-07,
      "loss": 0.9278,
      "step": 11041
    },
    {
      "epoch": 0.7966523574185636,
      "grad_norm": 3.027138664928335,
      "learning_rate": 4.1819017217306186e-07,
      "loss": 1.0328,
      "step": 11042
    },
    {
      "epoch": 0.7967245048879911,
      "grad_norm": 3.8183321510378603,
      "learning_rate": 4.1790421963873543e-07,
      "loss": 1.0344,
      "step": 11043
    },
    {
      "epoch": 0.7967966523574186,
      "grad_norm": 2.4973152049510365,
      "learning_rate": 4.176183534968034e-07,
      "loss": 0.963,
      "step": 11044
    },
    {
      "epoch": 0.7968687998268461,
      "grad_norm": 7.442878125022306,
      "learning_rate": 4.173325737628748e-07,
      "loss": 0.9223,
      "step": 11045
    },
    {
      "epoch": 0.7969409472962736,
      "grad_norm": 0.6538478620159482,
      "learning_rate": 4.170468804525553e-07,
      "loss": 0.8431,
      "step": 11046
    },
    {
      "epoch": 0.7970130947657011,
      "grad_norm": 3.277948807889775,
      "learning_rate": 4.167612735814465e-07,
      "loss": 1.1126,
      "step": 11047
    },
    {
      "epoch": 0.7970852422351286,
      "grad_norm": 3.0859691763109223,
      "learning_rate": 4.164757531651426e-07,
      "loss": 0.9757,
      "step": 11048
    },
    {
      "epoch": 0.7971573897045561,
      "grad_norm": 6.436023135355258,
      "learning_rate": 4.161903192192362e-07,
      "loss": 0.9766,
      "step": 11049
    },
    {
      "epoch": 0.7972295371739836,
      "grad_norm": 3.4266772430593146,
      "learning_rate": 4.159049717593135e-07,
      "loss": 0.9444,
      "step": 11050
    },
    {
      "epoch": 0.7973016846434111,
      "grad_norm": 6.244952796042952,
      "learning_rate": 4.156197108009567e-07,
      "loss": 0.9292,
      "step": 11051
    },
    {
      "epoch": 0.7973738321128386,
      "grad_norm": 5.0604976885756106,
      "learning_rate": 4.153345363597418e-07,
      "loss": 1.0657,
      "step": 11052
    },
    {
      "epoch": 0.7974459795822662,
      "grad_norm": 3.7383789559952225,
      "learning_rate": 4.1504944845124104e-07,
      "loss": 0.9193,
      "step": 11053
    },
    {
      "epoch": 0.7975181270516937,
      "grad_norm": 4.850872488295799,
      "learning_rate": 4.1476444709102367e-07,
      "loss": 0.9702,
      "step": 11054
    },
    {
      "epoch": 0.7975902745211212,
      "grad_norm": 3.8553554591701054,
      "learning_rate": 4.1447953229465104e-07,
      "loss": 0.9813,
      "step": 11055
    },
    {
      "epoch": 0.7976624219905487,
      "grad_norm": 2.6974580490343216,
      "learning_rate": 4.1419470407768165e-07,
      "loss": 0.905,
      "step": 11056
    },
    {
      "epoch": 0.7977345694599762,
      "grad_norm": 3.354269575037022,
      "learning_rate": 4.13909962455669e-07,
      "loss": 1.0205,
      "step": 11057
    },
    {
      "epoch": 0.7978067169294037,
      "grad_norm": 2.7748277215490953,
      "learning_rate": 4.136253074441625e-07,
      "loss": 0.8939,
      "step": 11058
    },
    {
      "epoch": 0.7978788643988312,
      "grad_norm": 2.9184203915452462,
      "learning_rate": 4.1334073905870426e-07,
      "loss": 1.0539,
      "step": 11059
    },
    {
      "epoch": 0.7979510118682587,
      "grad_norm": 4.760593847849203,
      "learning_rate": 4.1305625731483505e-07,
      "loss": 1.0132,
      "step": 11060
    },
    {
      "epoch": 0.7980231593376862,
      "grad_norm": 3.9207107822077014,
      "learning_rate": 4.127718622280898e-07,
      "loss": 0.9448,
      "step": 11061
    },
    {
      "epoch": 0.7980953068071137,
      "grad_norm": 3.2667072455811708,
      "learning_rate": 4.124875538139967e-07,
      "loss": 0.8811,
      "step": 11062
    },
    {
      "epoch": 0.7981674542765412,
      "grad_norm": 4.214518141167677,
      "learning_rate": 4.1220333208808154e-07,
      "loss": 0.9382,
      "step": 11063
    },
    {
      "epoch": 0.7982396017459688,
      "grad_norm": 3.935403780548394,
      "learning_rate": 4.1191919706586507e-07,
      "loss": 0.9848,
      "step": 11064
    },
    {
      "epoch": 0.7983117492153963,
      "grad_norm": 7.708149043449107,
      "learning_rate": 4.116351487628613e-07,
      "loss": 0.9632,
      "step": 11065
    },
    {
      "epoch": 0.7983838966848238,
      "grad_norm": 0.74586583196351,
      "learning_rate": 4.1135118719458274e-07,
      "loss": 0.9119,
      "step": 11066
    },
    {
      "epoch": 0.7984560441542513,
      "grad_norm": 3.557968872011342,
      "learning_rate": 4.110673123765349e-07,
      "loss": 0.981,
      "step": 11067
    },
    {
      "epoch": 0.7985281916236788,
      "grad_norm": 1.9676618445017073,
      "learning_rate": 4.107835243242197e-07,
      "loss": 0.9626,
      "step": 11068
    },
    {
      "epoch": 0.7986003390931063,
      "grad_norm": 4.14628841388102,
      "learning_rate": 4.1049982305313244e-07,
      "loss": 0.9132,
      "step": 11069
    },
    {
      "epoch": 0.7986724865625339,
      "grad_norm": 3.0874497197184523,
      "learning_rate": 4.102162085787657e-07,
      "loss": 1.0026,
      "step": 11070
    },
    {
      "epoch": 0.7987446340319613,
      "grad_norm": 3.8483007520571584,
      "learning_rate": 4.099326809166075e-07,
      "loss": 0.9983,
      "step": 11071
    },
    {
      "epoch": 0.7988167815013888,
      "grad_norm": 3.134163810027475,
      "learning_rate": 4.0964924008213786e-07,
      "loss": 0.8868,
      "step": 11072
    },
    {
      "epoch": 0.7988889289708163,
      "grad_norm": 3.859831130529051,
      "learning_rate": 4.093658860908367e-07,
      "loss": 1.0018,
      "step": 11073
    },
    {
      "epoch": 0.7989610764402438,
      "grad_norm": 4.098574525474575,
      "learning_rate": 4.090826189581762e-07,
      "loss": 0.9685,
      "step": 11074
    },
    {
      "epoch": 0.7990332239096714,
      "grad_norm": 2.985627236208718,
      "learning_rate": 4.0879943869962495e-07,
      "loss": 1.0348,
      "step": 11075
    },
    {
      "epoch": 0.7991053713790989,
      "grad_norm": 4.921223721481286,
      "learning_rate": 4.085163453306455e-07,
      "loss": 1.0541,
      "step": 11076
    },
    {
      "epoch": 0.7991775188485264,
      "grad_norm": 3.1726068864376473,
      "learning_rate": 4.082333388666963e-07,
      "loss": 0.8527,
      "step": 11077
    },
    {
      "epoch": 0.7992496663179539,
      "grad_norm": 3.2128388281824014,
      "learning_rate": 4.0795041932323305e-07,
      "loss": 0.983,
      "step": 11078
    },
    {
      "epoch": 0.7993218137873814,
      "grad_norm": 0.7412114601987823,
      "learning_rate": 4.076675867157033e-07,
      "loss": 0.8409,
      "step": 11079
    },
    {
      "epoch": 0.799393961256809,
      "grad_norm": 4.377066641837311,
      "learning_rate": 4.0738484105955216e-07,
      "loss": 1.0003,
      "step": 11080
    },
    {
      "epoch": 0.7994661087262365,
      "grad_norm": 3.907459041407641,
      "learning_rate": 4.0710218237021964e-07,
      "loss": 0.92,
      "step": 11081
    },
    {
      "epoch": 0.799538256195664,
      "grad_norm": 4.184450092168927,
      "learning_rate": 4.0681961066313966e-07,
      "loss": 0.9392,
      "step": 11082
    },
    {
      "epoch": 0.7996104036650914,
      "grad_norm": 18.916869757594178,
      "learning_rate": 4.0653712595374246e-07,
      "loss": 0.9967,
      "step": 11083
    },
    {
      "epoch": 0.7996825511345189,
      "grad_norm": 5.015792321001925,
      "learning_rate": 4.062547282574547e-07,
      "loss": 0.948,
      "step": 11084
    },
    {
      "epoch": 0.7997546986039464,
      "grad_norm": 4.047187704722965,
      "learning_rate": 4.0597241758969656e-07,
      "loss": 0.9313,
      "step": 11085
    },
    {
      "epoch": 0.799826846073374,
      "grad_norm": 3.357939515478475,
      "learning_rate": 4.056901939658837e-07,
      "loss": 0.9589,
      "step": 11086
    },
    {
      "epoch": 0.7998989935428015,
      "grad_norm": 5.6048640652727295,
      "learning_rate": 4.0540805740142715e-07,
      "loss": 0.9202,
      "step": 11087
    },
    {
      "epoch": 0.799971141012229,
      "grad_norm": 4.270742846515062,
      "learning_rate": 4.0512600791173424e-07,
      "loss": 0.9798,
      "step": 11088
    },
    {
      "epoch": 0.8000432884816565,
      "grad_norm": 4.243038759477541,
      "learning_rate": 4.0484404551220487e-07,
      "loss": 0.9776,
      "step": 11089
    },
    {
      "epoch": 0.800115435951084,
      "grad_norm": 3.0188284348433174,
      "learning_rate": 4.0456217021823756e-07,
      "loss": 0.964,
      "step": 11090
    },
    {
      "epoch": 0.8001875834205116,
      "grad_norm": 3.2765559453357485,
      "learning_rate": 4.0428038204522476e-07,
      "loss": 0.955,
      "step": 11091
    },
    {
      "epoch": 0.8002597308899391,
      "grad_norm": 3.5444106897270204,
      "learning_rate": 4.039986810085525e-07,
      "loss": 1.0048,
      "step": 11092
    },
    {
      "epoch": 0.8003318783593666,
      "grad_norm": 3.4131450336890308,
      "learning_rate": 4.037170671236041e-07,
      "loss": 0.9972,
      "step": 11093
    },
    {
      "epoch": 0.8004040258287941,
      "grad_norm": 3.1204220618362095,
      "learning_rate": 4.0343554040575746e-07,
      "loss": 1.0465,
      "step": 11094
    },
    {
      "epoch": 0.8004761732982215,
      "grad_norm": 5.01179505525748,
      "learning_rate": 4.031541008703863e-07,
      "loss": 0.8984,
      "step": 11095
    },
    {
      "epoch": 0.800548320767649,
      "grad_norm": 4.7763433274488385,
      "learning_rate": 4.028727485328571e-07,
      "loss": 0.9536,
      "step": 11096
    },
    {
      "epoch": 0.8006204682370766,
      "grad_norm": 5.7871083862427435,
      "learning_rate": 4.025914834085353e-07,
      "loss": 0.9427,
      "step": 11097
    },
    {
      "epoch": 0.8006926157065041,
      "grad_norm": 18.227438162098508,
      "learning_rate": 4.023103055127799e-07,
      "loss": 0.8168,
      "step": 11098
    },
    {
      "epoch": 0.8007647631759316,
      "grad_norm": 3.244532607969482,
      "learning_rate": 4.020292148609437e-07,
      "loss": 0.9174,
      "step": 11099
    },
    {
      "epoch": 0.8008369106453591,
      "grad_norm": 20.249393406954425,
      "learning_rate": 4.017482114683763e-07,
      "loss": 0.8525,
      "step": 11100
    },
    {
      "epoch": 0.8009090581147866,
      "grad_norm": 3.2850957223076653,
      "learning_rate": 4.0146729535042234e-07,
      "loss": 0.9538,
      "step": 11101
    },
    {
      "epoch": 0.8009812055842142,
      "grad_norm": 3.7307794731114496,
      "learning_rate": 4.011864665224227e-07,
      "loss": 0.9707,
      "step": 11102
    },
    {
      "epoch": 0.8010533530536417,
      "grad_norm": 3.0779736476808686,
      "learning_rate": 4.009057249997112e-07,
      "loss": 1.0182,
      "step": 11103
    },
    {
      "epoch": 0.8011255005230692,
      "grad_norm": 3.8719897729644015,
      "learning_rate": 4.0062507079761823e-07,
      "loss": 0.9758,
      "step": 11104
    },
    {
      "epoch": 0.8011976479924967,
      "grad_norm": 8.139322187616616,
      "learning_rate": 4.003445039314699e-07,
      "loss": 0.9715,
      "step": 11105
    },
    {
      "epoch": 0.8012697954619242,
      "grad_norm": 4.717577858157544,
      "learning_rate": 4.0006402441658624e-07,
      "loss": 0.9156,
      "step": 11106
    },
    {
      "epoch": 0.8013419429313516,
      "grad_norm": 6.1101888519420156,
      "learning_rate": 3.997836322682828e-07,
      "loss": 0.9989,
      "step": 11107
    },
    {
      "epoch": 0.8014140904007792,
      "grad_norm": 4.448452395383057,
      "learning_rate": 3.995033275018724e-07,
      "loss": 0.975,
      "step": 11108
    },
    {
      "epoch": 0.8014862378702067,
      "grad_norm": 4.291774760739877,
      "learning_rate": 3.9922311013266e-07,
      "loss": 0.866,
      "step": 11109
    },
    {
      "epoch": 0.8015583853396342,
      "grad_norm": 4.481602679091732,
      "learning_rate": 3.9894298017594765e-07,
      "loss": 0.9331,
      "step": 11110
    },
    {
      "epoch": 0.8016305328090617,
      "grad_norm": 5.69638172887188,
      "learning_rate": 3.9866293764703207e-07,
      "loss": 0.9662,
      "step": 11111
    },
    {
      "epoch": 0.8017026802784892,
      "grad_norm": 2.6768426947789727,
      "learning_rate": 3.983829825612062e-07,
      "loss": 0.9028,
      "step": 11112
    },
    {
      "epoch": 0.8017748277479168,
      "grad_norm": 5.0760788771420575,
      "learning_rate": 3.981031149337557e-07,
      "loss": 1.0051,
      "step": 11113
    },
    {
      "epoch": 0.8018469752173443,
      "grad_norm": 3.3799462983419266,
      "learning_rate": 3.978233347799644e-07,
      "loss": 0.9573,
      "step": 11114
    },
    {
      "epoch": 0.8019191226867718,
      "grad_norm": 4.455317014638547,
      "learning_rate": 3.975436421151104e-07,
      "loss": 0.9898,
      "step": 11115
    },
    {
      "epoch": 0.8019912701561993,
      "grad_norm": 2.895925085869161,
      "learning_rate": 3.9726403695446555e-07,
      "loss": 0.991,
      "step": 11116
    },
    {
      "epoch": 0.8020634176256268,
      "grad_norm": 6.410943521696233,
      "learning_rate": 3.9698451931329836e-07,
      "loss": 1.0078,
      "step": 11117
    },
    {
      "epoch": 0.8021355650950542,
      "grad_norm": 7.318056522289763,
      "learning_rate": 3.967050892068724e-07,
      "loss": 1.048,
      "step": 11118
    },
    {
      "epoch": 0.8022077125644818,
      "grad_norm": 7.042898608306438,
      "learning_rate": 3.96425746650447e-07,
      "loss": 0.947,
      "step": 11119
    },
    {
      "epoch": 0.8022798600339093,
      "grad_norm": 6.15238033223251,
      "learning_rate": 3.9614649165927426e-07,
      "loss": 0.8158,
      "step": 11120
    },
    {
      "epoch": 0.8023520075033368,
      "grad_norm": 4.73420337328149,
      "learning_rate": 3.958673242486048e-07,
      "loss": 1.0476,
      "step": 11121
    },
    {
      "epoch": 0.8024241549727643,
      "grad_norm": 3.3327960693984493,
      "learning_rate": 3.955882444336831e-07,
      "loss": 0.8444,
      "step": 11122
    },
    {
      "epoch": 0.8024963024421918,
      "grad_norm": 5.147104186170695,
      "learning_rate": 3.953092522297474e-07,
      "loss": 0.8608,
      "step": 11123
    },
    {
      "epoch": 0.8025684499116194,
      "grad_norm": 2.270624366968068,
      "learning_rate": 3.9503034765203315e-07,
      "loss": 0.8536,
      "step": 11124
    },
    {
      "epoch": 0.8026405973810469,
      "grad_norm": 6.521139541927696,
      "learning_rate": 3.9475153071577074e-07,
      "loss": 1.0585,
      "step": 11125
    },
    {
      "epoch": 0.8027127448504744,
      "grad_norm": 3.7826316137920672,
      "learning_rate": 3.9447280143618375e-07,
      "loss": 0.9397,
      "step": 11126
    },
    {
      "epoch": 0.8027848923199019,
      "grad_norm": 4.553830205151869,
      "learning_rate": 3.9419415982849416e-07,
      "loss": 0.9662,
      "step": 11127
    },
    {
      "epoch": 0.8028570397893294,
      "grad_norm": 0.9208353289989635,
      "learning_rate": 3.9391560590791695e-07,
      "loss": 0.7405,
      "step": 11128
    },
    {
      "epoch": 0.802929187258757,
      "grad_norm": 4.285951135539283,
      "learning_rate": 3.936371396896634e-07,
      "loss": 0.9817,
      "step": 11129
    },
    {
      "epoch": 0.8030013347281844,
      "grad_norm": 2.6887099625121724,
      "learning_rate": 3.9335876118893883e-07,
      "loss": 0.9985,
      "step": 11130
    },
    {
      "epoch": 0.8030734821976119,
      "grad_norm": 4.3569234975945585,
      "learning_rate": 3.9308047042094407e-07,
      "loss": 1.0472,
      "step": 11131
    },
    {
      "epoch": 0.8031456296670394,
      "grad_norm": 3.36600397468101,
      "learning_rate": 3.928022674008773e-07,
      "loss": 0.8618,
      "step": 11132
    },
    {
      "epoch": 0.8032177771364669,
      "grad_norm": 5.716572664931958,
      "learning_rate": 3.925241521439284e-07,
      "loss": 0.9916,
      "step": 11133
    },
    {
      "epoch": 0.8032899246058944,
      "grad_norm": 13.88624021239472,
      "learning_rate": 3.9224612466528504e-07,
      "loss": 1.0041,
      "step": 11134
    },
    {
      "epoch": 0.803362072075322,
      "grad_norm": 4.204047233062134,
      "learning_rate": 3.919681849801295e-07,
      "loss": 0.9743,
      "step": 11135
    },
    {
      "epoch": 0.8034342195447495,
      "grad_norm": 5.646532003755366,
      "learning_rate": 3.916903331036383e-07,
      "loss": 0.9611,
      "step": 11136
    },
    {
      "epoch": 0.803506367014177,
      "grad_norm": 3.644220631273498,
      "learning_rate": 3.9141256905098353e-07,
      "loss": 0.8878,
      "step": 11137
    },
    {
      "epoch": 0.8035785144836045,
      "grad_norm": 3.1380518150646846,
      "learning_rate": 3.9113489283733436e-07,
      "loss": 0.9606,
      "step": 11138
    },
    {
      "epoch": 0.803650661953032,
      "grad_norm": 4.553154977085815,
      "learning_rate": 3.9085730447785317e-07,
      "loss": 0.8832,
      "step": 11139
    },
    {
      "epoch": 0.8037228094224595,
      "grad_norm": 0.7244132874238616,
      "learning_rate": 3.905798039876973e-07,
      "loss": 0.8582,
      "step": 11140
    },
    {
      "epoch": 0.8037949568918871,
      "grad_norm": 3.5519861709566087,
      "learning_rate": 3.903023913820205e-07,
      "loss": 1.0076,
      "step": 11141
    },
    {
      "epoch": 0.8038671043613145,
      "grad_norm": 3.8066709389025157,
      "learning_rate": 3.900250666759717e-07,
      "loss": 1.0262,
      "step": 11142
    },
    {
      "epoch": 0.803939251830742,
      "grad_norm": 5.8584631451929425,
      "learning_rate": 3.897478298846935e-07,
      "loss": 0.9318,
      "step": 11143
    },
    {
      "epoch": 0.8040113993001695,
      "grad_norm": 4.708961084878634,
      "learning_rate": 3.8947068102332505e-07,
      "loss": 0.9863,
      "step": 11144
    },
    {
      "epoch": 0.804083546769597,
      "grad_norm": 2.4314858996331328,
      "learning_rate": 3.8919362010700097e-07,
      "loss": 1.0824,
      "step": 11145
    },
    {
      "epoch": 0.8041556942390246,
      "grad_norm": 0.6463171751352912,
      "learning_rate": 3.8891664715085093e-07,
      "loss": 0.7876,
      "step": 11146
    },
    {
      "epoch": 0.8042278417084521,
      "grad_norm": 5.277866340474165,
      "learning_rate": 3.886397621699984e-07,
      "loss": 0.9442,
      "step": 11147
    },
    {
      "epoch": 0.8042999891778796,
      "grad_norm": 6.044793453300651,
      "learning_rate": 3.8836296517956323e-07,
      "loss": 0.9337,
      "step": 11148
    },
    {
      "epoch": 0.8043721366473071,
      "grad_norm": 5.714389159764872,
      "learning_rate": 3.8808625619466093e-07,
      "loss": 0.8501,
      "step": 11149
    },
    {
      "epoch": 0.8044442841167346,
      "grad_norm": 3.3105002070825718,
      "learning_rate": 3.8780963523040014e-07,
      "loss": 0.912,
      "step": 11150
    },
    {
      "epoch": 0.8045164315861622,
      "grad_norm": 3.6815989201052997,
      "learning_rate": 3.875331023018875e-07,
      "loss": 0.971,
      "step": 11151
    },
    {
      "epoch": 0.8045885790555897,
      "grad_norm": 4.097755860026181,
      "learning_rate": 3.872566574242235e-07,
      "loss": 0.8281,
      "step": 11152
    },
    {
      "epoch": 0.8046607265250172,
      "grad_norm": 0.6718004983803597,
      "learning_rate": 3.86980300612503e-07,
      "loss": 0.7971,
      "step": 11153
    },
    {
      "epoch": 0.8047328739944446,
      "grad_norm": 3.98808050444375,
      "learning_rate": 3.867040318818169e-07,
      "loss": 0.9254,
      "step": 11154
    },
    {
      "epoch": 0.8048050214638721,
      "grad_norm": 5.428891262155074,
      "learning_rate": 3.8642785124725076e-07,
      "loss": 0.9712,
      "step": 11155
    },
    {
      "epoch": 0.8048771689332996,
      "grad_norm": 2.131063002272181,
      "learning_rate": 3.861517587238876e-07,
      "loss": 0.9878,
      "step": 11156
    },
    {
      "epoch": 0.8049493164027272,
      "grad_norm": 3.6733801820632586,
      "learning_rate": 3.858757543268021e-07,
      "loss": 0.9431,
      "step": 11157
    },
    {
      "epoch": 0.8050214638721547,
      "grad_norm": 5.035575381665781,
      "learning_rate": 3.8559983807106634e-07,
      "loss": 0.821,
      "step": 11158
    },
    {
      "epoch": 0.8050936113415822,
      "grad_norm": 4.625503924985751,
      "learning_rate": 3.8532400997174764e-07,
      "loss": 0.9785,
      "step": 11159
    },
    {
      "epoch": 0.8051657588110097,
      "grad_norm": 4.681728828500044,
      "learning_rate": 3.8504827004390706e-07,
      "loss": 0.9396,
      "step": 11160
    },
    {
      "epoch": 0.8052379062804372,
      "grad_norm": 6.749727596861664,
      "learning_rate": 3.8477261830260145e-07,
      "loss": 0.876,
      "step": 11161
    },
    {
      "epoch": 0.8053100537498648,
      "grad_norm": 3.5669150851243017,
      "learning_rate": 3.844970547628843e-07,
      "loss": 0.9963,
      "step": 11162
    },
    {
      "epoch": 0.8053822012192923,
      "grad_norm": 3.3345339679094725,
      "learning_rate": 3.8422157943980316e-07,
      "loss": 1.014,
      "step": 11163
    },
    {
      "epoch": 0.8054543486887198,
      "grad_norm": 3.8119805404230442,
      "learning_rate": 3.839461923483995e-07,
      "loss": 0.909,
      "step": 11164
    },
    {
      "epoch": 0.8055264961581472,
      "grad_norm": 4.495138084997862,
      "learning_rate": 3.836708935037121e-07,
      "loss": 1.0213,
      "step": 11165
    },
    {
      "epoch": 0.8055986436275747,
      "grad_norm": 4.825046977866956,
      "learning_rate": 3.8339568292077406e-07,
      "loss": 0.9144,
      "step": 11166
    },
    {
      "epoch": 0.8056707910970022,
      "grad_norm": 8.037452766433919,
      "learning_rate": 3.8312056061461285e-07,
      "loss": 0.9514,
      "step": 11167
    },
    {
      "epoch": 0.8057429385664298,
      "grad_norm": 4.5262865966733274,
      "learning_rate": 3.8284552660025195e-07,
      "loss": 0.9765,
      "step": 11168
    },
    {
      "epoch": 0.8058150860358573,
      "grad_norm": 2.3446912782708025,
      "learning_rate": 3.825705808927113e-07,
      "loss": 0.963,
      "step": 11169
    },
    {
      "epoch": 0.8058872335052848,
      "grad_norm": 4.049901589033908,
      "learning_rate": 3.8229572350700303e-07,
      "loss": 1.0868,
      "step": 11170
    },
    {
      "epoch": 0.8059593809747123,
      "grad_norm": 3.124938964247692,
      "learning_rate": 3.82020954458137e-07,
      "loss": 0.929,
      "step": 11171
    },
    {
      "epoch": 0.8060315284441398,
      "grad_norm": 5.576055062535233,
      "learning_rate": 3.81746273761117e-07,
      "loss": 0.9594,
      "step": 11172
    },
    {
      "epoch": 0.8061036759135674,
      "grad_norm": 6.929929159223453,
      "learning_rate": 3.814716814309429e-07,
      "loss": 0.9353,
      "step": 11173
    },
    {
      "epoch": 0.8061758233829949,
      "grad_norm": 0.796935097915228,
      "learning_rate": 3.811971774826077e-07,
      "loss": 0.8594,
      "step": 11174
    },
    {
      "epoch": 0.8062479708524224,
      "grad_norm": 4.825335539382412,
      "learning_rate": 3.809227619311026e-07,
      "loss": 0.9308,
      "step": 11175
    },
    {
      "epoch": 0.8063201183218499,
      "grad_norm": 3.13940284444774,
      "learning_rate": 3.806484347914123e-07,
      "loss": 0.9258,
      "step": 11176
    },
    {
      "epoch": 0.8063922657912773,
      "grad_norm": 21.32388708468488,
      "learning_rate": 3.8037419607851586e-07,
      "loss": 0.848,
      "step": 11177
    },
    {
      "epoch": 0.8064644132607048,
      "grad_norm": 3.994835141710887,
      "learning_rate": 3.8010004580738887e-07,
      "loss": 1.0306,
      "step": 11178
    },
    {
      "epoch": 0.8065365607301324,
      "grad_norm": 4.244334315213267,
      "learning_rate": 3.7982598399300115e-07,
      "loss": 0.883,
      "step": 11179
    },
    {
      "epoch": 0.8066087081995599,
      "grad_norm": 3.3179661976458865,
      "learning_rate": 3.7955201065032005e-07,
      "loss": 0.9578,
      "step": 11180
    },
    {
      "epoch": 0.8066808556689874,
      "grad_norm": 6.2167505471687265,
      "learning_rate": 3.7927812579430405e-07,
      "loss": 0.8783,
      "step": 11181
    },
    {
      "epoch": 0.8067530031384149,
      "grad_norm": 3.2562808454520638,
      "learning_rate": 3.7900432943991035e-07,
      "loss": 1.0794,
      "step": 11182
    },
    {
      "epoch": 0.8068251506078424,
      "grad_norm": 5.948922183083324,
      "learning_rate": 3.787306216020896e-07,
      "loss": 0.9951,
      "step": 11183
    },
    {
      "epoch": 0.80689729807727,
      "grad_norm": 2.996992033740693,
      "learning_rate": 3.784570022957876e-07,
      "loss": 0.8204,
      "step": 11184
    },
    {
      "epoch": 0.8069694455466975,
      "grad_norm": 3.790495108118839,
      "learning_rate": 3.7818347153594533e-07,
      "loss": 0.9459,
      "step": 11185
    },
    {
      "epoch": 0.807041593016125,
      "grad_norm": 5.175335881486926,
      "learning_rate": 3.7791002933750105e-07,
      "loss": 1.015,
      "step": 11186
    },
    {
      "epoch": 0.8071137404855525,
      "grad_norm": 4.616909477009105,
      "learning_rate": 3.776366757153846e-07,
      "loss": 0.9276,
      "step": 11187
    },
    {
      "epoch": 0.80718588795498,
      "grad_norm": 6.078085460387138,
      "learning_rate": 3.773634106845238e-07,
      "loss": 1.0037,
      "step": 11188
    },
    {
      "epoch": 0.8072580354244074,
      "grad_norm": 6.209637737427049,
      "learning_rate": 3.770902342598401e-07,
      "loss": 0.9713,
      "step": 11189
    },
    {
      "epoch": 0.807330182893835,
      "grad_norm": 0.7286504462211769,
      "learning_rate": 3.768171464562515e-07,
      "loss": 0.8373,
      "step": 11190
    },
    {
      "epoch": 0.8074023303632625,
      "grad_norm": 4.786916728133566,
      "learning_rate": 3.765441472886692e-07,
      "loss": 0.8515,
      "step": 11191
    },
    {
      "epoch": 0.80747447783269,
      "grad_norm": 3.6076095689192607,
      "learning_rate": 3.7627123677200066e-07,
      "loss": 0.9898,
      "step": 11192
    },
    {
      "epoch": 0.8075466253021175,
      "grad_norm": 2.868509512734095,
      "learning_rate": 3.759984149211501e-07,
      "loss": 1.0117,
      "step": 11193
    },
    {
      "epoch": 0.807618772771545,
      "grad_norm": 2.918610406514903,
      "learning_rate": 3.757256817510137e-07,
      "loss": 0.9223,
      "step": 11194
    },
    {
      "epoch": 0.8076909202409726,
      "grad_norm": 3.5648882707419505,
      "learning_rate": 3.7545303727648505e-07,
      "loss": 0.9148,
      "step": 11195
    },
    {
      "epoch": 0.8077630677104001,
      "grad_norm": 0.8283000617058001,
      "learning_rate": 3.751804815124526e-07,
      "loss": 0.9113,
      "step": 11196
    },
    {
      "epoch": 0.8078352151798276,
      "grad_norm": 3.689849363731041,
      "learning_rate": 3.749080144737986e-07,
      "loss": 0.906,
      "step": 11197
    },
    {
      "epoch": 0.8079073626492551,
      "grad_norm": 3.7191929913978896,
      "learning_rate": 3.746356361754017e-07,
      "loss": 0.9557,
      "step": 11198
    },
    {
      "epoch": 0.8079795101186826,
      "grad_norm": 5.193969185768236,
      "learning_rate": 3.7436334663213607e-07,
      "loss": 0.9121,
      "step": 11199
    },
    {
      "epoch": 0.8080516575881102,
      "grad_norm": 3.7091224488024066,
      "learning_rate": 3.7409114585887093e-07,
      "loss": 0.8251,
      "step": 11200
    },
    {
      "epoch": 0.8081238050575376,
      "grad_norm": 4.275827356776807,
      "learning_rate": 3.7381903387046874e-07,
      "loss": 0.9214,
      "step": 11201
    },
    {
      "epoch": 0.8081959525269651,
      "grad_norm": 3.2607993651863953,
      "learning_rate": 3.7354701068178927e-07,
      "loss": 0.9394,
      "step": 11202
    },
    {
      "epoch": 0.8082680999963926,
      "grad_norm": 3.7384548485766405,
      "learning_rate": 3.732750763076873e-07,
      "loss": 0.8942,
      "step": 11203
    },
    {
      "epoch": 0.8083402474658201,
      "grad_norm": 3.574036773885236,
      "learning_rate": 3.730032307630102e-07,
      "loss": 0.9844,
      "step": 11204
    },
    {
      "epoch": 0.8084123949352476,
      "grad_norm": 4.31896973863808,
      "learning_rate": 3.727314740626044e-07,
      "loss": 1.0006,
      "step": 11205
    },
    {
      "epoch": 0.8084845424046752,
      "grad_norm": 26.843539680127762,
      "learning_rate": 3.7245980622130893e-07,
      "loss": 0.9121,
      "step": 11206
    },
    {
      "epoch": 0.8085566898741027,
      "grad_norm": 5.135501801544043,
      "learning_rate": 3.721882272539587e-07,
      "loss": 0.994,
      "step": 11207
    },
    {
      "epoch": 0.8086288373435302,
      "grad_norm": 4.045480141641535,
      "learning_rate": 3.719167371753831e-07,
      "loss": 0.8062,
      "step": 11208
    },
    {
      "epoch": 0.8087009848129577,
      "grad_norm": 3.582842386606023,
      "learning_rate": 3.7164533600040704e-07,
      "loss": 0.9404,
      "step": 11209
    },
    {
      "epoch": 0.8087731322823852,
      "grad_norm": 3.38453528252762,
      "learning_rate": 3.7137402374385227e-07,
      "loss": 0.9438,
      "step": 11210
    },
    {
      "epoch": 0.8088452797518128,
      "grad_norm": 7.272436743048414,
      "learning_rate": 3.7110280042053256e-07,
      "loss": 1.0814,
      "step": 11211
    },
    {
      "epoch": 0.8089174272212402,
      "grad_norm": 2.914345571560386,
      "learning_rate": 3.708316660452588e-07,
      "loss": 0.9806,
      "step": 11212
    },
    {
      "epoch": 0.8089895746906677,
      "grad_norm": 3.2125039839070477,
      "learning_rate": 3.7056062063283756e-07,
      "loss": 1.0188,
      "step": 11213
    },
    {
      "epoch": 0.8090617221600952,
      "grad_norm": 5.286357869914654,
      "learning_rate": 3.702896641980682e-07,
      "loss": 0.8962,
      "step": 11214
    },
    {
      "epoch": 0.8091338696295227,
      "grad_norm": 0.774390894963002,
      "learning_rate": 3.700187967557471e-07,
      "loss": 0.8705,
      "step": 11215
    },
    {
      "epoch": 0.8092060170989502,
      "grad_norm": 3.5639815345140233,
      "learning_rate": 3.697480183206652e-07,
      "loss": 0.9231,
      "step": 11216
    },
    {
      "epoch": 0.8092781645683778,
      "grad_norm": 3.6138737373187553,
      "learning_rate": 3.694773289076101e-07,
      "loss": 0.9016,
      "step": 11217
    },
    {
      "epoch": 0.8093503120378053,
      "grad_norm": 3.177321845101758,
      "learning_rate": 3.692067285313616e-07,
      "loss": 0.9522,
      "step": 11218
    },
    {
      "epoch": 0.8094224595072328,
      "grad_norm": 4.312830373299815,
      "learning_rate": 3.6893621720669654e-07,
      "loss": 0.9552,
      "step": 11219
    },
    {
      "epoch": 0.8094946069766603,
      "grad_norm": 3.4745187988826687,
      "learning_rate": 3.6866579494838713e-07,
      "loss": 0.949,
      "step": 11220
    },
    {
      "epoch": 0.8095667544460878,
      "grad_norm": 4.38398887364166,
      "learning_rate": 3.683954617711993e-07,
      "loss": 1.0277,
      "step": 11221
    },
    {
      "epoch": 0.8096389019155154,
      "grad_norm": 10.522664994033601,
      "learning_rate": 3.681252176898948e-07,
      "loss": 0.9227,
      "step": 11222
    },
    {
      "epoch": 0.8097110493849429,
      "grad_norm": 3.865666130667399,
      "learning_rate": 3.678550627192316e-07,
      "loss": 0.9036,
      "step": 11223
    },
    {
      "epoch": 0.8097831968543703,
      "grad_norm": 5.437653462666935,
      "learning_rate": 3.6758499687396196e-07,
      "loss": 0.8712,
      "step": 11224
    },
    {
      "epoch": 0.8098553443237978,
      "grad_norm": 4.0509415744701265,
      "learning_rate": 3.673150201688322e-07,
      "loss": 0.9532,
      "step": 11225
    },
    {
      "epoch": 0.8099274917932253,
      "grad_norm": 4.282382119288351,
      "learning_rate": 3.6704513261858506e-07,
      "loss": 0.9439,
      "step": 11226
    },
    {
      "epoch": 0.8099996392626528,
      "grad_norm": 6.435184099221182,
      "learning_rate": 3.6677533423795895e-07,
      "loss": 0.9028,
      "step": 11227
    },
    {
      "epoch": 0.8100717867320804,
      "grad_norm": 6.467124550460728,
      "learning_rate": 3.665056250416847e-07,
      "loss": 0.9825,
      "step": 11228
    },
    {
      "epoch": 0.8101439342015079,
      "grad_norm": 4.435254670650861,
      "learning_rate": 3.662360050444917e-07,
      "loss": 0.8678,
      "step": 11229
    },
    {
      "epoch": 0.8102160816709354,
      "grad_norm": 13.11822044027052,
      "learning_rate": 3.6596647426110305e-07,
      "loss": 0.9022,
      "step": 11230
    },
    {
      "epoch": 0.8102882291403629,
      "grad_norm": 2.9484406310015836,
      "learning_rate": 3.656970327062359e-07,
      "loss": 1.0303,
      "step": 11231
    },
    {
      "epoch": 0.8103603766097904,
      "grad_norm": 5.473766358623988,
      "learning_rate": 3.654276803946037e-07,
      "loss": 0.967,
      "step": 11232
    },
    {
      "epoch": 0.810432524079218,
      "grad_norm": 3.2880005198011895,
      "learning_rate": 3.6515841734091435e-07,
      "loss": 1.0297,
      "step": 11233
    },
    {
      "epoch": 0.8105046715486455,
      "grad_norm": 7.744294065708758,
      "learning_rate": 3.648892435598727e-07,
      "loss": 1.003,
      "step": 11234
    },
    {
      "epoch": 0.810576819018073,
      "grad_norm": 10.63118357517094,
      "learning_rate": 3.646201590661762e-07,
      "loss": 0.861,
      "step": 11235
    },
    {
      "epoch": 0.8106489664875004,
      "grad_norm": 2.8975190303727203,
      "learning_rate": 3.6435116387451894e-07,
      "loss": 0.9926,
      "step": 11236
    },
    {
      "epoch": 0.8107211139569279,
      "grad_norm": 3.2617080779671968,
      "learning_rate": 3.6408225799958993e-07,
      "loss": 1.0228,
      "step": 11237
    },
    {
      "epoch": 0.8107932614263554,
      "grad_norm": 3.753457065347391,
      "learning_rate": 3.638134414560725e-07,
      "loss": 0.9296,
      "step": 11238
    },
    {
      "epoch": 0.810865408895783,
      "grad_norm": 0.7665749320488965,
      "learning_rate": 3.635447142586461e-07,
      "loss": 0.8734,
      "step": 11239
    },
    {
      "epoch": 0.8109375563652105,
      "grad_norm": 4.161950583183365,
      "learning_rate": 3.632760764219847e-07,
      "loss": 1.0262,
      "step": 11240
    },
    {
      "epoch": 0.811009703834638,
      "grad_norm": 6.979250809650966,
      "learning_rate": 3.630075279607581e-07,
      "loss": 0.9478,
      "step": 11241
    },
    {
      "epoch": 0.8110818513040655,
      "grad_norm": 3.569407408868912,
      "learning_rate": 3.627390688896301e-07,
      "loss": 0.8913,
      "step": 11242
    },
    {
      "epoch": 0.811153998773493,
      "grad_norm": 2.3913979870099813,
      "learning_rate": 3.624706992232609e-07,
      "loss": 0.9769,
      "step": 11243
    },
    {
      "epoch": 0.8112261462429206,
      "grad_norm": 2.880300528422894,
      "learning_rate": 3.622024189763053e-07,
      "loss": 0.9382,
      "step": 11244
    },
    {
      "epoch": 0.8112982937123481,
      "grad_norm": 5.347684783525046,
      "learning_rate": 3.6193422816341235e-07,
      "loss": 1.0411,
      "step": 11245
    },
    {
      "epoch": 0.8113704411817756,
      "grad_norm": 3.594740888571698,
      "learning_rate": 3.6166612679922647e-07,
      "loss": 0.9811,
      "step": 11246
    },
    {
      "epoch": 0.8114425886512031,
      "grad_norm": 2.6009411685601025,
      "learning_rate": 3.613981148983898e-07,
      "loss": 0.9947,
      "step": 11247
    },
    {
      "epoch": 0.8115147361206305,
      "grad_norm": 3.3334667338062736,
      "learning_rate": 3.6113019247553543e-07,
      "loss": 0.8304,
      "step": 11248
    },
    {
      "epoch": 0.811586883590058,
      "grad_norm": 5.353778909181667,
      "learning_rate": 3.6086235954529445e-07,
      "loss": 0.9045,
      "step": 11249
    },
    {
      "epoch": 0.8116590310594856,
      "grad_norm": 3.1926128296416834,
      "learning_rate": 3.605946161222921e-07,
      "loss": 0.9386,
      "step": 11250
    },
    {
      "epoch": 0.8117311785289131,
      "grad_norm": 4.058733558633233,
      "learning_rate": 3.6032696222114954e-07,
      "loss": 1.0204,
      "step": 11251
    },
    {
      "epoch": 0.8118033259983406,
      "grad_norm": 4.043730349839355,
      "learning_rate": 3.6005939785648075e-07,
      "loss": 0.8854,
      "step": 11252
    },
    {
      "epoch": 0.8118754734677681,
      "grad_norm": 2.8489709234337233,
      "learning_rate": 3.597919230428976e-07,
      "loss": 0.9665,
      "step": 11253
    },
    {
      "epoch": 0.8119476209371956,
      "grad_norm": 7.272070341351254,
      "learning_rate": 3.5952453779500646e-07,
      "loss": 1.0292,
      "step": 11254
    },
    {
      "epoch": 0.8120197684066232,
      "grad_norm": 0.7669203951705379,
      "learning_rate": 3.5925724212740694e-07,
      "loss": 0.8821,
      "step": 11255
    },
    {
      "epoch": 0.8120919158760507,
      "grad_norm": 6.115815731109364,
      "learning_rate": 3.589900360546956e-07,
      "loss": 1.0078,
      "step": 11256
    },
    {
      "epoch": 0.8121640633454782,
      "grad_norm": 0.7474486745481804,
      "learning_rate": 3.587229195914638e-07,
      "loss": 0.7878,
      "step": 11257
    },
    {
      "epoch": 0.8122362108149057,
      "grad_norm": 0.7544206435736656,
      "learning_rate": 3.5845589275229694e-07,
      "loss": 0.8067,
      "step": 11258
    },
    {
      "epoch": 0.8123083582843332,
      "grad_norm": 3.247976259896472,
      "learning_rate": 3.581889555517772e-07,
      "loss": 0.9854,
      "step": 11259
    },
    {
      "epoch": 0.8123805057537606,
      "grad_norm": 5.688858373050172,
      "learning_rate": 3.5792210800448095e-07,
      "loss": 0.9584,
      "step": 11260
    },
    {
      "epoch": 0.8124526532231882,
      "grad_norm": 3.5505900147897953,
      "learning_rate": 3.5765535012498014e-07,
      "loss": 0.9882,
      "step": 11261
    },
    {
      "epoch": 0.8125248006926157,
      "grad_norm": 0.6725720848036483,
      "learning_rate": 3.5738868192784045e-07,
      "loss": 0.795,
      "step": 11262
    },
    {
      "epoch": 0.8125969481620432,
      "grad_norm": 3.4174957820082157,
      "learning_rate": 3.571221034276235e-07,
      "loss": 1.0197,
      "step": 11263
    },
    {
      "epoch": 0.8126690956314707,
      "grad_norm": 4.285603558154442,
      "learning_rate": 3.568556146388879e-07,
      "loss": 0.9944,
      "step": 11264
    },
    {
      "epoch": 0.8127412431008982,
      "grad_norm": 3.829253734862002,
      "learning_rate": 3.5658921557618384e-07,
      "loss": 0.992,
      "step": 11265
    },
    {
      "epoch": 0.8128133905703258,
      "grad_norm": 3.314417158301466,
      "learning_rate": 3.563229062540589e-07,
      "loss": 0.9961,
      "step": 11266
    },
    {
      "epoch": 0.8128855380397533,
      "grad_norm": 3.2182271588309925,
      "learning_rate": 3.5605668668705555e-07,
      "loss": 0.8917,
      "step": 11267
    },
    {
      "epoch": 0.8129576855091808,
      "grad_norm": 6.2716993914596975,
      "learning_rate": 3.557905568897115e-07,
      "loss": 0.9756,
      "step": 11268
    },
    {
      "epoch": 0.8130298329786083,
      "grad_norm": 3.4173321114680677,
      "learning_rate": 3.5552451687655773e-07,
      "loss": 0.9443,
      "step": 11269
    },
    {
      "epoch": 0.8131019804480358,
      "grad_norm": 0.6896010718316292,
      "learning_rate": 3.5525856666212195e-07,
      "loss": 0.8067,
      "step": 11270
    },
    {
      "epoch": 0.8131741279174632,
      "grad_norm": 3.535135207982957,
      "learning_rate": 3.5499270626092834e-07,
      "loss": 0.9656,
      "step": 11271
    },
    {
      "epoch": 0.8132462753868908,
      "grad_norm": 3.966088070598059,
      "learning_rate": 3.547269356874929e-07,
      "loss": 0.8504,
      "step": 11272
    },
    {
      "epoch": 0.8133184228563183,
      "grad_norm": 6.577188094372835,
      "learning_rate": 3.5446125495632885e-07,
      "loss": 0.9428,
      "step": 11273
    },
    {
      "epoch": 0.8133905703257458,
      "grad_norm": 3.0239290532766288,
      "learning_rate": 3.541956640819448e-07,
      "loss": 1.0091,
      "step": 11274
    },
    {
      "epoch": 0.8134627177951733,
      "grad_norm": 4.048042276304924,
      "learning_rate": 3.5393016307884225e-07,
      "loss": 0.984,
      "step": 11275
    },
    {
      "epoch": 0.8135348652646008,
      "grad_norm": 4.993491132421805,
      "learning_rate": 3.5366475196151944e-07,
      "loss": 0.999,
      "step": 11276
    },
    {
      "epoch": 0.8136070127340284,
      "grad_norm": 3.3667287845986955,
      "learning_rate": 3.533994307444708e-07,
      "loss": 0.9563,
      "step": 11277
    },
    {
      "epoch": 0.8136791602034559,
      "grad_norm": 10.444984539307235,
      "learning_rate": 3.5313419944218394e-07,
      "loss": 1.074,
      "step": 11278
    },
    {
      "epoch": 0.8137513076728834,
      "grad_norm": 3.673791912575716,
      "learning_rate": 3.528690580691416e-07,
      "loss": 0.8934,
      "step": 11279
    },
    {
      "epoch": 0.8138234551423109,
      "grad_norm": 2.9276907257836924,
      "learning_rate": 3.5260400663982237e-07,
      "loss": 0.8568,
      "step": 11280
    },
    {
      "epoch": 0.8138956026117384,
      "grad_norm": 13.028314712691342,
      "learning_rate": 3.523390451687005e-07,
      "loss": 1.0573,
      "step": 11281
    },
    {
      "epoch": 0.813967750081166,
      "grad_norm": 3.001443674494751,
      "learning_rate": 3.5207417367024304e-07,
      "loss": 0.9488,
      "step": 11282
    },
    {
      "epoch": 0.8140398975505934,
      "grad_norm": 4.840446871785611,
      "learning_rate": 3.518093921589149e-07,
      "loss": 0.9749,
      "step": 11283
    },
    {
      "epoch": 0.8141120450200209,
      "grad_norm": 2.6945979401256586,
      "learning_rate": 3.51544700649175e-07,
      "loss": 0.9488,
      "step": 11284
    },
    {
      "epoch": 0.8141841924894484,
      "grad_norm": 3.9398936912794973,
      "learning_rate": 3.5128009915547607e-07,
      "loss": 0.9087,
      "step": 11285
    },
    {
      "epoch": 0.8142563399588759,
      "grad_norm": 3.8301484657376252,
      "learning_rate": 3.510155876922676e-07,
      "loss": 0.9351,
      "step": 11286
    },
    {
      "epoch": 0.8143284874283034,
      "grad_norm": 4.598580448211369,
      "learning_rate": 3.50751166273993e-07,
      "loss": 0.9685,
      "step": 11287
    },
    {
      "epoch": 0.814400634897731,
      "grad_norm": 6.336247560038184,
      "learning_rate": 3.504868349150931e-07,
      "loss": 0.8721,
      "step": 11288
    },
    {
      "epoch": 0.8144727823671585,
      "grad_norm": 2.8483775288959046,
      "learning_rate": 3.502225936300003e-07,
      "loss": 1.0037,
      "step": 11289
    },
    {
      "epoch": 0.814544929836586,
      "grad_norm": 3.432070241142614,
      "learning_rate": 3.4995844243314433e-07,
      "loss": 0.9794,
      "step": 11290
    },
    {
      "epoch": 0.8146170773060135,
      "grad_norm": 5.409105373739161,
      "learning_rate": 3.4969438133895015e-07,
      "loss": 1.0942,
      "step": 11291
    },
    {
      "epoch": 0.814689224775441,
      "grad_norm": 5.864024522446467,
      "learning_rate": 3.4943041036183594e-07,
      "loss": 1.0844,
      "step": 11292
    },
    {
      "epoch": 0.8147613722448686,
      "grad_norm": 0.7828335163394269,
      "learning_rate": 3.4916652951621693e-07,
      "loss": 0.8584,
      "step": 11293
    },
    {
      "epoch": 0.8148335197142961,
      "grad_norm": 3.1831404848918647,
      "learning_rate": 3.48902738816502e-07,
      "loss": 0.9026,
      "step": 11294
    },
    {
      "epoch": 0.8149056671837235,
      "grad_norm": 7.485417239702902,
      "learning_rate": 3.486390382770974e-07,
      "loss": 0.8981,
      "step": 11295
    },
    {
      "epoch": 0.814977814653151,
      "grad_norm": 4.116659575750959,
      "learning_rate": 3.4837542791240116e-07,
      "loss": 0.9456,
      "step": 11296
    },
    {
      "epoch": 0.8150499621225785,
      "grad_norm": 7.509968808207138,
      "learning_rate": 3.4811190773680886e-07,
      "loss": 0.8196,
      "step": 11297
    },
    {
      "epoch": 0.815122109592006,
      "grad_norm": 5.256697605955567,
      "learning_rate": 3.478484777647106e-07,
      "loss": 0.829,
      "step": 11298
    },
    {
      "epoch": 0.8151942570614336,
      "grad_norm": 5.817547647095012,
      "learning_rate": 3.475851380104904e-07,
      "loss": 1.1018,
      "step": 11299
    },
    {
      "epoch": 0.8152664045308611,
      "grad_norm": 0.9669705013462331,
      "learning_rate": 3.473218884885285e-07,
      "loss": 0.7718,
      "step": 11300
    },
    {
      "epoch": 0.8153385520002886,
      "grad_norm": 3.1887171796173637,
      "learning_rate": 3.4705872921320103e-07,
      "loss": 0.9154,
      "step": 11301
    },
    {
      "epoch": 0.8154106994697161,
      "grad_norm": 5.1901969792906595,
      "learning_rate": 3.467956601988771e-07,
      "loss": 0.9919,
      "step": 11302
    },
    {
      "epoch": 0.8154828469391436,
      "grad_norm": 4.224330799443268,
      "learning_rate": 3.4653268145992185e-07,
      "loss": 0.9725,
      "step": 11303
    },
    {
      "epoch": 0.8155549944085712,
      "grad_norm": 4.467212759388843,
      "learning_rate": 3.4626979301069615e-07,
      "loss": 1.0524,
      "step": 11304
    },
    {
      "epoch": 0.8156271418779987,
      "grad_norm": 2.843725518760523,
      "learning_rate": 3.460069948655557e-07,
      "loss": 0.9387,
      "step": 11305
    },
    {
      "epoch": 0.8156992893474262,
      "grad_norm": 4.27856829111142,
      "learning_rate": 3.457442870388494e-07,
      "loss": 1.0353,
      "step": 11306
    },
    {
      "epoch": 0.8157714368168536,
      "grad_norm": 2.8559364530722346,
      "learning_rate": 3.454816695449241e-07,
      "loss": 0.8871,
      "step": 11307
    },
    {
      "epoch": 0.8158435842862811,
      "grad_norm": 3.588557995822261,
      "learning_rate": 3.4521914239812076e-07,
      "loss": 1.028,
      "step": 11308
    },
    {
      "epoch": 0.8159157317557086,
      "grad_norm": 7.38131236826027,
      "learning_rate": 3.4495670561277354e-07,
      "loss": 0.974,
      "step": 11309
    },
    {
      "epoch": 0.8159878792251362,
      "grad_norm": 2.7374799266061802,
      "learning_rate": 3.446943592032139e-07,
      "loss": 0.989,
      "step": 11310
    },
    {
      "epoch": 0.8160600266945637,
      "grad_norm": 10.875414960990247,
      "learning_rate": 3.4443210318376716e-07,
      "loss": 0.9106,
      "step": 11311
    },
    {
      "epoch": 0.8161321741639912,
      "grad_norm": 3.7211174361271784,
      "learning_rate": 3.4416993756875544e-07,
      "loss": 0.9817,
      "step": 11312
    },
    {
      "epoch": 0.8162043216334187,
      "grad_norm": 4.643436902916019,
      "learning_rate": 3.4390786237249337e-07,
      "loss": 0.9962,
      "step": 11313
    },
    {
      "epoch": 0.8162764691028462,
      "grad_norm": 7.5289878138578965,
      "learning_rate": 3.436458776092923e-07,
      "loss": 0.9843,
      "step": 11314
    },
    {
      "epoch": 0.8163486165722738,
      "grad_norm": 3.2150855083574066,
      "learning_rate": 3.4338398329345863e-07,
      "loss": 0.9748,
      "step": 11315
    },
    {
      "epoch": 0.8164207640417013,
      "grad_norm": 8.77264308305072,
      "learning_rate": 3.4312217943929266e-07,
      "loss": 0.9334,
      "step": 11316
    },
    {
      "epoch": 0.8164929115111288,
      "grad_norm": 0.7902781498428956,
      "learning_rate": 3.4286046606109075e-07,
      "loss": 0.9049,
      "step": 11317
    },
    {
      "epoch": 0.8165650589805562,
      "grad_norm": 2.7509257752345135,
      "learning_rate": 3.425988431731448e-07,
      "loss": 0.9034,
      "step": 11318
    },
    {
      "epoch": 0.8166372064499837,
      "grad_norm": 4.079895103078512,
      "learning_rate": 3.423373107897396e-07,
      "loss": 0.962,
      "step": 11319
    },
    {
      "epoch": 0.8167093539194112,
      "grad_norm": 3.61105637794162,
      "learning_rate": 3.420758689251579e-07,
      "loss": 0.9834,
      "step": 11320
    },
    {
      "epoch": 0.8167815013888388,
      "grad_norm": 5.956905416191462,
      "learning_rate": 3.4181451759367575e-07,
      "loss": 0.8979,
      "step": 11321
    },
    {
      "epoch": 0.8168536488582663,
      "grad_norm": 4.500536356856775,
      "learning_rate": 3.415532568095647e-07,
      "loss": 1.0176,
      "step": 11322
    },
    {
      "epoch": 0.8169257963276938,
      "grad_norm": 2.694020279426399,
      "learning_rate": 3.412920865870905e-07,
      "loss": 0.9276,
      "step": 11323
    },
    {
      "epoch": 0.8169979437971213,
      "grad_norm": 6.147033298395416,
      "learning_rate": 3.4103100694051465e-07,
      "loss": 1.0196,
      "step": 11324
    },
    {
      "epoch": 0.8170700912665488,
      "grad_norm": 3.0299448566014218,
      "learning_rate": 3.407700178840951e-07,
      "loss": 0.9315,
      "step": 11325
    },
    {
      "epoch": 0.8171422387359764,
      "grad_norm": 3.5345735016925572,
      "learning_rate": 3.4050911943208215e-07,
      "loss": 0.9964,
      "step": 11326
    },
    {
      "epoch": 0.8172143862054039,
      "grad_norm": 5.761781315867929,
      "learning_rate": 3.402483115987229e-07,
      "loss": 1.0073,
      "step": 11327
    },
    {
      "epoch": 0.8172865336748314,
      "grad_norm": 4.166293521702999,
      "learning_rate": 3.3998759439826e-07,
      "loss": 1.015,
      "step": 11328
    },
    {
      "epoch": 0.8173586811442589,
      "grad_norm": 3.8107981401013107,
      "learning_rate": 3.397269678449286e-07,
      "loss": 0.9726,
      "step": 11329
    },
    {
      "epoch": 0.8174308286136863,
      "grad_norm": 3.9822027530601067,
      "learning_rate": 3.3946643195296096e-07,
      "loss": 0.9816,
      "step": 11330
    },
    {
      "epoch": 0.8175029760831138,
      "grad_norm": 2.8963711099735567,
      "learning_rate": 3.3920598673658486e-07,
      "loss": 0.9748,
      "step": 11331
    },
    {
      "epoch": 0.8175751235525414,
      "grad_norm": 0.7821468164987003,
      "learning_rate": 3.389456322100221e-07,
      "loss": 0.8496,
      "step": 11332
    },
    {
      "epoch": 0.8176472710219689,
      "grad_norm": 2.9958720417378157,
      "learning_rate": 3.3868536838748904e-07,
      "loss": 1.0071,
      "step": 11333
    },
    {
      "epoch": 0.8177194184913964,
      "grad_norm": 3.23870971937318,
      "learning_rate": 3.3842519528319825e-07,
      "loss": 0.9283,
      "step": 11334
    },
    {
      "epoch": 0.8177915659608239,
      "grad_norm": 5.7918155934605124,
      "learning_rate": 3.3816511291135676e-07,
      "loss": 1.0814,
      "step": 11335
    },
    {
      "epoch": 0.8178637134302514,
      "grad_norm": 5.048610140966398,
      "learning_rate": 3.3790512128616566e-07,
      "loss": 0.9767,
      "step": 11336
    },
    {
      "epoch": 0.817935860899679,
      "grad_norm": 3.9642308254688774,
      "learning_rate": 3.3764522042182366e-07,
      "loss": 0.9751,
      "step": 11337
    },
    {
      "epoch": 0.8180080083691065,
      "grad_norm": 3.325934607151404,
      "learning_rate": 3.373854103325222e-07,
      "loss": 0.942,
      "step": 11338
    },
    {
      "epoch": 0.818080155838534,
      "grad_norm": 0.8214572729699992,
      "learning_rate": 3.3712569103244913e-07,
      "loss": 0.864,
      "step": 11339
    },
    {
      "epoch": 0.8181523033079615,
      "grad_norm": 5.804277079513993,
      "learning_rate": 3.3686606253578596e-07,
      "loss": 0.892,
      "step": 11340
    },
    {
      "epoch": 0.818224450777389,
      "grad_norm": 3.839349724625695,
      "learning_rate": 3.366065248567103e-07,
      "loss": 1.0189,
      "step": 11341
    },
    {
      "epoch": 0.8182965982468164,
      "grad_norm": 3.4189974735381217,
      "learning_rate": 3.3634707800939535e-07,
      "loss": 0.9931,
      "step": 11342
    },
    {
      "epoch": 0.818368745716244,
      "grad_norm": 37.58635402362222,
      "learning_rate": 3.360877220080065e-07,
      "loss": 0.8737,
      "step": 11343
    },
    {
      "epoch": 0.8184408931856715,
      "grad_norm": 6.323693652440171,
      "learning_rate": 3.358284568667082e-07,
      "loss": 1.0276,
      "step": 11344
    },
    {
      "epoch": 0.818513040655099,
      "grad_norm": 0.8321172463833857,
      "learning_rate": 3.355692825996579e-07,
      "loss": 0.8759,
      "step": 11345
    },
    {
      "epoch": 0.8185851881245265,
      "grad_norm": 4.714674446460583,
      "learning_rate": 3.35310199221007e-07,
      "loss": 0.9431,
      "step": 11346
    },
    {
      "epoch": 0.818657335593954,
      "grad_norm": 2.8396294892091696,
      "learning_rate": 3.3505120674490386e-07,
      "loss": 1.0337,
      "step": 11347
    },
    {
      "epoch": 0.8187294830633816,
      "grad_norm": 4.566855037358968,
      "learning_rate": 3.347923051854902e-07,
      "loss": 0.9109,
      "step": 11348
    },
    {
      "epoch": 0.8188016305328091,
      "grad_norm": 5.466692030467918,
      "learning_rate": 3.3453349455690536e-07,
      "loss": 0.9318,
      "step": 11349
    },
    {
      "epoch": 0.8188737780022366,
      "grad_norm": 0.7452482055826732,
      "learning_rate": 3.342747748732806e-07,
      "loss": 0.8578,
      "step": 11350
    },
    {
      "epoch": 0.8189459254716641,
      "grad_norm": 4.440790165534905,
      "learning_rate": 3.340161461487439e-07,
      "loss": 0.9161,
      "step": 11351
    },
    {
      "epoch": 0.8190180729410916,
      "grad_norm": 2.258126735269538,
      "learning_rate": 3.3375760839741896e-07,
      "loss": 0.9692,
      "step": 11352
    },
    {
      "epoch": 0.8190902204105192,
      "grad_norm": 14.004002680107325,
      "learning_rate": 3.334991616334222e-07,
      "loss": 1.0143,
      "step": 11353
    },
    {
      "epoch": 0.8191623678799466,
      "grad_norm": 4.536223319703913,
      "learning_rate": 3.3324080587086644e-07,
      "loss": 1.0114,
      "step": 11354
    },
    {
      "epoch": 0.8192345153493741,
      "grad_norm": 0.7546819935473368,
      "learning_rate": 3.329825411238607e-07,
      "loss": 0.8219,
      "step": 11355
    },
    {
      "epoch": 0.8193066628188016,
      "grad_norm": 6.016349451609086,
      "learning_rate": 3.327243674065079e-07,
      "loss": 0.952,
      "step": 11356
    },
    {
      "epoch": 0.8193788102882291,
      "grad_norm": 4.327146564156951,
      "learning_rate": 3.324662847329047e-07,
      "loss": 0.9256,
      "step": 11357
    },
    {
      "epoch": 0.8194509577576566,
      "grad_norm": 4.9825715536366735,
      "learning_rate": 3.322082931171448e-07,
      "loss": 1.0138,
      "step": 11358
    },
    {
      "epoch": 0.8195231052270842,
      "grad_norm": 3.3429655777108693,
      "learning_rate": 3.3195039257331647e-07,
      "loss": 1.043,
      "step": 11359
    },
    {
      "epoch": 0.8195952526965117,
      "grad_norm": 0.6734751564702852,
      "learning_rate": 3.316925831155013e-07,
      "loss": 0.7443,
      "step": 11360
    },
    {
      "epoch": 0.8196674001659392,
      "grad_norm": 3.3108241322412986,
      "learning_rate": 3.3143486475777893e-07,
      "loss": 0.8482,
      "step": 11361
    },
    {
      "epoch": 0.8197395476353667,
      "grad_norm": 3.397283248443351,
      "learning_rate": 3.3117723751422233e-07,
      "loss": 0.954,
      "step": 11362
    },
    {
      "epoch": 0.8198116951047942,
      "grad_norm": 3.2098939411075844,
      "learning_rate": 3.3091970139889823e-07,
      "loss": 0.9348,
      "step": 11363
    },
    {
      "epoch": 0.8198838425742218,
      "grad_norm": 3.5821957815607512,
      "learning_rate": 3.3066225642587075e-07,
      "loss": 0.9719,
      "step": 11364
    },
    {
      "epoch": 0.8199559900436493,
      "grad_norm": 6.05774179515257,
      "learning_rate": 3.3040490260919773e-07,
      "loss": 0.9647,
      "step": 11365
    },
    {
      "epoch": 0.8200281375130767,
      "grad_norm": 2.781644128883747,
      "learning_rate": 3.301476399629328e-07,
      "loss": 0.8794,
      "step": 11366
    },
    {
      "epoch": 0.8201002849825042,
      "grad_norm": 0.7158677734628384,
      "learning_rate": 3.298904685011228e-07,
      "loss": 0.8229,
      "step": 11367
    },
    {
      "epoch": 0.8201724324519317,
      "grad_norm": 9.718285650708644,
      "learning_rate": 3.2963338823781194e-07,
      "loss": 1.0515,
      "step": 11368
    },
    {
      "epoch": 0.8202445799213592,
      "grad_norm": 2.8287086516833786,
      "learning_rate": 3.293763991870391e-07,
      "loss": 0.9419,
      "step": 11369
    },
    {
      "epoch": 0.8203167273907868,
      "grad_norm": 6.691195598789484,
      "learning_rate": 3.2911950136283605e-07,
      "loss": 0.9072,
      "step": 11370
    },
    {
      "epoch": 0.8203888748602143,
      "grad_norm": 3.0627011602859326,
      "learning_rate": 3.288626947792317e-07,
      "loss": 0.9168,
      "step": 11371
    },
    {
      "epoch": 0.8204610223296418,
      "grad_norm": 3.305516925732081,
      "learning_rate": 3.286059794502496e-07,
      "loss": 1.0026,
      "step": 11372
    },
    {
      "epoch": 0.8205331697990693,
      "grad_norm": 3.7457380871797517,
      "learning_rate": 3.2834935538990683e-07,
      "loss": 0.9047,
      "step": 11373
    },
    {
      "epoch": 0.8206053172684968,
      "grad_norm": 5.125155935008641,
      "learning_rate": 3.2809282261221793e-07,
      "loss": 1.0202,
      "step": 11374
    },
    {
      "epoch": 0.8206774647379244,
      "grad_norm": 8.339977311576884,
      "learning_rate": 3.2783638113119085e-07,
      "loss": 1.0244,
      "step": 11375
    },
    {
      "epoch": 0.8207496122073519,
      "grad_norm": 5.277628904328784,
      "learning_rate": 3.275800309608294e-07,
      "loss": 1.0029,
      "step": 11376
    },
    {
      "epoch": 0.8208217596767793,
      "grad_norm": 3.0725456056865026,
      "learning_rate": 3.2732377211513075e-07,
      "loss": 0.9634,
      "step": 11377
    },
    {
      "epoch": 0.8208939071462068,
      "grad_norm": 5.107071009996577,
      "learning_rate": 3.2706760460808847e-07,
      "loss": 0.9664,
      "step": 11378
    },
    {
      "epoch": 0.8209660546156343,
      "grad_norm": 2.8537520689921285,
      "learning_rate": 3.268115284536923e-07,
      "loss": 1.0169,
      "step": 11379
    },
    {
      "epoch": 0.8210382020850618,
      "grad_norm": 6.1504768372808085,
      "learning_rate": 3.26555543665924e-07,
      "loss": 0.8899,
      "step": 11380
    },
    {
      "epoch": 0.8211103495544894,
      "grad_norm": 4.22120854830419,
      "learning_rate": 3.262996502587627e-07,
      "loss": 0.9421,
      "step": 11381
    },
    {
      "epoch": 0.8211824970239169,
      "grad_norm": 3.408878816885107,
      "learning_rate": 3.260438482461818e-07,
      "loss": 0.8722,
      "step": 11382
    },
    {
      "epoch": 0.8212546444933444,
      "grad_norm": 2.7941088930234983,
      "learning_rate": 3.2578813764214984e-07,
      "loss": 0.9743,
      "step": 11383
    },
    {
      "epoch": 0.8213267919627719,
      "grad_norm": 6.081071232739054,
      "learning_rate": 3.25532518460629e-07,
      "loss": 0.8879,
      "step": 11384
    },
    {
      "epoch": 0.8213989394321994,
      "grad_norm": 4.3996070946483705,
      "learning_rate": 3.2527699071557926e-07,
      "loss": 0.959,
      "step": 11385
    },
    {
      "epoch": 0.821471086901627,
      "grad_norm": 0.6121038079997599,
      "learning_rate": 3.2502155442095405e-07,
      "loss": 0.7551,
      "step": 11386
    },
    {
      "epoch": 0.8215432343710545,
      "grad_norm": 3.528411669779759,
      "learning_rate": 3.247662095907004e-07,
      "loss": 0.9888,
      "step": 11387
    },
    {
      "epoch": 0.821615381840482,
      "grad_norm": 4.766707500647435,
      "learning_rate": 3.2451095623876267e-07,
      "loss": 0.9458,
      "step": 11388
    },
    {
      "epoch": 0.8216875293099094,
      "grad_norm": 5.199780004689426,
      "learning_rate": 3.242557943790798e-07,
      "loss": 1.0803,
      "step": 11389
    },
    {
      "epoch": 0.8217596767793369,
      "grad_norm": 7.060638992742785,
      "learning_rate": 3.2400072402558396e-07,
      "loss": 0.9889,
      "step": 11390
    },
    {
      "epoch": 0.8218318242487644,
      "grad_norm": 2.7897661106968408,
      "learning_rate": 3.237457451922039e-07,
      "loss": 0.9339,
      "step": 11391
    },
    {
      "epoch": 0.821903971718192,
      "grad_norm": 4.609652442987605,
      "learning_rate": 3.2349085789286366e-07,
      "loss": 0.94,
      "step": 11392
    },
    {
      "epoch": 0.8219761191876195,
      "grad_norm": 3.3639890699084294,
      "learning_rate": 3.232360621414822e-07,
      "loss": 0.8581,
      "step": 11393
    },
    {
      "epoch": 0.822048266657047,
      "grad_norm": 4.446910282726881,
      "learning_rate": 3.229813579519716e-07,
      "loss": 0.8409,
      "step": 11394
    },
    {
      "epoch": 0.8221204141264745,
      "grad_norm": 0.6990625857070509,
      "learning_rate": 3.2272674533824093e-07,
      "loss": 0.8217,
      "step": 11395
    },
    {
      "epoch": 0.822192561595902,
      "grad_norm": 0.8040835141557882,
      "learning_rate": 3.224722243141942e-07,
      "loss": 0.8856,
      "step": 11396
    },
    {
      "epoch": 0.8222647090653296,
      "grad_norm": 4.094418070785995,
      "learning_rate": 3.2221779489372815e-07,
      "loss": 0.8654,
      "step": 11397
    },
    {
      "epoch": 0.8223368565347571,
      "grad_norm": 3.2937956254900755,
      "learning_rate": 3.2196345709073793e-07,
      "loss": 0.9061,
      "step": 11398
    },
    {
      "epoch": 0.8224090040041846,
      "grad_norm": 4.196424851807218,
      "learning_rate": 3.217092109191115e-07,
      "loss": 1.0185,
      "step": 11399
    },
    {
      "epoch": 0.8224811514736121,
      "grad_norm": 12.08921078209758,
      "learning_rate": 3.2145505639273274e-07,
      "loss": 0.9169,
      "step": 11400
    },
    {
      "epoch": 0.8225532989430395,
      "grad_norm": 27.773727582873285,
      "learning_rate": 3.212009935254791e-07,
      "loss": 0.8968,
      "step": 11401
    },
    {
      "epoch": 0.822625446412467,
      "grad_norm": 3.4449293827520524,
      "learning_rate": 3.2094702233122405e-07,
      "loss": 0.87,
      "step": 11402
    },
    {
      "epoch": 0.8226975938818946,
      "grad_norm": 5.771816165378436,
      "learning_rate": 3.2069314282383754e-07,
      "loss": 0.907,
      "step": 11403
    },
    {
      "epoch": 0.8227697413513221,
      "grad_norm": 5.465750258478842,
      "learning_rate": 3.2043935501718113e-07,
      "loss": 0.8301,
      "step": 11404
    },
    {
      "epoch": 0.8228418888207496,
      "grad_norm": 3.0361840117104744,
      "learning_rate": 3.201856589251142e-07,
      "loss": 0.9688,
      "step": 11405
    },
    {
      "epoch": 0.8229140362901771,
      "grad_norm": 3.745976005668801,
      "learning_rate": 3.1993205456149054e-07,
      "loss": 0.9941,
      "step": 11406
    },
    {
      "epoch": 0.8229861837596046,
      "grad_norm": 6.034914158777124,
      "learning_rate": 3.196785419401573e-07,
      "loss": 0.9379,
      "step": 11407
    },
    {
      "epoch": 0.8230583312290322,
      "grad_norm": 3.1882775050325654,
      "learning_rate": 3.194251210749581e-07,
      "loss": 0.9401,
      "step": 11408
    },
    {
      "epoch": 0.8231304786984597,
      "grad_norm": 4.778155353816478,
      "learning_rate": 3.191717919797323e-07,
      "loss": 0.913,
      "step": 11409
    },
    {
      "epoch": 0.8232026261678872,
      "grad_norm": 3.3351262198244678,
      "learning_rate": 3.189185546683133e-07,
      "loss": 0.9357,
      "step": 11410
    },
    {
      "epoch": 0.8232747736373147,
      "grad_norm": 3.2572348835022185,
      "learning_rate": 3.1866540915452846e-07,
      "loss": 0.9102,
      "step": 11411
    },
    {
      "epoch": 0.8233469211067422,
      "grad_norm": 4.7825126227428125,
      "learning_rate": 3.1841235545220137e-07,
      "loss": 0.9109,
      "step": 11412
    },
    {
      "epoch": 0.8234190685761696,
      "grad_norm": 3.0933120880000273,
      "learning_rate": 3.18159393575151e-07,
      "loss": 0.8617,
      "step": 11413
    },
    {
      "epoch": 0.8234912160455972,
      "grad_norm": 2.713931757929893,
      "learning_rate": 3.179065235371896e-07,
      "loss": 0.8536,
      "step": 11414
    },
    {
      "epoch": 0.8235633635150247,
      "grad_norm": 3.100739415492783,
      "learning_rate": 3.1765374535212575e-07,
      "loss": 0.9624,
      "step": 11415
    },
    {
      "epoch": 0.8236355109844522,
      "grad_norm": 3.4299822356984637,
      "learning_rate": 3.174010590337641e-07,
      "loss": 0.9639,
      "step": 11416
    },
    {
      "epoch": 0.8237076584538797,
      "grad_norm": 3.3411471494977842,
      "learning_rate": 3.1714846459590106e-07,
      "loss": 0.9351,
      "step": 11417
    },
    {
      "epoch": 0.8237798059233072,
      "grad_norm": 2.750874553680904,
      "learning_rate": 3.168959620523308e-07,
      "loss": 0.9327,
      "step": 11418
    },
    {
      "epoch": 0.8238519533927348,
      "grad_norm": 6.940654690338483,
      "learning_rate": 3.166435514168415e-07,
      "loss": 0.9223,
      "step": 11419
    },
    {
      "epoch": 0.8239241008621623,
      "grad_norm": 3.093635325762976,
      "learning_rate": 3.1639123270321677e-07,
      "loss": 1.0076,
      "step": 11420
    },
    {
      "epoch": 0.8239962483315898,
      "grad_norm": 5.021299390675762,
      "learning_rate": 3.1613900592523337e-07,
      "loss": 1.0425,
      "step": 11421
    },
    {
      "epoch": 0.8240683958010173,
      "grad_norm": 7.1660499603122645,
      "learning_rate": 3.1588687109666577e-07,
      "loss": 1.0593,
      "step": 11422
    },
    {
      "epoch": 0.8241405432704448,
      "grad_norm": 4.654567747442005,
      "learning_rate": 3.156348282312822e-07,
      "loss": 0.9972,
      "step": 11423
    },
    {
      "epoch": 0.8242126907398722,
      "grad_norm": 3.726816228709927,
      "learning_rate": 3.1538287734284507e-07,
      "loss": 0.9398,
      "step": 11424
    },
    {
      "epoch": 0.8242848382092998,
      "grad_norm": 21.300091036749777,
      "learning_rate": 3.1513101844511256e-07,
      "loss": 0.9853,
      "step": 11425
    },
    {
      "epoch": 0.8243569856787273,
      "grad_norm": 3.1651606408402433,
      "learning_rate": 3.148792515518377e-07,
      "loss": 0.9333,
      "step": 11426
    },
    {
      "epoch": 0.8244291331481548,
      "grad_norm": 5.6996079561029,
      "learning_rate": 3.1462757667676964e-07,
      "loss": 1.0347,
      "step": 11427
    },
    {
      "epoch": 0.8245012806175823,
      "grad_norm": 3.5786724088293966,
      "learning_rate": 3.143759938336501e-07,
      "loss": 0.9855,
      "step": 11428
    },
    {
      "epoch": 0.8245734280870098,
      "grad_norm": 3.0768877825179937,
      "learning_rate": 3.1412450303621763e-07,
      "loss": 0.8876,
      "step": 11429
    },
    {
      "epoch": 0.8246455755564374,
      "grad_norm": 0.7549496481696742,
      "learning_rate": 3.1387310429820544e-07,
      "loss": 0.8178,
      "step": 11430
    },
    {
      "epoch": 0.8247177230258649,
      "grad_norm": 5.220611605714384,
      "learning_rate": 3.1362179763334107e-07,
      "loss": 0.8907,
      "step": 11431
    },
    {
      "epoch": 0.8247898704952924,
      "grad_norm": 2.7269713524270855,
      "learning_rate": 3.1337058305534683e-07,
      "loss": 0.9515,
      "step": 11432
    },
    {
      "epoch": 0.8248620179647199,
      "grad_norm": 3.437663542585144,
      "learning_rate": 3.1311946057794237e-07,
      "loss": 1.0568,
      "step": 11433
    },
    {
      "epoch": 0.8249341654341474,
      "grad_norm": 3.597834645838236,
      "learning_rate": 3.12868430214839e-07,
      "loss": 0.9442,
      "step": 11434
    },
    {
      "epoch": 0.825006312903575,
      "grad_norm": 3.0746116144799522,
      "learning_rate": 3.126174919797453e-07,
      "loss": 0.9462,
      "step": 11435
    },
    {
      "epoch": 0.8250784603730024,
      "grad_norm": 4.625890697594762,
      "learning_rate": 3.1236664588636385e-07,
      "loss": 0.9962,
      "step": 11436
    },
    {
      "epoch": 0.8251506078424299,
      "grad_norm": 3.5177891830507924,
      "learning_rate": 3.121158919483928e-07,
      "loss": 0.9917,
      "step": 11437
    },
    {
      "epoch": 0.8252227553118574,
      "grad_norm": 4.067701794777409,
      "learning_rate": 3.1186523017952414e-07,
      "loss": 0.8447,
      "step": 11438
    },
    {
      "epoch": 0.8252949027812849,
      "grad_norm": 4.509885737627404,
      "learning_rate": 3.1161466059344556e-07,
      "loss": 0.964,
      "step": 11439
    },
    {
      "epoch": 0.8253670502507124,
      "grad_norm": 5.747426659645846,
      "learning_rate": 3.1136418320384116e-07,
      "loss": 0.9503,
      "step": 11440
    },
    {
      "epoch": 0.82543919772014,
      "grad_norm": 3.0300183497713626,
      "learning_rate": 3.11113798024387e-07,
      "loss": 1.0318,
      "step": 11441
    },
    {
      "epoch": 0.8255113451895675,
      "grad_norm": 4.2624553599774835,
      "learning_rate": 3.1086350506875623e-07,
      "loss": 1.0422,
      "step": 11442
    },
    {
      "epoch": 0.825583492658995,
      "grad_norm": 4.036399450385367,
      "learning_rate": 3.106133043506167e-07,
      "loss": 0.9043,
      "step": 11443
    },
    {
      "epoch": 0.8256556401284225,
      "grad_norm": 5.239562057947404,
      "learning_rate": 3.1036319588363105e-07,
      "loss": 0.9661,
      "step": 11444
    },
    {
      "epoch": 0.82572778759785,
      "grad_norm": 4.897299797168306,
      "learning_rate": 3.1011317968145554e-07,
      "loss": 1.0403,
      "step": 11445
    },
    {
      "epoch": 0.8257999350672776,
      "grad_norm": 4.340188082298011,
      "learning_rate": 3.098632557577441e-07,
      "loss": 0.8496,
      "step": 11446
    },
    {
      "epoch": 0.8258720825367051,
      "grad_norm": 4.013089221699208,
      "learning_rate": 3.0961342412614386e-07,
      "loss": 0.9532,
      "step": 11447
    },
    {
      "epoch": 0.8259442300061325,
      "grad_norm": 3.1106376193026093,
      "learning_rate": 3.093636848002967e-07,
      "loss": 0.9592,
      "step": 11448
    },
    {
      "epoch": 0.82601637747556,
      "grad_norm": 6.114530061644628,
      "learning_rate": 3.0911403779384015e-07,
      "loss": 0.8742,
      "step": 11449
    },
    {
      "epoch": 0.8260885249449875,
      "grad_norm": 3.564260482807989,
      "learning_rate": 3.0886448312040704e-07,
      "loss": 1.053,
      "step": 11450
    },
    {
      "epoch": 0.826160672414415,
      "grad_norm": 5.920659003307594,
      "learning_rate": 3.0861502079362287e-07,
      "loss": 0.945,
      "step": 11451
    },
    {
      "epoch": 0.8262328198838426,
      "grad_norm": 4.69102853847419,
      "learning_rate": 3.0836565082711195e-07,
      "loss": 0.9674,
      "step": 11452
    },
    {
      "epoch": 0.8263049673532701,
      "grad_norm": 5.565386762451809,
      "learning_rate": 3.081163732344905e-07,
      "loss": 1.1233,
      "step": 11453
    },
    {
      "epoch": 0.8263771148226976,
      "grad_norm": 3.0896837118300553,
      "learning_rate": 3.0786718802937127e-07,
      "loss": 0.9535,
      "step": 11454
    },
    {
      "epoch": 0.8264492622921251,
      "grad_norm": 5.313487959605724,
      "learning_rate": 3.0761809522536044e-07,
      "loss": 1.0035,
      "step": 11455
    },
    {
      "epoch": 0.8265214097615526,
      "grad_norm": 3.9653241848313114,
      "learning_rate": 3.0736909483605966e-07,
      "loss": 0.8588,
      "step": 11456
    },
    {
      "epoch": 0.8265935572309802,
      "grad_norm": 6.279021365942693,
      "learning_rate": 3.071201868750679e-07,
      "loss": 0.9891,
      "step": 11457
    },
    {
      "epoch": 0.8266657047004077,
      "grad_norm": 3.8074171860543613,
      "learning_rate": 3.068713713559754e-07,
      "loss": 0.983,
      "step": 11458
    },
    {
      "epoch": 0.8267378521698352,
      "grad_norm": 4.49404576884087,
      "learning_rate": 3.066226482923697e-07,
      "loss": 0.8933,
      "step": 11459
    },
    {
      "epoch": 0.8268099996392626,
      "grad_norm": 0.7339898174707373,
      "learning_rate": 3.063740176978329e-07,
      "loss": 0.8153,
      "step": 11460
    },
    {
      "epoch": 0.8268821471086901,
      "grad_norm": 0.8064376538745196,
      "learning_rate": 3.061254795859407e-07,
      "loss": 0.8751,
      "step": 11461
    },
    {
      "epoch": 0.8269542945781176,
      "grad_norm": 4.15593058570558,
      "learning_rate": 3.058770339702659e-07,
      "loss": 0.9975,
      "step": 11462
    },
    {
      "epoch": 0.8270264420475452,
      "grad_norm": 3.9361419757885328,
      "learning_rate": 3.056286808643742e-07,
      "loss": 0.8981,
      "step": 11463
    },
    {
      "epoch": 0.8270985895169727,
      "grad_norm": 9.2774428448655,
      "learning_rate": 3.0538042028182865e-07,
      "loss": 0.9863,
      "step": 11464
    },
    {
      "epoch": 0.8271707369864002,
      "grad_norm": 2.8962102594394494,
      "learning_rate": 3.0513225223618497e-07,
      "loss": 1.1257,
      "step": 11465
    },
    {
      "epoch": 0.8272428844558277,
      "grad_norm": 8.527140651211687,
      "learning_rate": 3.0488417674099466e-07,
      "loss": 0.9772,
      "step": 11466
    },
    {
      "epoch": 0.8273150319252552,
      "grad_norm": 5.287604848124264,
      "learning_rate": 3.0463619380980475e-07,
      "loss": 0.9533,
      "step": 11467
    },
    {
      "epoch": 0.8273871793946828,
      "grad_norm": 5.148174273371,
      "learning_rate": 3.0438830345615586e-07,
      "loss": 0.9755,
      "step": 11468
    },
    {
      "epoch": 0.8274593268641103,
      "grad_norm": 3.3809135340116123,
      "learning_rate": 3.041405056935842e-07,
      "loss": 1.1132,
      "step": 11469
    },
    {
      "epoch": 0.8275314743335378,
      "grad_norm": 4.157739587464906,
      "learning_rate": 3.038928005356223e-07,
      "loss": 0.9028,
      "step": 11470
    },
    {
      "epoch": 0.8276036218029652,
      "grad_norm": 3.4872994915601168,
      "learning_rate": 3.036451879957962e-07,
      "loss": 1.045,
      "step": 11471
    },
    {
      "epoch": 0.8276757692723927,
      "grad_norm": 3.209172787187898,
      "learning_rate": 3.0339766808762623e-07,
      "loss": 0.8997,
      "step": 11472
    },
    {
      "epoch": 0.8277479167418202,
      "grad_norm": 4.532671475246651,
      "learning_rate": 3.031502408246289e-07,
      "loss": 0.8567,
      "step": 11473
    },
    {
      "epoch": 0.8278200642112478,
      "grad_norm": 3.0416634616225267,
      "learning_rate": 3.0290290622031613e-07,
      "loss": 1.0053,
      "step": 11474
    },
    {
      "epoch": 0.8278922116806753,
      "grad_norm": 2.976632665573103,
      "learning_rate": 3.0265566428819214e-07,
      "loss": 0.9729,
      "step": 11475
    },
    {
      "epoch": 0.8279643591501028,
      "grad_norm": 4.26110645234805,
      "learning_rate": 3.0240851504175923e-07,
      "loss": 0.9096,
      "step": 11476
    },
    {
      "epoch": 0.8280365066195303,
      "grad_norm": 4.920694845491963,
      "learning_rate": 3.021614584945138e-07,
      "loss": 0.7963,
      "step": 11477
    },
    {
      "epoch": 0.8281086540889578,
      "grad_norm": 3.86967225059518,
      "learning_rate": 3.019144946599452e-07,
      "loss": 1.0541,
      "step": 11478
    },
    {
      "epoch": 0.8281808015583854,
      "grad_norm": 5.734330096575723,
      "learning_rate": 3.016676235515403e-07,
      "loss": 1.0264,
      "step": 11479
    },
    {
      "epoch": 0.8282529490278129,
      "grad_norm": 4.53363458436558,
      "learning_rate": 3.0142084518277865e-07,
      "loss": 0.9904,
      "step": 11480
    },
    {
      "epoch": 0.8283250964972404,
      "grad_norm": 5.433468255706166,
      "learning_rate": 3.011741595671378e-07,
      "loss": 1.0133,
      "step": 11481
    },
    {
      "epoch": 0.8283972439666679,
      "grad_norm": 2.7453446431414115,
      "learning_rate": 3.0092756671808684e-07,
      "loss": 1.0238,
      "step": 11482
    },
    {
      "epoch": 0.8284693914360953,
      "grad_norm": 6.113888685376741,
      "learning_rate": 3.0068106664909174e-07,
      "loss": 1.0683,
      "step": 11483
    },
    {
      "epoch": 0.8285415389055228,
      "grad_norm": 4.547369966079122,
      "learning_rate": 3.004346593736133e-07,
      "loss": 1.0777,
      "step": 11484
    },
    {
      "epoch": 0.8286136863749504,
      "grad_norm": 5.176716270200935,
      "learning_rate": 3.0018834490510613e-07,
      "loss": 0.9401,
      "step": 11485
    },
    {
      "epoch": 0.8286858338443779,
      "grad_norm": 8.672157406719164,
      "learning_rate": 2.99942123257021e-07,
      "loss": 0.8327,
      "step": 11486
    },
    {
      "epoch": 0.8287579813138054,
      "grad_norm": 3.7891254223190414,
      "learning_rate": 2.9969599444280235e-07,
      "loss": 0.967,
      "step": 11487
    },
    {
      "epoch": 0.8288301287832329,
      "grad_norm": 3.6225600087715426,
      "learning_rate": 2.994499584758925e-07,
      "loss": 0.9377,
      "step": 11488
    },
    {
      "epoch": 0.8289022762526604,
      "grad_norm": 3.547731943001344,
      "learning_rate": 2.992040153697244e-07,
      "loss": 0.9962,
      "step": 11489
    },
    {
      "epoch": 0.828974423722088,
      "grad_norm": 2.691265130184324,
      "learning_rate": 2.989581651377291e-07,
      "loss": 0.8816,
      "step": 11490
    },
    {
      "epoch": 0.8290465711915155,
      "grad_norm": 4.31681917036535,
      "learning_rate": 2.9871240779333164e-07,
      "loss": 0.9131,
      "step": 11491
    },
    {
      "epoch": 0.829118718660943,
      "grad_norm": 0.6872423296015022,
      "learning_rate": 2.984667433499513e-07,
      "loss": 0.7694,
      "step": 11492
    },
    {
      "epoch": 0.8291908661303705,
      "grad_norm": 5.813202846240872,
      "learning_rate": 2.982211718210028e-07,
      "loss": 0.9691,
      "step": 11493
    },
    {
      "epoch": 0.829263013599798,
      "grad_norm": 5.293948104736628,
      "learning_rate": 2.9797569321989713e-07,
      "loss": 1.0449,
      "step": 11494
    },
    {
      "epoch": 0.8293351610692254,
      "grad_norm": 4.829901211125184,
      "learning_rate": 2.977303075600379e-07,
      "loss": 1.0255,
      "step": 11495
    },
    {
      "epoch": 0.829407308538653,
      "grad_norm": 4.633830532776759,
      "learning_rate": 2.9748501485482493e-07,
      "loss": 0.9569,
      "step": 11496
    },
    {
      "epoch": 0.8294794560080805,
      "grad_norm": 2.868445679058346,
      "learning_rate": 2.972398151176527e-07,
      "loss": 0.9432,
      "step": 11497
    },
    {
      "epoch": 0.829551603477508,
      "grad_norm": 3.3373898777313817,
      "learning_rate": 2.9699470836191134e-07,
      "loss": 0.9827,
      "step": 11498
    },
    {
      "epoch": 0.8296237509469355,
      "grad_norm": 3.4533172139659336,
      "learning_rate": 2.9674969460098354e-07,
      "loss": 0.823,
      "step": 11499
    },
    {
      "epoch": 0.829695898416363,
      "grad_norm": 5.304843366577462,
      "learning_rate": 2.9650477384825023e-07,
      "loss": 0.9198,
      "step": 11500
    },
    {
      "epoch": 0.8297680458857906,
      "grad_norm": 4.380694797399037,
      "learning_rate": 2.962599461170856e-07,
      "loss": 0.9303,
      "step": 11501
    },
    {
      "epoch": 0.8298401933552181,
      "grad_norm": 4.263608350992654,
      "learning_rate": 2.960152114208578e-07,
      "loss": 0.9025,
      "step": 11502
    },
    {
      "epoch": 0.8299123408246456,
      "grad_norm": 17.9511641185994,
      "learning_rate": 2.957705697729316e-07,
      "loss": 0.9945,
      "step": 11503
    },
    {
      "epoch": 0.8299844882940731,
      "grad_norm": 0.7502408435673363,
      "learning_rate": 2.955260211866659e-07,
      "loss": 0.8968,
      "step": 11504
    },
    {
      "epoch": 0.8300566357635006,
      "grad_norm": 0.660596193066275,
      "learning_rate": 2.9528156567541374e-07,
      "loss": 0.8209,
      "step": 11505
    },
    {
      "epoch": 0.8301287832329282,
      "grad_norm": 0.6969622424816082,
      "learning_rate": 2.95037203252525e-07,
      "loss": 0.8398,
      "step": 11506
    },
    {
      "epoch": 0.8302009307023556,
      "grad_norm": 3.6336672853990093,
      "learning_rate": 2.9479293393134285e-07,
      "loss": 0.8722,
      "step": 11507
    },
    {
      "epoch": 0.8302730781717831,
      "grad_norm": 4.482235920026236,
      "learning_rate": 2.9454875772520684e-07,
      "loss": 0.9044,
      "step": 11508
    },
    {
      "epoch": 0.8303452256412106,
      "grad_norm": 7.109389219165038,
      "learning_rate": 2.943046746474491e-07,
      "loss": 0.994,
      "step": 11509
    },
    {
      "epoch": 0.8304173731106381,
      "grad_norm": 5.17172102684557,
      "learning_rate": 2.940606847113991e-07,
      "loss": 0.8431,
      "step": 11510
    },
    {
      "epoch": 0.8304895205800656,
      "grad_norm": 3.4862315616003814,
      "learning_rate": 2.938167879303797e-07,
      "loss": 0.9383,
      "step": 11511
    },
    {
      "epoch": 0.8305616680494932,
      "grad_norm": 12.985696993847332,
      "learning_rate": 2.935729843177095e-07,
      "loss": 0.9163,
      "step": 11512
    },
    {
      "epoch": 0.8306338155189207,
      "grad_norm": 3.824849373837347,
      "learning_rate": 2.933292738867017e-07,
      "loss": 0.9955,
      "step": 11513
    },
    {
      "epoch": 0.8307059629883482,
      "grad_norm": 9.544271410655394,
      "learning_rate": 2.9308565665066456e-07,
      "loss": 0.9706,
      "step": 11514
    },
    {
      "epoch": 0.8307781104577757,
      "grad_norm": 2.525336245064163,
      "learning_rate": 2.928421326229011e-07,
      "loss": 1.0174,
      "step": 11515
    },
    {
      "epoch": 0.8308502579272032,
      "grad_norm": 3.856866801439719,
      "learning_rate": 2.9259870181670865e-07,
      "loss": 0.9592,
      "step": 11516
    },
    {
      "epoch": 0.8309224053966308,
      "grad_norm": 2.76421427981204,
      "learning_rate": 2.9235536424538023e-07,
      "loss": 0.9896,
      "step": 11517
    },
    {
      "epoch": 0.8309945528660583,
      "grad_norm": 5.168161955182434,
      "learning_rate": 2.921121199222045e-07,
      "loss": 0.9374,
      "step": 11518
    },
    {
      "epoch": 0.8310667003354857,
      "grad_norm": 6.032215100487937,
      "learning_rate": 2.918689688604634e-07,
      "loss": 1.0265,
      "step": 11519
    },
    {
      "epoch": 0.8311388478049132,
      "grad_norm": 3.0823920247889336,
      "learning_rate": 2.916259110734345e-07,
      "loss": 0.9354,
      "step": 11520
    },
    {
      "epoch": 0.8312109952743407,
      "grad_norm": 4.675823265658137,
      "learning_rate": 2.9138294657439067e-07,
      "loss": 0.8649,
      "step": 11521
    },
    {
      "epoch": 0.8312831427437682,
      "grad_norm": 4.441481187870991,
      "learning_rate": 2.911400753765989e-07,
      "loss": 1.034,
      "step": 11522
    },
    {
      "epoch": 0.8313552902131958,
      "grad_norm": 3.4026384718702434,
      "learning_rate": 2.90897297493321e-07,
      "loss": 0.9576,
      "step": 11523
    },
    {
      "epoch": 0.8314274376826233,
      "grad_norm": 4.420445511058286,
      "learning_rate": 2.9065461293781537e-07,
      "loss": 0.9203,
      "step": 11524
    },
    {
      "epoch": 0.8314995851520508,
      "grad_norm": 4.888087775358517,
      "learning_rate": 2.904120217233339e-07,
      "loss": 0.9918,
      "step": 11525
    },
    {
      "epoch": 0.8315717326214783,
      "grad_norm": 4.317913472356844,
      "learning_rate": 2.901695238631228e-07,
      "loss": 1.0004,
      "step": 11526
    },
    {
      "epoch": 0.8316438800909058,
      "grad_norm": 3.3896882481546493,
      "learning_rate": 2.899271193704245e-07,
      "loss": 0.897,
      "step": 11527
    },
    {
      "epoch": 0.8317160275603334,
      "grad_norm": 3.0907078437772046,
      "learning_rate": 2.89684808258476e-07,
      "loss": 0.936,
      "step": 11528
    },
    {
      "epoch": 0.8317881750297609,
      "grad_norm": 4.715189013287832,
      "learning_rate": 2.8944259054050803e-07,
      "loss": 0.9061,
      "step": 11529
    },
    {
      "epoch": 0.8318603224991883,
      "grad_norm": 7.228779220825429,
      "learning_rate": 2.8920046622974823e-07,
      "loss": 0.9608,
      "step": 11530
    },
    {
      "epoch": 0.8319324699686158,
      "grad_norm": 3.04093502773333,
      "learning_rate": 2.8895843533941786e-07,
      "loss": 0.9493,
      "step": 11531
    },
    {
      "epoch": 0.8320046174380433,
      "grad_norm": 3.7633254758371564,
      "learning_rate": 2.887164978827339e-07,
      "loss": 0.938,
      "step": 11532
    },
    {
      "epoch": 0.8320767649074708,
      "grad_norm": 5.379609194320547,
      "learning_rate": 2.884746538729064e-07,
      "loss": 1.0383,
      "step": 11533
    },
    {
      "epoch": 0.8321489123768984,
      "grad_norm": 3.309817973776088,
      "learning_rate": 2.882329033231423e-07,
      "loss": 0.8902,
      "step": 11534
    },
    {
      "epoch": 0.8322210598463259,
      "grad_norm": 3.139959767806719,
      "learning_rate": 2.879912462466425e-07,
      "loss": 0.9753,
      "step": 11535
    },
    {
      "epoch": 0.8322932073157534,
      "grad_norm": 4.973842098881847,
      "learning_rate": 2.8774968265660324e-07,
      "loss": 0.9009,
      "step": 11536
    },
    {
      "epoch": 0.8323653547851809,
      "grad_norm": 0.807039065475509,
      "learning_rate": 2.8750821256621515e-07,
      "loss": 0.8263,
      "step": 11537
    },
    {
      "epoch": 0.8324375022546084,
      "grad_norm": 5.751282590372879,
      "learning_rate": 2.8726683598866476e-07,
      "loss": 0.9345,
      "step": 11538
    },
    {
      "epoch": 0.832509649724036,
      "grad_norm": 3.9468020809200186,
      "learning_rate": 2.870255529371315e-07,
      "loss": 0.95,
      "step": 11539
    },
    {
      "epoch": 0.8325817971934635,
      "grad_norm": 4.83344832097218,
      "learning_rate": 2.867843634247917e-07,
      "loss": 0.9514,
      "step": 11540
    },
    {
      "epoch": 0.832653944662891,
      "grad_norm": 3.486257549180604,
      "learning_rate": 2.8654326746481517e-07,
      "loss": 0.9743,
      "step": 11541
    },
    {
      "epoch": 0.8327260921323184,
      "grad_norm": 5.815824152894633,
      "learning_rate": 2.863022650703688e-07,
      "loss": 0.9205,
      "step": 11542
    },
    {
      "epoch": 0.8327982396017459,
      "grad_norm": 5.966509813062007,
      "learning_rate": 2.860613562546113e-07,
      "loss": 0.9193,
      "step": 11543
    },
    {
      "epoch": 0.8328703870711734,
      "grad_norm": 3.6415422454692337,
      "learning_rate": 2.8582054103069844e-07,
      "loss": 0.9112,
      "step": 11544
    },
    {
      "epoch": 0.832942534540601,
      "grad_norm": 3.475500600420456,
      "learning_rate": 2.8557981941178043e-07,
      "loss": 0.9523,
      "step": 11545
    },
    {
      "epoch": 0.8330146820100285,
      "grad_norm": 5.248018708346911,
      "learning_rate": 2.8533919141100174e-07,
      "loss": 0.957,
      "step": 11546
    },
    {
      "epoch": 0.833086829479456,
      "grad_norm": 3.2530042994201374,
      "learning_rate": 2.8509865704150173e-07,
      "loss": 0.9238,
      "step": 11547
    },
    {
      "epoch": 0.8331589769488835,
      "grad_norm": 4.060114468243384,
      "learning_rate": 2.848582163164166e-07,
      "loss": 0.9082,
      "step": 11548
    },
    {
      "epoch": 0.833231124418311,
      "grad_norm": 2.3750235907738215,
      "learning_rate": 2.8461786924887455e-07,
      "loss": 0.8789,
      "step": 11549
    },
    {
      "epoch": 0.8333032718877386,
      "grad_norm": 3.4223069567187947,
      "learning_rate": 2.8437761585200057e-07,
      "loss": 0.8625,
      "step": 11550
    },
    {
      "epoch": 0.8333754193571661,
      "grad_norm": 13.957984001381163,
      "learning_rate": 2.84137456138914e-07,
      "loss": 0.9122,
      "step": 11551
    },
    {
      "epoch": 0.8334475668265936,
      "grad_norm": 4.004338296049546,
      "learning_rate": 2.838973901227295e-07,
      "loss": 0.9364,
      "step": 11552
    },
    {
      "epoch": 0.8335197142960211,
      "grad_norm": 5.086156321797507,
      "learning_rate": 2.8365741781655494e-07,
      "loss": 1.0367,
      "step": 11553
    },
    {
      "epoch": 0.8335918617654485,
      "grad_norm": 3.640249994307689,
      "learning_rate": 2.8341753923349543e-07,
      "loss": 0.9846,
      "step": 11554
    },
    {
      "epoch": 0.833664009234876,
      "grad_norm": 2.762418277964371,
      "learning_rate": 2.8317775438665003e-07,
      "loss": 0.9971,
      "step": 11555
    },
    {
      "epoch": 0.8337361567043036,
      "grad_norm": 3.696762993890142,
      "learning_rate": 2.829380632891114e-07,
      "loss": 0.934,
      "step": 11556
    },
    {
      "epoch": 0.8338083041737311,
      "grad_norm": 3.186313183810133,
      "learning_rate": 2.8269846595396906e-07,
      "loss": 0.9413,
      "step": 11557
    },
    {
      "epoch": 0.8338804516431586,
      "grad_norm": 8.8857747609889,
      "learning_rate": 2.824589623943061e-07,
      "loss": 0.9213,
      "step": 11558
    },
    {
      "epoch": 0.8339525991125861,
      "grad_norm": 3.8184855018331496,
      "learning_rate": 2.8221955262320096e-07,
      "loss": 0.9753,
      "step": 11559
    },
    {
      "epoch": 0.8340247465820136,
      "grad_norm": 6.252856402941042,
      "learning_rate": 2.8198023665372716e-07,
      "loss": 0.909,
      "step": 11560
    },
    {
      "epoch": 0.8340968940514412,
      "grad_norm": 5.7145430302586515,
      "learning_rate": 2.817410144989529e-07,
      "loss": 0.7795,
      "step": 11561
    },
    {
      "epoch": 0.8341690415208687,
      "grad_norm": 5.857504258652888,
      "learning_rate": 2.815018861719414e-07,
      "loss": 1.0137,
      "step": 11562
    },
    {
      "epoch": 0.8342411889902962,
      "grad_norm": 3.181190079460087,
      "learning_rate": 2.812628516857498e-07,
      "loss": 0.8859,
      "step": 11563
    },
    {
      "epoch": 0.8343133364597237,
      "grad_norm": 2.640759255873062,
      "learning_rate": 2.8102391105343117e-07,
      "loss": 0.9806,
      "step": 11564
    },
    {
      "epoch": 0.8343854839291512,
      "grad_norm": 3.735907332020539,
      "learning_rate": 2.807850642880338e-07,
      "loss": 0.9676,
      "step": 11565
    },
    {
      "epoch": 0.8344576313985786,
      "grad_norm": 4.035426142807316,
      "learning_rate": 2.80546311402599e-07,
      "loss": 1.0014,
      "step": 11566
    },
    {
      "epoch": 0.8345297788680062,
      "grad_norm": 3.1979011924609497,
      "learning_rate": 2.803076524101655e-07,
      "loss": 1.0276,
      "step": 11567
    },
    {
      "epoch": 0.8346019263374337,
      "grad_norm": 4.63686560227115,
      "learning_rate": 2.8006908732376475e-07,
      "loss": 0.8738,
      "step": 11568
    },
    {
      "epoch": 0.8346740738068612,
      "grad_norm": 3.9518401579713816,
      "learning_rate": 2.798306161564248e-07,
      "loss": 0.8201,
      "step": 11569
    },
    {
      "epoch": 0.8347462212762887,
      "grad_norm": 3.1812326487009925,
      "learning_rate": 2.7959223892116643e-07,
      "loss": 1.0618,
      "step": 11570
    },
    {
      "epoch": 0.8348183687457162,
      "grad_norm": 5.233363176167597,
      "learning_rate": 2.7935395563100696e-07,
      "loss": 0.8876,
      "step": 11571
    },
    {
      "epoch": 0.8348905162151438,
      "grad_norm": 4.824496291358511,
      "learning_rate": 2.791157662989589e-07,
      "loss": 1.0032,
      "step": 11572
    },
    {
      "epoch": 0.8349626636845713,
      "grad_norm": 3.564330851926778,
      "learning_rate": 2.7887767093802825e-07,
      "loss": 0.9581,
      "step": 11573
    },
    {
      "epoch": 0.8350348111539988,
      "grad_norm": 7.445178003687474,
      "learning_rate": 2.7863966956121653e-07,
      "loss": 1.0121,
      "step": 11574
    },
    {
      "epoch": 0.8351069586234263,
      "grad_norm": 8.82044700337548,
      "learning_rate": 2.7840176218152e-07,
      "loss": 1.0342,
      "step": 11575
    },
    {
      "epoch": 0.8351791060928538,
      "grad_norm": 4.6579013430809155,
      "learning_rate": 2.781639488119307e-07,
      "loss": 0.9939,
      "step": 11576
    },
    {
      "epoch": 0.8352512535622812,
      "grad_norm": 4.405904336992963,
      "learning_rate": 2.779262294654332e-07,
      "loss": 0.9672,
      "step": 11577
    },
    {
      "epoch": 0.8353234010317088,
      "grad_norm": 3.6250741885255886,
      "learning_rate": 2.7768860415500995e-07,
      "loss": 0.932,
      "step": 11578
    },
    {
      "epoch": 0.8353955485011363,
      "grad_norm": 4.650214668158981,
      "learning_rate": 2.7745107289363655e-07,
      "loss": 0.9696,
      "step": 11579
    },
    {
      "epoch": 0.8354676959705638,
      "grad_norm": 6.375186992221563,
      "learning_rate": 2.7721363569428314e-07,
      "loss": 0.9671,
      "step": 11580
    },
    {
      "epoch": 0.8355398434399913,
      "grad_norm": 3.0506172868771415,
      "learning_rate": 2.769762925699155e-07,
      "loss": 0.8927,
      "step": 11581
    },
    {
      "epoch": 0.8356119909094188,
      "grad_norm": 3.303013731671258,
      "learning_rate": 2.7673904353349444e-07,
      "loss": 0.9091,
      "step": 11582
    },
    {
      "epoch": 0.8356841383788464,
      "grad_norm": 4.318157300012515,
      "learning_rate": 2.7650188859797396e-07,
      "loss": 0.9356,
      "step": 11583
    },
    {
      "epoch": 0.8357562858482739,
      "grad_norm": 3.236576381593544,
      "learning_rate": 2.7626482777630575e-07,
      "loss": 0.8631,
      "step": 11584
    },
    {
      "epoch": 0.8358284333177014,
      "grad_norm": 2.7848518280115897,
      "learning_rate": 2.7602786108143437e-07,
      "loss": 1.0527,
      "step": 11585
    },
    {
      "epoch": 0.8359005807871289,
      "grad_norm": 4.290775367007934,
      "learning_rate": 2.757909885262999e-07,
      "loss": 0.9854,
      "step": 11586
    },
    {
      "epoch": 0.8359727282565564,
      "grad_norm": 5.557322975458998,
      "learning_rate": 2.755542101238362e-07,
      "loss": 0.899,
      "step": 11587
    },
    {
      "epoch": 0.836044875725984,
      "grad_norm": 4.326152696416293,
      "learning_rate": 2.7531752588697354e-07,
      "loss": 0.9725,
      "step": 11588
    },
    {
      "epoch": 0.8361170231954114,
      "grad_norm": 3.8962922395454864,
      "learning_rate": 2.7508093582863656e-07,
      "loss": 0.931,
      "step": 11589
    },
    {
      "epoch": 0.8361891706648389,
      "grad_norm": 3.419255198752614,
      "learning_rate": 2.7484443996174334e-07,
      "loss": 0.9475,
      "step": 11590
    },
    {
      "epoch": 0.8362613181342664,
      "grad_norm": 3.977203857023764,
      "learning_rate": 2.7460803829920954e-07,
      "loss": 0.8885,
      "step": 11591
    },
    {
      "epoch": 0.8363334656036939,
      "grad_norm": 3.0667374547445383,
      "learning_rate": 2.7437173085394396e-07,
      "loss": 0.9439,
      "step": 11592
    },
    {
      "epoch": 0.8364056130731214,
      "grad_norm": 2.7297062716523715,
      "learning_rate": 2.7413551763884954e-07,
      "loss": 0.9146,
      "step": 11593
    },
    {
      "epoch": 0.836477760542549,
      "grad_norm": 3.1360523850970274,
      "learning_rate": 2.7389939866682586e-07,
      "loss": 0.8399,
      "step": 11594
    },
    {
      "epoch": 0.8365499080119765,
      "grad_norm": 3.2481709248428765,
      "learning_rate": 2.7366337395076546e-07,
      "loss": 0.9229,
      "step": 11595
    },
    {
      "epoch": 0.836622055481404,
      "grad_norm": 5.8402431797973895,
      "learning_rate": 2.734274435035586e-07,
      "loss": 0.9011,
      "step": 11596
    },
    {
      "epoch": 0.8366942029508315,
      "grad_norm": 4.287427913849407,
      "learning_rate": 2.731916073380871e-07,
      "loss": 0.9324,
      "step": 11597
    },
    {
      "epoch": 0.836766350420259,
      "grad_norm": 4.519794374025403,
      "learning_rate": 2.7295586546722926e-07,
      "loss": 0.9238,
      "step": 11598
    },
    {
      "epoch": 0.8368384978896866,
      "grad_norm": 4.677204065385933,
      "learning_rate": 2.7272021790385903e-07,
      "loss": 0.9768,
      "step": 11599
    },
    {
      "epoch": 0.8369106453591141,
      "grad_norm": 2.8920202238818797,
      "learning_rate": 2.72484664660843e-07,
      "loss": 1.0038,
      "step": 11600
    },
    {
      "epoch": 0.8369827928285415,
      "grad_norm": 4.8536987515754335,
      "learning_rate": 2.7224920575104373e-07,
      "loss": 0.9624,
      "step": 11601
    },
    {
      "epoch": 0.837054940297969,
      "grad_norm": 5.965858597975245,
      "learning_rate": 2.720138411873201e-07,
      "loss": 0.9322,
      "step": 11602
    },
    {
      "epoch": 0.8371270877673965,
      "grad_norm": 6.178078308083392,
      "learning_rate": 2.717785709825242e-07,
      "loss": 0.7967,
      "step": 11603
    },
    {
      "epoch": 0.837199235236824,
      "grad_norm": 3.1900730593455697,
      "learning_rate": 2.715433951495023e-07,
      "loss": 1.0496,
      "step": 11604
    },
    {
      "epoch": 0.8372713827062516,
      "grad_norm": 3.4356501370211827,
      "learning_rate": 2.713083137010974e-07,
      "loss": 0.921,
      "step": 11605
    },
    {
      "epoch": 0.8373435301756791,
      "grad_norm": 0.6836139566977227,
      "learning_rate": 2.7107332665014617e-07,
      "loss": 0.8218,
      "step": 11606
    },
    {
      "epoch": 0.8374156776451066,
      "grad_norm": 0.6672532560655681,
      "learning_rate": 2.708384340094796e-07,
      "loss": 0.7668,
      "step": 11607
    },
    {
      "epoch": 0.8374878251145341,
      "grad_norm": 6.5048595749331595,
      "learning_rate": 2.7060363579192547e-07,
      "loss": 0.9156,
      "step": 11608
    },
    {
      "epoch": 0.8375599725839616,
      "grad_norm": 3.9528046127579626,
      "learning_rate": 2.703689320103051e-07,
      "loss": 0.9386,
      "step": 11609
    },
    {
      "epoch": 0.8376321200533892,
      "grad_norm": 5.764338652235506,
      "learning_rate": 2.701343226774342e-07,
      "loss": 0.894,
      "step": 11610
    },
    {
      "epoch": 0.8377042675228167,
      "grad_norm": 4.906597538198489,
      "learning_rate": 2.698998078061241e-07,
      "loss": 0.9719,
      "step": 11611
    },
    {
      "epoch": 0.8377764149922442,
      "grad_norm": 1.02880859375,
      "learning_rate": 2.6966538740918097e-07,
      "loss": 0.9039,
      "step": 11612
    },
    {
      "epoch": 0.8378485624616716,
      "grad_norm": 5.272183372283852,
      "learning_rate": 2.69431061499406e-07,
      "loss": 0.9419,
      "step": 11613
    },
    {
      "epoch": 0.8379207099310991,
      "grad_norm": 4.470299298807951,
      "learning_rate": 2.6919683008959326e-07,
      "loss": 0.9037,
      "step": 11614
    },
    {
      "epoch": 0.8379928574005266,
      "grad_norm": 3.6415080689929176,
      "learning_rate": 2.6896269319253506e-07,
      "loss": 0.9304,
      "step": 11615
    },
    {
      "epoch": 0.8380650048699542,
      "grad_norm": 4.06861675179037,
      "learning_rate": 2.687286508210167e-07,
      "loss": 0.8458,
      "step": 11616
    },
    {
      "epoch": 0.8381371523393817,
      "grad_norm": 2.7034687859984023,
      "learning_rate": 2.684947029878171e-07,
      "loss": 0.8681,
      "step": 11617
    },
    {
      "epoch": 0.8382092998088092,
      "grad_norm": 3.909363505272619,
      "learning_rate": 2.6826084970571196e-07,
      "loss": 0.9679,
      "step": 11618
    },
    {
      "epoch": 0.8382814472782367,
      "grad_norm": 3.06590051064318,
      "learning_rate": 2.680270909874709e-07,
      "loss": 0.951,
      "step": 11619
    },
    {
      "epoch": 0.8383535947476642,
      "grad_norm": 4.092896350643316,
      "learning_rate": 2.6779342684585947e-07,
      "loss": 0.9713,
      "step": 11620
    },
    {
      "epoch": 0.8384257422170918,
      "grad_norm": 0.6977908509753223,
      "learning_rate": 2.675598572936364e-07,
      "loss": 0.8362,
      "step": 11621
    },
    {
      "epoch": 0.8384978896865193,
      "grad_norm": 0.6759636472952343,
      "learning_rate": 2.673263823435561e-07,
      "loss": 0.8258,
      "step": 11622
    },
    {
      "epoch": 0.8385700371559468,
      "grad_norm": 0.7930956959702995,
      "learning_rate": 2.670930020083684e-07,
      "loss": 0.832,
      "step": 11623
    },
    {
      "epoch": 0.8386421846253742,
      "grad_norm": 3.2314999098213346,
      "learning_rate": 2.6685971630081636e-07,
      "loss": 1.0124,
      "step": 11624
    },
    {
      "epoch": 0.8387143320948017,
      "grad_norm": 2.3989257315572203,
      "learning_rate": 2.6662652523363905e-07,
      "loss": 1.0608,
      "step": 11625
    },
    {
      "epoch": 0.8387864795642292,
      "grad_norm": 4.105471072935832,
      "learning_rate": 2.663934288195713e-07,
      "loss": 1.0068,
      "step": 11626
    },
    {
      "epoch": 0.8388586270336568,
      "grad_norm": 3.181896145596186,
      "learning_rate": 2.6616042707134025e-07,
      "loss": 1.0488,
      "step": 11627
    },
    {
      "epoch": 0.8389307745030843,
      "grad_norm": 9.467764925940605,
      "learning_rate": 2.659275200016697e-07,
      "loss": 0.8998,
      "step": 11628
    },
    {
      "epoch": 0.8390029219725118,
      "grad_norm": 4.136135215054256,
      "learning_rate": 2.6569470762327826e-07,
      "loss": 0.9784,
      "step": 11629
    },
    {
      "epoch": 0.8390750694419393,
      "grad_norm": 4.628016957033524,
      "learning_rate": 2.6546198994887904e-07,
      "loss": 0.9626,
      "step": 11630
    },
    {
      "epoch": 0.8391472169113668,
      "grad_norm": 4.66952179806796,
      "learning_rate": 2.652293669911787e-07,
      "loss": 1.0065,
      "step": 11631
    },
    {
      "epoch": 0.8392193643807944,
      "grad_norm": 3.5630226421123914,
      "learning_rate": 2.6499683876288093e-07,
      "loss": 0.8746,
      "step": 11632
    },
    {
      "epoch": 0.8392915118502219,
      "grad_norm": 5.123118915989377,
      "learning_rate": 2.647644052766838e-07,
      "loss": 1.0395,
      "step": 11633
    },
    {
      "epoch": 0.8393636593196494,
      "grad_norm": 0.9321830338138675,
      "learning_rate": 2.645320665452782e-07,
      "loss": 0.9369,
      "step": 11634
    },
    {
      "epoch": 0.8394358067890769,
      "grad_norm": 2.7089760140067893,
      "learning_rate": 2.6429982258135195e-07,
      "loss": 0.963,
      "step": 11635
    },
    {
      "epoch": 0.8395079542585043,
      "grad_norm": 6.4547480650299445,
      "learning_rate": 2.640676733975875e-07,
      "loss": 0.8354,
      "step": 11636
    },
    {
      "epoch": 0.8395801017279318,
      "grad_norm": 2.845109048908865,
      "learning_rate": 2.6383561900666086e-07,
      "loss": 1.0101,
      "step": 11637
    },
    {
      "epoch": 0.8396522491973594,
      "grad_norm": 3.0977062784126086,
      "learning_rate": 2.636036594212434e-07,
      "loss": 0.9401,
      "step": 11638
    },
    {
      "epoch": 0.8397243966667869,
      "grad_norm": 4.244455423279877,
      "learning_rate": 2.633717946540028e-07,
      "loss": 0.9408,
      "step": 11639
    },
    {
      "epoch": 0.8397965441362144,
      "grad_norm": 4.741449389761935,
      "learning_rate": 2.631400247176001e-07,
      "loss": 0.9471,
      "step": 11640
    },
    {
      "epoch": 0.8398686916056419,
      "grad_norm": 6.538962451164874,
      "learning_rate": 2.6290834962469046e-07,
      "loss": 0.9791,
      "step": 11641
    },
    {
      "epoch": 0.8399408390750694,
      "grad_norm": 3.5373327478560985,
      "learning_rate": 2.6267676938792547e-07,
      "loss": 0.9293,
      "step": 11642
    },
    {
      "epoch": 0.840012986544497,
      "grad_norm": 3.9287984460410104,
      "learning_rate": 2.6244528401995114e-07,
      "loss": 0.8963,
      "step": 11643
    },
    {
      "epoch": 0.8400851340139245,
      "grad_norm": 3.579349491540715,
      "learning_rate": 2.622138935334068e-07,
      "loss": 0.9407,
      "step": 11644
    },
    {
      "epoch": 0.840157281483352,
      "grad_norm": 4.230375195540455,
      "learning_rate": 2.6198259794092916e-07,
      "loss": 0.9173,
      "step": 11645
    },
    {
      "epoch": 0.8402294289527795,
      "grad_norm": 3.0639392818281324,
      "learning_rate": 2.6175139725514793e-07,
      "loss": 0.9191,
      "step": 11646
    },
    {
      "epoch": 0.840301576422207,
      "grad_norm": 3.5970266248891085,
      "learning_rate": 2.615202914886885e-07,
      "loss": 1.0512,
      "step": 11647
    },
    {
      "epoch": 0.8403737238916344,
      "grad_norm": 6.507175886104148,
      "learning_rate": 2.6128928065416975e-07,
      "loss": 0.9738,
      "step": 11648
    },
    {
      "epoch": 0.840445871361062,
      "grad_norm": 3.152457196204473,
      "learning_rate": 2.610583647642066e-07,
      "loss": 0.9299,
      "step": 11649
    },
    {
      "epoch": 0.8405180188304895,
      "grad_norm": 4.807754951655804,
      "learning_rate": 2.608275438314098e-07,
      "loss": 0.9657,
      "step": 11650
    },
    {
      "epoch": 0.840590166299917,
      "grad_norm": 4.4113726767098544,
      "learning_rate": 2.605968178683822e-07,
      "loss": 0.9266,
      "step": 11651
    },
    {
      "epoch": 0.8406623137693445,
      "grad_norm": 6.456160969595788,
      "learning_rate": 2.6036618688772317e-07,
      "loss": 0.9403,
      "step": 11652
    },
    {
      "epoch": 0.840734461238772,
      "grad_norm": 2.895903186275782,
      "learning_rate": 2.6013565090202714e-07,
      "loss": 0.9201,
      "step": 11653
    },
    {
      "epoch": 0.8408066087081996,
      "grad_norm": 2.6993918687700087,
      "learning_rate": 2.599052099238821e-07,
      "loss": 0.9643,
      "step": 11654
    },
    {
      "epoch": 0.8408787561776271,
      "grad_norm": 0.6748409233908469,
      "learning_rate": 2.5967486396587146e-07,
      "loss": 0.8257,
      "step": 11655
    },
    {
      "epoch": 0.8409509036470546,
      "grad_norm": 4.127625525707092,
      "learning_rate": 2.594446130405745e-07,
      "loss": 1.0494,
      "step": 11656
    },
    {
      "epoch": 0.8410230511164821,
      "grad_norm": 0.6789252073321149,
      "learning_rate": 2.5921445716056454e-07,
      "loss": 0.845,
      "step": 11657
    },
    {
      "epoch": 0.8410951985859096,
      "grad_norm": 15.445027481405372,
      "learning_rate": 2.589843963384082e-07,
      "loss": 1.0165,
      "step": 11658
    },
    {
      "epoch": 0.8411673460553372,
      "grad_norm": 5.151794358029237,
      "learning_rate": 2.5875443058666914e-07,
      "loss": 0.9942,
      "step": 11659
    },
    {
      "epoch": 0.8412394935247646,
      "grad_norm": 3.2157323821204353,
      "learning_rate": 2.5852455991790513e-07,
      "loss": 0.8872,
      "step": 11660
    },
    {
      "epoch": 0.8413116409941921,
      "grad_norm": 7.028705594314146,
      "learning_rate": 2.5829478434466767e-07,
      "loss": 1.0426,
      "step": 11661
    },
    {
      "epoch": 0.8413837884636196,
      "grad_norm": 8.866088691555225,
      "learning_rate": 2.5806510387950407e-07,
      "loss": 0.7948,
      "step": 11662
    },
    {
      "epoch": 0.8414559359330471,
      "grad_norm": 4.762705825002495,
      "learning_rate": 2.5783551853495724e-07,
      "loss": 1.0054,
      "step": 11663
    },
    {
      "epoch": 0.8415280834024746,
      "grad_norm": 4.75706097660434,
      "learning_rate": 2.576060283235639e-07,
      "loss": 0.9688,
      "step": 11664
    },
    {
      "epoch": 0.8416002308719022,
      "grad_norm": 8.114277968988233,
      "learning_rate": 2.573766332578546e-07,
      "loss": 0.9724,
      "step": 11665
    },
    {
      "epoch": 0.8416723783413297,
      "grad_norm": 2.962833652941873,
      "learning_rate": 2.5714733335035666e-07,
      "loss": 0.8938,
      "step": 11666
    },
    {
      "epoch": 0.8417445258107572,
      "grad_norm": 4.3591680443426775,
      "learning_rate": 2.569181286135913e-07,
      "loss": 0.9706,
      "step": 11667
    },
    {
      "epoch": 0.8418166732801847,
      "grad_norm": 3.8225801130922425,
      "learning_rate": 2.566890190600735e-07,
      "loss": 1.0283,
      "step": 11668
    },
    {
      "epoch": 0.8418888207496122,
      "grad_norm": 2.5005142636654143,
      "learning_rate": 2.564600047023151e-07,
      "loss": 0.9576,
      "step": 11669
    },
    {
      "epoch": 0.8419609682190398,
      "grad_norm": 2.8328997523341526,
      "learning_rate": 2.56231085552822e-07,
      "loss": 0.9647,
      "step": 11670
    },
    {
      "epoch": 0.8420331156884673,
      "grad_norm": 3.2123058215877034,
      "learning_rate": 2.560022616240938e-07,
      "loss": 1.0191,
      "step": 11671
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 5.495634600690501,
      "learning_rate": 2.557735329286259e-07,
      "loss": 0.9586,
      "step": 11672
    },
    {
      "epoch": 0.8421774106273222,
      "grad_norm": 2.4461684024010433,
      "learning_rate": 2.55544899478908e-07,
      "loss": 0.992,
      "step": 11673
    },
    {
      "epoch": 0.8422495580967497,
      "grad_norm": 3.7674704180836316,
      "learning_rate": 2.553163612874263e-07,
      "loss": 0.8582,
      "step": 11674
    },
    {
      "epoch": 0.8423217055661772,
      "grad_norm": 3.67323881753481,
      "learning_rate": 2.5508791836665924e-07,
      "loss": 0.939,
      "step": 11675
    },
    {
      "epoch": 0.8423938530356048,
      "grad_norm": 3.543760570799655,
      "learning_rate": 2.5485957072908125e-07,
      "loss": 1.0019,
      "step": 11676
    },
    {
      "epoch": 0.8424660005050323,
      "grad_norm": 6.719887712539594,
      "learning_rate": 2.5463131838716247e-07,
      "loss": 0.9392,
      "step": 11677
    },
    {
      "epoch": 0.8425381479744598,
      "grad_norm": 3.221640168890315,
      "learning_rate": 2.544031613533657e-07,
      "loss": 0.9899,
      "step": 11678
    },
    {
      "epoch": 0.8426102954438873,
      "grad_norm": 4.2518510714239905,
      "learning_rate": 2.5417509964014994e-07,
      "loss": 0.9021,
      "step": 11679
    },
    {
      "epoch": 0.8426824429133148,
      "grad_norm": 3.8370303832743375,
      "learning_rate": 2.5394713325996987e-07,
      "loss": 0.9796,
      "step": 11680
    },
    {
      "epoch": 0.8427545903827424,
      "grad_norm": 4.162982049013096,
      "learning_rate": 2.5371926222527285e-07,
      "loss": 1.0704,
      "step": 11681
    },
    {
      "epoch": 0.8428267378521699,
      "grad_norm": 2.79782557184154,
      "learning_rate": 2.534914865485025e-07,
      "loss": 0.9459,
      "step": 11682
    },
    {
      "epoch": 0.8428988853215973,
      "grad_norm": 5.560596440385433,
      "learning_rate": 2.5326380624209665e-07,
      "loss": 0.9433,
      "step": 11683
    },
    {
      "epoch": 0.8429710327910248,
      "grad_norm": 4.1026054291599925,
      "learning_rate": 2.530362213184887e-07,
      "loss": 1.0031,
      "step": 11684
    },
    {
      "epoch": 0.8430431802604523,
      "grad_norm": 2.5613143783230696,
      "learning_rate": 2.528087317901051e-07,
      "loss": 0.9793,
      "step": 11685
    },
    {
      "epoch": 0.8431153277298798,
      "grad_norm": 3.5032276529862374,
      "learning_rate": 2.525813376693682e-07,
      "loss": 0.9662,
      "step": 11686
    },
    {
      "epoch": 0.8431874751993074,
      "grad_norm": 7.231497206620856,
      "learning_rate": 2.523540389686969e-07,
      "loss": 1.1003,
      "step": 11687
    },
    {
      "epoch": 0.8432596226687349,
      "grad_norm": 4.292085176573721,
      "learning_rate": 2.521268357005013e-07,
      "loss": 0.9648,
      "step": 11688
    },
    {
      "epoch": 0.8433317701381624,
      "grad_norm": 0.7296517438929533,
      "learning_rate": 2.5189972787718884e-07,
      "loss": 0.9048,
      "step": 11689
    },
    {
      "epoch": 0.8434039176075899,
      "grad_norm": 4.0100203413221465,
      "learning_rate": 2.5167271551116107e-07,
      "loss": 0.8631,
      "step": 11690
    },
    {
      "epoch": 0.8434760650770174,
      "grad_norm": 2.7116169806275527,
      "learning_rate": 2.514457986148146e-07,
      "loss": 0.9921,
      "step": 11691
    },
    {
      "epoch": 0.843548212546445,
      "grad_norm": 4.356521164542242,
      "learning_rate": 2.5121897720053933e-07,
      "loss": 0.8826,
      "step": 11692
    },
    {
      "epoch": 0.8436203600158725,
      "grad_norm": 3.202913454094785,
      "learning_rate": 2.5099225128072235e-07,
      "loss": 0.8307,
      "step": 11693
    },
    {
      "epoch": 0.8436925074853,
      "grad_norm": 2.894083789005288,
      "learning_rate": 2.5076562086774445e-07,
      "loss": 0.9968,
      "step": 11694
    },
    {
      "epoch": 0.8437646549547274,
      "grad_norm": 3.9160925802912407,
      "learning_rate": 2.5053908597398e-07,
      "loss": 0.9833,
      "step": 11695
    },
    {
      "epoch": 0.8438368024241549,
      "grad_norm": 4.054606824597817,
      "learning_rate": 2.5031264661179976e-07,
      "loss": 0.939,
      "step": 11696
    },
    {
      "epoch": 0.8439089498935824,
      "grad_norm": 6.344505828780188,
      "learning_rate": 2.5008630279356914e-07,
      "loss": 0.9707,
      "step": 11697
    },
    {
      "epoch": 0.84398109736301,
      "grad_norm": 4.744270885150789,
      "learning_rate": 2.498600545316469e-07,
      "loss": 0.9323,
      "step": 11698
    },
    {
      "epoch": 0.8440532448324375,
      "grad_norm": 4.972834989177687,
      "learning_rate": 2.496339018383886e-07,
      "loss": 1.0236,
      "step": 11699
    },
    {
      "epoch": 0.844125392301865,
      "grad_norm": 4.674145201190952,
      "learning_rate": 2.494078447261434e-07,
      "loss": 0.9247,
      "step": 11700
    },
    {
      "epoch": 0.8441975397712925,
      "grad_norm": 3.7114346141130423,
      "learning_rate": 2.4918188320725563e-07,
      "loss": 0.9118,
      "step": 11701
    },
    {
      "epoch": 0.84426968724072,
      "grad_norm": 2.551818636706922,
      "learning_rate": 2.4895601729406345e-07,
      "loss": 1.0212,
      "step": 11702
    },
    {
      "epoch": 0.8443418347101476,
      "grad_norm": 6.390756022782016,
      "learning_rate": 2.487302469989008e-07,
      "loss": 0.8026,
      "step": 11703
    },
    {
      "epoch": 0.8444139821795751,
      "grad_norm": 3.230478787204782,
      "learning_rate": 2.4850457233409705e-07,
      "loss": 0.9041,
      "step": 11704
    },
    {
      "epoch": 0.8444861296490026,
      "grad_norm": 5.3733549927558135,
      "learning_rate": 2.4827899331197446e-07,
      "loss": 1.011,
      "step": 11705
    },
    {
      "epoch": 0.8445582771184301,
      "grad_norm": 13.452991728615435,
      "learning_rate": 2.480535099448513e-07,
      "loss": 0.9706,
      "step": 11706
    },
    {
      "epoch": 0.8446304245878575,
      "grad_norm": 11.325954129276552,
      "learning_rate": 2.4782812224504047e-07,
      "loss": 0.9265,
      "step": 11707
    },
    {
      "epoch": 0.844702572057285,
      "grad_norm": 5.58878843740932,
      "learning_rate": 2.4760283022485005e-07,
      "loss": 0.923,
      "step": 11708
    },
    {
      "epoch": 0.8447747195267126,
      "grad_norm": 6.314782060621295,
      "learning_rate": 2.473776338965816e-07,
      "loss": 0.9367,
      "step": 11709
    },
    {
      "epoch": 0.8448468669961401,
      "grad_norm": 3.511950798133183,
      "learning_rate": 2.471525332725319e-07,
      "loss": 0.8946,
      "step": 11710
    },
    {
      "epoch": 0.8449190144655676,
      "grad_norm": 3.878118552203355,
      "learning_rate": 2.469275283649945e-07,
      "loss": 1.038,
      "step": 11711
    },
    {
      "epoch": 0.8449911619349951,
      "grad_norm": 4.100460463927845,
      "learning_rate": 2.467026191862549e-07,
      "loss": 0.9146,
      "step": 11712
    },
    {
      "epoch": 0.8450633094044226,
      "grad_norm": 4.228093629707974,
      "learning_rate": 2.4647780574859454e-07,
      "loss": 0.8946,
      "step": 11713
    },
    {
      "epoch": 0.8451354568738502,
      "grad_norm": 4.008056633189396,
      "learning_rate": 2.4625308806429033e-07,
      "loss": 0.919,
      "step": 11714
    },
    {
      "epoch": 0.8452076043432777,
      "grad_norm": 2.5128481685373147,
      "learning_rate": 2.4602846614561246e-07,
      "loss": 1.1221,
      "step": 11715
    },
    {
      "epoch": 0.8452797518127052,
      "grad_norm": 5.332714621100827,
      "learning_rate": 2.4580394000482686e-07,
      "loss": 0.9188,
      "step": 11716
    },
    {
      "epoch": 0.8453518992821327,
      "grad_norm": 5.568164298856277,
      "learning_rate": 2.455795096541944e-07,
      "loss": 1.0493,
      "step": 11717
    },
    {
      "epoch": 0.8454240467515602,
      "grad_norm": 4.574094280370462,
      "learning_rate": 2.453551751059708e-07,
      "loss": 0.8823,
      "step": 11718
    },
    {
      "epoch": 0.8454961942209877,
      "grad_norm": 3.282584218835453,
      "learning_rate": 2.451309363724052e-07,
      "loss": 1.0035,
      "step": 11719
    },
    {
      "epoch": 0.8455683416904152,
      "grad_norm": 2.2927052543049826,
      "learning_rate": 2.449067934657429e-07,
      "loss": 0.9793,
      "step": 11720
    },
    {
      "epoch": 0.8456404891598427,
      "grad_norm": 13.696699003724687,
      "learning_rate": 2.4468274639822375e-07,
      "loss": 0.8755,
      "step": 11721
    },
    {
      "epoch": 0.8457126366292702,
      "grad_norm": 6.662643299386759,
      "learning_rate": 2.4445879518208114e-07,
      "loss": 0.9958,
      "step": 11722
    },
    {
      "epoch": 0.8457847840986977,
      "grad_norm": 4.87212658832627,
      "learning_rate": 2.4423493982954535e-07,
      "loss": 0.9682,
      "step": 11723
    },
    {
      "epoch": 0.8458569315681252,
      "grad_norm": 4.086941247482764,
      "learning_rate": 2.4401118035283996e-07,
      "loss": 0.9674,
      "step": 11724
    },
    {
      "epoch": 0.8459290790375528,
      "grad_norm": 77.79423417640764,
      "learning_rate": 2.4378751676418384e-07,
      "loss": 1.0067,
      "step": 11725
    },
    {
      "epoch": 0.8460012265069803,
      "grad_norm": 8.339235377400467,
      "learning_rate": 2.4356394907578967e-07,
      "loss": 0.8843,
      "step": 11726
    },
    {
      "epoch": 0.8460733739764078,
      "grad_norm": 3.4689416231030856,
      "learning_rate": 2.4334047729986596e-07,
      "loss": 0.9867,
      "step": 11727
    },
    {
      "epoch": 0.8461455214458353,
      "grad_norm": 5.180493322321927,
      "learning_rate": 2.4311710144861664e-07,
      "loss": 1.0249,
      "step": 11728
    },
    {
      "epoch": 0.8462176689152628,
      "grad_norm": 0.7010623772005414,
      "learning_rate": 2.42893821534238e-07,
      "loss": 0.8464,
      "step": 11729
    },
    {
      "epoch": 0.8462898163846903,
      "grad_norm": 5.391628713734615,
      "learning_rate": 2.4267063756892336e-07,
      "loss": 0.9808,
      "step": 11730
    },
    {
      "epoch": 0.8463619638541178,
      "grad_norm": 3.8882837566469592,
      "learning_rate": 2.4244754956486013e-07,
      "loss": 0.9231,
      "step": 11731
    },
    {
      "epoch": 0.8464341113235453,
      "grad_norm": 3.619294378943586,
      "learning_rate": 2.4222455753422963e-07,
      "loss": 0.8648,
      "step": 11732
    },
    {
      "epoch": 0.8465062587929728,
      "grad_norm": 5.135124626393099,
      "learning_rate": 2.420016614892091e-07,
      "loss": 0.9347,
      "step": 11733
    },
    {
      "epoch": 0.8465784062624003,
      "grad_norm": 3.5470943971241273,
      "learning_rate": 2.4177886144196933e-07,
      "loss": 0.8965,
      "step": 11734
    },
    {
      "epoch": 0.8466505537318278,
      "grad_norm": 3.558018190840246,
      "learning_rate": 2.4155615740467807e-07,
      "loss": 1.0506,
      "step": 11735
    },
    {
      "epoch": 0.8467227012012554,
      "grad_norm": 3.45484211893575,
      "learning_rate": 2.4133354938949525e-07,
      "loss": 0.9427,
      "step": 11736
    },
    {
      "epoch": 0.8467948486706829,
      "grad_norm": 5.032964189310688,
      "learning_rate": 2.4111103740857674e-07,
      "loss": 1.0662,
      "step": 11737
    },
    {
      "epoch": 0.8468669961401104,
      "grad_norm": 4.762472542110397,
      "learning_rate": 2.4088862147407397e-07,
      "loss": 0.9991,
      "step": 11738
    },
    {
      "epoch": 0.8469391436095379,
      "grad_norm": 0.7094794612382266,
      "learning_rate": 2.4066630159813096e-07,
      "loss": 0.7835,
      "step": 11739
    },
    {
      "epoch": 0.8470112910789654,
      "grad_norm": 2.914832782954777,
      "learning_rate": 2.40444077792888e-07,
      "loss": 0.978,
      "step": 11740
    },
    {
      "epoch": 0.847083438548393,
      "grad_norm": 2.5288824618272208,
      "learning_rate": 2.402219500704812e-07,
      "loss": 0.9244,
      "step": 11741
    },
    {
      "epoch": 0.8471555860178204,
      "grad_norm": 3.7333576939560063,
      "learning_rate": 2.399999184430388e-07,
      "loss": 0.9787,
      "step": 11742
    },
    {
      "epoch": 0.8472277334872479,
      "grad_norm": 3.2939125961909212,
      "learning_rate": 2.3977798292268536e-07,
      "loss": 0.9606,
      "step": 11743
    },
    {
      "epoch": 0.8472998809566754,
      "grad_norm": 4.58001232628642,
      "learning_rate": 2.395561435215405e-07,
      "loss": 1.0375,
      "step": 11744
    },
    {
      "epoch": 0.8473720284261029,
      "grad_norm": 0.6645103066680573,
      "learning_rate": 2.393344002517179e-07,
      "loss": 0.8002,
      "step": 11745
    },
    {
      "epoch": 0.8474441758955304,
      "grad_norm": 0.6949651257595286,
      "learning_rate": 2.391127531253252e-07,
      "loss": 0.7766,
      "step": 11746
    },
    {
      "epoch": 0.847516323364958,
      "grad_norm": 4.704412442531305,
      "learning_rate": 2.388912021544667e-07,
      "loss": 0.9775,
      "step": 11747
    },
    {
      "epoch": 0.8475884708343855,
      "grad_norm": 4.573749000150952,
      "learning_rate": 2.386697473512409e-07,
      "loss": 0.9167,
      "step": 11748
    },
    {
      "epoch": 0.847660618303813,
      "grad_norm": 8.159032821931499,
      "learning_rate": 2.384483887277393e-07,
      "loss": 1.0244,
      "step": 11749
    },
    {
      "epoch": 0.8477327657732405,
      "grad_norm": 3.025383532462232,
      "learning_rate": 2.3822712629605047e-07,
      "loss": 0.9502,
      "step": 11750
    },
    {
      "epoch": 0.847804913242668,
      "grad_norm": 3.913991474413273,
      "learning_rate": 2.380059600682558e-07,
      "loss": 1.0087,
      "step": 11751
    },
    {
      "epoch": 0.8478770607120956,
      "grad_norm": 0.6423115046709572,
      "learning_rate": 2.3778489005643387e-07,
      "loss": 0.7947,
      "step": 11752
    },
    {
      "epoch": 0.8479492081815231,
      "grad_norm": 29.604791243154903,
      "learning_rate": 2.3756391627265504e-07,
      "loss": 0.9216,
      "step": 11753
    },
    {
      "epoch": 0.8480213556509505,
      "grad_norm": 4.330843846157444,
      "learning_rate": 2.3734303872898653e-07,
      "loss": 0.9918,
      "step": 11754
    },
    {
      "epoch": 0.848093503120378,
      "grad_norm": 5.736042164944186,
      "learning_rate": 2.3712225743749003e-07,
      "loss": 0.9838,
      "step": 11755
    },
    {
      "epoch": 0.8481656505898055,
      "grad_norm": 5.578646595394237,
      "learning_rate": 2.3690157241022057e-07,
      "loss": 0.7642,
      "step": 11756
    },
    {
      "epoch": 0.848237798059233,
      "grad_norm": 3.6282641743096797,
      "learning_rate": 2.3668098365922961e-07,
      "loss": 0.9089,
      "step": 11757
    },
    {
      "epoch": 0.8483099455286606,
      "grad_norm": 4.4272613609272025,
      "learning_rate": 2.3646049119656286e-07,
      "loss": 0.8667,
      "step": 11758
    },
    {
      "epoch": 0.8483820929980881,
      "grad_norm": 5.933314725582362,
      "learning_rate": 2.3624009503425934e-07,
      "loss": 0.9558,
      "step": 11759
    },
    {
      "epoch": 0.8484542404675156,
      "grad_norm": 2.4401188011639814,
      "learning_rate": 2.3601979518435565e-07,
      "loss": 0.9675,
      "step": 11760
    },
    {
      "epoch": 0.8485263879369431,
      "grad_norm": 3.236560617497733,
      "learning_rate": 2.3579959165888086e-07,
      "loss": 0.9598,
      "step": 11761
    },
    {
      "epoch": 0.8485985354063706,
      "grad_norm": 4.051601874899056,
      "learning_rate": 2.3557948446985977e-07,
      "loss": 0.8976,
      "step": 11762
    },
    {
      "epoch": 0.8486706828757982,
      "grad_norm": 3.1420442780908417,
      "learning_rate": 2.3535947362931118e-07,
      "loss": 0.9106,
      "step": 11763
    },
    {
      "epoch": 0.8487428303452257,
      "grad_norm": 5.7166310538173395,
      "learning_rate": 2.3513955914924865e-07,
      "loss": 0.9794,
      "step": 11764
    },
    {
      "epoch": 0.8488149778146532,
      "grad_norm": 4.1917443472386795,
      "learning_rate": 2.3491974104168255e-07,
      "loss": 1.1,
      "step": 11765
    },
    {
      "epoch": 0.8488871252840806,
      "grad_norm": 3.598235805356523,
      "learning_rate": 2.3470001931861482e-07,
      "loss": 0.8372,
      "step": 11766
    },
    {
      "epoch": 0.8489592727535081,
      "grad_norm": 4.404299256859112,
      "learning_rate": 2.344803939920439e-07,
      "loss": 1.0748,
      "step": 11767
    },
    {
      "epoch": 0.8490314202229357,
      "grad_norm": 7.547841138582044,
      "learning_rate": 2.3426086507396326e-07,
      "loss": 0.9604,
      "step": 11768
    },
    {
      "epoch": 0.8491035676923632,
      "grad_norm": 6.098309029260277,
      "learning_rate": 2.3404143257636045e-07,
      "loss": 0.8603,
      "step": 11769
    },
    {
      "epoch": 0.8491757151617907,
      "grad_norm": 4.487980366872999,
      "learning_rate": 2.3382209651121675e-07,
      "loss": 0.924,
      "step": 11770
    },
    {
      "epoch": 0.8492478626312182,
      "grad_norm": 3.6446258434518795,
      "learning_rate": 2.3360285689051084e-07,
      "loss": 0.8575,
      "step": 11771
    },
    {
      "epoch": 0.8493200101006457,
      "grad_norm": 0.6905491360039223,
      "learning_rate": 2.3338371372621424e-07,
      "loss": 0.8176,
      "step": 11772
    },
    {
      "epoch": 0.8493921575700732,
      "grad_norm": 5.0999807619218185,
      "learning_rate": 2.331646670302927e-07,
      "loss": 0.9755,
      "step": 11773
    },
    {
      "epoch": 0.8494643050395008,
      "grad_norm": 2.750323276591852,
      "learning_rate": 2.3294571681470821e-07,
      "loss": 1.0973,
      "step": 11774
    },
    {
      "epoch": 0.8495364525089283,
      "grad_norm": 4.107540518896909,
      "learning_rate": 2.327268630914172e-07,
      "loss": 0.7856,
      "step": 11775
    },
    {
      "epoch": 0.8496085999783558,
      "grad_norm": 3.827086389519986,
      "learning_rate": 2.3250810587236902e-07,
      "loss": 0.9383,
      "step": 11776
    },
    {
      "epoch": 0.8496807474477833,
      "grad_norm": 4.0274247347277,
      "learning_rate": 2.3228944516951077e-07,
      "loss": 0.9579,
      "step": 11777
    },
    {
      "epoch": 0.8497528949172107,
      "grad_norm": 2.7990505345761596,
      "learning_rate": 2.3207088099478178e-07,
      "loss": 0.9369,
      "step": 11778
    },
    {
      "epoch": 0.8498250423866383,
      "grad_norm": 0.7579365265019701,
      "learning_rate": 2.3185241336011785e-07,
      "loss": 0.8356,
      "step": 11779
    },
    {
      "epoch": 0.8498971898560658,
      "grad_norm": 3.310176682381033,
      "learning_rate": 2.3163404227744788e-07,
      "loss": 0.8996,
      "step": 11780
    },
    {
      "epoch": 0.8499693373254933,
      "grad_norm": 4.302138776937615,
      "learning_rate": 2.3141576775869653e-07,
      "loss": 0.9845,
      "step": 11781
    },
    {
      "epoch": 0.8500414847949208,
      "grad_norm": 4.022009379432398,
      "learning_rate": 2.311975898157834e-07,
      "loss": 1.0214,
      "step": 11782
    },
    {
      "epoch": 0.8501136322643483,
      "grad_norm": 3.652088372589192,
      "learning_rate": 2.3097950846062119e-07,
      "loss": 0.9175,
      "step": 11783
    },
    {
      "epoch": 0.8501857797337758,
      "grad_norm": 4.455872447751381,
      "learning_rate": 2.307615237051197e-07,
      "loss": 0.9889,
      "step": 11784
    },
    {
      "epoch": 0.8502579272032034,
      "grad_norm": 3.622595285409022,
      "learning_rate": 2.305436355611825e-07,
      "loss": 0.8692,
      "step": 11785
    },
    {
      "epoch": 0.8503300746726309,
      "grad_norm": 5.365050399842175,
      "learning_rate": 2.3032584404070653e-07,
      "loss": 0.988,
      "step": 11786
    },
    {
      "epoch": 0.8504022221420584,
      "grad_norm": 7.293992055032481,
      "learning_rate": 2.3010814915558497e-07,
      "loss": 0.9954,
      "step": 11787
    },
    {
      "epoch": 0.8504743696114859,
      "grad_norm": 6.81041944738475,
      "learning_rate": 2.2989055091770515e-07,
      "loss": 0.8954,
      "step": 11788
    },
    {
      "epoch": 0.8505465170809133,
      "grad_norm": 2.988656213709619,
      "learning_rate": 2.296730493389505e-07,
      "loss": 1.1152,
      "step": 11789
    },
    {
      "epoch": 0.8506186645503409,
      "grad_norm": 2.480416076581481,
      "learning_rate": 2.294556444311968e-07,
      "loss": 1.0311,
      "step": 11790
    },
    {
      "epoch": 0.8506908120197684,
      "grad_norm": 3.5877940040831304,
      "learning_rate": 2.292383362063157e-07,
      "loss": 0.8731,
      "step": 11791
    },
    {
      "epoch": 0.8507629594891959,
      "grad_norm": 4.208116028466089,
      "learning_rate": 2.2902112467617462e-07,
      "loss": 1.0389,
      "step": 11792
    },
    {
      "epoch": 0.8508351069586234,
      "grad_norm": 3.3887035357400634,
      "learning_rate": 2.288040098526336e-07,
      "loss": 0.8932,
      "step": 11793
    },
    {
      "epoch": 0.8509072544280509,
      "grad_norm": 2.737652803101739,
      "learning_rate": 2.2858699174754824e-07,
      "loss": 0.8983,
      "step": 11794
    },
    {
      "epoch": 0.8509794018974784,
      "grad_norm": 0.707965212888686,
      "learning_rate": 2.2837007037277024e-07,
      "loss": 0.8325,
      "step": 11795
    },
    {
      "epoch": 0.851051549366906,
      "grad_norm": 3.0303256798389038,
      "learning_rate": 2.2815324574014473e-07,
      "loss": 0.9321,
      "step": 11796
    },
    {
      "epoch": 0.8511236968363335,
      "grad_norm": 3.1423470250347316,
      "learning_rate": 2.2793651786151092e-07,
      "loss": 0.9768,
      "step": 11797
    },
    {
      "epoch": 0.851195844305761,
      "grad_norm": 5.628245794767528,
      "learning_rate": 2.277198867487038e-07,
      "loss": 0.9262,
      "step": 11798
    },
    {
      "epoch": 0.8512679917751885,
      "grad_norm": 7.334043410631235,
      "learning_rate": 2.2750335241355346e-07,
      "loss": 0.9693,
      "step": 11799
    },
    {
      "epoch": 0.851340139244616,
      "grad_norm": 4.848365209288253,
      "learning_rate": 2.2728691486788266e-07,
      "loss": 0.8428,
      "step": 11800
    },
    {
      "epoch": 0.8514122867140435,
      "grad_norm": 3.333716195689261,
      "learning_rate": 2.2707057412351148e-07,
      "loss": 1.0087,
      "step": 11801
    },
    {
      "epoch": 0.851484434183471,
      "grad_norm": 6.747744147967214,
      "learning_rate": 2.2685433019225343e-07,
      "loss": 0.8861,
      "step": 11802
    },
    {
      "epoch": 0.8515565816528985,
      "grad_norm": 4.766440800115006,
      "learning_rate": 2.2663818308591608e-07,
      "loss": 0.9489,
      "step": 11803
    },
    {
      "epoch": 0.851628729122326,
      "grad_norm": 36.86143156089777,
      "learning_rate": 2.2642213281630295e-07,
      "loss": 0.9686,
      "step": 11804
    },
    {
      "epoch": 0.8517008765917535,
      "grad_norm": 2.8925077338350214,
      "learning_rate": 2.2620617939521125e-07,
      "loss": 0.9549,
      "step": 11805
    },
    {
      "epoch": 0.851773024061181,
      "grad_norm": 3.7150080471028675,
      "learning_rate": 2.2599032283443397e-07,
      "loss": 0.9509,
      "step": 11806
    },
    {
      "epoch": 0.8518451715306086,
      "grad_norm": 3.2294405093007663,
      "learning_rate": 2.2577456314575793e-07,
      "loss": 0.9634,
      "step": 11807
    },
    {
      "epoch": 0.8519173190000361,
      "grad_norm": 5.991590965050975,
      "learning_rate": 2.2555890034096504e-07,
      "loss": 0.9205,
      "step": 11808
    },
    {
      "epoch": 0.8519894664694636,
      "grad_norm": 6.122252801879157,
      "learning_rate": 2.253433344318323e-07,
      "loss": 1.045,
      "step": 11809
    },
    {
      "epoch": 0.8520616139388911,
      "grad_norm": 3.4474966171723853,
      "learning_rate": 2.2512786543013007e-07,
      "loss": 0.9782,
      "step": 11810
    },
    {
      "epoch": 0.8521337614083186,
      "grad_norm": 2.8916785717223488,
      "learning_rate": 2.2491249334762474e-07,
      "loss": 1.0051,
      "step": 11811
    },
    {
      "epoch": 0.8522059088777462,
      "grad_norm": 4.736486385886843,
      "learning_rate": 2.2469721819607668e-07,
      "loss": 0.9657,
      "step": 11812
    },
    {
      "epoch": 0.8522780563471736,
      "grad_norm": 3.3985275782781845,
      "learning_rate": 2.2448203998724225e-07,
      "loss": 0.8896,
      "step": 11813
    },
    {
      "epoch": 0.8523502038166011,
      "grad_norm": 4.10395252166114,
      "learning_rate": 2.242669587328707e-07,
      "loss": 0.9794,
      "step": 11814
    },
    {
      "epoch": 0.8524223512860286,
      "grad_norm": 6.341600476728671,
      "learning_rate": 2.2405197444470668e-07,
      "loss": 0.9558,
      "step": 11815
    },
    {
      "epoch": 0.8524944987554561,
      "grad_norm": 16.988443373421624,
      "learning_rate": 2.238370871344908e-07,
      "loss": 0.9726,
      "step": 11816
    },
    {
      "epoch": 0.8525666462248837,
      "grad_norm": 4.977753165918389,
      "learning_rate": 2.2362229681395563e-07,
      "loss": 0.8764,
      "step": 11817
    },
    {
      "epoch": 0.8526387936943112,
      "grad_norm": 5.712673606989133,
      "learning_rate": 2.234076034948309e-07,
      "loss": 0.978,
      "step": 11818
    },
    {
      "epoch": 0.8527109411637387,
      "grad_norm": 4.105475486510292,
      "learning_rate": 2.231930071888408e-07,
      "loss": 0.9169,
      "step": 11819
    },
    {
      "epoch": 0.8527830886331662,
      "grad_norm": 3.9968066581746973,
      "learning_rate": 2.2297850790770267e-07,
      "loss": 0.8997,
      "step": 11820
    },
    {
      "epoch": 0.8528552361025937,
      "grad_norm": 4.376637833425972,
      "learning_rate": 2.2276410566312998e-07,
      "loss": 0.8891,
      "step": 11821
    },
    {
      "epoch": 0.8529273835720212,
      "grad_norm": 3.843957414690885,
      "learning_rate": 2.2254980046683025e-07,
      "loss": 0.9241,
      "step": 11822
    },
    {
      "epoch": 0.8529995310414488,
      "grad_norm": 3.506460222276202,
      "learning_rate": 2.2233559233050658e-07,
      "loss": 1.0108,
      "step": 11823
    },
    {
      "epoch": 0.8530716785108763,
      "grad_norm": 5.941064025934572,
      "learning_rate": 2.221214812658545e-07,
      "loss": 0.9625,
      "step": 11824
    },
    {
      "epoch": 0.8531438259803037,
      "grad_norm": 9.746245223647549,
      "learning_rate": 2.2190746728456755e-07,
      "loss": 0.9131,
      "step": 11825
    },
    {
      "epoch": 0.8532159734497312,
      "grad_norm": 5.191091006935359,
      "learning_rate": 2.216935503983317e-07,
      "loss": 0.8754,
      "step": 11826
    },
    {
      "epoch": 0.8532881209191587,
      "grad_norm": 4.618286208636578,
      "learning_rate": 2.2147973061882763e-07,
      "loss": 0.9316,
      "step": 11827
    },
    {
      "epoch": 0.8533602683885863,
      "grad_norm": 3.347022051033937,
      "learning_rate": 2.2126600795773175e-07,
      "loss": 0.9938,
      "step": 11828
    },
    {
      "epoch": 0.8534324158580138,
      "grad_norm": 3.717933629289495,
      "learning_rate": 2.2105238242671476e-07,
      "loss": 0.9418,
      "step": 11829
    },
    {
      "epoch": 0.8535045633274413,
      "grad_norm": 5.87583170746019,
      "learning_rate": 2.2083885403744106e-07,
      "loss": 1.0399,
      "step": 11830
    },
    {
      "epoch": 0.8535767107968688,
      "grad_norm": 3.7653124667698044,
      "learning_rate": 2.206254228015716e-07,
      "loss": 0.9794,
      "step": 11831
    },
    {
      "epoch": 0.8536488582662963,
      "grad_norm": 6.801661557668711,
      "learning_rate": 2.20412088730761e-07,
      "loss": 0.9001,
      "step": 11832
    },
    {
      "epoch": 0.8537210057357238,
      "grad_norm": 9.284209219557663,
      "learning_rate": 2.2019885183665866e-07,
      "loss": 1.0789,
      "step": 11833
    },
    {
      "epoch": 0.8537931532051514,
      "grad_norm": 5.0481601649537025,
      "learning_rate": 2.199857121309079e-07,
      "loss": 0.9956,
      "step": 11834
    },
    {
      "epoch": 0.8538653006745789,
      "grad_norm": 3.6357235637785164,
      "learning_rate": 2.1977266962514828e-07,
      "loss": 0.9928,
      "step": 11835
    },
    {
      "epoch": 0.8539374481440063,
      "grad_norm": 4.804316908027487,
      "learning_rate": 2.1955972433101322e-07,
      "loss": 1.0377,
      "step": 11836
    },
    {
      "epoch": 0.8540095956134338,
      "grad_norm": 3.9563009297978255,
      "learning_rate": 2.193468762601298e-07,
      "loss": 0.9872,
      "step": 11837
    },
    {
      "epoch": 0.8540817430828613,
      "grad_norm": 4.070389818273561,
      "learning_rate": 2.1913412542412235e-07,
      "loss": 1.0416,
      "step": 11838
    },
    {
      "epoch": 0.8541538905522889,
      "grad_norm": 4.324542494373772,
      "learning_rate": 2.1892147183460752e-07,
      "loss": 0.8736,
      "step": 11839
    },
    {
      "epoch": 0.8542260380217164,
      "grad_norm": 5.438953424646027,
      "learning_rate": 2.1870891550319846e-07,
      "loss": 0.9166,
      "step": 11840
    },
    {
      "epoch": 0.8542981854911439,
      "grad_norm": 5.190362531168324,
      "learning_rate": 2.1849645644150082e-07,
      "loss": 0.9448,
      "step": 11841
    },
    {
      "epoch": 0.8543703329605714,
      "grad_norm": 2.6207389761996653,
      "learning_rate": 2.182840946611164e-07,
      "loss": 1.042,
      "step": 11842
    },
    {
      "epoch": 0.8544424804299989,
      "grad_norm": 3.06076280797366,
      "learning_rate": 2.1807183017364262e-07,
      "loss": 1.0306,
      "step": 11843
    },
    {
      "epoch": 0.8545146278994264,
      "grad_norm": 2.9922510842522048,
      "learning_rate": 2.178596629906695e-07,
      "loss": 0.8965,
      "step": 11844
    },
    {
      "epoch": 0.854586775368854,
      "grad_norm": 9.902367634340031,
      "learning_rate": 2.176475931237829e-07,
      "loss": 0.8526,
      "step": 11845
    },
    {
      "epoch": 0.8546589228382815,
      "grad_norm": 3.688212180465593,
      "learning_rate": 2.1743562058456334e-07,
      "loss": 0.9654,
      "step": 11846
    },
    {
      "epoch": 0.854731070307709,
      "grad_norm": 3.5995563869539993,
      "learning_rate": 2.1722374538458553e-07,
      "loss": 1.0028,
      "step": 11847
    },
    {
      "epoch": 0.8548032177771364,
      "grad_norm": 4.632904103291784,
      "learning_rate": 2.1701196753541895e-07,
      "loss": 0.9305,
      "step": 11848
    },
    {
      "epoch": 0.8548753652465639,
      "grad_norm": 6.641441241195611,
      "learning_rate": 2.1680028704862897e-07,
      "loss": 1.0533,
      "step": 11849
    },
    {
      "epoch": 0.8549475127159915,
      "grad_norm": 5.118976123179578,
      "learning_rate": 2.1658870393577433e-07,
      "loss": 1.0216,
      "step": 11850
    },
    {
      "epoch": 0.855019660185419,
      "grad_norm": 3.168334789272444,
      "learning_rate": 2.1637721820840825e-07,
      "loss": 0.8887,
      "step": 11851
    },
    {
      "epoch": 0.8550918076548465,
      "grad_norm": 0.8003487139890444,
      "learning_rate": 2.1616582987807953e-07,
      "loss": 0.8343,
      "step": 11852
    },
    {
      "epoch": 0.855163955124274,
      "grad_norm": 3.6890687353524974,
      "learning_rate": 2.1595453895633174e-07,
      "loss": 0.9022,
      "step": 11853
    },
    {
      "epoch": 0.8552361025937015,
      "grad_norm": 4.888161133053544,
      "learning_rate": 2.157433454547015e-07,
      "loss": 1.0418,
      "step": 11854
    },
    {
      "epoch": 0.855308250063129,
      "grad_norm": 2.7601335392311053,
      "learning_rate": 2.1553224938472247e-07,
      "loss": 0.8834,
      "step": 11855
    },
    {
      "epoch": 0.8553803975325566,
      "grad_norm": 2.9072487048304416,
      "learning_rate": 2.153212507579214e-07,
      "loss": 0.8265,
      "step": 11856
    },
    {
      "epoch": 0.8554525450019841,
      "grad_norm": 3.3523208720507367,
      "learning_rate": 2.1511034958582042e-07,
      "loss": 1.0472,
      "step": 11857
    },
    {
      "epoch": 0.8555246924714116,
      "grad_norm": 0.7610207905705155,
      "learning_rate": 2.1489954587993564e-07,
      "loss": 0.8632,
      "step": 11858
    },
    {
      "epoch": 0.8555968399408391,
      "grad_norm": 6.948891574647782,
      "learning_rate": 2.146888396517783e-07,
      "loss": 1.0252,
      "step": 11859
    },
    {
      "epoch": 0.8556689874102665,
      "grad_norm": 7.01408631033286,
      "learning_rate": 2.14478230912855e-07,
      "loss": 0.8817,
      "step": 11860
    },
    {
      "epoch": 0.855741134879694,
      "grad_norm": 0.8030238314188435,
      "learning_rate": 2.1426771967466472e-07,
      "loss": 0.8437,
      "step": 11861
    },
    {
      "epoch": 0.8558132823491216,
      "grad_norm": 3.866277414219157,
      "learning_rate": 2.1405730594870431e-07,
      "loss": 0.9063,
      "step": 11862
    },
    {
      "epoch": 0.8558854298185491,
      "grad_norm": 2.1346194259964824,
      "learning_rate": 2.1384698974646344e-07,
      "loss": 1.0431,
      "step": 11863
    },
    {
      "epoch": 0.8559575772879766,
      "grad_norm": 4.34070156351212,
      "learning_rate": 2.136367710794258e-07,
      "loss": 1.0479,
      "step": 11864
    },
    {
      "epoch": 0.8560297247574041,
      "grad_norm": 3.3549487384910717,
      "learning_rate": 2.1342664995907134e-07,
      "loss": 0.9758,
      "step": 11865
    },
    {
      "epoch": 0.8561018722268317,
      "grad_norm": 0.809763444749443,
      "learning_rate": 2.1321662639687333e-07,
      "loss": 0.8617,
      "step": 11866
    },
    {
      "epoch": 0.8561740196962592,
      "grad_norm": 3.4660632751553124,
      "learning_rate": 2.1300670040430192e-07,
      "loss": 0.9765,
      "step": 11867
    },
    {
      "epoch": 0.8562461671656867,
      "grad_norm": 6.470692218618185,
      "learning_rate": 2.127968719928188e-07,
      "loss": 1.044,
      "step": 11868
    },
    {
      "epoch": 0.8563183146351142,
      "grad_norm": 5.683448773924185,
      "learning_rate": 2.1258714117388266e-07,
      "loss": 0.8903,
      "step": 11869
    },
    {
      "epoch": 0.8563904621045417,
      "grad_norm": 9.789829379378567,
      "learning_rate": 2.1237750795894605e-07,
      "loss": 0.8671,
      "step": 11870
    },
    {
      "epoch": 0.8564626095739692,
      "grad_norm": 5.521341878633044,
      "learning_rate": 2.1216797235945606e-07,
      "loss": 0.864,
      "step": 11871
    },
    {
      "epoch": 0.8565347570433967,
      "grad_norm": 15.634431723675332,
      "learning_rate": 2.119585343868544e-07,
      "loss": 0.9702,
      "step": 11872
    },
    {
      "epoch": 0.8566069045128242,
      "grad_norm": 3.264345201198907,
      "learning_rate": 2.1174919405257863e-07,
      "loss": 0.9365,
      "step": 11873
    },
    {
      "epoch": 0.8566790519822517,
      "grad_norm": 3.2074988299610743,
      "learning_rate": 2.1153995136805936e-07,
      "loss": 0.9013,
      "step": 11874
    },
    {
      "epoch": 0.8567511994516792,
      "grad_norm": 4.1961712233643444,
      "learning_rate": 2.1133080634472254e-07,
      "loss": 0.9145,
      "step": 11875
    },
    {
      "epoch": 0.8568233469211067,
      "grad_norm": 4.530455631474584,
      "learning_rate": 2.1112175899398909e-07,
      "loss": 1.0358,
      "step": 11876
    },
    {
      "epoch": 0.8568954943905343,
      "grad_norm": 0.7425991322266522,
      "learning_rate": 2.109128093272745e-07,
      "loss": 0.8682,
      "step": 11877
    },
    {
      "epoch": 0.8569676418599618,
      "grad_norm": 5.835347545377378,
      "learning_rate": 2.1070395735598744e-07,
      "loss": 0.9057,
      "step": 11878
    },
    {
      "epoch": 0.8570397893293893,
      "grad_norm": 4.353515625,
      "learning_rate": 2.1049520309153411e-07,
      "loss": 0.8763,
      "step": 11879
    },
    {
      "epoch": 0.8571119367988168,
      "grad_norm": 8.006154077507688,
      "learning_rate": 2.1028654654531365e-07,
      "loss": 0.9658,
      "step": 11880
    },
    {
      "epoch": 0.8571840842682443,
      "grad_norm": 6.394839662118137,
      "learning_rate": 2.100779877287191e-07,
      "loss": 0.9511,
      "step": 11881
    },
    {
      "epoch": 0.8572562317376718,
      "grad_norm": 5.446047324208589,
      "learning_rate": 2.0986952665313963e-07,
      "loss": 0.9025,
      "step": 11882
    },
    {
      "epoch": 0.8573283792070993,
      "grad_norm": 7.909098115955605,
      "learning_rate": 2.0966116332995853e-07,
      "loss": 0.8763,
      "step": 11883
    },
    {
      "epoch": 0.8574005266765268,
      "grad_norm": 3.4321531847666145,
      "learning_rate": 2.0945289777055385e-07,
      "loss": 0.8874,
      "step": 11884
    },
    {
      "epoch": 0.8574726741459543,
      "grad_norm": 6.220107653995622,
      "learning_rate": 2.0924472998629762e-07,
      "loss": 0.9336,
      "step": 11885
    },
    {
      "epoch": 0.8575448216153818,
      "grad_norm": 2.841243100038366,
      "learning_rate": 2.0903665998855757e-07,
      "loss": 0.9608,
      "step": 11886
    },
    {
      "epoch": 0.8576169690848093,
      "grad_norm": 3.3127331921530563,
      "learning_rate": 2.0882868778869623e-07,
      "loss": 1.0402,
      "step": 11887
    },
    {
      "epoch": 0.8576891165542369,
      "grad_norm": 4.508101482152845,
      "learning_rate": 2.0862081339806915e-07,
      "loss": 1.0076,
      "step": 11888
    },
    {
      "epoch": 0.8577612640236644,
      "grad_norm": 4.266526123553431,
      "learning_rate": 2.084130368280279e-07,
      "loss": 0.9489,
      "step": 11889
    },
    {
      "epoch": 0.8578334114930919,
      "grad_norm": 4.364047919719199,
      "learning_rate": 2.082053580899188e-07,
      "loss": 0.8961,
      "step": 11890
    },
    {
      "epoch": 0.8579055589625194,
      "grad_norm": 3.330393989327015,
      "learning_rate": 2.079977771950816e-07,
      "loss": 0.8716,
      "step": 11891
    },
    {
      "epoch": 0.8579777064319469,
      "grad_norm": 4.322543625266723,
      "learning_rate": 2.0779029415485217e-07,
      "loss": 1.0134,
      "step": 11892
    },
    {
      "epoch": 0.8580498539013744,
      "grad_norm": 4.309790603259666,
      "learning_rate": 2.075829089805603e-07,
      "loss": 0.8753,
      "step": 11893
    },
    {
      "epoch": 0.858122001370802,
      "grad_norm": 0.6652837406821223,
      "learning_rate": 2.0737562168353073e-07,
      "loss": 0.8301,
      "step": 11894
    },
    {
      "epoch": 0.8581941488402294,
      "grad_norm": 4.030990237361184,
      "learning_rate": 2.0716843227508218e-07,
      "loss": 0.9711,
      "step": 11895
    },
    {
      "epoch": 0.8582662963096569,
      "grad_norm": 4.557987228355575,
      "learning_rate": 2.0696134076652827e-07,
      "loss": 0.8472,
      "step": 11896
    },
    {
      "epoch": 0.8583384437790844,
      "grad_norm": 3.3105618546858735,
      "learning_rate": 2.0675434716917883e-07,
      "loss": 0.8937,
      "step": 11897
    },
    {
      "epoch": 0.8584105912485119,
      "grad_norm": 3.1756084249657386,
      "learning_rate": 2.065474514943355e-07,
      "loss": 0.9192,
      "step": 11898
    },
    {
      "epoch": 0.8584827387179395,
      "grad_norm": 3.1698224421693357,
      "learning_rate": 2.0634065375329702e-07,
      "loss": 0.914,
      "step": 11899
    },
    {
      "epoch": 0.858554886187367,
      "grad_norm": 2.688445346372533,
      "learning_rate": 2.0613395395735523e-07,
      "loss": 0.9569,
      "step": 11900
    },
    {
      "epoch": 0.8586270336567945,
      "grad_norm": 4.583184788926609,
      "learning_rate": 2.05927352117798e-07,
      "loss": 0.9522,
      "step": 11901
    },
    {
      "epoch": 0.858699181126222,
      "grad_norm": 7.213488490233451,
      "learning_rate": 2.0572084824590608e-07,
      "loss": 0.9772,
      "step": 11902
    },
    {
      "epoch": 0.8587713285956495,
      "grad_norm": 3.3495665255550087,
      "learning_rate": 2.0551444235295668e-07,
      "loss": 0.9644,
      "step": 11903
    },
    {
      "epoch": 0.858843476065077,
      "grad_norm": 3.779509971584303,
      "learning_rate": 2.05308134450221e-07,
      "loss": 0.9567,
      "step": 11904
    },
    {
      "epoch": 0.8589156235345046,
      "grad_norm": 9.204236207983978,
      "learning_rate": 2.0510192454896403e-07,
      "loss": 0.9866,
      "step": 11905
    },
    {
      "epoch": 0.8589877710039321,
      "grad_norm": 3.572758410686736,
      "learning_rate": 2.0489581266044632e-07,
      "loss": 1.0154,
      "step": 11906
    },
    {
      "epoch": 0.8590599184733595,
      "grad_norm": 35.025832751855276,
      "learning_rate": 2.046897987959235e-07,
      "loss": 0.8099,
      "step": 11907
    },
    {
      "epoch": 0.859132065942787,
      "grad_norm": 3.677804517760809,
      "learning_rate": 2.0448388296664442e-07,
      "loss": 0.8861,
      "step": 11908
    },
    {
      "epoch": 0.8592042134122145,
      "grad_norm": 10.90218559541324,
      "learning_rate": 2.0427806518385337e-07,
      "loss": 0.9582,
      "step": 11909
    },
    {
      "epoch": 0.859276360881642,
      "grad_norm": 3.5136596611620683,
      "learning_rate": 2.0407234545879003e-07,
      "loss": 0.8452,
      "step": 11910
    },
    {
      "epoch": 0.8593485083510696,
      "grad_norm": 3.7902627518696446,
      "learning_rate": 2.0386672380268787e-07,
      "loss": 0.9203,
      "step": 11911
    },
    {
      "epoch": 0.8594206558204971,
      "grad_norm": 5.073235514161877,
      "learning_rate": 2.0366120022677457e-07,
      "loss": 1.0551,
      "step": 11912
    },
    {
      "epoch": 0.8594928032899246,
      "grad_norm": 3.3482699638460245,
      "learning_rate": 2.034557747422734e-07,
      "loss": 0.9179,
      "step": 11913
    },
    {
      "epoch": 0.8595649507593521,
      "grad_norm": 3.984769076138406,
      "learning_rate": 2.03250447360402e-07,
      "loss": 1.0266,
      "step": 11914
    },
    {
      "epoch": 0.8596370982287797,
      "grad_norm": 8.809421732150346,
      "learning_rate": 2.0304521809237162e-07,
      "loss": 0.9324,
      "step": 11915
    },
    {
      "epoch": 0.8597092456982072,
      "grad_norm": 4.923412845686645,
      "learning_rate": 2.0284008694939026e-07,
      "loss": 0.9399,
      "step": 11916
    },
    {
      "epoch": 0.8597813931676347,
      "grad_norm": 4.328937550644825,
      "learning_rate": 2.0263505394265912e-07,
      "loss": 0.9328,
      "step": 11917
    },
    {
      "epoch": 0.8598535406370622,
      "grad_norm": 3.2891480350925115,
      "learning_rate": 2.0243011908337393e-07,
      "loss": 0.9238,
      "step": 11918
    },
    {
      "epoch": 0.8599256881064896,
      "grad_norm": 4.004309716714361,
      "learning_rate": 2.022252823827253e-07,
      "loss": 0.9068,
      "step": 11919
    },
    {
      "epoch": 0.8599978355759171,
      "grad_norm": 2.7059557430992953,
      "learning_rate": 2.0202054385189848e-07,
      "loss": 1.0223,
      "step": 11920
    },
    {
      "epoch": 0.8600699830453447,
      "grad_norm": 4.9220776712984655,
      "learning_rate": 2.0181590350207479e-07,
      "loss": 1.0445,
      "step": 11921
    },
    {
      "epoch": 0.8601421305147722,
      "grad_norm": 3.4418277200943064,
      "learning_rate": 2.0161136134442748e-07,
      "loss": 0.983,
      "step": 11922
    },
    {
      "epoch": 0.8602142779841997,
      "grad_norm": 5.3549895291266125,
      "learning_rate": 2.014069173901265e-07,
      "loss": 0.9034,
      "step": 11923
    },
    {
      "epoch": 0.8602864254536272,
      "grad_norm": 3.712467946129313,
      "learning_rate": 2.0120257165033584e-07,
      "loss": 0.9436,
      "step": 11924
    },
    {
      "epoch": 0.8603585729230547,
      "grad_norm": 4.85507590928926,
      "learning_rate": 2.0099832413621343e-07,
      "loss": 0.9164,
      "step": 11925
    },
    {
      "epoch": 0.8604307203924823,
      "grad_norm": 10.962062960082847,
      "learning_rate": 2.0079417485891236e-07,
      "loss": 0.977,
      "step": 11926
    },
    {
      "epoch": 0.8605028678619098,
      "grad_norm": 5.21294642304579,
      "learning_rate": 2.0059012382958152e-07,
      "loss": 0.8851,
      "step": 11927
    },
    {
      "epoch": 0.8605750153313373,
      "grad_norm": 0.7757340062441529,
      "learning_rate": 2.003861710593633e-07,
      "loss": 0.8083,
      "step": 11928
    },
    {
      "epoch": 0.8606471628007648,
      "grad_norm": 3.718467894039682,
      "learning_rate": 2.0018231655939365e-07,
      "loss": 0.8463,
      "step": 11929
    },
    {
      "epoch": 0.8607193102701923,
      "grad_norm": 3.642139169052386,
      "learning_rate": 1.9997856034080528e-07,
      "loss": 0.974,
      "step": 11930
    },
    {
      "epoch": 0.8607914577396197,
      "grad_norm": 3.8004862875734324,
      "learning_rate": 1.997749024147246e-07,
      "loss": 1.0461,
      "step": 11931
    },
    {
      "epoch": 0.8608636052090473,
      "grad_norm": 2.8878965955910147,
      "learning_rate": 1.9957134279227185e-07,
      "loss": 0.9783,
      "step": 11932
    },
    {
      "epoch": 0.8609357526784748,
      "grad_norm": 4.893768762623176,
      "learning_rate": 1.9936788148456273e-07,
      "loss": 0.9605,
      "step": 11933
    },
    {
      "epoch": 0.8610079001479023,
      "grad_norm": 4.118134536257384,
      "learning_rate": 1.9916451850270865e-07,
      "loss": 1.0842,
      "step": 11934
    },
    {
      "epoch": 0.8610800476173298,
      "grad_norm": 13.420805172366126,
      "learning_rate": 1.9896125385781337e-07,
      "loss": 0.9966,
      "step": 11935
    },
    {
      "epoch": 0.8611521950867573,
      "grad_norm": 3.9215392922725387,
      "learning_rate": 1.987580875609769e-07,
      "loss": 0.9813,
      "step": 11936
    },
    {
      "epoch": 0.8612243425561849,
      "grad_norm": 3.1760968453850156,
      "learning_rate": 1.9855501962329302e-07,
      "loss": 1.0131,
      "step": 11937
    },
    {
      "epoch": 0.8612964900256124,
      "grad_norm": 3.3084396812840375,
      "learning_rate": 1.9835205005585155e-07,
      "loss": 0.9801,
      "step": 11938
    },
    {
      "epoch": 0.8613686374950399,
      "grad_norm": 3.8185343280331625,
      "learning_rate": 1.9814917886973403e-07,
      "loss": 0.8975,
      "step": 11939
    },
    {
      "epoch": 0.8614407849644674,
      "grad_norm": 5.906261948669296,
      "learning_rate": 1.9794640607602032e-07,
      "loss": 1.0295,
      "step": 11940
    },
    {
      "epoch": 0.8615129324338949,
      "grad_norm": 0.7953016793764645,
      "learning_rate": 1.9774373168578263e-07,
      "loss": 0.8528,
      "step": 11941
    },
    {
      "epoch": 0.8615850799033223,
      "grad_norm": 3.067608834397969,
      "learning_rate": 1.9754115571008767e-07,
      "loss": 0.9533,
      "step": 11942
    },
    {
      "epoch": 0.8616572273727499,
      "grad_norm": 4.161523900412691,
      "learning_rate": 1.973386781599977e-07,
      "loss": 1.0503,
      "step": 11943
    },
    {
      "epoch": 0.8617293748421774,
      "grad_norm": 6.631510864074041,
      "learning_rate": 1.9713629904656902e-07,
      "loss": 0.8782,
      "step": 11944
    },
    {
      "epoch": 0.8618015223116049,
      "grad_norm": 4.172876359297667,
      "learning_rate": 1.9693401838085367e-07,
      "loss": 1.1231,
      "step": 11945
    },
    {
      "epoch": 0.8618736697810324,
      "grad_norm": 7.56871521340904,
      "learning_rate": 1.9673183617389656e-07,
      "loss": 0.9293,
      "step": 11946
    },
    {
      "epoch": 0.8619458172504599,
      "grad_norm": 3.18209605180264,
      "learning_rate": 1.9652975243673842e-07,
      "loss": 0.935,
      "step": 11947
    },
    {
      "epoch": 0.8620179647198875,
      "grad_norm": 10.220255796213667,
      "learning_rate": 1.9632776718041467e-07,
      "loss": 0.9414,
      "step": 11948
    },
    {
      "epoch": 0.862090112189315,
      "grad_norm": 2.6887417076050357,
      "learning_rate": 1.9612588041595425e-07,
      "loss": 0.9933,
      "step": 11949
    },
    {
      "epoch": 0.8621622596587425,
      "grad_norm": 2.6852681851134843,
      "learning_rate": 1.9592409215438122e-07,
      "loss": 0.8383,
      "step": 11950
    },
    {
      "epoch": 0.86223440712817,
      "grad_norm": 3.4454240499763267,
      "learning_rate": 1.9572240240671588e-07,
      "loss": 0.8708,
      "step": 11951
    },
    {
      "epoch": 0.8623065545975975,
      "grad_norm": 13.254166433859403,
      "learning_rate": 1.955208111839708e-07,
      "loss": 0.9245,
      "step": 11952
    },
    {
      "epoch": 0.862378702067025,
      "grad_norm": 3.061527798541551,
      "learning_rate": 1.95319318497154e-07,
      "loss": 0.9319,
      "step": 11953
    },
    {
      "epoch": 0.8624508495364525,
      "grad_norm": 2.2184028891251932,
      "learning_rate": 1.951179243572687e-07,
      "loss": 0.9375,
      "step": 11954
    },
    {
      "epoch": 0.86252299700588,
      "grad_norm": 6.034463767411739,
      "learning_rate": 1.9491662877531234e-07,
      "loss": 0.9658,
      "step": 11955
    },
    {
      "epoch": 0.8625951444753075,
      "grad_norm": 6.2292217477598015,
      "learning_rate": 1.9471543176227635e-07,
      "loss": 0.9397,
      "step": 11956
    },
    {
      "epoch": 0.862667291944735,
      "grad_norm": 3.134244139840577,
      "learning_rate": 1.9451433332914724e-07,
      "loss": 0.9306,
      "step": 11957
    },
    {
      "epoch": 0.8627394394141625,
      "grad_norm": 3.7094633688037777,
      "learning_rate": 1.943133334869076e-07,
      "loss": 0.9086,
      "step": 11958
    },
    {
      "epoch": 0.86281158688359,
      "grad_norm": 0.834358506629032,
      "learning_rate": 1.9411243224653195e-07,
      "loss": 0.8713,
      "step": 11959
    },
    {
      "epoch": 0.8628837343530176,
      "grad_norm": 5.30969092207713,
      "learning_rate": 1.9391162961899132e-07,
      "loss": 0.9177,
      "step": 11960
    },
    {
      "epoch": 0.8629558818224451,
      "grad_norm": 4.527107818485806,
      "learning_rate": 1.9371092561525093e-07,
      "loss": 0.9365,
      "step": 11961
    },
    {
      "epoch": 0.8630280292918726,
      "grad_norm": 4.771700828786154,
      "learning_rate": 1.9351032024626979e-07,
      "loss": 0.89,
      "step": 11962
    },
    {
      "epoch": 0.8631001767613001,
      "grad_norm": 3.2741775586562056,
      "learning_rate": 1.9330981352300246e-07,
      "loss": 0.9405,
      "step": 11963
    },
    {
      "epoch": 0.8631723242307277,
      "grad_norm": 2.933249299695237,
      "learning_rate": 1.9310940545639842e-07,
      "loss": 1.0225,
      "step": 11964
    },
    {
      "epoch": 0.8632444717001552,
      "grad_norm": 6.397953373778749,
      "learning_rate": 1.9290909605740112e-07,
      "loss": 0.9132,
      "step": 11965
    },
    {
      "epoch": 0.8633166191695826,
      "grad_norm": 4.103262993844688,
      "learning_rate": 1.927088853369483e-07,
      "loss": 0.9874,
      "step": 11966
    },
    {
      "epoch": 0.8633887666390101,
      "grad_norm": 0.7713138440136428,
      "learning_rate": 1.925087733059727e-07,
      "loss": 0.8916,
      "step": 11967
    },
    {
      "epoch": 0.8634609141084376,
      "grad_norm": 6.757256457434702,
      "learning_rate": 1.923087599754023e-07,
      "loss": 1.0271,
      "step": 11968
    },
    {
      "epoch": 0.8635330615778651,
      "grad_norm": 9.291277776086686,
      "learning_rate": 1.9210884535615812e-07,
      "loss": 1.0627,
      "step": 11969
    },
    {
      "epoch": 0.8636052090472927,
      "grad_norm": 4.465606223855078,
      "learning_rate": 1.9190902945915742e-07,
      "loss": 0.8735,
      "step": 11970
    },
    {
      "epoch": 0.8636773565167202,
      "grad_norm": 0.7015710401121441,
      "learning_rate": 1.917093122953115e-07,
      "loss": 0.7727,
      "step": 11971
    },
    {
      "epoch": 0.8637495039861477,
      "grad_norm": 4.2678288506153255,
      "learning_rate": 1.9150969387552629e-07,
      "loss": 0.8988,
      "step": 11972
    },
    {
      "epoch": 0.8638216514555752,
      "grad_norm": 4.3898668822524085,
      "learning_rate": 1.9131017421070128e-07,
      "loss": 0.9286,
      "step": 11973
    },
    {
      "epoch": 0.8638937989250027,
      "grad_norm": 6.5238038468427355,
      "learning_rate": 1.9111075331173197e-07,
      "loss": 0.9506,
      "step": 11974
    },
    {
      "epoch": 0.8639659463944303,
      "grad_norm": 4.070472289479911,
      "learning_rate": 1.9091143118950902e-07,
      "loss": 1.035,
      "step": 11975
    },
    {
      "epoch": 0.8640380938638578,
      "grad_norm": 4.663308138521387,
      "learning_rate": 1.9071220785491525e-07,
      "loss": 1.0071,
      "step": 11976
    },
    {
      "epoch": 0.8641102413332853,
      "grad_norm": 9.70342732079854,
      "learning_rate": 1.9051308331883e-07,
      "loss": 1.029,
      "step": 11977
    },
    {
      "epoch": 0.8641823888027127,
      "grad_norm": 4.152466392679655,
      "learning_rate": 1.9031405759212716e-07,
      "loss": 0.9593,
      "step": 11978
    },
    {
      "epoch": 0.8642545362721402,
      "grad_norm": 3.8390178548959577,
      "learning_rate": 1.901151306856741e-07,
      "loss": 0.9522,
      "step": 11979
    },
    {
      "epoch": 0.8643266837415677,
      "grad_norm": 10.816428115949368,
      "learning_rate": 1.8991630261033388e-07,
      "loss": 1.0234,
      "step": 11980
    },
    {
      "epoch": 0.8643988312109953,
      "grad_norm": 4.172880701578872,
      "learning_rate": 1.8971757337696314e-07,
      "loss": 0.8848,
      "step": 11981
    },
    {
      "epoch": 0.8644709786804228,
      "grad_norm": 2.4199331066249643,
      "learning_rate": 1.89518942996415e-07,
      "loss": 0.9962,
      "step": 11982
    },
    {
      "epoch": 0.8645431261498503,
      "grad_norm": 3.3298777947899025,
      "learning_rate": 1.8932041147953504e-07,
      "loss": 0.9377,
      "step": 11983
    },
    {
      "epoch": 0.8646152736192778,
      "grad_norm": 13.661374315688718,
      "learning_rate": 1.891219788371643e-07,
      "loss": 0.8534,
      "step": 11984
    },
    {
      "epoch": 0.8646874210887053,
      "grad_norm": 5.582361146162663,
      "learning_rate": 1.8892364508013926e-07,
      "loss": 0.9836,
      "step": 11985
    },
    {
      "epoch": 0.8647595685581329,
      "grad_norm": 4.284041782278689,
      "learning_rate": 1.8872541021928902e-07,
      "loss": 0.9551,
      "step": 11986
    },
    {
      "epoch": 0.8648317160275604,
      "grad_norm": 4.0499392657965405,
      "learning_rate": 1.8852727426543868e-07,
      "loss": 0.9445,
      "step": 11987
    },
    {
      "epoch": 0.8649038634969879,
      "grad_norm": 4.488060901833813,
      "learning_rate": 1.8832923722940853e-07,
      "loss": 0.97,
      "step": 11988
    },
    {
      "epoch": 0.8649760109664153,
      "grad_norm": 4.231198854573989,
      "learning_rate": 1.8813129912201232e-07,
      "loss": 0.8943,
      "step": 11989
    },
    {
      "epoch": 0.8650481584358428,
      "grad_norm": 4.214877237181614,
      "learning_rate": 1.879334599540583e-07,
      "loss": 0.9709,
      "step": 11990
    },
    {
      "epoch": 0.8651203059052703,
      "grad_norm": 3.660649057797405,
      "learning_rate": 1.8773571973635006e-07,
      "loss": 0.8618,
      "step": 11991
    },
    {
      "epoch": 0.8651924533746979,
      "grad_norm": 4.155265748927963,
      "learning_rate": 1.8753807847968584e-07,
      "loss": 0.9409,
      "step": 11992
    },
    {
      "epoch": 0.8652646008441254,
      "grad_norm": 3.429223518008472,
      "learning_rate": 1.873405361948568e-07,
      "loss": 0.9354,
      "step": 11993
    },
    {
      "epoch": 0.8653367483135529,
      "grad_norm": 7.468100036686639,
      "learning_rate": 1.871430928926512e-07,
      "loss": 0.8972,
      "step": 11994
    },
    {
      "epoch": 0.8654088957829804,
      "grad_norm": 4.454923567710295,
      "learning_rate": 1.869457485838506e-07,
      "loss": 0.9707,
      "step": 11995
    },
    {
      "epoch": 0.8654810432524079,
      "grad_norm": 3.7249450372314903,
      "learning_rate": 1.8674850327923063e-07,
      "loss": 0.9648,
      "step": 11996
    },
    {
      "epoch": 0.8655531907218355,
      "grad_norm": 2.692685115475539,
      "learning_rate": 1.8655135698956247e-07,
      "loss": 0.8843,
      "step": 11997
    },
    {
      "epoch": 0.865625338191263,
      "grad_norm": 5.065493605155098,
      "learning_rate": 1.863543097256113e-07,
      "loss": 0.9827,
      "step": 11998
    },
    {
      "epoch": 0.8656974856606905,
      "grad_norm": 3.7882328344245817,
      "learning_rate": 1.86157361498138e-07,
      "loss": 0.9392,
      "step": 11999
    },
    {
      "epoch": 0.865769633130118,
      "grad_norm": 2.838266808504325,
      "learning_rate": 1.8596051231789645e-07,
      "loss": 0.909,
      "step": 12000
    },
    {
      "epoch": 0.8658417805995454,
      "grad_norm": 3.5561446672068904,
      "learning_rate": 1.857637621956356e-07,
      "loss": 0.9448,
      "step": 12001
    },
    {
      "epoch": 0.8659139280689729,
      "grad_norm": 4.149586089301487,
      "learning_rate": 1.8556711114210022e-07,
      "loss": 0.8826,
      "step": 12002
    },
    {
      "epoch": 0.8659860755384005,
      "grad_norm": 3.8016964789635774,
      "learning_rate": 1.853705591680277e-07,
      "loss": 1.0038,
      "step": 12003
    },
    {
      "epoch": 0.866058223007828,
      "grad_norm": 4.80870481187746,
      "learning_rate": 1.8517410628415142e-07,
      "loss": 0.9681,
      "step": 12004
    },
    {
      "epoch": 0.8661303704772555,
      "grad_norm": 0.7014992249974642,
      "learning_rate": 1.849777525011993e-07,
      "loss": 0.8153,
      "step": 12005
    },
    {
      "epoch": 0.866202517946683,
      "grad_norm": 2.426381109155203,
      "learning_rate": 1.847814978298925e-07,
      "loss": 1.0514,
      "step": 12006
    },
    {
      "epoch": 0.8662746654161105,
      "grad_norm": 4.241399702403308,
      "learning_rate": 1.8458534228094868e-07,
      "loss": 0.9575,
      "step": 12007
    },
    {
      "epoch": 0.866346812885538,
      "grad_norm": 6.038949114091614,
      "learning_rate": 1.8438928586507885e-07,
      "loss": 0.982,
      "step": 12008
    },
    {
      "epoch": 0.8664189603549656,
      "grad_norm": 3.8302000066261916,
      "learning_rate": 1.841933285929893e-07,
      "loss": 0.8961,
      "step": 12009
    },
    {
      "epoch": 0.8664911078243931,
      "grad_norm": 4.335672896275845,
      "learning_rate": 1.8399747047537994e-07,
      "loss": 0.8897,
      "step": 12010
    },
    {
      "epoch": 0.8665632552938206,
      "grad_norm": 3.3311633358215675,
      "learning_rate": 1.8380171152294555e-07,
      "loss": 1.0791,
      "step": 12011
    },
    {
      "epoch": 0.8666354027632481,
      "grad_norm": 6.4725758024383575,
      "learning_rate": 1.8360605174637733e-07,
      "loss": 0.9269,
      "step": 12012
    },
    {
      "epoch": 0.8667075502326755,
      "grad_norm": 5.168431729131796,
      "learning_rate": 1.834104911563581e-07,
      "loss": 0.9444,
      "step": 12013
    },
    {
      "epoch": 0.8667796977021031,
      "grad_norm": 3.175310651344833,
      "learning_rate": 1.8321502976356705e-07,
      "loss": 0.9672,
      "step": 12014
    },
    {
      "epoch": 0.8668518451715306,
      "grad_norm": 9.70377916474427,
      "learning_rate": 1.830196675786777e-07,
      "loss": 0.9943,
      "step": 12015
    },
    {
      "epoch": 0.8669239926409581,
      "grad_norm": 3.942023812069673,
      "learning_rate": 1.8282440461235858e-07,
      "loss": 0.8488,
      "step": 12016
    },
    {
      "epoch": 0.8669961401103856,
      "grad_norm": 4.248645791113871,
      "learning_rate": 1.8262924087527097e-07,
      "loss": 0.8958,
      "step": 12017
    },
    {
      "epoch": 0.8670682875798131,
      "grad_norm": 2.5476103113191866,
      "learning_rate": 1.8243417637807345e-07,
      "loss": 0.9416,
      "step": 12018
    },
    {
      "epoch": 0.8671404350492407,
      "grad_norm": 3.49499085148172,
      "learning_rate": 1.8223921113141726e-07,
      "loss": 0.9592,
      "step": 12019
    },
    {
      "epoch": 0.8672125825186682,
      "grad_norm": 2.5135731355352138,
      "learning_rate": 1.8204434514594835e-07,
      "loss": 0.9139,
      "step": 12020
    },
    {
      "epoch": 0.8672847299880957,
      "grad_norm": 2.669599311280269,
      "learning_rate": 1.8184957843230776e-07,
      "loss": 0.9403,
      "step": 12021
    },
    {
      "epoch": 0.8673568774575232,
      "grad_norm": 3.5835802895072986,
      "learning_rate": 1.8165491100113163e-07,
      "loss": 0.9553,
      "step": 12022
    },
    {
      "epoch": 0.8674290249269507,
      "grad_norm": 0.6266462817030096,
      "learning_rate": 1.8146034286304879e-07,
      "loss": 0.8209,
      "step": 12023
    },
    {
      "epoch": 0.8675011723963783,
      "grad_norm": 2.8805592732991534,
      "learning_rate": 1.8126587402868498e-07,
      "loss": 0.9949,
      "step": 12024
    },
    {
      "epoch": 0.8675733198658057,
      "grad_norm": 7.514626545716163,
      "learning_rate": 1.8107150450865906e-07,
      "loss": 0.8224,
      "step": 12025
    },
    {
      "epoch": 0.8676454673352332,
      "grad_norm": 3.8568325548644458,
      "learning_rate": 1.8087723431358514e-07,
      "loss": 0.9764,
      "step": 12026
    },
    {
      "epoch": 0.8677176148046607,
      "grad_norm": 8.361093390844866,
      "learning_rate": 1.8068306345407104e-07,
      "loss": 0.915,
      "step": 12027
    },
    {
      "epoch": 0.8677897622740882,
      "grad_norm": 3.9696421183746065,
      "learning_rate": 1.8048899194071997e-07,
      "loss": 0.9421,
      "step": 12028
    },
    {
      "epoch": 0.8678619097435157,
      "grad_norm": 3.5146541356927967,
      "learning_rate": 1.8029501978412974e-07,
      "loss": 0.9703,
      "step": 12029
    },
    {
      "epoch": 0.8679340572129433,
      "grad_norm": 3.2958161158700747,
      "learning_rate": 1.8010114699489142e-07,
      "loss": 0.8996,
      "step": 12030
    },
    {
      "epoch": 0.8680062046823708,
      "grad_norm": 8.185931106393046,
      "learning_rate": 1.7990737358359276e-07,
      "loss": 0.9867,
      "step": 12031
    },
    {
      "epoch": 0.8680783521517983,
      "grad_norm": 3.6008960827232683,
      "learning_rate": 1.7971369956081483e-07,
      "loss": 0.9738,
      "step": 12032
    },
    {
      "epoch": 0.8681504996212258,
      "grad_norm": 5.790003402788022,
      "learning_rate": 1.7952012493713365e-07,
      "loss": 0.9837,
      "step": 12033
    },
    {
      "epoch": 0.8682226470906533,
      "grad_norm": 3.9939136692472337,
      "learning_rate": 1.7932664972311872e-07,
      "loss": 0.9989,
      "step": 12034
    },
    {
      "epoch": 0.8682947945600809,
      "grad_norm": 2.70828750767208,
      "learning_rate": 1.791332739293354e-07,
      "loss": 0.9532,
      "step": 12035
    },
    {
      "epoch": 0.8683669420295083,
      "grad_norm": 9.739515487244194,
      "learning_rate": 1.789399975663439e-07,
      "loss": 0.9573,
      "step": 12036
    },
    {
      "epoch": 0.8684390894989358,
      "grad_norm": 3.9904725334657094,
      "learning_rate": 1.787468206446976e-07,
      "loss": 0.9786,
      "step": 12037
    },
    {
      "epoch": 0.8685112369683633,
      "grad_norm": 3.1722390853810714,
      "learning_rate": 1.785537431749453e-07,
      "loss": 0.9212,
      "step": 12038
    },
    {
      "epoch": 0.8685833844377908,
      "grad_norm": 4.26916692214142,
      "learning_rate": 1.7836076516763067e-07,
      "loss": 0.916,
      "step": 12039
    },
    {
      "epoch": 0.8686555319072183,
      "grad_norm": 5.606692381961256,
      "learning_rate": 1.7816788663329074e-07,
      "loss": 0.8729,
      "step": 12040
    },
    {
      "epoch": 0.8687276793766459,
      "grad_norm": 3.128052403296427,
      "learning_rate": 1.7797510758245805e-07,
      "loss": 1.0055,
      "step": 12041
    },
    {
      "epoch": 0.8687998268460734,
      "grad_norm": 2.8294259972649973,
      "learning_rate": 1.7778242802566014e-07,
      "loss": 1.003,
      "step": 12042
    },
    {
      "epoch": 0.8688719743155009,
      "grad_norm": 6.552585050925274,
      "learning_rate": 1.775898479734188e-07,
      "loss": 0.8795,
      "step": 12043
    },
    {
      "epoch": 0.8689441217849284,
      "grad_norm": 3.232153677978708,
      "learning_rate": 1.77397367436249e-07,
      "loss": 0.9005,
      "step": 12044
    },
    {
      "epoch": 0.8690162692543559,
      "grad_norm": 6.1837091103816,
      "learning_rate": 1.7720498642466185e-07,
      "loss": 0.8907,
      "step": 12045
    },
    {
      "epoch": 0.8690884167237835,
      "grad_norm": 3.20973067796746,
      "learning_rate": 1.7701270494916298e-07,
      "loss": 1.0214,
      "step": 12046
    },
    {
      "epoch": 0.869160564193211,
      "grad_norm": 4.550560575054596,
      "learning_rate": 1.768205230202513e-07,
      "loss": 0.9484,
      "step": 12047
    },
    {
      "epoch": 0.8692327116626384,
      "grad_norm": 3.0285629475415696,
      "learning_rate": 1.7662844064842198e-07,
      "loss": 0.9677,
      "step": 12048
    },
    {
      "epoch": 0.8693048591320659,
      "grad_norm": 6.357360970810419,
      "learning_rate": 1.7643645784416394e-07,
      "loss": 0.9355,
      "step": 12049
    },
    {
      "epoch": 0.8693770066014934,
      "grad_norm": 3.416670372813626,
      "learning_rate": 1.7624457461795994e-07,
      "loss": 0.892,
      "step": 12050
    },
    {
      "epoch": 0.8694491540709209,
      "grad_norm": 3.3275640310407657,
      "learning_rate": 1.7605279098028846e-07,
      "loss": 0.9545,
      "step": 12051
    },
    {
      "epoch": 0.8695213015403485,
      "grad_norm": 3.796302532476939,
      "learning_rate": 1.7586110694162205e-07,
      "loss": 0.9815,
      "step": 12052
    },
    {
      "epoch": 0.869593449009776,
      "grad_norm": 4.506379585677935,
      "learning_rate": 1.7566952251242806e-07,
      "loss": 0.9987,
      "step": 12053
    },
    {
      "epoch": 0.8696655964792035,
      "grad_norm": 3.811797655793573,
      "learning_rate": 1.754780377031675e-07,
      "loss": 0.8938,
      "step": 12054
    },
    {
      "epoch": 0.869737743948631,
      "grad_norm": 0.7774426957996504,
      "learning_rate": 1.752866525242973e-07,
      "loss": 0.8878,
      "step": 12055
    },
    {
      "epoch": 0.8698098914180585,
      "grad_norm": 3.169275581065843,
      "learning_rate": 1.750953669862687e-07,
      "loss": 1.0666,
      "step": 12056
    },
    {
      "epoch": 0.869882038887486,
      "grad_norm": 3.061436215364388,
      "learning_rate": 1.749041810995262e-07,
      "loss": 0.9482,
      "step": 12057
    },
    {
      "epoch": 0.8699541863569136,
      "grad_norm": 3.7222327584897243,
      "learning_rate": 1.747130948745099e-07,
      "loss": 0.8209,
      "step": 12058
    },
    {
      "epoch": 0.8700263338263411,
      "grad_norm": 3.0437729914930705,
      "learning_rate": 1.7452210832165438e-07,
      "loss": 0.9319,
      "step": 12059
    },
    {
      "epoch": 0.8700984812957685,
      "grad_norm": 3.1965859261009673,
      "learning_rate": 1.7433122145138945e-07,
      "loss": 0.9291,
      "step": 12060
    },
    {
      "epoch": 0.870170628765196,
      "grad_norm": 4.714602434988036,
      "learning_rate": 1.7414043427413772e-07,
      "loss": 0.9454,
      "step": 12061
    },
    {
      "epoch": 0.8702427762346235,
      "grad_norm": 3.7554156298656944,
      "learning_rate": 1.739497468003177e-07,
      "loss": 0.8998,
      "step": 12062
    },
    {
      "epoch": 0.8703149237040511,
      "grad_norm": 5.349077212886251,
      "learning_rate": 1.7375915904034265e-07,
      "loss": 0.9685,
      "step": 12063
    },
    {
      "epoch": 0.8703870711734786,
      "grad_norm": 7.169472285686346,
      "learning_rate": 1.7356867100461914e-07,
      "loss": 0.9333,
      "step": 12064
    },
    {
      "epoch": 0.8704592186429061,
      "grad_norm": 4.9509582593834525,
      "learning_rate": 1.7337828270354882e-07,
      "loss": 0.9169,
      "step": 12065
    },
    {
      "epoch": 0.8705313661123336,
      "grad_norm": 5.177754451406287,
      "learning_rate": 1.7318799414752938e-07,
      "loss": 0.9213,
      "step": 12066
    },
    {
      "epoch": 0.8706035135817611,
      "grad_norm": 10.250364343634137,
      "learning_rate": 1.7299780534695052e-07,
      "loss": 0.8574,
      "step": 12067
    },
    {
      "epoch": 0.8706756610511887,
      "grad_norm": 2.2287554733029444,
      "learning_rate": 1.7280771631219816e-07,
      "loss": 0.9841,
      "step": 12068
    },
    {
      "epoch": 0.8707478085206162,
      "grad_norm": 3.9533354812496153,
      "learning_rate": 1.726177270536524e-07,
      "loss": 0.983,
      "step": 12069
    },
    {
      "epoch": 0.8708199559900437,
      "grad_norm": 3.1114015349057045,
      "learning_rate": 1.7242783758168834e-07,
      "loss": 1.0023,
      "step": 12070
    },
    {
      "epoch": 0.8708921034594712,
      "grad_norm": 3.1769487362893116,
      "learning_rate": 1.7223804790667362e-07,
      "loss": 1.0414,
      "step": 12071
    },
    {
      "epoch": 0.8709642509288986,
      "grad_norm": 4.591836543786008,
      "learning_rate": 1.7204835803897356e-07,
      "loss": 0.9323,
      "step": 12072
    },
    {
      "epoch": 0.8710363983983261,
      "grad_norm": 7.522951991814289,
      "learning_rate": 1.7185876798894604e-07,
      "loss": 0.8847,
      "step": 12073
    },
    {
      "epoch": 0.8711085458677537,
      "grad_norm": 4.303238363441846,
      "learning_rate": 1.7166927776694328e-07,
      "loss": 0.8937,
      "step": 12074
    },
    {
      "epoch": 0.8711806933371812,
      "grad_norm": 6.8214550830950715,
      "learning_rate": 1.714798873833132e-07,
      "loss": 0.9106,
      "step": 12075
    },
    {
      "epoch": 0.8712528408066087,
      "grad_norm": 9.79322350212125,
      "learning_rate": 1.7129059684839732e-07,
      "loss": 1.0608,
      "step": 12076
    },
    {
      "epoch": 0.8713249882760362,
      "grad_norm": 3.11893469620978,
      "learning_rate": 1.711014061725329e-07,
      "loss": 1.1079,
      "step": 12077
    },
    {
      "epoch": 0.8713971357454637,
      "grad_norm": 3.7606634479958183,
      "learning_rate": 1.7091231536604923e-07,
      "loss": 0.9913,
      "step": 12078
    },
    {
      "epoch": 0.8714692832148913,
      "grad_norm": 7.059010141374312,
      "learning_rate": 1.707233244392736e-07,
      "loss": 0.9067,
      "step": 12079
    },
    {
      "epoch": 0.8715414306843188,
      "grad_norm": 0.9206202097843889,
      "learning_rate": 1.7053443340252583e-07,
      "loss": 0.8505,
      "step": 12080
    },
    {
      "epoch": 0.8716135781537463,
      "grad_norm": 4.817732901295748,
      "learning_rate": 1.703456422661198e-07,
      "loss": 0.876,
      "step": 12081
    },
    {
      "epoch": 0.8716857256231738,
      "grad_norm": 7.079433968541652,
      "learning_rate": 1.701569510403651e-07,
      "loss": 0.9954,
      "step": 12082
    },
    {
      "epoch": 0.8717578730926013,
      "grad_norm": 6.010521564416715,
      "learning_rate": 1.6996835973556588e-07,
      "loss": 0.9845,
      "step": 12083
    },
    {
      "epoch": 0.8718300205620287,
      "grad_norm": 4.100058085332522,
      "learning_rate": 1.6977986836201908e-07,
      "loss": 0.9509,
      "step": 12084
    },
    {
      "epoch": 0.8719021680314563,
      "grad_norm": 4.891716490994903,
      "learning_rate": 1.6959147693001886e-07,
      "loss": 0.8783,
      "step": 12085
    },
    {
      "epoch": 0.8719743155008838,
      "grad_norm": 3.4818964469275167,
      "learning_rate": 1.6940318544985232e-07,
      "loss": 0.9375,
      "step": 12086
    },
    {
      "epoch": 0.8720464629703113,
      "grad_norm": 3.141553447379837,
      "learning_rate": 1.6921499393180128e-07,
      "loss": 1.0023,
      "step": 12087
    },
    {
      "epoch": 0.8721186104397388,
      "grad_norm": 3.002143570884417,
      "learning_rate": 1.6902690238614192e-07,
      "loss": 0.937,
      "step": 12088
    },
    {
      "epoch": 0.8721907579091663,
      "grad_norm": 4.804234925331341,
      "learning_rate": 1.6883891082314494e-07,
      "loss": 0.8871,
      "step": 12089
    },
    {
      "epoch": 0.8722629053785939,
      "grad_norm": 6.258393716668982,
      "learning_rate": 1.6865101925307702e-07,
      "loss": 1.0489,
      "step": 12090
    },
    {
      "epoch": 0.8723350528480214,
      "grad_norm": 4.076655916172262,
      "learning_rate": 1.6846322768619724e-07,
      "loss": 0.9405,
      "step": 12091
    },
    {
      "epoch": 0.8724072003174489,
      "grad_norm": 11.998506453075407,
      "learning_rate": 1.6827553613276035e-07,
      "loss": 1.0333,
      "step": 12092
    },
    {
      "epoch": 0.8724793477868764,
      "grad_norm": 2.857676684373106,
      "learning_rate": 1.6808794460301612e-07,
      "loss": 0.8859,
      "step": 12093
    },
    {
      "epoch": 0.8725514952563039,
      "grad_norm": 2.797264796537765,
      "learning_rate": 1.6790045310720747e-07,
      "loss": 0.8945,
      "step": 12094
    },
    {
      "epoch": 0.8726236427257313,
      "grad_norm": 2.9608561723556357,
      "learning_rate": 1.6771306165557218e-07,
      "loss": 0.9824,
      "step": 12095
    },
    {
      "epoch": 0.8726957901951589,
      "grad_norm": 2.4418640195835475,
      "learning_rate": 1.6752577025834436e-07,
      "loss": 0.9371,
      "step": 12096
    },
    {
      "epoch": 0.8727679376645864,
      "grad_norm": 3.0594072879883756,
      "learning_rate": 1.6733857892575086e-07,
      "loss": 0.9496,
      "step": 12097
    },
    {
      "epoch": 0.8728400851340139,
      "grad_norm": 3.8842024282235816,
      "learning_rate": 1.671514876680129e-07,
      "loss": 0.9464,
      "step": 12098
    },
    {
      "epoch": 0.8729122326034414,
      "grad_norm": 3.154158399630706,
      "learning_rate": 1.669644964953474e-07,
      "loss": 0.9765,
      "step": 12099
    },
    {
      "epoch": 0.8729843800728689,
      "grad_norm": 3.3017427928710434,
      "learning_rate": 1.6677760541796527e-07,
      "loss": 0.8546,
      "step": 12100
    },
    {
      "epoch": 0.8730565275422965,
      "grad_norm": 5.598764910055289,
      "learning_rate": 1.6659081444607127e-07,
      "loss": 1.0376,
      "step": 12101
    },
    {
      "epoch": 0.873128675011724,
      "grad_norm": 5.386096412166907,
      "learning_rate": 1.6640412358986565e-07,
      "loss": 0.9064,
      "step": 12102
    },
    {
      "epoch": 0.8732008224811515,
      "grad_norm": 2.8447969626330027,
      "learning_rate": 1.6621753285954366e-07,
      "loss": 0.9767,
      "step": 12103
    },
    {
      "epoch": 0.873272969950579,
      "grad_norm": 2.932377020687326,
      "learning_rate": 1.6603104226529374e-07,
      "loss": 0.9785,
      "step": 12104
    },
    {
      "epoch": 0.8733451174200065,
      "grad_norm": 3.730026969914435,
      "learning_rate": 1.6584465181729933e-07,
      "loss": 0.9509,
      "step": 12105
    },
    {
      "epoch": 0.873417264889434,
      "grad_norm": 3.6613592981842675,
      "learning_rate": 1.6565836152573876e-07,
      "loss": 0.9403,
      "step": 12106
    },
    {
      "epoch": 0.8734894123588615,
      "grad_norm": 4.531897557783402,
      "learning_rate": 1.6547217140078473e-07,
      "loss": 0.9506,
      "step": 12107
    },
    {
      "epoch": 0.873561559828289,
      "grad_norm": 3.9662086580359825,
      "learning_rate": 1.6528608145260358e-07,
      "loss": 0.9838,
      "step": 12108
    },
    {
      "epoch": 0.8736337072977165,
      "grad_norm": 4.663293823082083,
      "learning_rate": 1.651000916913581e-07,
      "loss": 0.9776,
      "step": 12109
    },
    {
      "epoch": 0.873705854767144,
      "grad_norm": 6.616920006399787,
      "learning_rate": 1.649142021272043e-07,
      "loss": 0.8705,
      "step": 12110
    },
    {
      "epoch": 0.8737780022365715,
      "grad_norm": 4.091802235602097,
      "learning_rate": 1.6472841277029237e-07,
      "loss": 0.9777,
      "step": 12111
    },
    {
      "epoch": 0.8738501497059991,
      "grad_norm": 5.326016990723334,
      "learning_rate": 1.6454272363076772e-07,
      "loss": 0.9417,
      "step": 12112
    },
    {
      "epoch": 0.8739222971754266,
      "grad_norm": 3.978053324998392,
      "learning_rate": 1.6435713471877022e-07,
      "loss": 0.9832,
      "step": 12113
    },
    {
      "epoch": 0.8739944446448541,
      "grad_norm": 2.6490497612835338,
      "learning_rate": 1.6417164604443467e-07,
      "loss": 0.9114,
      "step": 12114
    },
    {
      "epoch": 0.8740665921142816,
      "grad_norm": 3.851857388796726,
      "learning_rate": 1.6398625761788941e-07,
      "loss": 0.8576,
      "step": 12115
    },
    {
      "epoch": 0.8741387395837091,
      "grad_norm": 3.53915910399506,
      "learning_rate": 1.6380096944925793e-07,
      "loss": 0.925,
      "step": 12116
    },
    {
      "epoch": 0.8742108870531367,
      "grad_norm": 3.7862967868154658,
      "learning_rate": 1.6361578154865829e-07,
      "loss": 0.8883,
      "step": 12117
    },
    {
      "epoch": 0.8742830345225642,
      "grad_norm": 4.3501353363381705,
      "learning_rate": 1.6343069392620268e-07,
      "loss": 0.9096,
      "step": 12118
    },
    {
      "epoch": 0.8743551819919916,
      "grad_norm": 6.3868337265071204,
      "learning_rate": 1.6324570659199765e-07,
      "loss": 1.0153,
      "step": 12119
    },
    {
      "epoch": 0.8744273294614191,
      "grad_norm": 4.550769304906534,
      "learning_rate": 1.6306081955614538e-07,
      "loss": 0.9912,
      "step": 12120
    },
    {
      "epoch": 0.8744994769308466,
      "grad_norm": 7.293046424750391,
      "learning_rate": 1.628760328287422e-07,
      "loss": 0.8733,
      "step": 12121
    },
    {
      "epoch": 0.8745716244002741,
      "grad_norm": 3.421029600459733,
      "learning_rate": 1.626913464198776e-07,
      "loss": 1.0073,
      "step": 12122
    },
    {
      "epoch": 0.8746437718697017,
      "grad_norm": 0.7403227821238296,
      "learning_rate": 1.6250676033963707e-07,
      "loss": 0.8657,
      "step": 12123
    },
    {
      "epoch": 0.8747159193391292,
      "grad_norm": 0.78101295689765,
      "learning_rate": 1.6232227459810056e-07,
      "loss": 0.8596,
      "step": 12124
    },
    {
      "epoch": 0.8747880668085567,
      "grad_norm": 10.721299293692061,
      "learning_rate": 1.621378892053409e-07,
      "loss": 0.9149,
      "step": 12125
    },
    {
      "epoch": 0.8748602142779842,
      "grad_norm": 3.7412726253560793,
      "learning_rate": 1.6195360417142823e-07,
      "loss": 0.9368,
      "step": 12126
    },
    {
      "epoch": 0.8749323617474117,
      "grad_norm": 16.671390830305576,
      "learning_rate": 1.617694195064252e-07,
      "loss": 0.9768,
      "step": 12127
    },
    {
      "epoch": 0.8750045092168393,
      "grad_norm": 3.38421249492632,
      "learning_rate": 1.6158533522038908e-07,
      "loss": 0.8881,
      "step": 12128
    },
    {
      "epoch": 0.8750766566862668,
      "grad_norm": 2.9142971404570224,
      "learning_rate": 1.6140135132337207e-07,
      "loss": 0.8886,
      "step": 12129
    },
    {
      "epoch": 0.8751488041556943,
      "grad_norm": 3.26396421701788,
      "learning_rate": 1.6121746782542101e-07,
      "loss": 0.928,
      "step": 12130
    },
    {
      "epoch": 0.8752209516251217,
      "grad_norm": 3.9401891093690327,
      "learning_rate": 1.6103368473657742e-07,
      "loss": 0.9544,
      "step": 12131
    },
    {
      "epoch": 0.8752930990945492,
      "grad_norm": 3.3239698664266797,
      "learning_rate": 1.6085000206687616e-07,
      "loss": 0.9252,
      "step": 12132
    },
    {
      "epoch": 0.8753652465639767,
      "grad_norm": 8.950101917901288,
      "learning_rate": 1.6066641982634833e-07,
      "loss": 0.9657,
      "step": 12133
    },
    {
      "epoch": 0.8754373940334043,
      "grad_norm": 3.988113384842679,
      "learning_rate": 1.6048293802501856e-07,
      "loss": 0.8835,
      "step": 12134
    },
    {
      "epoch": 0.8755095415028318,
      "grad_norm": 5.279888090334866,
      "learning_rate": 1.602995566729055e-07,
      "loss": 0.9351,
      "step": 12135
    },
    {
      "epoch": 0.8755816889722593,
      "grad_norm": 4.400158506486118,
      "learning_rate": 1.601162757800234e-07,
      "loss": 0.8901,
      "step": 12136
    },
    {
      "epoch": 0.8756538364416868,
      "grad_norm": 2.4489479219406527,
      "learning_rate": 1.5993309535638068e-07,
      "loss": 0.8908,
      "step": 12137
    },
    {
      "epoch": 0.8757259839111143,
      "grad_norm": 8.455395518774166,
      "learning_rate": 1.597500154119793e-07,
      "loss": 0.8829,
      "step": 12138
    },
    {
      "epoch": 0.8757981313805419,
      "grad_norm": 4.294253905821827,
      "learning_rate": 1.5956703595681774e-07,
      "loss": 0.9835,
      "step": 12139
    },
    {
      "epoch": 0.8758702788499694,
      "grad_norm": 2.967063022953462,
      "learning_rate": 1.5938415700088714e-07,
      "loss": 0.9351,
      "step": 12140
    },
    {
      "epoch": 0.8759424263193969,
      "grad_norm": 4.935838250698715,
      "learning_rate": 1.5920137855417437e-07,
      "loss": 0.9465,
      "step": 12141
    },
    {
      "epoch": 0.8760145737888243,
      "grad_norm": 2.9158759725643217,
      "learning_rate": 1.5901870062665946e-07,
      "loss": 0.8748,
      "step": 12142
    },
    {
      "epoch": 0.8760867212582518,
      "grad_norm": 0.6884359794180861,
      "learning_rate": 1.5883612322831797e-07,
      "loss": 0.7846,
      "step": 12143
    },
    {
      "epoch": 0.8761588687276793,
      "grad_norm": 6.012994681646068,
      "learning_rate": 1.586536463691208e-07,
      "loss": 0.9251,
      "step": 12144
    },
    {
      "epoch": 0.8762310161971069,
      "grad_norm": 5.27873269304333,
      "learning_rate": 1.5847127005903117e-07,
      "loss": 0.9343,
      "step": 12145
    },
    {
      "epoch": 0.8763031636665344,
      "grad_norm": 3.6735599628838673,
      "learning_rate": 1.5828899430800834e-07,
      "loss": 0.9845,
      "step": 12146
    },
    {
      "epoch": 0.8763753111359619,
      "grad_norm": 4.333258163582671,
      "learning_rate": 1.5810681912600577e-07,
      "loss": 0.968,
      "step": 12147
    },
    {
      "epoch": 0.8764474586053894,
      "grad_norm": 12.35410278372413,
      "learning_rate": 1.5792474452297167e-07,
      "loss": 1.0141,
      "step": 12148
    },
    {
      "epoch": 0.8765196060748169,
      "grad_norm": 8.930836574417052,
      "learning_rate": 1.5774277050884744e-07,
      "loss": 0.8747,
      "step": 12149
    },
    {
      "epoch": 0.8765917535442445,
      "grad_norm": 4.315238856913012,
      "learning_rate": 1.575608970935711e-07,
      "loss": 1.016,
      "step": 12150
    },
    {
      "epoch": 0.876663901013672,
      "grad_norm": 6.263821996466532,
      "learning_rate": 1.5737912428707413e-07,
      "loss": 0.9269,
      "step": 12151
    },
    {
      "epoch": 0.8767360484830995,
      "grad_norm": 3.2943245663488754,
      "learning_rate": 1.5719745209928158e-07,
      "loss": 0.8803,
      "step": 12152
    },
    {
      "epoch": 0.876808195952527,
      "grad_norm": 4.277419555549726,
      "learning_rate": 1.570158805401143e-07,
      "loss": 0.8853,
      "step": 12153
    },
    {
      "epoch": 0.8768803434219544,
      "grad_norm": 3.248658123295029,
      "learning_rate": 1.568344096194878e-07,
      "loss": 0.8458,
      "step": 12154
    },
    {
      "epoch": 0.8769524908913819,
      "grad_norm": 4.30881021739203,
      "learning_rate": 1.5665303934731046e-07,
      "loss": 1.0125,
      "step": 12155
    },
    {
      "epoch": 0.8770246383608095,
      "grad_norm": 2.4401893451579055,
      "learning_rate": 1.5647176973348676e-07,
      "loss": 0.9533,
      "step": 12156
    },
    {
      "epoch": 0.877096785830237,
      "grad_norm": 5.825881157890979,
      "learning_rate": 1.5629060078791522e-07,
      "loss": 0.8765,
      "step": 12157
    },
    {
      "epoch": 0.8771689332996645,
      "grad_norm": 0.6879906637395538,
      "learning_rate": 1.5610953252048931e-07,
      "loss": 0.7732,
      "step": 12158
    },
    {
      "epoch": 0.877241080769092,
      "grad_norm": 5.409735289353714,
      "learning_rate": 1.5592856494109574e-07,
      "loss": 0.8297,
      "step": 12159
    },
    {
      "epoch": 0.8773132282385195,
      "grad_norm": 4.4214028689396505,
      "learning_rate": 1.5574769805961642e-07,
      "loss": 0.9619,
      "step": 12160
    },
    {
      "epoch": 0.8773853757079471,
      "grad_norm": 5.777593002462151,
      "learning_rate": 1.5556693188592872e-07,
      "loss": 0.9082,
      "step": 12161
    },
    {
      "epoch": 0.8774575231773746,
      "grad_norm": 4.5973453076273785,
      "learning_rate": 1.5538626642990215e-07,
      "loss": 0.9959,
      "step": 12162
    },
    {
      "epoch": 0.8775296706468021,
      "grad_norm": 4.818744719783201,
      "learning_rate": 1.5520570170140368e-07,
      "loss": 1.0475,
      "step": 12163
    },
    {
      "epoch": 0.8776018181162296,
      "grad_norm": 2.3430230602801667,
      "learning_rate": 1.550252377102923e-07,
      "loss": 0.8777,
      "step": 12164
    },
    {
      "epoch": 0.8776739655856571,
      "grad_norm": 3.3661420929644157,
      "learning_rate": 1.5484487446642346e-07,
      "loss": 0.9387,
      "step": 12165
    },
    {
      "epoch": 0.8777461130550845,
      "grad_norm": 4.171568441378902,
      "learning_rate": 1.5466461197964486e-07,
      "loss": 0.9629,
      "step": 12166
    },
    {
      "epoch": 0.8778182605245121,
      "grad_norm": 6.565289004633038,
      "learning_rate": 1.544844502598004e-07,
      "loss": 1.007,
      "step": 12167
    },
    {
      "epoch": 0.8778904079939396,
      "grad_norm": 3.757104374710756,
      "learning_rate": 1.5430438931672906e-07,
      "loss": 0.99,
      "step": 12168
    },
    {
      "epoch": 0.8779625554633671,
      "grad_norm": 5.0948900690470085,
      "learning_rate": 1.5412442916026192e-07,
      "loss": 0.9418,
      "step": 12169
    },
    {
      "epoch": 0.8780347029327946,
      "grad_norm": 3.701307792482925,
      "learning_rate": 1.5394456980022664e-07,
      "loss": 1.0167,
      "step": 12170
    },
    {
      "epoch": 0.8781068504022221,
      "grad_norm": 11.993627763674915,
      "learning_rate": 1.537648112464449e-07,
      "loss": 0.9045,
      "step": 12171
    },
    {
      "epoch": 0.8781789978716497,
      "grad_norm": 3.190177391405571,
      "learning_rate": 1.535851535087318e-07,
      "loss": 0.9406,
      "step": 12172
    },
    {
      "epoch": 0.8782511453410772,
      "grad_norm": 3.190548355422399,
      "learning_rate": 1.5340559659689812e-07,
      "loss": 0.868,
      "step": 12173
    },
    {
      "epoch": 0.8783232928105047,
      "grad_norm": 3.0588743585359293,
      "learning_rate": 1.532261405207491e-07,
      "loss": 0.9144,
      "step": 12174
    },
    {
      "epoch": 0.8783954402799322,
      "grad_norm": 4.156545069266277,
      "learning_rate": 1.5304678529008453e-07,
      "loss": 1.0308,
      "step": 12175
    },
    {
      "epoch": 0.8784675877493597,
      "grad_norm": 3.7345118796828305,
      "learning_rate": 1.5286753091469739e-07,
      "loss": 0.8756,
      "step": 12176
    },
    {
      "epoch": 0.8785397352187873,
      "grad_norm": 4.972696716191888,
      "learning_rate": 1.5268837740437656e-07,
      "loss": 0.9182,
      "step": 12177
    },
    {
      "epoch": 0.8786118826882147,
      "grad_norm": 5.234437492339177,
      "learning_rate": 1.5250932476890532e-07,
      "loss": 0.9803,
      "step": 12178
    },
    {
      "epoch": 0.8786840301576422,
      "grad_norm": 2.7261853326109216,
      "learning_rate": 1.523303730180603e-07,
      "loss": 0.9337,
      "step": 12179
    },
    {
      "epoch": 0.8787561776270697,
      "grad_norm": 0.7088993326261211,
      "learning_rate": 1.5215152216161342e-07,
      "loss": 0.8212,
      "step": 12180
    },
    {
      "epoch": 0.8788283250964972,
      "grad_norm": 4.550518450700537,
      "learning_rate": 1.5197277220933226e-07,
      "loss": 0.9199,
      "step": 12181
    },
    {
      "epoch": 0.8789004725659247,
      "grad_norm": 2.9868178031634995,
      "learning_rate": 1.5179412317097629e-07,
      "loss": 0.9123,
      "step": 12182
    },
    {
      "epoch": 0.8789726200353523,
      "grad_norm": 5.218337539546071,
      "learning_rate": 1.516155750563015e-07,
      "loss": 0.9964,
      "step": 12183
    },
    {
      "epoch": 0.8790447675047798,
      "grad_norm": 4.531546536969871,
      "learning_rate": 1.5143712787505769e-07,
      "loss": 0.8703,
      "step": 12184
    },
    {
      "epoch": 0.8791169149742073,
      "grad_norm": 4.580730438955136,
      "learning_rate": 1.5125878163698947e-07,
      "loss": 0.9342,
      "step": 12185
    },
    {
      "epoch": 0.8791890624436348,
      "grad_norm": 4.713689857246921,
      "learning_rate": 1.510805363518346e-07,
      "loss": 0.8485,
      "step": 12186
    },
    {
      "epoch": 0.8792612099130623,
      "grad_norm": 3.595406589992748,
      "learning_rate": 1.5090239202932775e-07,
      "loss": 0.9422,
      "step": 12187
    },
    {
      "epoch": 0.8793333573824899,
      "grad_norm": 3.190161995915045,
      "learning_rate": 1.5072434867919627e-07,
      "loss": 0.9709,
      "step": 12188
    },
    {
      "epoch": 0.8794055048519174,
      "grad_norm": 3.982263821042104,
      "learning_rate": 1.505464063111621e-07,
      "loss": 1.0105,
      "step": 12189
    },
    {
      "epoch": 0.8794776523213448,
      "grad_norm": 4.182516520319917,
      "learning_rate": 1.503685649349422e-07,
      "loss": 0.9821,
      "step": 12190
    },
    {
      "epoch": 0.8795497997907723,
      "grad_norm": 6.165306903659265,
      "learning_rate": 1.5019082456024768e-07,
      "loss": 0.9641,
      "step": 12191
    },
    {
      "epoch": 0.8796219472601998,
      "grad_norm": 0.7403584076933236,
      "learning_rate": 1.5001318519678496e-07,
      "loss": 0.8769,
      "step": 12192
    },
    {
      "epoch": 0.8796940947296273,
      "grad_norm": 3.243094517330607,
      "learning_rate": 1.4983564685425343e-07,
      "loss": 0.9435,
      "step": 12193
    },
    {
      "epoch": 0.8797662421990549,
      "grad_norm": 4.1337594171583465,
      "learning_rate": 1.496582095423482e-07,
      "loss": 0.9977,
      "step": 12194
    },
    {
      "epoch": 0.8798383896684824,
      "grad_norm": 5.751822141441558,
      "learning_rate": 1.4948087327075888e-07,
      "loss": 0.9324,
      "step": 12195
    },
    {
      "epoch": 0.8799105371379099,
      "grad_norm": 4.538832640613879,
      "learning_rate": 1.4930363804916813e-07,
      "loss": 0.9192,
      "step": 12196
    },
    {
      "epoch": 0.8799826846073374,
      "grad_norm": 6.219497712390735,
      "learning_rate": 1.4912650388725445e-07,
      "loss": 0.9691,
      "step": 12197
    },
    {
      "epoch": 0.8800548320767649,
      "grad_norm": 4.84969472022668,
      "learning_rate": 1.4894947079469145e-07,
      "loss": 0.8277,
      "step": 12198
    },
    {
      "epoch": 0.8801269795461925,
      "grad_norm": 4.733019744980162,
      "learning_rate": 1.4877253878114516e-07,
      "loss": 0.9806,
      "step": 12199
    },
    {
      "epoch": 0.88019912701562,
      "grad_norm": 6.663035770939134,
      "learning_rate": 1.485957078562774e-07,
      "loss": 0.8856,
      "step": 12200
    },
    {
      "epoch": 0.8802712744850474,
      "grad_norm": 5.161601485786295,
      "learning_rate": 1.4841897802974422e-07,
      "loss": 0.9662,
      "step": 12201
    },
    {
      "epoch": 0.8803434219544749,
      "grad_norm": 2.7762034203865475,
      "learning_rate": 1.4824234931119684e-07,
      "loss": 0.9109,
      "step": 12202
    },
    {
      "epoch": 0.8804155694239024,
      "grad_norm": 4.446473625335662,
      "learning_rate": 1.480658217102795e-07,
      "loss": 1.0447,
      "step": 12203
    },
    {
      "epoch": 0.8804877168933299,
      "grad_norm": 2.8379823652798915,
      "learning_rate": 1.4788939523663158e-07,
      "loss": 0.93,
      "step": 12204
    },
    {
      "epoch": 0.8805598643627575,
      "grad_norm": 4.367473122088786,
      "learning_rate": 1.4771306989988829e-07,
      "loss": 0.8271,
      "step": 12205
    },
    {
      "epoch": 0.880632011832185,
      "grad_norm": 3.298731276794008,
      "learning_rate": 1.4753684570967705e-07,
      "loss": 0.9812,
      "step": 12206
    },
    {
      "epoch": 0.8807041593016125,
      "grad_norm": 3.383807240887318,
      "learning_rate": 1.4736072267562084e-07,
      "loss": 0.8988,
      "step": 12207
    },
    {
      "epoch": 0.88077630677104,
      "grad_norm": 7.531207452550977,
      "learning_rate": 1.4718470080733746e-07,
      "loss": 0.9609,
      "step": 12208
    },
    {
      "epoch": 0.8808484542404675,
      "grad_norm": 3.7385301018840105,
      "learning_rate": 1.4700878011443906e-07,
      "loss": 0.8405,
      "step": 12209
    },
    {
      "epoch": 0.8809206017098951,
      "grad_norm": 4.694465243401935,
      "learning_rate": 1.4683296060653082e-07,
      "loss": 0.9226,
      "step": 12210
    },
    {
      "epoch": 0.8809927491793226,
      "grad_norm": 4.7012213722986305,
      "learning_rate": 1.4665724229321485e-07,
      "loss": 0.9842,
      "step": 12211
    },
    {
      "epoch": 0.8810648966487501,
      "grad_norm": 3.3347287277319926,
      "learning_rate": 1.4648162518408635e-07,
      "loss": 0.997,
      "step": 12212
    },
    {
      "epoch": 0.8811370441181775,
      "grad_norm": 2.7151885108767617,
      "learning_rate": 1.4630610928873454e-07,
      "loss": 0.9192,
      "step": 12213
    },
    {
      "epoch": 0.881209191587605,
      "grad_norm": 4.181792283637629,
      "learning_rate": 1.46130694616744e-07,
      "loss": 1.0468,
      "step": 12214
    },
    {
      "epoch": 0.8812813390570325,
      "grad_norm": 4.520301799319775,
      "learning_rate": 1.4595538117769369e-07,
      "loss": 0.9102,
      "step": 12215
    },
    {
      "epoch": 0.8813534865264601,
      "grad_norm": 3.0489857722695186,
      "learning_rate": 1.4578016898115574e-07,
      "loss": 1.025,
      "step": 12216
    },
    {
      "epoch": 0.8814256339958876,
      "grad_norm": 3.279826845622809,
      "learning_rate": 1.456050580366992e-07,
      "loss": 0.8727,
      "step": 12217
    },
    {
      "epoch": 0.8814977814653151,
      "grad_norm": 4.274132820204128,
      "learning_rate": 1.4543004835388572e-07,
      "loss": 0.9466,
      "step": 12218
    },
    {
      "epoch": 0.8815699289347426,
      "grad_norm": 3.999568439091224,
      "learning_rate": 1.4525513994227213e-07,
      "loss": 0.946,
      "step": 12219
    },
    {
      "epoch": 0.8816420764041701,
      "grad_norm": 3.340608636371727,
      "learning_rate": 1.450803328114092e-07,
      "loss": 1.0252,
      "step": 12220
    },
    {
      "epoch": 0.8817142238735977,
      "grad_norm": 4.717779602918503,
      "learning_rate": 1.4490562697084196e-07,
      "loss": 0.8729,
      "step": 12221
    },
    {
      "epoch": 0.8817863713430252,
      "grad_norm": 3.407797505635642,
      "learning_rate": 1.4473102243011192e-07,
      "loss": 0.9868,
      "step": 12222
    },
    {
      "epoch": 0.8818585188124527,
      "grad_norm": 3.430079270495054,
      "learning_rate": 1.4455651919875257e-07,
      "loss": 0.9419,
      "step": 12223
    },
    {
      "epoch": 0.8819306662818802,
      "grad_norm": 2.9711887983374954,
      "learning_rate": 1.443821172862929e-07,
      "loss": 0.9763,
      "step": 12224
    },
    {
      "epoch": 0.8820028137513076,
      "grad_norm": 5.198970384634128,
      "learning_rate": 1.4420781670225668e-07,
      "loss": 0.8907,
      "step": 12225
    },
    {
      "epoch": 0.8820749612207351,
      "grad_norm": 4.200648829888189,
      "learning_rate": 1.4403361745616204e-07,
      "loss": 0.9492,
      "step": 12226
    },
    {
      "epoch": 0.8821471086901627,
      "grad_norm": 3.6071660932685465,
      "learning_rate": 1.4385951955752052e-07,
      "loss": 0.9819,
      "step": 12227
    },
    {
      "epoch": 0.8822192561595902,
      "grad_norm": 3.3818888118971557,
      "learning_rate": 1.4368552301583935e-07,
      "loss": 0.9307,
      "step": 12228
    },
    {
      "epoch": 0.8822914036290177,
      "grad_norm": 3.7823881848052294,
      "learning_rate": 1.4351162784062032e-07,
      "loss": 0.8732,
      "step": 12229
    },
    {
      "epoch": 0.8823635510984452,
      "grad_norm": 4.059378260845375,
      "learning_rate": 1.4333783404135868e-07,
      "loss": 0.9594,
      "step": 12230
    },
    {
      "epoch": 0.8824356985678727,
      "grad_norm": 3.8140820832083797,
      "learning_rate": 1.4316414162754464e-07,
      "loss": 0.9814,
      "step": 12231
    },
    {
      "epoch": 0.8825078460373003,
      "grad_norm": 3.946510299299936,
      "learning_rate": 1.429905506086635e-07,
      "loss": 0.873,
      "step": 12232
    },
    {
      "epoch": 0.8825799935067278,
      "grad_norm": 7.197145044359658,
      "learning_rate": 1.4281706099419344e-07,
      "loss": 1.0093,
      "step": 12233
    },
    {
      "epoch": 0.8826521409761553,
      "grad_norm": 4.550839508076668,
      "learning_rate": 1.4264367279360845e-07,
      "loss": 0.8271,
      "step": 12234
    },
    {
      "epoch": 0.8827242884455828,
      "grad_norm": 4.827826925909636,
      "learning_rate": 1.4247038601637717e-07,
      "loss": 0.8414,
      "step": 12235
    },
    {
      "epoch": 0.8827964359150103,
      "grad_norm": 2.7364437451873465,
      "learning_rate": 1.4229720067196204e-07,
      "loss": 0.9487,
      "step": 12236
    },
    {
      "epoch": 0.8828685833844377,
      "grad_norm": 3.5196888133692363,
      "learning_rate": 1.421241167698195e-07,
      "loss": 0.9232,
      "step": 12237
    },
    {
      "epoch": 0.8829407308538653,
      "grad_norm": 5.12900442594537,
      "learning_rate": 1.419511343194013e-07,
      "loss": 1.0152,
      "step": 12238
    },
    {
      "epoch": 0.8830128783232928,
      "grad_norm": 4.372875242626211,
      "learning_rate": 1.417782533301537e-07,
      "loss": 0.9158,
      "step": 12239
    },
    {
      "epoch": 0.8830850257927203,
      "grad_norm": 3.996038143781615,
      "learning_rate": 1.4160547381151622e-07,
      "loss": 0.9592,
      "step": 12240
    },
    {
      "epoch": 0.8831571732621478,
      "grad_norm": 3.8543784968269215,
      "learning_rate": 1.4143279577292444e-07,
      "loss": 0.8655,
      "step": 12241
    },
    {
      "epoch": 0.8832293207315753,
      "grad_norm": 3.952658644598313,
      "learning_rate": 1.4126021922380793e-07,
      "loss": 0.9188,
      "step": 12242
    },
    {
      "epoch": 0.8833014682010029,
      "grad_norm": 11.40615635663615,
      "learning_rate": 1.4108774417358982e-07,
      "loss": 0.9567,
      "step": 12243
    },
    {
      "epoch": 0.8833736156704304,
      "grad_norm": 3.811632401601892,
      "learning_rate": 1.4091537063168835e-07,
      "loss": 1.0614,
      "step": 12244
    },
    {
      "epoch": 0.8834457631398579,
      "grad_norm": 3.05263253089073,
      "learning_rate": 1.407430986075162e-07,
      "loss": 0.9439,
      "step": 12245
    },
    {
      "epoch": 0.8835179106092854,
      "grad_norm": 2.71069932930265,
      "learning_rate": 1.405709281104812e-07,
      "loss": 0.885,
      "step": 12246
    },
    {
      "epoch": 0.8835900580787129,
      "grad_norm": 4.59542504736687,
      "learning_rate": 1.4039885914998429e-07,
      "loss": 1.0482,
      "step": 12247
    },
    {
      "epoch": 0.8836622055481403,
      "grad_norm": 3.7283425867963156,
      "learning_rate": 1.4022689173542146e-07,
      "loss": 1.0549,
      "step": 12248
    },
    {
      "epoch": 0.8837343530175679,
      "grad_norm": 4.1828060886817,
      "learning_rate": 1.400550258761839e-07,
      "loss": 0.9388,
      "step": 12249
    },
    {
      "epoch": 0.8838065004869954,
      "grad_norm": 4.093102090829418,
      "learning_rate": 1.3988326158165586e-07,
      "loss": 1.0368,
      "step": 12250
    },
    {
      "epoch": 0.8838786479564229,
      "grad_norm": 3.1118500793157913,
      "learning_rate": 1.3971159886121675e-07,
      "loss": 0.9639,
      "step": 12251
    },
    {
      "epoch": 0.8839507954258504,
      "grad_norm": 4.9213987089734506,
      "learning_rate": 1.395400377242406e-07,
      "loss": 0.8953,
      "step": 12252
    },
    {
      "epoch": 0.8840229428952779,
      "grad_norm": 3.797906931608149,
      "learning_rate": 1.393685781800964e-07,
      "loss": 1.0244,
      "step": 12253
    },
    {
      "epoch": 0.8840950903647055,
      "grad_norm": 3.3940097772640643,
      "learning_rate": 1.3919722023814572e-07,
      "loss": 0.9379,
      "step": 12254
    },
    {
      "epoch": 0.884167237834133,
      "grad_norm": 3.55372717325056,
      "learning_rate": 1.3902596390774667e-07,
      "loss": 0.9874,
      "step": 12255
    },
    {
      "epoch": 0.8842393853035605,
      "grad_norm": 4.22299816854023,
      "learning_rate": 1.3885480919825087e-07,
      "loss": 1.0241,
      "step": 12256
    },
    {
      "epoch": 0.884311532772988,
      "grad_norm": 2.377236367419536,
      "learning_rate": 1.3868375611900396e-07,
      "loss": 1.0161,
      "step": 12257
    },
    {
      "epoch": 0.8843836802424155,
      "grad_norm": 3.931032470118751,
      "learning_rate": 1.3851280467934644e-07,
      "loss": 1.0276,
      "step": 12258
    },
    {
      "epoch": 0.8844558277118431,
      "grad_norm": 3.0533720730596405,
      "learning_rate": 1.3834195488861422e-07,
      "loss": 0.9664,
      "step": 12259
    },
    {
      "epoch": 0.8845279751812705,
      "grad_norm": 4.710408296483446,
      "learning_rate": 1.3817120675613602e-07,
      "loss": 0.8901,
      "step": 12260
    },
    {
      "epoch": 0.884600122650698,
      "grad_norm": 4.435839275093597,
      "learning_rate": 1.3800056029123596e-07,
      "loss": 1.005,
      "step": 12261
    },
    {
      "epoch": 0.8846722701201255,
      "grad_norm": 4.182677267437108,
      "learning_rate": 1.3783001550323237e-07,
      "loss": 0.9337,
      "step": 12262
    },
    {
      "epoch": 0.884744417589553,
      "grad_norm": 3.981655853520802,
      "learning_rate": 1.3765957240143844e-07,
      "loss": 1.0374,
      "step": 12263
    },
    {
      "epoch": 0.8848165650589805,
      "grad_norm": 2.841854763424571,
      "learning_rate": 1.374892309951603e-07,
      "loss": 0.8951,
      "step": 12264
    },
    {
      "epoch": 0.8848887125284081,
      "grad_norm": 2.3829741876495234,
      "learning_rate": 1.3731899129370095e-07,
      "loss": 0.9254,
      "step": 12265
    },
    {
      "epoch": 0.8849608599978356,
      "grad_norm": 4.212321717425989,
      "learning_rate": 1.371488533063565e-07,
      "loss": 0.9496,
      "step": 12266
    },
    {
      "epoch": 0.8850330074672631,
      "grad_norm": 6.5993423249831835,
      "learning_rate": 1.3697881704241665e-07,
      "loss": 0.9259,
      "step": 12267
    },
    {
      "epoch": 0.8851051549366906,
      "grad_norm": 3.5889316267599214,
      "learning_rate": 1.3680888251116706e-07,
      "loss": 0.944,
      "step": 12268
    },
    {
      "epoch": 0.8851773024061181,
      "grad_norm": 4.395005616758401,
      "learning_rate": 1.3663904972188677e-07,
      "loss": 1.0649,
      "step": 12269
    },
    {
      "epoch": 0.8852494498755457,
      "grad_norm": 3.839543343451494,
      "learning_rate": 1.3646931868385059e-07,
      "loss": 0.991,
      "step": 12270
    },
    {
      "epoch": 0.8853215973449732,
      "grad_norm": 4.359566633052802,
      "learning_rate": 1.3629968940632597e-07,
      "loss": 0.9107,
      "step": 12271
    },
    {
      "epoch": 0.8853937448144006,
      "grad_norm": 3.0561733680900294,
      "learning_rate": 1.361301618985764e-07,
      "loss": 0.9102,
      "step": 12272
    },
    {
      "epoch": 0.8854658922838281,
      "grad_norm": 3.2087631102867364,
      "learning_rate": 1.3596073616985894e-07,
      "loss": 0.9539,
      "step": 12273
    },
    {
      "epoch": 0.8855380397532556,
      "grad_norm": 0.7916979072496885,
      "learning_rate": 1.3579141222942503e-07,
      "loss": 0.8765,
      "step": 12274
    },
    {
      "epoch": 0.8856101872226831,
      "grad_norm": 2.423196837959481,
      "learning_rate": 1.356221900865211e-07,
      "loss": 0.9212,
      "step": 12275
    },
    {
      "epoch": 0.8856823346921107,
      "grad_norm": 6.353066804678432,
      "learning_rate": 1.3545306975038794e-07,
      "loss": 0.881,
      "step": 12276
    },
    {
      "epoch": 0.8857544821615382,
      "grad_norm": 3.6368758252699003,
      "learning_rate": 1.3528405123025954e-07,
      "loss": 0.8655,
      "step": 12277
    },
    {
      "epoch": 0.8858266296309657,
      "grad_norm": 5.4913386724039395,
      "learning_rate": 1.351151345353667e-07,
      "loss": 0.9991,
      "step": 12278
    },
    {
      "epoch": 0.8858987771003932,
      "grad_norm": 6.002825389655392,
      "learning_rate": 1.3494631967493276e-07,
      "loss": 0.9013,
      "step": 12279
    },
    {
      "epoch": 0.8859709245698207,
      "grad_norm": 6.444792829428873,
      "learning_rate": 1.3477760665817628e-07,
      "loss": 0.9497,
      "step": 12280
    },
    {
      "epoch": 0.8860430720392483,
      "grad_norm": 3.1686816748000073,
      "learning_rate": 1.346089954943097e-07,
      "loss": 0.954,
      "step": 12281
    },
    {
      "epoch": 0.8861152195086758,
      "grad_norm": 27.189152248645048,
      "learning_rate": 1.344404861925401e-07,
      "loss": 0.8668,
      "step": 12282
    },
    {
      "epoch": 0.8861873669781033,
      "grad_norm": 5.8031077641984545,
      "learning_rate": 1.3427207876207013e-07,
      "loss": 0.9163,
      "step": 12283
    },
    {
      "epoch": 0.8862595144475307,
      "grad_norm": 5.257760987164073,
      "learning_rate": 1.3410377321209487e-07,
      "loss": 0.9537,
      "step": 12284
    },
    {
      "epoch": 0.8863316619169582,
      "grad_norm": 5.9334208078401245,
      "learning_rate": 1.3393556955180519e-07,
      "loss": 1.077,
      "step": 12285
    },
    {
      "epoch": 0.8864038093863857,
      "grad_norm": 2.9771165707554514,
      "learning_rate": 1.3376746779038662e-07,
      "loss": 0.9643,
      "step": 12286
    },
    {
      "epoch": 0.8864759568558133,
      "grad_norm": 0.8489663683427197,
      "learning_rate": 1.335994679370176e-07,
      "loss": 0.9843,
      "step": 12287
    },
    {
      "epoch": 0.8865481043252408,
      "grad_norm": 3.917334371155908,
      "learning_rate": 1.3343157000087234e-07,
      "loss": 0.9419,
      "step": 12288
    },
    {
      "epoch": 0.8866202517946683,
      "grad_norm": 4.2508741489572985,
      "learning_rate": 1.3326377399111933e-07,
      "loss": 0.9483,
      "step": 12289
    },
    {
      "epoch": 0.8866923992640958,
      "grad_norm": 5.487497858902985,
      "learning_rate": 1.3309607991692162e-07,
      "loss": 0.9214,
      "step": 12290
    },
    {
      "epoch": 0.8867645467335233,
      "grad_norm": 5.479235205082657,
      "learning_rate": 1.329284877874357e-07,
      "loss": 1.0682,
      "step": 12291
    },
    {
      "epoch": 0.8868366942029509,
      "grad_norm": 3.448257951854364,
      "learning_rate": 1.3276099761181314e-07,
      "loss": 0.9116,
      "step": 12292
    },
    {
      "epoch": 0.8869088416723784,
      "grad_norm": 8.440481153757386,
      "learning_rate": 1.3259360939920084e-07,
      "loss": 1.018,
      "step": 12293
    },
    {
      "epoch": 0.8869809891418059,
      "grad_norm": 0.6778335059056071,
      "learning_rate": 1.3242632315873792e-07,
      "loss": 0.8631,
      "step": 12294
    },
    {
      "epoch": 0.8870531366112333,
      "grad_norm": 0.8287356662229896,
      "learning_rate": 1.322591388995602e-07,
      "loss": 0.8246,
      "step": 12295
    },
    {
      "epoch": 0.8871252840806608,
      "grad_norm": 3.0575270467390028,
      "learning_rate": 1.320920566307968e-07,
      "loss": 0.9621,
      "step": 12296
    },
    {
      "epoch": 0.8871974315500883,
      "grad_norm": 5.419798351632954,
      "learning_rate": 1.3192507636157179e-07,
      "loss": 0.9671,
      "step": 12297
    },
    {
      "epoch": 0.8872695790195159,
      "grad_norm": 6.304721685645392,
      "learning_rate": 1.3175819810100252e-07,
      "loss": 0.9148,
      "step": 12298
    },
    {
      "epoch": 0.8873417264889434,
      "grad_norm": 4.365380310759689,
      "learning_rate": 1.3159142185820216e-07,
      "loss": 1.0129,
      "step": 12299
    },
    {
      "epoch": 0.8874138739583709,
      "grad_norm": 5.2114468029666545,
      "learning_rate": 1.3142474764227807e-07,
      "loss": 0.892,
      "step": 12300
    },
    {
      "epoch": 0.8874860214277984,
      "grad_norm": 6.142460750070945,
      "learning_rate": 1.3125817546233054e-07,
      "loss": 1.0305,
      "step": 12301
    },
    {
      "epoch": 0.8875581688972259,
      "grad_norm": 2.3774137779349416,
      "learning_rate": 1.3109170532745651e-07,
      "loss": 0.9354,
      "step": 12302
    },
    {
      "epoch": 0.8876303163666535,
      "grad_norm": 4.753039341342198,
      "learning_rate": 1.3092533724674625e-07,
      "loss": 0.9006,
      "step": 12303
    },
    {
      "epoch": 0.887702463836081,
      "grad_norm": 7.348152042516635,
      "learning_rate": 1.307590712292841e-07,
      "loss": 0.8748,
      "step": 12304
    },
    {
      "epoch": 0.8877746113055085,
      "grad_norm": 3.1834749726212945,
      "learning_rate": 1.3059290728414918e-07,
      "loss": 0.916,
      "step": 12305
    },
    {
      "epoch": 0.887846758774936,
      "grad_norm": 7.357821095984706,
      "learning_rate": 1.3042684542041514e-07,
      "loss": 0.9643,
      "step": 12306
    },
    {
      "epoch": 0.8879189062443634,
      "grad_norm": 3.579460594415401,
      "learning_rate": 1.3026088564715055e-07,
      "loss": 0.8931,
      "step": 12307
    },
    {
      "epoch": 0.887991053713791,
      "grad_norm": 6.016743503971075,
      "learning_rate": 1.3009502797341743e-07,
      "loss": 0.9841,
      "step": 12308
    },
    {
      "epoch": 0.8880632011832185,
      "grad_norm": 3.731410042606802,
      "learning_rate": 1.2992927240827256e-07,
      "loss": 0.9617,
      "step": 12309
    },
    {
      "epoch": 0.888135348652646,
      "grad_norm": 3.129569717910546,
      "learning_rate": 1.297636189607676e-07,
      "loss": 0.9352,
      "step": 12310
    },
    {
      "epoch": 0.8882074961220735,
      "grad_norm": 3.120086622013823,
      "learning_rate": 1.2959806763994774e-07,
      "loss": 0.9217,
      "step": 12311
    },
    {
      "epoch": 0.888279643591501,
      "grad_norm": 4.851886830569154,
      "learning_rate": 1.294326184548531e-07,
      "loss": 0.9798,
      "step": 12312
    },
    {
      "epoch": 0.8883517910609285,
      "grad_norm": 2.7963545330973627,
      "learning_rate": 1.2926727141451889e-07,
      "loss": 0.9681,
      "step": 12313
    },
    {
      "epoch": 0.8884239385303561,
      "grad_norm": 3.3587381868257613,
      "learning_rate": 1.291020265279741e-07,
      "loss": 0.962,
      "step": 12314
    },
    {
      "epoch": 0.8884960859997836,
      "grad_norm": 4.074506652549474,
      "learning_rate": 1.2893688380424152e-07,
      "loss": 1.0821,
      "step": 12315
    },
    {
      "epoch": 0.8885682334692111,
      "grad_norm": 5.365910496167529,
      "learning_rate": 1.2877184325233926e-07,
      "loss": 0.9369,
      "step": 12316
    },
    {
      "epoch": 0.8886403809386386,
      "grad_norm": 3.1806352782125544,
      "learning_rate": 1.286069048812799e-07,
      "loss": 1.0857,
      "step": 12317
    },
    {
      "epoch": 0.8887125284080661,
      "grad_norm": 5.771537912188238,
      "learning_rate": 1.2844206870006936e-07,
      "loss": 0.9165,
      "step": 12318
    },
    {
      "epoch": 0.8887846758774935,
      "grad_norm": 3.74680942862994,
      "learning_rate": 1.282773347177093e-07,
      "loss": 0.857,
      "step": 12319
    },
    {
      "epoch": 0.8888568233469211,
      "grad_norm": 2.8955624863092804,
      "learning_rate": 1.2811270294319541e-07,
      "loss": 0.9999,
      "step": 12320
    },
    {
      "epoch": 0.8889289708163486,
      "grad_norm": 4.888185520341307,
      "learning_rate": 1.27948173385517e-07,
      "loss": 0.9802,
      "step": 12321
    },
    {
      "epoch": 0.8890011182857761,
      "grad_norm": 6.097943706689262,
      "learning_rate": 1.2778374605365883e-07,
      "loss": 1.0376,
      "step": 12322
    },
    {
      "epoch": 0.8890732657552036,
      "grad_norm": 4.957039907461412,
      "learning_rate": 1.276194209565995e-07,
      "loss": 0.9941,
      "step": 12323
    },
    {
      "epoch": 0.8891454132246311,
      "grad_norm": 3.6092970538289473,
      "learning_rate": 1.274551981033125e-07,
      "loss": 0.9447,
      "step": 12324
    },
    {
      "epoch": 0.8892175606940587,
      "grad_norm": 6.042667946113077,
      "learning_rate": 1.2729107750276492e-07,
      "loss": 0.8716,
      "step": 12325
    },
    {
      "epoch": 0.8892897081634862,
      "grad_norm": 2.3050203761857175,
      "learning_rate": 1.2712705916391909e-07,
      "loss": 1.0104,
      "step": 12326
    },
    {
      "epoch": 0.8893618556329137,
      "grad_norm": 3.4717884674413404,
      "learning_rate": 1.2696314309573186e-07,
      "loss": 1.0043,
      "step": 12327
    },
    {
      "epoch": 0.8894340031023412,
      "grad_norm": 3.9094393718002873,
      "learning_rate": 1.267993293071532e-07,
      "loss": 0.9697,
      "step": 12328
    },
    {
      "epoch": 0.8895061505717687,
      "grad_norm": 4.246928732155257,
      "learning_rate": 1.2663561780712884e-07,
      "loss": 0.9573,
      "step": 12329
    },
    {
      "epoch": 0.8895782980411963,
      "grad_norm": 6.732823876977507,
      "learning_rate": 1.2647200860459893e-07,
      "loss": 0.9339,
      "step": 12330
    },
    {
      "epoch": 0.8896504455106237,
      "grad_norm": 3.8603469921946765,
      "learning_rate": 1.2630850170849617e-07,
      "loss": 0.9082,
      "step": 12331
    },
    {
      "epoch": 0.8897225929800512,
      "grad_norm": 2.791833179878852,
      "learning_rate": 1.2614509712775047e-07,
      "loss": 0.9482,
      "step": 12332
    },
    {
      "epoch": 0.8897947404494787,
      "grad_norm": 2.67626521056769,
      "learning_rate": 1.2598179487128423e-07,
      "loss": 0.9331,
      "step": 12333
    },
    {
      "epoch": 0.8898668879189062,
      "grad_norm": 3.1860496362817505,
      "learning_rate": 1.2581859494801506e-07,
      "loss": 0.931,
      "step": 12334
    },
    {
      "epoch": 0.8899390353883337,
      "grad_norm": 5.342152139123245,
      "learning_rate": 1.256554973668542e-07,
      "loss": 0.8461,
      "step": 12335
    },
    {
      "epoch": 0.8900111828577613,
      "grad_norm": 3.5701308590282705,
      "learning_rate": 1.2549250213670748e-07,
      "loss": 0.9415,
      "step": 12336
    },
    {
      "epoch": 0.8900833303271888,
      "grad_norm": 3.7618197767213326,
      "learning_rate": 1.2532960926647684e-07,
      "loss": 1.0312,
      "step": 12337
    },
    {
      "epoch": 0.8901554777966163,
      "grad_norm": 2.6194083285367453,
      "learning_rate": 1.251668187650563e-07,
      "loss": 0.9802,
      "step": 12338
    },
    {
      "epoch": 0.8902276252660438,
      "grad_norm": 4.123554323195907,
      "learning_rate": 1.250041306413352e-07,
      "loss": 0.9805,
      "step": 12339
    },
    {
      "epoch": 0.8902997727354713,
      "grad_norm": 3.773010988002603,
      "learning_rate": 1.2484154490419752e-07,
      "loss": 1.0486,
      "step": 12340
    },
    {
      "epoch": 0.8903719202048989,
      "grad_norm": 18.391725155602415,
      "learning_rate": 1.2467906156252174e-07,
      "loss": 0.9489,
      "step": 12341
    },
    {
      "epoch": 0.8904440676743264,
      "grad_norm": 4.562981567051307,
      "learning_rate": 1.2451668062517962e-07,
      "loss": 0.921,
      "step": 12342
    },
    {
      "epoch": 0.8905162151437538,
      "grad_norm": 10.367800869077291,
      "learning_rate": 1.243544021010392e-07,
      "loss": 0.9406,
      "step": 12343
    },
    {
      "epoch": 0.8905883626131813,
      "grad_norm": 7.5463513178814186,
      "learning_rate": 1.2419222599896162e-07,
      "loss": 0.9825,
      "step": 12344
    },
    {
      "epoch": 0.8906605100826088,
      "grad_norm": 0.8091064154772643,
      "learning_rate": 1.2403015232780244e-07,
      "loss": 0.8573,
      "step": 12345
    },
    {
      "epoch": 0.8907326575520363,
      "grad_norm": 7.5673880425077105,
      "learning_rate": 1.2386818109641194e-07,
      "loss": 0.9595,
      "step": 12346
    },
    {
      "epoch": 0.8908048050214639,
      "grad_norm": 3.0300809828004764,
      "learning_rate": 1.2370631231363526e-07,
      "loss": 0.8857,
      "step": 12347
    },
    {
      "epoch": 0.8908769524908914,
      "grad_norm": 5.650662354661329,
      "learning_rate": 1.2354454598831066e-07,
      "loss": 0.9021,
      "step": 12348
    },
    {
      "epoch": 0.8909490999603189,
      "grad_norm": 2.6171784813568957,
      "learning_rate": 1.2338288212927172e-07,
      "loss": 0.9208,
      "step": 12349
    },
    {
      "epoch": 0.8910212474297464,
      "grad_norm": 8.78760057285975,
      "learning_rate": 1.2322132074534697e-07,
      "loss": 0.9702,
      "step": 12350
    },
    {
      "epoch": 0.8910933948991739,
      "grad_norm": 3.4591310492617646,
      "learning_rate": 1.2305986184535843e-07,
      "loss": 0.9354,
      "step": 12351
    },
    {
      "epoch": 0.8911655423686015,
      "grad_norm": 2.8130979114339345,
      "learning_rate": 1.2289850543812242e-07,
      "loss": 1.0264,
      "step": 12352
    },
    {
      "epoch": 0.891237689838029,
      "grad_norm": 4.1450172017181135,
      "learning_rate": 1.227372515324503e-07,
      "loss": 0.9609,
      "step": 12353
    },
    {
      "epoch": 0.8913098373074564,
      "grad_norm": 3.2792346396342764,
      "learning_rate": 1.225761001371477e-07,
      "loss": 0.9062,
      "step": 12354
    },
    {
      "epoch": 0.8913819847768839,
      "grad_norm": 3.814760257292413,
      "learning_rate": 1.224150512610138e-07,
      "loss": 0.9584,
      "step": 12355
    },
    {
      "epoch": 0.8914541322463114,
      "grad_norm": 4.456901581839684,
      "learning_rate": 1.222541049128436e-07,
      "loss": 0.9847,
      "step": 12356
    },
    {
      "epoch": 0.891526279715739,
      "grad_norm": 3.3420501381145358,
      "learning_rate": 1.2209326110142537e-07,
      "loss": 0.927,
      "step": 12357
    },
    {
      "epoch": 0.8915984271851665,
      "grad_norm": 2.963943767029626,
      "learning_rate": 1.2193251983554298e-07,
      "loss": 0.9212,
      "step": 12358
    },
    {
      "epoch": 0.891670574654594,
      "grad_norm": 5.288513245229769,
      "learning_rate": 1.2177188112397296e-07,
      "loss": 0.9805,
      "step": 12359
    },
    {
      "epoch": 0.8917427221240215,
      "grad_norm": 3.432507721849884,
      "learning_rate": 1.2161134497548697e-07,
      "loss": 0.982,
      "step": 12360
    },
    {
      "epoch": 0.891814869593449,
      "grad_norm": 0.6822788780896967,
      "learning_rate": 1.2145091139885287e-07,
      "loss": 0.8495,
      "step": 12361
    },
    {
      "epoch": 0.8918870170628765,
      "grad_norm": 5.17143445861409,
      "learning_rate": 1.212905804028297e-07,
      "loss": 0.9619,
      "step": 12362
    },
    {
      "epoch": 0.8919591645323041,
      "grad_norm": 3.261161564731459,
      "learning_rate": 1.211303519961735e-07,
      "loss": 0.86,
      "step": 12363
    },
    {
      "epoch": 0.8920313120017316,
      "grad_norm": 2.354893451232241,
      "learning_rate": 1.2097022618763352e-07,
      "loss": 0.9226,
      "step": 12364
    },
    {
      "epoch": 0.8921034594711591,
      "grad_norm": 6.888374220514988,
      "learning_rate": 1.2081020298595323e-07,
      "loss": 0.9429,
      "step": 12365
    },
    {
      "epoch": 0.8921756069405865,
      "grad_norm": 2.4375805963738633,
      "learning_rate": 1.2065028239987118e-07,
      "loss": 1.0702,
      "step": 12366
    },
    {
      "epoch": 0.892247754410014,
      "grad_norm": 0.9155682933550386,
      "learning_rate": 1.2049046443812017e-07,
      "loss": 0.9096,
      "step": 12367
    },
    {
      "epoch": 0.8923199018794415,
      "grad_norm": 4.122299697664312,
      "learning_rate": 1.2033074910942764e-07,
      "loss": 0.9793,
      "step": 12368
    },
    {
      "epoch": 0.8923920493488691,
      "grad_norm": 5.422903529515318,
      "learning_rate": 1.2017113642251418e-07,
      "loss": 0.9384,
      "step": 12369
    },
    {
      "epoch": 0.8924641968182966,
      "grad_norm": 4.410632179603609,
      "learning_rate": 1.200116263860962e-07,
      "loss": 0.9458,
      "step": 12370
    },
    {
      "epoch": 0.8925363442877241,
      "grad_norm": 3.1628720151399223,
      "learning_rate": 1.1985221900888398e-07,
      "loss": 0.9309,
      "step": 12371
    },
    {
      "epoch": 0.8926084917571516,
      "grad_norm": 13.825599500133292,
      "learning_rate": 1.1969291429958173e-07,
      "loss": 0.909,
      "step": 12372
    },
    {
      "epoch": 0.8926806392265791,
      "grad_norm": 3.1771938283914856,
      "learning_rate": 1.1953371226688847e-07,
      "loss": 0.941,
      "step": 12373
    },
    {
      "epoch": 0.8927527866960067,
      "grad_norm": 6.14794826863551,
      "learning_rate": 1.1937461291949857e-07,
      "loss": 0.9255,
      "step": 12374
    },
    {
      "epoch": 0.8928249341654342,
      "grad_norm": 19.02350968928808,
      "learning_rate": 1.1921561626609888e-07,
      "loss": 0.8781,
      "step": 12375
    },
    {
      "epoch": 0.8928970816348617,
      "grad_norm": 3.4201944483306925,
      "learning_rate": 1.1905672231537223e-07,
      "loss": 0.909,
      "step": 12376
    },
    {
      "epoch": 0.8929692291042892,
      "grad_norm": 4.042537765751164,
      "learning_rate": 1.1889793107599455e-07,
      "loss": 0.929,
      "step": 12377
    },
    {
      "epoch": 0.8930413765737166,
      "grad_norm": 6.142468513032342,
      "learning_rate": 1.187392425566378e-07,
      "loss": 1.0372,
      "step": 12378
    },
    {
      "epoch": 0.8931135240431441,
      "grad_norm": 3.0713628083852975,
      "learning_rate": 1.1858065676596618e-07,
      "loss": 1.0326,
      "step": 12379
    },
    {
      "epoch": 0.8931856715125717,
      "grad_norm": 5.970205718670186,
      "learning_rate": 1.1842217371264052e-07,
      "loss": 0.9904,
      "step": 12380
    },
    {
      "epoch": 0.8932578189819992,
      "grad_norm": 4.215226121586304,
      "learning_rate": 1.1826379340531478e-07,
      "loss": 0.8676,
      "step": 12381
    },
    {
      "epoch": 0.8933299664514267,
      "grad_norm": 4.367232702905038,
      "learning_rate": 1.1810551585263694e-07,
      "loss": 1.0464,
      "step": 12382
    },
    {
      "epoch": 0.8934021139208542,
      "grad_norm": 6.5399555795076765,
      "learning_rate": 1.1794734106325055e-07,
      "loss": 0.7562,
      "step": 12383
    },
    {
      "epoch": 0.8934742613902817,
      "grad_norm": 0.747902998272491,
      "learning_rate": 1.1778926904579222e-07,
      "loss": 0.8574,
      "step": 12384
    },
    {
      "epoch": 0.8935464088597093,
      "grad_norm": 6.078222906505805,
      "learning_rate": 1.1763129980889485e-07,
      "loss": 0.9442,
      "step": 12385
    },
    {
      "epoch": 0.8936185563291368,
      "grad_norm": 5.040988004210459,
      "learning_rate": 1.1747343336118376e-07,
      "loss": 0.835,
      "step": 12386
    },
    {
      "epoch": 0.8936907037985643,
      "grad_norm": 5.22984404818256,
      "learning_rate": 1.1731566971127959e-07,
      "loss": 0.9301,
      "step": 12387
    },
    {
      "epoch": 0.8937628512679918,
      "grad_norm": 3.275919626426099,
      "learning_rate": 1.1715800886779726e-07,
      "loss": 0.7866,
      "step": 12388
    },
    {
      "epoch": 0.8938349987374193,
      "grad_norm": 4.220767068797786,
      "learning_rate": 1.1700045083934606e-07,
      "loss": 0.957,
      "step": 12389
    },
    {
      "epoch": 0.8939071462068467,
      "grad_norm": 4.0239025254625425,
      "learning_rate": 1.1684299563452915e-07,
      "loss": 0.9885,
      "step": 12390
    },
    {
      "epoch": 0.8939792936762743,
      "grad_norm": 3.1763205357765916,
      "learning_rate": 1.1668564326194563e-07,
      "loss": 0.9502,
      "step": 12391
    },
    {
      "epoch": 0.8940514411457018,
      "grad_norm": 5.833107135565099,
      "learning_rate": 1.1652839373018707e-07,
      "loss": 1.0067,
      "step": 12392
    },
    {
      "epoch": 0.8941235886151293,
      "grad_norm": 2.7590925579743133,
      "learning_rate": 1.1637124704784063e-07,
      "loss": 1.0206,
      "step": 12393
    },
    {
      "epoch": 0.8941957360845568,
      "grad_norm": 4.072030493811324,
      "learning_rate": 1.1621420322348741e-07,
      "loss": 0.9072,
      "step": 12394
    },
    {
      "epoch": 0.8942678835539843,
      "grad_norm": 9.942731907561651,
      "learning_rate": 1.1605726226570323e-07,
      "loss": 0.9739,
      "step": 12395
    },
    {
      "epoch": 0.8943400310234119,
      "grad_norm": 3.6948738724961623,
      "learning_rate": 1.1590042418305745e-07,
      "loss": 1.0102,
      "step": 12396
    },
    {
      "epoch": 0.8944121784928394,
      "grad_norm": 44.43518420917286,
      "learning_rate": 1.157436889841148e-07,
      "loss": 0.9228,
      "step": 12397
    },
    {
      "epoch": 0.8944843259622669,
      "grad_norm": 4.311761765249058,
      "learning_rate": 1.155870566774344e-07,
      "loss": 0.9059,
      "step": 12398
    },
    {
      "epoch": 0.8945564734316944,
      "grad_norm": 4.67667145250057,
      "learning_rate": 1.1543052727156877e-07,
      "loss": 0.9706,
      "step": 12399
    },
    {
      "epoch": 0.8946286209011219,
      "grad_norm": 5.4236980088822,
      "learning_rate": 1.152741007750655e-07,
      "loss": 0.9601,
      "step": 12400
    },
    {
      "epoch": 0.8947007683705493,
      "grad_norm": 7.6956949361663956,
      "learning_rate": 1.1511777719646664e-07,
      "loss": 0.9734,
      "step": 12401
    },
    {
      "epoch": 0.8947729158399769,
      "grad_norm": 0.7963418486004022,
      "learning_rate": 1.1496155654430873e-07,
      "loss": 0.8402,
      "step": 12402
    },
    {
      "epoch": 0.8948450633094044,
      "grad_norm": 6.690008530775256,
      "learning_rate": 1.1480543882712135e-07,
      "loss": 0.8413,
      "step": 12403
    },
    {
      "epoch": 0.8949172107788319,
      "grad_norm": 3.8661797337478476,
      "learning_rate": 1.1464942405343059e-07,
      "loss": 0.9523,
      "step": 12404
    },
    {
      "epoch": 0.8949893582482594,
      "grad_norm": 3.224808259927008,
      "learning_rate": 1.1449351223175564e-07,
      "loss": 0.857,
      "step": 12405
    },
    {
      "epoch": 0.895061505717687,
      "grad_norm": 5.796719037448779,
      "learning_rate": 1.1433770337060988e-07,
      "loss": 1.005,
      "step": 12406
    },
    {
      "epoch": 0.8951336531871145,
      "grad_norm": 4.663938996742135,
      "learning_rate": 1.1418199747850166e-07,
      "loss": 1.0235,
      "step": 12407
    },
    {
      "epoch": 0.895205800656542,
      "grad_norm": 4.5568204043912,
      "learning_rate": 1.1402639456393371e-07,
      "loss": 1.0077,
      "step": 12408
    },
    {
      "epoch": 0.8952779481259695,
      "grad_norm": 3.872870690844259,
      "learning_rate": 1.1387089463540211e-07,
      "loss": 0.9405,
      "step": 12409
    },
    {
      "epoch": 0.895350095595397,
      "grad_norm": 3.5237905659527717,
      "learning_rate": 1.1371549770139921e-07,
      "loss": 1.0287,
      "step": 12410
    },
    {
      "epoch": 0.8954222430648245,
      "grad_norm": 4.341738667443271,
      "learning_rate": 1.1356020377040998e-07,
      "loss": 0.8775,
      "step": 12411
    },
    {
      "epoch": 0.8954943905342521,
      "grad_norm": 7.751309438033705,
      "learning_rate": 1.1340501285091497e-07,
      "loss": 0.8844,
      "step": 12412
    },
    {
      "epoch": 0.8955665380036795,
      "grad_norm": 15.461619450283681,
      "learning_rate": 1.1324992495138808e-07,
      "loss": 0.9388,
      "step": 12413
    },
    {
      "epoch": 0.895638685473107,
      "grad_norm": 3.994483004121415,
      "learning_rate": 1.1309494008029808e-07,
      "loss": 0.9042,
      "step": 12414
    },
    {
      "epoch": 0.8957108329425345,
      "grad_norm": 5.720047933737777,
      "learning_rate": 1.1294005824610887e-07,
      "loss": 0.947,
      "step": 12415
    },
    {
      "epoch": 0.895782980411962,
      "grad_norm": 3.3034918315238535,
      "learning_rate": 1.12785279457277e-07,
      "loss": 0.8796,
      "step": 12416
    },
    {
      "epoch": 0.8958551278813895,
      "grad_norm": 3.7910153734382965,
      "learning_rate": 1.1263060372225486e-07,
      "loss": 0.9668,
      "step": 12417
    },
    {
      "epoch": 0.8959272753508171,
      "grad_norm": 8.013010888914314,
      "learning_rate": 1.12476031049489e-07,
      "loss": 0.9845,
      "step": 12418
    },
    {
      "epoch": 0.8959994228202446,
      "grad_norm": 0.7212249190202494,
      "learning_rate": 1.1232156144741933e-07,
      "loss": 0.8877,
      "step": 12419
    },
    {
      "epoch": 0.8960715702896721,
      "grad_norm": 4.47575054161496,
      "learning_rate": 1.1216719492448112e-07,
      "loss": 0.8969,
      "step": 12420
    },
    {
      "epoch": 0.8961437177590996,
      "grad_norm": 7.471038340110064,
      "learning_rate": 1.1201293148910385e-07,
      "loss": 0.9396,
      "step": 12421
    },
    {
      "epoch": 0.8962158652285271,
      "grad_norm": 6.59821591394752,
      "learning_rate": 1.1185877114971166e-07,
      "loss": 0.9702,
      "step": 12422
    },
    {
      "epoch": 0.8962880126979547,
      "grad_norm": 8.360870284635679,
      "learning_rate": 1.1170471391472225e-07,
      "loss": 0.9867,
      "step": 12423
    },
    {
      "epoch": 0.8963601601673822,
      "grad_norm": 4.880888878899923,
      "learning_rate": 1.1155075979254802e-07,
      "loss": 0.9526,
      "step": 12424
    },
    {
      "epoch": 0.8964323076368096,
      "grad_norm": 4.82004164270164,
      "learning_rate": 1.1139690879159624e-07,
      "loss": 0.9952,
      "step": 12425
    },
    {
      "epoch": 0.8965044551062371,
      "grad_norm": 4.048295526580566,
      "learning_rate": 1.1124316092026776e-07,
      "loss": 0.9654,
      "step": 12426
    },
    {
      "epoch": 0.8965766025756646,
      "grad_norm": 2.563117720419893,
      "learning_rate": 1.1108951618695783e-07,
      "loss": 1.007,
      "step": 12427
    },
    {
      "epoch": 0.8966487500450921,
      "grad_norm": 11.491315796944978,
      "learning_rate": 1.1093597460005755e-07,
      "loss": 0.9639,
      "step": 12428
    },
    {
      "epoch": 0.8967208975145197,
      "grad_norm": 0.8379474967382422,
      "learning_rate": 1.1078253616795064e-07,
      "loss": 0.9291,
      "step": 12429
    },
    {
      "epoch": 0.8967930449839472,
      "grad_norm": 5.036376426475217,
      "learning_rate": 1.106292008990155e-07,
      "loss": 0.9868,
      "step": 12430
    },
    {
      "epoch": 0.8968651924533747,
      "grad_norm": 11.244430519701847,
      "learning_rate": 1.1047596880162569e-07,
      "loss": 0.9399,
      "step": 12431
    },
    {
      "epoch": 0.8969373399228022,
      "grad_norm": 10.866436974217834,
      "learning_rate": 1.1032283988414847e-07,
      "loss": 0.8838,
      "step": 12432
    },
    {
      "epoch": 0.8970094873922297,
      "grad_norm": 5.528466165300204,
      "learning_rate": 1.101698141549452e-07,
      "loss": 1.0006,
      "step": 12433
    },
    {
      "epoch": 0.8970816348616573,
      "grad_norm": 2.8674954864812316,
      "learning_rate": 1.100168916223727e-07,
      "loss": 0.8374,
      "step": 12434
    },
    {
      "epoch": 0.8971537823310848,
      "grad_norm": 5.7856276974934975,
      "learning_rate": 1.0986407229478167e-07,
      "loss": 1.046,
      "step": 12435
    },
    {
      "epoch": 0.8972259298005123,
      "grad_norm": 4.723937562043754,
      "learning_rate": 1.0971135618051608e-07,
      "loss": 0.8814,
      "step": 12436
    },
    {
      "epoch": 0.8972980772699397,
      "grad_norm": 0.7916084987619795,
      "learning_rate": 1.0955874328791592e-07,
      "loss": 0.9482,
      "step": 12437
    },
    {
      "epoch": 0.8973702247393672,
      "grad_norm": 6.5839067362301655,
      "learning_rate": 1.0940623362531431e-07,
      "loss": 1.016,
      "step": 12438
    },
    {
      "epoch": 0.8974423722087947,
      "grad_norm": 5.96102229266773,
      "learning_rate": 1.0925382720104015e-07,
      "loss": 0.9039,
      "step": 12439
    },
    {
      "epoch": 0.8975145196782223,
      "grad_norm": 3.9805188236699522,
      "learning_rate": 1.0910152402341477e-07,
      "loss": 0.8655,
      "step": 12440
    },
    {
      "epoch": 0.8975866671476498,
      "grad_norm": 6.026298110855323,
      "learning_rate": 1.089493241007553e-07,
      "loss": 1.0628,
      "step": 12441
    },
    {
      "epoch": 0.8976588146170773,
      "grad_norm": 4.385923862198249,
      "learning_rate": 1.0879722744137331e-07,
      "loss": 0.9228,
      "step": 12442
    },
    {
      "epoch": 0.8977309620865048,
      "grad_norm": 2.4047672607995314,
      "learning_rate": 1.0864523405357329e-07,
      "loss": 0.9735,
      "step": 12443
    },
    {
      "epoch": 0.8978031095559323,
      "grad_norm": 3.1675033133353208,
      "learning_rate": 1.0849334394565501e-07,
      "loss": 0.9349,
      "step": 12444
    },
    {
      "epoch": 0.8978752570253599,
      "grad_norm": 4.192887896687028,
      "learning_rate": 1.0834155712591364e-07,
      "loss": 1.0233,
      "step": 12445
    },
    {
      "epoch": 0.8979474044947874,
      "grad_norm": 2.9555616871890504,
      "learning_rate": 1.0818987360263743e-07,
      "loss": 0.9369,
      "step": 12446
    },
    {
      "epoch": 0.8980195519642149,
      "grad_norm": 5.836554845715152,
      "learning_rate": 1.0803829338410863e-07,
      "loss": 0.9782,
      "step": 12447
    },
    {
      "epoch": 0.8980916994336423,
      "grad_norm": 4.289544973005815,
      "learning_rate": 1.0788681647860488e-07,
      "loss": 0.8479,
      "step": 12448
    },
    {
      "epoch": 0.8981638469030698,
      "grad_norm": 4.255723577313723,
      "learning_rate": 1.0773544289439795e-07,
      "loss": 0.9265,
      "step": 12449
    },
    {
      "epoch": 0.8982359943724973,
      "grad_norm": 4.30954209765152,
      "learning_rate": 1.0758417263975327e-07,
      "loss": 1.0024,
      "step": 12450
    },
    {
      "epoch": 0.8983081418419249,
      "grad_norm": 3.345211529622227,
      "learning_rate": 1.0743300572293113e-07,
      "loss": 0.8852,
      "step": 12451
    },
    {
      "epoch": 0.8983802893113524,
      "grad_norm": 4.191213982003823,
      "learning_rate": 1.072819421521871e-07,
      "loss": 0.915,
      "step": 12452
    },
    {
      "epoch": 0.8984524367807799,
      "grad_norm": 5.33472291407401,
      "learning_rate": 1.0713098193576953e-07,
      "loss": 0.8756,
      "step": 12453
    },
    {
      "epoch": 0.8985245842502074,
      "grad_norm": 3.565426076543981,
      "learning_rate": 1.0698012508192156e-07,
      "loss": 1.0674,
      "step": 12454
    },
    {
      "epoch": 0.898596731719635,
      "grad_norm": 3.6001258351268164,
      "learning_rate": 1.0682937159888128e-07,
      "loss": 1.0643,
      "step": 12455
    },
    {
      "epoch": 0.8986688791890625,
      "grad_norm": 6.280212316629379,
      "learning_rate": 1.0667872149488122e-07,
      "loss": 0.9789,
      "step": 12456
    },
    {
      "epoch": 0.89874102665849,
      "grad_norm": 5.354775994011787,
      "learning_rate": 1.0652817477814657e-07,
      "loss": 0.992,
      "step": 12457
    },
    {
      "epoch": 0.8988131741279175,
      "grad_norm": 0.7930965978227925,
      "learning_rate": 1.0637773145689944e-07,
      "loss": 0.8469,
      "step": 12458
    },
    {
      "epoch": 0.898885321597345,
      "grad_norm": 5.025893303718873,
      "learning_rate": 1.0622739153935456e-07,
      "loss": 1.0266,
      "step": 12459
    },
    {
      "epoch": 0.8989574690667724,
      "grad_norm": 2.2052967378761985,
      "learning_rate": 1.0607715503372095e-07,
      "loss": 0.9795,
      "step": 12460
    },
    {
      "epoch": 0.8990296165362,
      "grad_norm": 4.481178766213821,
      "learning_rate": 1.0592702194820291e-07,
      "loss": 1.0225,
      "step": 12461
    },
    {
      "epoch": 0.8991017640056275,
      "grad_norm": 6.36232289965256,
      "learning_rate": 1.0577699229099901e-07,
      "loss": 0.9701,
      "step": 12462
    },
    {
      "epoch": 0.899173911475055,
      "grad_norm": 10.574107997780734,
      "learning_rate": 1.056270660703007e-07,
      "loss": 0.9769,
      "step": 12463
    },
    {
      "epoch": 0.8992460589444825,
      "grad_norm": 3.999198714107544,
      "learning_rate": 1.0547724329429586e-07,
      "loss": 0.9675,
      "step": 12464
    },
    {
      "epoch": 0.89931820641391,
      "grad_norm": 7.0546158526089044,
      "learning_rate": 1.0532752397116551e-07,
      "loss": 0.9224,
      "step": 12465
    },
    {
      "epoch": 0.8993903538833375,
      "grad_norm": 4.943604956427037,
      "learning_rate": 1.0517790810908555e-07,
      "loss": 1.1002,
      "step": 12466
    },
    {
      "epoch": 0.8994625013527651,
      "grad_norm": 0.6292242348040918,
      "learning_rate": 1.0502839571622524e-07,
      "loss": 0.8166,
      "step": 12467
    },
    {
      "epoch": 0.8995346488221926,
      "grad_norm": 4.795931347404138,
      "learning_rate": 1.0487898680074913e-07,
      "loss": 0.9633,
      "step": 12468
    },
    {
      "epoch": 0.8996067962916201,
      "grad_norm": 4.274156471597778,
      "learning_rate": 1.047296813708165e-07,
      "loss": 1.0425,
      "step": 12469
    },
    {
      "epoch": 0.8996789437610476,
      "grad_norm": 6.091254246532354,
      "learning_rate": 1.0458047943457971e-07,
      "loss": 0.9879,
      "step": 12470
    },
    {
      "epoch": 0.8997510912304751,
      "grad_norm": 4.258838662897644,
      "learning_rate": 1.0443138100018622e-07,
      "loss": 0.8831,
      "step": 12471
    },
    {
      "epoch": 0.8998232386999026,
      "grad_norm": 3.147583043318699,
      "learning_rate": 1.04282386075778e-07,
      "loss": 1.0654,
      "step": 12472
    },
    {
      "epoch": 0.8998953861693301,
      "grad_norm": 4.033797294686983,
      "learning_rate": 1.0413349466949139e-07,
      "loss": 0.9222,
      "step": 12473
    },
    {
      "epoch": 0.8999675336387576,
      "grad_norm": 5.439871785952994,
      "learning_rate": 1.0398470678945593e-07,
      "loss": 1.0577,
      "step": 12474
    },
    {
      "epoch": 0.9000396811081851,
      "grad_norm": 4.48772430284381,
      "learning_rate": 1.038360224437964e-07,
      "loss": 1.0348,
      "step": 12475
    },
    {
      "epoch": 0.9001118285776126,
      "grad_norm": 3.819383878902002,
      "learning_rate": 1.0368744164063303e-07,
      "loss": 0.9288,
      "step": 12476
    },
    {
      "epoch": 0.9001839760470401,
      "grad_norm": 4.254210462105018,
      "learning_rate": 1.0353896438807842e-07,
      "loss": 0.9954,
      "step": 12477
    },
    {
      "epoch": 0.9002561235164677,
      "grad_norm": 3.455431965876449,
      "learning_rate": 1.0339059069424027e-07,
      "loss": 0.9712,
      "step": 12478
    },
    {
      "epoch": 0.9003282709858952,
      "grad_norm": 4.43458805446104,
      "learning_rate": 1.0324232056722127e-07,
      "loss": 0.8924,
      "step": 12479
    },
    {
      "epoch": 0.9004004184553227,
      "grad_norm": 4.057874658084286,
      "learning_rate": 1.0309415401511757e-07,
      "loss": 0.9882,
      "step": 12480
    },
    {
      "epoch": 0.9004725659247502,
      "grad_norm": 4.011321496656199,
      "learning_rate": 1.0294609104601937e-07,
      "loss": 0.9847,
      "step": 12481
    },
    {
      "epoch": 0.9005447133941777,
      "grad_norm": 4.754548956400448,
      "learning_rate": 1.0279813166801288e-07,
      "loss": 0.962,
      "step": 12482
    },
    {
      "epoch": 0.9006168608636053,
      "grad_norm": 5.457865318831253,
      "learning_rate": 1.0265027588917763e-07,
      "loss": 0.9776,
      "step": 12483
    },
    {
      "epoch": 0.9006890083330327,
      "grad_norm": 3.177994710976657,
      "learning_rate": 1.0250252371758672e-07,
      "loss": 0.9768,
      "step": 12484
    },
    {
      "epoch": 0.9007611558024602,
      "grad_norm": 6.715533386997268,
      "learning_rate": 1.0235487516130859e-07,
      "loss": 0.9174,
      "step": 12485
    },
    {
      "epoch": 0.9008333032718877,
      "grad_norm": 6.381186083138206,
      "learning_rate": 1.0220733022840633e-07,
      "loss": 0.8884,
      "step": 12486
    },
    {
      "epoch": 0.9009054507413152,
      "grad_norm": 4.0074933912159665,
      "learning_rate": 1.0205988892693573e-07,
      "loss": 0.9738,
      "step": 12487
    },
    {
      "epoch": 0.9009775982107427,
      "grad_norm": 3.2790863169447992,
      "learning_rate": 1.0191255126494901e-07,
      "loss": 0.9732,
      "step": 12488
    },
    {
      "epoch": 0.9010497456801703,
      "grad_norm": 3.2538419169904613,
      "learning_rate": 1.0176531725049132e-07,
      "loss": 0.9705,
      "step": 12489
    },
    {
      "epoch": 0.9011218931495978,
      "grad_norm": 5.127107698562722,
      "learning_rate": 1.0161818689160285e-07,
      "loss": 0.9042,
      "step": 12490
    },
    {
      "epoch": 0.9011940406190253,
      "grad_norm": 5.046438664804303,
      "learning_rate": 1.014711601963174e-07,
      "loss": 0.9903,
      "step": 12491
    },
    {
      "epoch": 0.9012661880884528,
      "grad_norm": 10.681288671005005,
      "learning_rate": 1.013242371726637e-07,
      "loss": 0.9672,
      "step": 12492
    },
    {
      "epoch": 0.9013383355578803,
      "grad_norm": 5.023519512355419,
      "learning_rate": 1.0117741782866506e-07,
      "loss": 1.0105,
      "step": 12493
    },
    {
      "epoch": 0.9014104830273079,
      "grad_norm": 2.9852084449052616,
      "learning_rate": 1.0103070217233823e-07,
      "loss": 0.969,
      "step": 12494
    },
    {
      "epoch": 0.9014826304967354,
      "grad_norm": 3.48460791113025,
      "learning_rate": 1.0088409021169497e-07,
      "loss": 0.9165,
      "step": 12495
    },
    {
      "epoch": 0.9015547779661628,
      "grad_norm": 3.9662008434078353,
      "learning_rate": 1.0073758195474158e-07,
      "loss": 0.9497,
      "step": 12496
    },
    {
      "epoch": 0.9016269254355903,
      "grad_norm": 2.862299856326953,
      "learning_rate": 1.0059117740947764e-07,
      "loss": 0.9434,
      "step": 12497
    },
    {
      "epoch": 0.9016990729050178,
      "grad_norm": 16.044532790664338,
      "learning_rate": 1.0044487658389833e-07,
      "loss": 0.96,
      "step": 12498
    },
    {
      "epoch": 0.9017712203744453,
      "grad_norm": 5.183448772228429,
      "learning_rate": 1.0029867948599191e-07,
      "loss": 0.959,
      "step": 12499
    },
    {
      "epoch": 0.9018433678438729,
      "grad_norm": 3.650072274733585,
      "learning_rate": 1.0015258612374289e-07,
      "loss": 0.9075,
      "step": 12500
    },
    {
      "epoch": 0.9019155153133004,
      "grad_norm": 8.095528543428214,
      "learning_rate": 1.0000659650512777e-07,
      "loss": 0.8391,
      "step": 12501
    },
    {
      "epoch": 0.9019876627827279,
      "grad_norm": 3.9847042172122933,
      "learning_rate": 9.986071063811885e-08,
      "loss": 1.0606,
      "step": 12502
    },
    {
      "epoch": 0.9020598102521554,
      "grad_norm": 3.7358972487441124,
      "learning_rate": 9.971492853068264e-08,
      "loss": 1.0197,
      "step": 12503
    },
    {
      "epoch": 0.9021319577215829,
      "grad_norm": 0.8761948193468136,
      "learning_rate": 9.956925019077944e-08,
      "loss": 0.8917,
      "step": 12504
    },
    {
      "epoch": 0.9022041051910105,
      "grad_norm": 6.122245324828368,
      "learning_rate": 9.9423675626364e-08,
      "loss": 1.0831,
      "step": 12505
    },
    {
      "epoch": 0.902276252660438,
      "grad_norm": 3.0758647850190384,
      "learning_rate": 9.927820484538663e-08,
      "loss": 0.94,
      "step": 12506
    },
    {
      "epoch": 0.9023484001298654,
      "grad_norm": 4.175509923072364,
      "learning_rate": 9.913283785578985e-08,
      "loss": 0.9596,
      "step": 12507
    },
    {
      "epoch": 0.9024205475992929,
      "grad_norm": 6.518622497891678,
      "learning_rate": 9.8987574665512e-08,
      "loss": 0.8433,
      "step": 12508
    },
    {
      "epoch": 0.9024926950687204,
      "grad_norm": 4.201838526910046,
      "learning_rate": 9.884241528248538e-08,
      "loss": 0.8929,
      "step": 12509
    },
    {
      "epoch": 0.902564842538148,
      "grad_norm": 2.4244642629414592,
      "learning_rate": 9.869735971463699e-08,
      "loss": 0.9559,
      "step": 12510
    },
    {
      "epoch": 0.9026369900075755,
      "grad_norm": 19.70792204069494,
      "learning_rate": 9.855240796988695e-08,
      "loss": 0.9605,
      "step": 12511
    },
    {
      "epoch": 0.902709137477003,
      "grad_norm": 5.216223190681165,
      "learning_rate": 9.840756005615115e-08,
      "loss": 0.8878,
      "step": 12512
    },
    {
      "epoch": 0.9027812849464305,
      "grad_norm": 5.502747976286697,
      "learning_rate": 9.826281598133923e-08,
      "loss": 0.975,
      "step": 12513
    },
    {
      "epoch": 0.902853432415858,
      "grad_norm": 2.7114551061043652,
      "learning_rate": 9.811817575335468e-08,
      "loss": 0.9061,
      "step": 12514
    },
    {
      "epoch": 0.9029255798852855,
      "grad_norm": 5.286957675545804,
      "learning_rate": 9.797363938009606e-08,
      "loss": 0.9469,
      "step": 12515
    },
    {
      "epoch": 0.9029977273547131,
      "grad_norm": 4.065627082982508,
      "learning_rate": 9.782920686945572e-08,
      "loss": 1.0848,
      "step": 12516
    },
    {
      "epoch": 0.9030698748241406,
      "grad_norm": 3.57432267501747,
      "learning_rate": 9.768487822932137e-08,
      "loss": 0.772,
      "step": 12517
    },
    {
      "epoch": 0.9031420222935681,
      "grad_norm": 5.072475824124559,
      "learning_rate": 9.754065346757334e-08,
      "loss": 0.9791,
      "step": 12518
    },
    {
      "epoch": 0.9032141697629955,
      "grad_norm": 3.6271825336794667,
      "learning_rate": 9.739653259208758e-08,
      "loss": 0.9769,
      "step": 12519
    },
    {
      "epoch": 0.903286317232423,
      "grad_norm": 3.3700380937817362,
      "learning_rate": 9.725251561073423e-08,
      "loss": 0.914,
      "step": 12520
    },
    {
      "epoch": 0.9033584647018505,
      "grad_norm": 2.838724915541864,
      "learning_rate": 9.710860253137699e-08,
      "loss": 0.9239,
      "step": 12521
    },
    {
      "epoch": 0.9034306121712781,
      "grad_norm": 0.8866260496913408,
      "learning_rate": 9.696479336187468e-08,
      "loss": 0.9262,
      "step": 12522
    },
    {
      "epoch": 0.9035027596407056,
      "grad_norm": 5.1018920605051905,
      "learning_rate": 9.68210881100806e-08,
      "loss": 0.9107,
      "step": 12523
    },
    {
      "epoch": 0.9035749071101331,
      "grad_norm": 3.807908967885477,
      "learning_rate": 9.667748678384092e-08,
      "loss": 0.9851,
      "step": 12524
    },
    {
      "epoch": 0.9036470545795606,
      "grad_norm": 3.6586418479414995,
      "learning_rate": 9.653398939099821e-08,
      "loss": 0.915,
      "step": 12525
    },
    {
      "epoch": 0.9037192020489881,
      "grad_norm": 3.6759261488858814,
      "learning_rate": 9.639059593938781e-08,
      "loss": 0.9691,
      "step": 12526
    },
    {
      "epoch": 0.9037913495184157,
      "grad_norm": 3.481112196983993,
      "learning_rate": 9.624730643684031e-08,
      "loss": 0.9711,
      "step": 12527
    },
    {
      "epoch": 0.9038634969878432,
      "grad_norm": 5.431194288453119,
      "learning_rate": 9.610412089117969e-08,
      "loss": 0.8425,
      "step": 12528
    },
    {
      "epoch": 0.9039356444572707,
      "grad_norm": 0.8006518003458393,
      "learning_rate": 9.596103931022504e-08,
      "loss": 0.8594,
      "step": 12529
    },
    {
      "epoch": 0.9040077919266982,
      "grad_norm": 4.306129504817042,
      "learning_rate": 9.581806170178985e-08,
      "loss": 0.9955,
      "step": 12530
    },
    {
      "epoch": 0.9040799393961256,
      "grad_norm": 0.7476599267691807,
      "learning_rate": 9.567518807368102e-08,
      "loss": 0.8014,
      "step": 12531
    },
    {
      "epoch": 0.9041520868655532,
      "grad_norm": 3.603051926466569,
      "learning_rate": 9.553241843370053e-08,
      "loss": 0.9236,
      "step": 12532
    },
    {
      "epoch": 0.9042242343349807,
      "grad_norm": 6.110841228806525,
      "learning_rate": 9.538975278964456e-08,
      "loss": 0.8802,
      "step": 12533
    },
    {
      "epoch": 0.9042963818044082,
      "grad_norm": 3.6901240952877243,
      "learning_rate": 9.524719114930402e-08,
      "loss": 0.93,
      "step": 12534
    },
    {
      "epoch": 0.9043685292738357,
      "grad_norm": 3.187234456090166,
      "learning_rate": 9.510473352046244e-08,
      "loss": 0.9203,
      "step": 12535
    },
    {
      "epoch": 0.9044406767432632,
      "grad_norm": 4.353744535045715,
      "learning_rate": 9.496237991090006e-08,
      "loss": 1.0199,
      "step": 12536
    },
    {
      "epoch": 0.9045128242126907,
      "grad_norm": 3.533066004110919,
      "learning_rate": 9.482013032839021e-08,
      "loss": 0.8651,
      "step": 12537
    },
    {
      "epoch": 0.9045849716821183,
      "grad_norm": 4.311651173964247,
      "learning_rate": 9.467798478069999e-08,
      "loss": 0.8837,
      "step": 12538
    },
    {
      "epoch": 0.9046571191515458,
      "grad_norm": 15.214325749365711,
      "learning_rate": 9.45359432755919e-08,
      "loss": 0.9899,
      "step": 12539
    },
    {
      "epoch": 0.9047292666209733,
      "grad_norm": 4.4331755722390325,
      "learning_rate": 9.439400582082236e-08,
      "loss": 0.8554,
      "step": 12540
    },
    {
      "epoch": 0.9048014140904008,
      "grad_norm": 5.215598611096363,
      "learning_rate": 9.425217242414119e-08,
      "loss": 0.9213,
      "step": 12541
    },
    {
      "epoch": 0.9048735615598283,
      "grad_norm": 3.9380724051867877,
      "learning_rate": 9.411044309329441e-08,
      "loss": 0.907,
      "step": 12542
    },
    {
      "epoch": 0.9049457090292558,
      "grad_norm": 3.6143239099518745,
      "learning_rate": 9.396881783602095e-08,
      "loss": 0.9518,
      "step": 12543
    },
    {
      "epoch": 0.9050178564986833,
      "grad_norm": 3.7026607689220405,
      "learning_rate": 9.382729666005484e-08,
      "loss": 1.0282,
      "step": 12544
    },
    {
      "epoch": 0.9050900039681108,
      "grad_norm": 3.642908518025552,
      "learning_rate": 9.368587957312324e-08,
      "loss": 0.9181,
      "step": 12545
    },
    {
      "epoch": 0.9051621514375383,
      "grad_norm": 7.5191908405091095,
      "learning_rate": 9.354456658294885e-08,
      "loss": 1.0756,
      "step": 12546
    },
    {
      "epoch": 0.9052342989069658,
      "grad_norm": 4.208401569279071,
      "learning_rate": 9.340335769724861e-08,
      "loss": 0.9697,
      "step": 12547
    },
    {
      "epoch": 0.9053064463763933,
      "grad_norm": 0.754373711851169,
      "learning_rate": 9.326225292373236e-08,
      "loss": 0.8101,
      "step": 12548
    },
    {
      "epoch": 0.9053785938458209,
      "grad_norm": 4.154195865897073,
      "learning_rate": 9.312125227010615e-08,
      "loss": 0.9221,
      "step": 12549
    },
    {
      "epoch": 0.9054507413152484,
      "grad_norm": 2.6416239457230994,
      "learning_rate": 9.298035574406981e-08,
      "loss": 0.9767,
      "step": 12550
    },
    {
      "epoch": 0.9055228887846759,
      "grad_norm": 4.196623698512764,
      "learning_rate": 9.283956335331633e-08,
      "loss": 0.9098,
      "step": 12551
    },
    {
      "epoch": 0.9055950362541034,
      "grad_norm": 3.2848791488643907,
      "learning_rate": 9.26988751055342e-08,
      "loss": 0.9616,
      "step": 12552
    },
    {
      "epoch": 0.9056671837235309,
      "grad_norm": 2.8834662794506554,
      "learning_rate": 9.255829100840573e-08,
      "loss": 0.9529,
      "step": 12553
    },
    {
      "epoch": 0.9057393311929584,
      "grad_norm": 2.6344651827868986,
      "learning_rate": 9.241781106960834e-08,
      "loss": 0.9658,
      "step": 12554
    },
    {
      "epoch": 0.9058114786623859,
      "grad_norm": 3.1691917758548467,
      "learning_rate": 9.227743529681254e-08,
      "loss": 0.9375,
      "step": 12555
    },
    {
      "epoch": 0.9058836261318134,
      "grad_norm": 5.6933258736149845,
      "learning_rate": 9.213716369768354e-08,
      "loss": 0.9547,
      "step": 12556
    },
    {
      "epoch": 0.9059557736012409,
      "grad_norm": 2.9598451073818683,
      "learning_rate": 9.199699627988189e-08,
      "loss": 0.9786,
      "step": 12557
    },
    {
      "epoch": 0.9060279210706684,
      "grad_norm": 3.3947647067124445,
      "learning_rate": 9.18569330510608e-08,
      "loss": 1.0112,
      "step": 12558
    },
    {
      "epoch": 0.906100068540096,
      "grad_norm": 0.6552265224347841,
      "learning_rate": 9.171697401886857e-08,
      "loss": 0.8036,
      "step": 12559
    },
    {
      "epoch": 0.9061722160095235,
      "grad_norm": 3.5474569372637523,
      "learning_rate": 9.157711919094846e-08,
      "loss": 0.8438,
      "step": 12560
    },
    {
      "epoch": 0.906244363478951,
      "grad_norm": 4.53536884292577,
      "learning_rate": 9.143736857493744e-08,
      "loss": 0.9305,
      "step": 12561
    },
    {
      "epoch": 0.9063165109483785,
      "grad_norm": 0.68088019071659,
      "learning_rate": 9.12977221784661e-08,
      "loss": 0.7886,
      "step": 12562
    },
    {
      "epoch": 0.906388658417806,
      "grad_norm": 7.020429499602337,
      "learning_rate": 9.115818000916031e-08,
      "loss": 1.0744,
      "step": 12563
    },
    {
      "epoch": 0.9064608058872335,
      "grad_norm": 3.872962415950639,
      "learning_rate": 9.101874207464044e-08,
      "loss": 0.9429,
      "step": 12564
    },
    {
      "epoch": 0.9065329533566611,
      "grad_norm": 3.6209895871025073,
      "learning_rate": 9.087940838251974e-08,
      "loss": 0.9708,
      "step": 12565
    },
    {
      "epoch": 0.9066051008260885,
      "grad_norm": 3.676478839640614,
      "learning_rate": 9.074017894040743e-08,
      "loss": 0.8871,
      "step": 12566
    },
    {
      "epoch": 0.906677248295516,
      "grad_norm": 8.20879268490501,
      "learning_rate": 9.060105375590632e-08,
      "loss": 0.9516,
      "step": 12567
    },
    {
      "epoch": 0.9067493957649435,
      "grad_norm": 18.376400018412713,
      "learning_rate": 9.046203283661324e-08,
      "loss": 0.9294,
      "step": 12568
    },
    {
      "epoch": 0.906821543234371,
      "grad_norm": 3.608766314552567,
      "learning_rate": 9.032311619011945e-08,
      "loss": 0.9427,
      "step": 12569
    },
    {
      "epoch": 0.9068936907037985,
      "grad_norm": 2.3638470437449888,
      "learning_rate": 9.01843038240111e-08,
      "loss": 0.9335,
      "step": 12570
    },
    {
      "epoch": 0.9069658381732261,
      "grad_norm": 4.277357573465106,
      "learning_rate": 9.004559574586812e-08,
      "loss": 0.9253,
      "step": 12571
    },
    {
      "epoch": 0.9070379856426536,
      "grad_norm": 8.24811093059131,
      "learning_rate": 8.990699196326423e-08,
      "loss": 0.9311,
      "step": 12572
    },
    {
      "epoch": 0.9071101331120811,
      "grad_norm": 4.107252145580781,
      "learning_rate": 8.976849248376894e-08,
      "loss": 0.8313,
      "step": 12573
    },
    {
      "epoch": 0.9071822805815086,
      "grad_norm": 3.0497482446045856,
      "learning_rate": 8.963009731494508e-08,
      "loss": 0.8404,
      "step": 12574
    },
    {
      "epoch": 0.9072544280509361,
      "grad_norm": 3.50500375831827,
      "learning_rate": 8.949180646434929e-08,
      "loss": 0.9399,
      "step": 12575
    },
    {
      "epoch": 0.9073265755203637,
      "grad_norm": 11.743269270903452,
      "learning_rate": 8.935361993953372e-08,
      "loss": 1.0027,
      "step": 12576
    },
    {
      "epoch": 0.9073987229897912,
      "grad_norm": 3.0812554102109257,
      "learning_rate": 8.921553774804347e-08,
      "loss": 0.9541,
      "step": 12577
    },
    {
      "epoch": 0.9074708704592186,
      "grad_norm": 3.7415638445777932,
      "learning_rate": 8.907755989742005e-08,
      "loss": 0.9783,
      "step": 12578
    },
    {
      "epoch": 0.9075430179286461,
      "grad_norm": 3.968259285453988,
      "learning_rate": 8.893968639519656e-08,
      "loss": 0.9789,
      "step": 12579
    },
    {
      "epoch": 0.9076151653980736,
      "grad_norm": 2.631598218044194,
      "learning_rate": 8.880191724890229e-08,
      "loss": 0.8931,
      "step": 12580
    },
    {
      "epoch": 0.9076873128675012,
      "grad_norm": 4.135513778558413,
      "learning_rate": 8.86642524660608e-08,
      "loss": 0.9517,
      "step": 12581
    },
    {
      "epoch": 0.9077594603369287,
      "grad_norm": 3.9412136427790037,
      "learning_rate": 8.85266920541885e-08,
      "loss": 0.9361,
      "step": 12582
    },
    {
      "epoch": 0.9078316078063562,
      "grad_norm": 4.666441957422399,
      "learning_rate": 8.838923602079739e-08,
      "loss": 0.9656,
      "step": 12583
    },
    {
      "epoch": 0.9079037552757837,
      "grad_norm": 3.1822505434233435,
      "learning_rate": 8.825188437339393e-08,
      "loss": 0.9186,
      "step": 12584
    },
    {
      "epoch": 0.9079759027452112,
      "grad_norm": 3.405886849459806,
      "learning_rate": 8.811463711947786e-08,
      "loss": 0.9639,
      "step": 12585
    },
    {
      "epoch": 0.9080480502146387,
      "grad_norm": 3.7104795153244785,
      "learning_rate": 8.797749426654388e-08,
      "loss": 0.9595,
      "step": 12586
    },
    {
      "epoch": 0.9081201976840663,
      "grad_norm": 2.366613639541944,
      "learning_rate": 8.784045582208089e-08,
      "loss": 0.9658,
      "step": 12587
    },
    {
      "epoch": 0.9081923451534938,
      "grad_norm": 3.388659773483435,
      "learning_rate": 8.770352179357221e-08,
      "loss": 0.8698,
      "step": 12588
    },
    {
      "epoch": 0.9082644926229213,
      "grad_norm": 4.364996454296442,
      "learning_rate": 8.756669218849456e-08,
      "loss": 1.0386,
      "step": 12589
    },
    {
      "epoch": 0.9083366400923487,
      "grad_norm": 11.031974790282902,
      "learning_rate": 8.742996701432082e-08,
      "loss": 0.9447,
      "step": 12590
    },
    {
      "epoch": 0.9084087875617762,
      "grad_norm": 2.5316891701329967,
      "learning_rate": 8.729334627851637e-08,
      "loss": 0.8928,
      "step": 12591
    },
    {
      "epoch": 0.9084809350312038,
      "grad_norm": 6.624814012903789,
      "learning_rate": 8.715682998854168e-08,
      "loss": 0.8167,
      "step": 12592
    },
    {
      "epoch": 0.9085530825006313,
      "grad_norm": 3.4008426968302996,
      "learning_rate": 8.702041815185124e-08,
      "loss": 1.0292,
      "step": 12593
    },
    {
      "epoch": 0.9086252299700588,
      "grad_norm": 4.93525606180975,
      "learning_rate": 8.688411077589463e-08,
      "loss": 1.0531,
      "step": 12594
    },
    {
      "epoch": 0.9086973774394863,
      "grad_norm": 4.088467983221723,
      "learning_rate": 8.674790786811414e-08,
      "loss": 1.013,
      "step": 12595
    },
    {
      "epoch": 0.9087695249089138,
      "grad_norm": 6.520058712841203,
      "learning_rate": 8.661180943594759e-08,
      "loss": 0.946,
      "step": 12596
    },
    {
      "epoch": 0.9088416723783413,
      "grad_norm": 3.4032737958621624,
      "learning_rate": 8.647581548682748e-08,
      "loss": 1.0801,
      "step": 12597
    },
    {
      "epoch": 0.9089138198477689,
      "grad_norm": 3.2009971435972684,
      "learning_rate": 8.633992602817941e-08,
      "loss": 0.9078,
      "step": 12598
    },
    {
      "epoch": 0.9089859673171964,
      "grad_norm": 2.257998026580894,
      "learning_rate": 8.620414106742368e-08,
      "loss": 0.9417,
      "step": 12599
    },
    {
      "epoch": 0.9090581147866239,
      "grad_norm": 2.5817117421217226,
      "learning_rate": 8.606846061197526e-08,
      "loss": 0.9624,
      "step": 12600
    },
    {
      "epoch": 0.9091302622560513,
      "grad_norm": 3.211419804866056,
      "learning_rate": 8.593288466924331e-08,
      "loss": 1.1732,
      "step": 12601
    },
    {
      "epoch": 0.9092024097254788,
      "grad_norm": 4.7042851332058895,
      "learning_rate": 8.579741324663037e-08,
      "loss": 1.0282,
      "step": 12602
    },
    {
      "epoch": 0.9092745571949064,
      "grad_norm": 9.052851730293808,
      "learning_rate": 8.566204635153474e-08,
      "loss": 0.9986,
      "step": 12603
    },
    {
      "epoch": 0.9093467046643339,
      "grad_norm": 11.993316696633263,
      "learning_rate": 8.552678399134827e-08,
      "loss": 0.8446,
      "step": 12604
    },
    {
      "epoch": 0.9094188521337614,
      "grad_norm": 4.294500853340899,
      "learning_rate": 8.539162617345708e-08,
      "loss": 0.9508,
      "step": 12605
    },
    {
      "epoch": 0.9094909996031889,
      "grad_norm": 3.7812018273177554,
      "learning_rate": 8.525657290524146e-08,
      "loss": 0.8406,
      "step": 12606
    },
    {
      "epoch": 0.9095631470726164,
      "grad_norm": 7.651278878817927,
      "learning_rate": 8.512162419407599e-08,
      "loss": 1.0597,
      "step": 12607
    },
    {
      "epoch": 0.909635294542044,
      "grad_norm": 0.6230332902246472,
      "learning_rate": 8.498678004733028e-08,
      "loss": 0.8277,
      "step": 12608
    },
    {
      "epoch": 0.9097074420114715,
      "grad_norm": 3.134102039869391,
      "learning_rate": 8.485204047236738e-08,
      "loss": 0.8259,
      "step": 12609
    },
    {
      "epoch": 0.909779589480899,
      "grad_norm": 2.523574116790908,
      "learning_rate": 8.471740547654471e-08,
      "loss": 0.9947,
      "step": 12610
    },
    {
      "epoch": 0.9098517369503265,
      "grad_norm": 5.828734337910212,
      "learning_rate": 8.458287506721463e-08,
      "loss": 0.9852,
      "step": 12611
    },
    {
      "epoch": 0.909923884419754,
      "grad_norm": 3.6653385502470637,
      "learning_rate": 8.4448449251723e-08,
      "loss": 0.9164,
      "step": 12612
    },
    {
      "epoch": 0.9099960318891814,
      "grad_norm": 7.241323079985485,
      "learning_rate": 8.431412803741023e-08,
      "loss": 1.0115,
      "step": 12613
    },
    {
      "epoch": 0.910068179358609,
      "grad_norm": 2.4350543346256868,
      "learning_rate": 8.417991143161151e-08,
      "loss": 0.8084,
      "step": 12614
    },
    {
      "epoch": 0.9101403268280365,
      "grad_norm": 3.7887613393122415,
      "learning_rate": 8.40457994416559e-08,
      "loss": 0.9467,
      "step": 12615
    },
    {
      "epoch": 0.910212474297464,
      "grad_norm": 4.074680788608199,
      "learning_rate": 8.39117920748662e-08,
      "loss": 0.9802,
      "step": 12616
    },
    {
      "epoch": 0.9102846217668915,
      "grad_norm": 4.239872478150899,
      "learning_rate": 8.377788933856056e-08,
      "loss": 0.922,
      "step": 12617
    },
    {
      "epoch": 0.910356769236319,
      "grad_norm": 6.040656625790207,
      "learning_rate": 8.36440912400509e-08,
      "loss": 0.9164,
      "step": 12618
    },
    {
      "epoch": 0.9104289167057465,
      "grad_norm": 3.825928474383842,
      "learning_rate": 8.351039778664315e-08,
      "loss": 0.905,
      "step": 12619
    },
    {
      "epoch": 0.9105010641751741,
      "grad_norm": 0.7539792593751917,
      "learning_rate": 8.337680898563792e-08,
      "loss": 0.829,
      "step": 12620
    },
    {
      "epoch": 0.9105732116446016,
      "grad_norm": 5.534240776170471,
      "learning_rate": 8.324332484433005e-08,
      "loss": 0.9758,
      "step": 12621
    },
    {
      "epoch": 0.9106453591140291,
      "grad_norm": 5.626288880928109,
      "learning_rate": 8.310994537000881e-08,
      "loss": 1.0181,
      "step": 12622
    },
    {
      "epoch": 0.9107175065834566,
      "grad_norm": 2.900030556879524,
      "learning_rate": 8.297667056995727e-08,
      "loss": 0.8768,
      "step": 12623
    },
    {
      "epoch": 0.9107896540528841,
      "grad_norm": 4.94970315370674,
      "learning_rate": 8.284350045145317e-08,
      "loss": 0.9476,
      "step": 12624
    },
    {
      "epoch": 0.9108618015223116,
      "grad_norm": 4.218385129620889,
      "learning_rate": 8.271043502176866e-08,
      "loss": 0.9587,
      "step": 12625
    },
    {
      "epoch": 0.9109339489917391,
      "grad_norm": 8.383525096401405,
      "learning_rate": 8.25774742881693e-08,
      "loss": 1.042,
      "step": 12626
    },
    {
      "epoch": 0.9110060964611666,
      "grad_norm": 0.725124289283608,
      "learning_rate": 8.244461825791616e-08,
      "loss": 0.8677,
      "step": 12627
    },
    {
      "epoch": 0.9110782439305941,
      "grad_norm": 8.595429523380798,
      "learning_rate": 8.231186693826408e-08,
      "loss": 0.9287,
      "step": 12628
    },
    {
      "epoch": 0.9111503914000216,
      "grad_norm": 4.39423458200009,
      "learning_rate": 8.217922033646174e-08,
      "loss": 0.9512,
      "step": 12629
    },
    {
      "epoch": 0.9112225388694492,
      "grad_norm": 4.5509815875967385,
      "learning_rate": 8.204667845975267e-08,
      "loss": 0.9116,
      "step": 12630
    },
    {
      "epoch": 0.9112946863388767,
      "grad_norm": 5.110814069786986,
      "learning_rate": 8.191424131537416e-08,
      "loss": 0.9761,
      "step": 12631
    },
    {
      "epoch": 0.9113668338083042,
      "grad_norm": 6.256057855206849,
      "learning_rate": 8.178190891055914e-08,
      "loss": 0.9955,
      "step": 12632
    },
    {
      "epoch": 0.9114389812777317,
      "grad_norm": 6.015656389427034,
      "learning_rate": 8.164968125253247e-08,
      "loss": 0.8886,
      "step": 12633
    },
    {
      "epoch": 0.9115111287471592,
      "grad_norm": 10.185009985550417,
      "learning_rate": 8.151755834851548e-08,
      "loss": 0.9327,
      "step": 12634
    },
    {
      "epoch": 0.9115832762165867,
      "grad_norm": 5.20965679392822,
      "learning_rate": 8.138554020572286e-08,
      "loss": 0.9509,
      "step": 12635
    },
    {
      "epoch": 0.9116554236860143,
      "grad_norm": 3.390430497598411,
      "learning_rate": 8.125362683136305e-08,
      "loss": 0.9343,
      "step": 12636
    },
    {
      "epoch": 0.9117275711554417,
      "grad_norm": 3.6557348165519605,
      "learning_rate": 8.112181823263964e-08,
      "loss": 0.9298,
      "step": 12637
    },
    {
      "epoch": 0.9117997186248692,
      "grad_norm": 3.5246314755693566,
      "learning_rate": 8.099011441675063e-08,
      "loss": 1.0514,
      "step": 12638
    },
    {
      "epoch": 0.9118718660942967,
      "grad_norm": 2.614587568505088,
      "learning_rate": 8.08585153908874e-08,
      "loss": 0.8866,
      "step": 12639
    },
    {
      "epoch": 0.9119440135637242,
      "grad_norm": 6.821996107437513,
      "learning_rate": 8.072702116223618e-08,
      "loss": 0.9307,
      "step": 12640
    },
    {
      "epoch": 0.9120161610331518,
      "grad_norm": 3.470855898438397,
      "learning_rate": 8.059563173797746e-08,
      "loss": 1.0845,
      "step": 12641
    },
    {
      "epoch": 0.9120883085025793,
      "grad_norm": 3.234185991892143,
      "learning_rate": 8.046434712528594e-08,
      "loss": 0.94,
      "step": 12642
    },
    {
      "epoch": 0.9121604559720068,
      "grad_norm": 0.7801634671629774,
      "learning_rate": 8.033316733133055e-08,
      "loss": 0.8562,
      "step": 12643
    },
    {
      "epoch": 0.9122326034414343,
      "grad_norm": 4.431552393485937,
      "learning_rate": 8.020209236327424e-08,
      "loss": 0.9369,
      "step": 12644
    },
    {
      "epoch": 0.9123047509108618,
      "grad_norm": 3.5177638350651104,
      "learning_rate": 8.007112222827528e-08,
      "loss": 0.9726,
      "step": 12645
    },
    {
      "epoch": 0.9123768983802893,
      "grad_norm": 6.145434488404342,
      "learning_rate": 7.994025693348482e-08,
      "loss": 1.1089,
      "step": 12646
    },
    {
      "epoch": 0.9124490458497169,
      "grad_norm": 2.750006675712112,
      "learning_rate": 7.980949648604896e-08,
      "loss": 0.9888,
      "step": 12647
    },
    {
      "epoch": 0.9125211933191444,
      "grad_norm": 7.088286459479514,
      "learning_rate": 7.967884089310839e-08,
      "loss": 0.8902,
      "step": 12648
    },
    {
      "epoch": 0.9125933407885718,
      "grad_norm": 3.7246699295762506,
      "learning_rate": 7.954829016179765e-08,
      "loss": 0.9481,
      "step": 12649
    },
    {
      "epoch": 0.9126654882579993,
      "grad_norm": 6.156583670466525,
      "learning_rate": 7.941784429924503e-08,
      "loss": 0.8115,
      "step": 12650
    },
    {
      "epoch": 0.9127376357274268,
      "grad_norm": 3.8983442948997142,
      "learning_rate": 7.928750331257439e-08,
      "loss": 0.9922,
      "step": 12651
    },
    {
      "epoch": 0.9128097831968544,
      "grad_norm": 6.67056897711999,
      "learning_rate": 7.915726720890314e-08,
      "loss": 0.9188,
      "step": 12652
    },
    {
      "epoch": 0.9128819306662819,
      "grad_norm": 5.097295732282818,
      "learning_rate": 7.902713599534272e-08,
      "loss": 0.9487,
      "step": 12653
    },
    {
      "epoch": 0.9129540781357094,
      "grad_norm": 4.53253597558752,
      "learning_rate": 7.88971096789992e-08,
      "loss": 0.9942,
      "step": 12654
    },
    {
      "epoch": 0.9130262256051369,
      "grad_norm": 4.109912445546571,
      "learning_rate": 7.876718826697315e-08,
      "loss": 0.9442,
      "step": 12655
    },
    {
      "epoch": 0.9130983730745644,
      "grad_norm": 2.714383883601159,
      "learning_rate": 7.863737176635821e-08,
      "loss": 0.8728,
      "step": 12656
    },
    {
      "epoch": 0.913170520543992,
      "grad_norm": 4.6997037591303545,
      "learning_rate": 7.850766018424382e-08,
      "loss": 0.9249,
      "step": 12657
    },
    {
      "epoch": 0.9132426680134195,
      "grad_norm": 3.7310294643640427,
      "learning_rate": 7.837805352771321e-08,
      "loss": 0.9457,
      "step": 12658
    },
    {
      "epoch": 0.913314815482847,
      "grad_norm": 3.4754191029254304,
      "learning_rate": 7.824855180384382e-08,
      "loss": 0.8812,
      "step": 12659
    },
    {
      "epoch": 0.9133869629522744,
      "grad_norm": 4.042643451964557,
      "learning_rate": 7.811915501970645e-08,
      "loss": 0.9494,
      "step": 12660
    },
    {
      "epoch": 0.9134591104217019,
      "grad_norm": 3.369336356406849,
      "learning_rate": 7.798986318236744e-08,
      "loss": 1.062,
      "step": 12661
    },
    {
      "epoch": 0.9135312578911294,
      "grad_norm": 3.0578849429134314,
      "learning_rate": 7.786067629888716e-08,
      "loss": 0.9447,
      "step": 12662
    },
    {
      "epoch": 0.913603405360557,
      "grad_norm": 3.4946556159578166,
      "learning_rate": 7.773159437631994e-08,
      "loss": 0.9225,
      "step": 12663
    },
    {
      "epoch": 0.9136755528299845,
      "grad_norm": 6.241299486076786,
      "learning_rate": 7.760261742171414e-08,
      "loss": 0.9637,
      "step": 12664
    },
    {
      "epoch": 0.913747700299412,
      "grad_norm": 0.7580543604031729,
      "learning_rate": 7.747374544211305e-08,
      "loss": 0.8618,
      "step": 12665
    },
    {
      "epoch": 0.9138198477688395,
      "grad_norm": 3.2504844304492173,
      "learning_rate": 7.73449784445539e-08,
      "loss": 0.9243,
      "step": 12666
    },
    {
      "epoch": 0.913891995238267,
      "grad_norm": 2.722769473552548,
      "learning_rate": 7.721631643606774e-08,
      "loss": 0.9357,
      "step": 12667
    },
    {
      "epoch": 0.9139641427076945,
      "grad_norm": 5.783115065502315,
      "learning_rate": 7.708775942368051e-08,
      "loss": 1.0425,
      "step": 12668
    },
    {
      "epoch": 0.9140362901771221,
      "grad_norm": 5.614018890185998,
      "learning_rate": 7.695930741441281e-08,
      "loss": 0.9561,
      "step": 12669
    },
    {
      "epoch": 0.9141084376465496,
      "grad_norm": 3.8339123634108985,
      "learning_rate": 7.683096041527837e-08,
      "loss": 0.9079,
      "step": 12670
    },
    {
      "epoch": 0.9141805851159771,
      "grad_norm": 11.349520474021551,
      "learning_rate": 7.67027184332858e-08,
      "loss": 0.9832,
      "step": 12671
    },
    {
      "epoch": 0.9142527325854045,
      "grad_norm": 3.07035070497461,
      "learning_rate": 7.657458147543816e-08,
      "loss": 0.9285,
      "step": 12672
    },
    {
      "epoch": 0.914324880054832,
      "grad_norm": 3.1885238199981822,
      "learning_rate": 7.644654954873209e-08,
      "loss": 0.9835,
      "step": 12673
    },
    {
      "epoch": 0.9143970275242596,
      "grad_norm": 0.6807845240302569,
      "learning_rate": 7.631862266015887e-08,
      "loss": 0.8169,
      "step": 12674
    },
    {
      "epoch": 0.9144691749936871,
      "grad_norm": 4.540160579782389,
      "learning_rate": 7.619080081670492e-08,
      "loss": 0.9257,
      "step": 12675
    },
    {
      "epoch": 0.9145413224631146,
      "grad_norm": 4.953176985506964,
      "learning_rate": 7.606308402534977e-08,
      "loss": 0.9576,
      "step": 12676
    },
    {
      "epoch": 0.9146134699325421,
      "grad_norm": 3.6894200385626013,
      "learning_rate": 7.593547229306718e-08,
      "loss": 0.9499,
      "step": 12677
    },
    {
      "epoch": 0.9146856174019696,
      "grad_norm": 3.9779446041801263,
      "learning_rate": 7.580796562682579e-08,
      "loss": 0.9204,
      "step": 12678
    },
    {
      "epoch": 0.9147577648713971,
      "grad_norm": 4.231447678967432,
      "learning_rate": 7.568056403358846e-08,
      "loss": 0.9634,
      "step": 12679
    },
    {
      "epoch": 0.9148299123408247,
      "grad_norm": 2.9176525266340945,
      "learning_rate": 7.555326752031166e-08,
      "loss": 0.9424,
      "step": 12680
    },
    {
      "epoch": 0.9149020598102522,
      "grad_norm": 9.311463163887908,
      "learning_rate": 7.542607609394691e-08,
      "loss": 0.9313,
      "step": 12681
    },
    {
      "epoch": 0.9149742072796797,
      "grad_norm": 0.7077302796565057,
      "learning_rate": 7.529898976143978e-08,
      "loss": 0.8662,
      "step": 12682
    },
    {
      "epoch": 0.9150463547491072,
      "grad_norm": 4.018437332163996,
      "learning_rate": 7.517200852972982e-08,
      "loss": 0.9268,
      "step": 12683
    },
    {
      "epoch": 0.9151185022185346,
      "grad_norm": 3.135681664979619,
      "learning_rate": 7.504513240575083e-08,
      "loss": 0.9637,
      "step": 12684
    },
    {
      "epoch": 0.9151906496879622,
      "grad_norm": 5.9272369025190805,
      "learning_rate": 7.491836139643104e-08,
      "loss": 0.9844,
      "step": 12685
    },
    {
      "epoch": 0.9152627971573897,
      "grad_norm": 3.8012204519250528,
      "learning_rate": 7.47916955086938e-08,
      "loss": 0.8974,
      "step": 12686
    },
    {
      "epoch": 0.9153349446268172,
      "grad_norm": 8.934233808919565,
      "learning_rate": 7.466513474945468e-08,
      "loss": 0.9825,
      "step": 12687
    },
    {
      "epoch": 0.9154070920962447,
      "grad_norm": 2.904953534120708,
      "learning_rate": 7.453867912562529e-08,
      "loss": 0.9828,
      "step": 12688
    },
    {
      "epoch": 0.9154792395656722,
      "grad_norm": 2.319068446746098,
      "learning_rate": 7.441232864411117e-08,
      "loss": 1.0517,
      "step": 12689
    },
    {
      "epoch": 0.9155513870350998,
      "grad_norm": 7.39986044262407,
      "learning_rate": 7.428608331181107e-08,
      "loss": 0.934,
      "step": 12690
    },
    {
      "epoch": 0.9156235345045273,
      "grad_norm": 3.4021279127506063,
      "learning_rate": 7.415994313561924e-08,
      "loss": 1.0359,
      "step": 12691
    },
    {
      "epoch": 0.9156956819739548,
      "grad_norm": 5.688718895723408,
      "learning_rate": 7.40339081224235e-08,
      "loss": 0.9711,
      "step": 12692
    },
    {
      "epoch": 0.9157678294433823,
      "grad_norm": 5.987683688792457,
      "learning_rate": 7.390797827910699e-08,
      "loss": 0.939,
      "step": 12693
    },
    {
      "epoch": 0.9158399769128098,
      "grad_norm": 7.402811403021231,
      "learning_rate": 7.378215361254514e-08,
      "loss": 0.946,
      "step": 12694
    },
    {
      "epoch": 0.9159121243822373,
      "grad_norm": 3.946066846587784,
      "learning_rate": 7.365643412960953e-08,
      "loss": 0.9669,
      "step": 12695
    },
    {
      "epoch": 0.9159842718516648,
      "grad_norm": 3.9952349891349574,
      "learning_rate": 7.353081983716491e-08,
      "loss": 0.918,
      "step": 12696
    },
    {
      "epoch": 0.9160564193210923,
      "grad_norm": 10.847247963935981,
      "learning_rate": 7.340531074207068e-08,
      "loss": 1.0151,
      "step": 12697
    },
    {
      "epoch": 0.9161285667905198,
      "grad_norm": 3.402063859711366,
      "learning_rate": 7.327990685118024e-08,
      "loss": 1.012,
      "step": 12698
    },
    {
      "epoch": 0.9162007142599473,
      "grad_norm": 3.4985768285746497,
      "learning_rate": 7.315460817134189e-08,
      "loss": 1.058,
      "step": 12699
    },
    {
      "epoch": 0.9162728617293748,
      "grad_norm": 4.041904062332094,
      "learning_rate": 7.302941470939727e-08,
      "loss": 0.9532,
      "step": 12700
    },
    {
      "epoch": 0.9163450091988024,
      "grad_norm": 2.9337447492034823,
      "learning_rate": 7.290432647218292e-08,
      "loss": 1.0463,
      "step": 12701
    },
    {
      "epoch": 0.9164171566682299,
      "grad_norm": 2.903394383147387,
      "learning_rate": 7.277934346652937e-08,
      "loss": 0.8592,
      "step": 12702
    },
    {
      "epoch": 0.9164893041376574,
      "grad_norm": 22.178070041630864,
      "learning_rate": 7.265446569926182e-08,
      "loss": 0.9944,
      "step": 12703
    },
    {
      "epoch": 0.9165614516070849,
      "grad_norm": 3.6374335318978144,
      "learning_rate": 7.252969317719859e-08,
      "loss": 0.86,
      "step": 12704
    },
    {
      "epoch": 0.9166335990765124,
      "grad_norm": 2.9142457633392254,
      "learning_rate": 7.240502590715403e-08,
      "loss": 0.8802,
      "step": 12705
    },
    {
      "epoch": 0.91670574654594,
      "grad_norm": 3.626248900904155,
      "learning_rate": 7.228046389593534e-08,
      "loss": 0.9071,
      "step": 12706
    },
    {
      "epoch": 0.9167778940153674,
      "grad_norm": 5.025283403138608,
      "learning_rate": 7.21560071503442e-08,
      "loss": 0.854,
      "step": 12707
    },
    {
      "epoch": 0.9168500414847949,
      "grad_norm": 4.974752003991828,
      "learning_rate": 7.203165567717673e-08,
      "loss": 0.9182,
      "step": 12708
    },
    {
      "epoch": 0.9169221889542224,
      "grad_norm": 3.985682493611398,
      "learning_rate": 7.190740948322327e-08,
      "loss": 1.0025,
      "step": 12709
    },
    {
      "epoch": 0.9169943364236499,
      "grad_norm": 0.7482211156560026,
      "learning_rate": 7.178326857526906e-08,
      "loss": 0.8207,
      "step": 12710
    },
    {
      "epoch": 0.9170664838930774,
      "grad_norm": 3.905151823172042,
      "learning_rate": 7.165923296009246e-08,
      "loss": 0.9062,
      "step": 12711
    },
    {
      "epoch": 0.917138631362505,
      "grad_norm": 7.177633396398544,
      "learning_rate": 7.15353026444665e-08,
      "loss": 0.9826,
      "step": 12712
    },
    {
      "epoch": 0.9172107788319325,
      "grad_norm": 5.938668226459117,
      "learning_rate": 7.141147763515909e-08,
      "loss": 1.0201,
      "step": 12713
    },
    {
      "epoch": 0.91728292630136,
      "grad_norm": 6.386136967003458,
      "learning_rate": 7.128775793893127e-08,
      "loss": 0.8659,
      "step": 12714
    },
    {
      "epoch": 0.9173550737707875,
      "grad_norm": 3.604399980405893,
      "learning_rate": 7.116414356253898e-08,
      "loss": 1.0273,
      "step": 12715
    },
    {
      "epoch": 0.917427221240215,
      "grad_norm": 3.347587878531673,
      "learning_rate": 7.104063451273279e-08,
      "loss": 0.9198,
      "step": 12716
    },
    {
      "epoch": 0.9174993687096425,
      "grad_norm": 4.05213968716671,
      "learning_rate": 7.091723079625645e-08,
      "loss": 0.9716,
      "step": 12717
    },
    {
      "epoch": 0.9175715161790701,
      "grad_norm": 4.215277252672062,
      "learning_rate": 7.079393241984921e-08,
      "loss": 1.0711,
      "step": 12718
    },
    {
      "epoch": 0.9176436636484975,
      "grad_norm": 3.030980048989608,
      "learning_rate": 7.067073939024349e-08,
      "loss": 1.0826,
      "step": 12719
    },
    {
      "epoch": 0.917715811117925,
      "grad_norm": 2.981949544601292,
      "learning_rate": 7.054765171416677e-08,
      "loss": 0.9954,
      "step": 12720
    },
    {
      "epoch": 0.9177879585873525,
      "grad_norm": 3.6209387555741195,
      "learning_rate": 7.04246693983399e-08,
      "loss": 0.9707,
      "step": 12721
    },
    {
      "epoch": 0.91786010605678,
      "grad_norm": 0.7588458537054638,
      "learning_rate": 7.030179244947887e-08,
      "loss": 0.8592,
      "step": 12722
    },
    {
      "epoch": 0.9179322535262076,
      "grad_norm": 4.251086152353472,
      "learning_rate": 7.017902087429362e-08,
      "loss": 1.0875,
      "step": 12723
    },
    {
      "epoch": 0.9180044009956351,
      "grad_norm": 2.4665658681052207,
      "learning_rate": 7.00563546794879e-08,
      "loss": 0.9337,
      "step": 12724
    },
    {
      "epoch": 0.9180765484650626,
      "grad_norm": 2.888361854243648,
      "learning_rate": 6.993379387176035e-08,
      "loss": 0.9726,
      "step": 12725
    },
    {
      "epoch": 0.9181486959344901,
      "grad_norm": 3.086149647820513,
      "learning_rate": 6.98113384578034e-08,
      "loss": 0.9184,
      "step": 12726
    },
    {
      "epoch": 0.9182208434039176,
      "grad_norm": 2.654652822177078,
      "learning_rate": 6.968898844430415e-08,
      "loss": 0.8619,
      "step": 12727
    },
    {
      "epoch": 0.9182929908733451,
      "grad_norm": 3.3163001377127426,
      "learning_rate": 6.956674383794304e-08,
      "loss": 0.9136,
      "step": 12728
    },
    {
      "epoch": 0.9183651383427727,
      "grad_norm": 4.308659488179939,
      "learning_rate": 6.944460464539603e-08,
      "loss": 0.8752,
      "step": 12729
    },
    {
      "epoch": 0.9184372858122002,
      "grad_norm": 3.312809479849433,
      "learning_rate": 6.93225708733327e-08,
      "loss": 1.0421,
      "step": 12730
    },
    {
      "epoch": 0.9185094332816276,
      "grad_norm": 26.392650899149462,
      "learning_rate": 6.920064252841662e-08,
      "loss": 0.9825,
      "step": 12731
    },
    {
      "epoch": 0.9185815807510551,
      "grad_norm": 5.187052328257264,
      "learning_rate": 6.907881961730578e-08,
      "loss": 0.9448,
      "step": 12732
    },
    {
      "epoch": 0.9186537282204826,
      "grad_norm": 5.457650741333663,
      "learning_rate": 6.895710214665262e-08,
      "loss": 0.871,
      "step": 12733
    },
    {
      "epoch": 0.9187258756899102,
      "grad_norm": 2.8937280443977595,
      "learning_rate": 6.883549012310341e-08,
      "loss": 1.0226,
      "step": 12734
    },
    {
      "epoch": 0.9187980231593377,
      "grad_norm": 0.6991493787397587,
      "learning_rate": 6.871398355329927e-08,
      "loss": 0.7836,
      "step": 12735
    },
    {
      "epoch": 0.9188701706287652,
      "grad_norm": 0.6767093481460189,
      "learning_rate": 6.859258244387511e-08,
      "loss": 0.8322,
      "step": 12736
    },
    {
      "epoch": 0.9189423180981927,
      "grad_norm": 7.039560247471656,
      "learning_rate": 6.847128680146052e-08,
      "loss": 0.961,
      "step": 12737
    },
    {
      "epoch": 0.9190144655676202,
      "grad_norm": 2.9462377577257453,
      "learning_rate": 6.835009663267821e-08,
      "loss": 0.9539,
      "step": 12738
    },
    {
      "epoch": 0.9190866130370478,
      "grad_norm": 2.86809357091193,
      "learning_rate": 6.822901194414665e-08,
      "loss": 1.0006,
      "step": 12739
    },
    {
      "epoch": 0.9191587605064753,
      "grad_norm": 0.686402072733137,
      "learning_rate": 6.810803274247745e-08,
      "loss": 0.8735,
      "step": 12740
    },
    {
      "epoch": 0.9192309079759028,
      "grad_norm": 2.778027328301864,
      "learning_rate": 6.79871590342771e-08,
      "loss": 0.9253,
      "step": 12741
    },
    {
      "epoch": 0.9193030554453303,
      "grad_norm": 38.859901170040466,
      "learning_rate": 6.786639082614587e-08,
      "loss": 1.0675,
      "step": 12742
    },
    {
      "epoch": 0.9193752029147577,
      "grad_norm": 0.8150662097482302,
      "learning_rate": 6.774572812467872e-08,
      "loss": 0.9109,
      "step": 12743
    },
    {
      "epoch": 0.9194473503841852,
      "grad_norm": 3.7361781663009914,
      "learning_rate": 6.762517093646413e-08,
      "loss": 0.9879,
      "step": 12744
    },
    {
      "epoch": 0.9195194978536128,
      "grad_norm": 5.744800704805995,
      "learning_rate": 6.750471926808554e-08,
      "loss": 0.9504,
      "step": 12745
    },
    {
      "epoch": 0.9195916453230403,
      "grad_norm": 4.664649504714267,
      "learning_rate": 6.738437312612012e-08,
      "loss": 1.019,
      "step": 12746
    },
    {
      "epoch": 0.9196637927924678,
      "grad_norm": 2.7165386809486685,
      "learning_rate": 6.726413251714036e-08,
      "loss": 1.0231,
      "step": 12747
    },
    {
      "epoch": 0.9197359402618953,
      "grad_norm": 6.210225919831081,
      "learning_rate": 6.714399744771127e-08,
      "loss": 0.9719,
      "step": 12748
    },
    {
      "epoch": 0.9198080877313228,
      "grad_norm": 3.8274543018964273,
      "learning_rate": 6.702396792439336e-08,
      "loss": 0.9486,
      "step": 12749
    },
    {
      "epoch": 0.9198802352007504,
      "grad_norm": 6.848196875496253,
      "learning_rate": 6.690404395374116e-08,
      "loss": 0.9496,
      "step": 12750
    },
    {
      "epoch": 0.9199523826701779,
      "grad_norm": 3.670287872790475,
      "learning_rate": 6.678422554230278e-08,
      "loss": 0.9734,
      "step": 12751
    },
    {
      "epoch": 0.9200245301396054,
      "grad_norm": 0.7967003649669954,
      "learning_rate": 6.666451269662122e-08,
      "loss": 0.8962,
      "step": 12752
    },
    {
      "epoch": 0.9200966776090329,
      "grad_norm": 3.6487869650787124,
      "learning_rate": 6.654490542323366e-08,
      "loss": 1.0145,
      "step": 12753
    },
    {
      "epoch": 0.9201688250784604,
      "grad_norm": 2.6633882536336126,
      "learning_rate": 6.642540372867179e-08,
      "loss": 1.0,
      "step": 12754
    },
    {
      "epoch": 0.9202409725478878,
      "grad_norm": 3.1047236478378166,
      "learning_rate": 6.630600761946036e-08,
      "loss": 1.0795,
      "step": 12755
    },
    {
      "epoch": 0.9203131200173154,
      "grad_norm": 5.4225577768261335,
      "learning_rate": 6.618671710211976e-08,
      "loss": 0.8771,
      "step": 12756
    },
    {
      "epoch": 0.9203852674867429,
      "grad_norm": 3.255021983598344,
      "learning_rate": 6.606753218316407e-08,
      "loss": 0.8957,
      "step": 12757
    },
    {
      "epoch": 0.9204574149561704,
      "grad_norm": 3.6949858894130334,
      "learning_rate": 6.594845286910078e-08,
      "loss": 0.837,
      "step": 12758
    },
    {
      "epoch": 0.9205295624255979,
      "grad_norm": 7.149381714877818,
      "learning_rate": 6.582947916643289e-08,
      "loss": 0.8319,
      "step": 12759
    },
    {
      "epoch": 0.9206017098950254,
      "grad_norm": 6.149485943407307,
      "learning_rate": 6.571061108165743e-08,
      "loss": 0.8533,
      "step": 12760
    },
    {
      "epoch": 0.920673857364453,
      "grad_norm": 5.027257057515446,
      "learning_rate": 6.559184862126498e-08,
      "loss": 1.0342,
      "step": 12761
    },
    {
      "epoch": 0.9207460048338805,
      "grad_norm": 3.579117949527629,
      "learning_rate": 6.54731917917406e-08,
      "loss": 0.8436,
      "step": 12762
    },
    {
      "epoch": 0.920818152303308,
      "grad_norm": 4.873949964852706,
      "learning_rate": 6.535464059956352e-08,
      "loss": 0.9594,
      "step": 12763
    },
    {
      "epoch": 0.9208902997727355,
      "grad_norm": 2.8213603854778375,
      "learning_rate": 6.523619505120814e-08,
      "loss": 1.0064,
      "step": 12764
    },
    {
      "epoch": 0.920962447242163,
      "grad_norm": 4.482234856188403,
      "learning_rate": 6.511785515314172e-08,
      "loss": 0.8849,
      "step": 12765
    },
    {
      "epoch": 0.9210345947115904,
      "grad_norm": 6.007981395951686,
      "learning_rate": 6.499962091182665e-08,
      "loss": 0.9648,
      "step": 12766
    },
    {
      "epoch": 0.921106742181018,
      "grad_norm": 6.603994790456971,
      "learning_rate": 6.488149233371932e-08,
      "loss": 0.9703,
      "step": 12767
    },
    {
      "epoch": 0.9211788896504455,
      "grad_norm": 3.68299771904441,
      "learning_rate": 6.476346942526989e-08,
      "loss": 0.9845,
      "step": 12768
    },
    {
      "epoch": 0.921251037119873,
      "grad_norm": 4.163447293467099,
      "learning_rate": 6.464555219292345e-08,
      "loss": 0.9773,
      "step": 12769
    },
    {
      "epoch": 0.9213231845893005,
      "grad_norm": 4.659588736858144,
      "learning_rate": 6.452774064311861e-08,
      "loss": 0.991,
      "step": 12770
    },
    {
      "epoch": 0.921395332058728,
      "grad_norm": 4.9452046989756155,
      "learning_rate": 6.441003478228957e-08,
      "loss": 0.9578,
      "step": 12771
    },
    {
      "epoch": 0.9214674795281556,
      "grad_norm": 3.3345060828630064,
      "learning_rate": 6.429243461686296e-08,
      "loss": 0.9245,
      "step": 12772
    },
    {
      "epoch": 0.9215396269975831,
      "grad_norm": 3.951131325483969,
      "learning_rate": 6.417494015326075e-08,
      "loss": 1.0277,
      "step": 12773
    },
    {
      "epoch": 0.9216117744670106,
      "grad_norm": 8.668938656867024,
      "learning_rate": 6.405755139789914e-08,
      "loss": 0.9958,
      "step": 12774
    },
    {
      "epoch": 0.9216839219364381,
      "grad_norm": 3.182688653453136,
      "learning_rate": 6.394026835718769e-08,
      "loss": 1.0043,
      "step": 12775
    },
    {
      "epoch": 0.9217560694058656,
      "grad_norm": 7.913032607824274,
      "learning_rate": 6.382309103753103e-08,
      "loss": 0.9514,
      "step": 12776
    },
    {
      "epoch": 0.9218282168752931,
      "grad_norm": 6.038463962597433,
      "learning_rate": 6.370601944532828e-08,
      "loss": 0.9169,
      "step": 12777
    },
    {
      "epoch": 0.9219003643447206,
      "grad_norm": 3.059299431433516,
      "learning_rate": 6.358905358697165e-08,
      "loss": 0.9754,
      "step": 12778
    },
    {
      "epoch": 0.9219725118141481,
      "grad_norm": 5.174396004190123,
      "learning_rate": 6.347219346884847e-08,
      "loss": 0.9673,
      "step": 12779
    },
    {
      "epoch": 0.9220446592835756,
      "grad_norm": 12.506877990542344,
      "learning_rate": 6.335543909734009e-08,
      "loss": 0.9484,
      "step": 12780
    },
    {
      "epoch": 0.9221168067530031,
      "grad_norm": 3.6327865107437294,
      "learning_rate": 6.323879047882208e-08,
      "loss": 0.9243,
      "step": 12781
    },
    {
      "epoch": 0.9221889542224306,
      "grad_norm": 4.45073370724838,
      "learning_rate": 6.312224761966378e-08,
      "loss": 0.9796,
      "step": 12782
    },
    {
      "epoch": 0.9222611016918582,
      "grad_norm": 4.420845478027997,
      "learning_rate": 6.300581052622944e-08,
      "loss": 0.8706,
      "step": 12783
    },
    {
      "epoch": 0.9223332491612857,
      "grad_norm": 3.4797255624882797,
      "learning_rate": 6.288947920487775e-08,
      "loss": 1.0037,
      "step": 12784
    },
    {
      "epoch": 0.9224053966307132,
      "grad_norm": 6.413240517138279,
      "learning_rate": 6.277325366196029e-08,
      "loss": 1.0024,
      "step": 12785
    },
    {
      "epoch": 0.9224775441001407,
      "grad_norm": 2.6734342431249796,
      "learning_rate": 6.265713390382421e-08,
      "loss": 0.9727,
      "step": 12786
    },
    {
      "epoch": 0.9225496915695682,
      "grad_norm": 2.607914681601227,
      "learning_rate": 6.254111993681044e-08,
      "loss": 1.0458,
      "step": 12787
    },
    {
      "epoch": 0.9226218390389958,
      "grad_norm": 3.215812008890737,
      "learning_rate": 6.242521176725323e-08,
      "loss": 0.9819,
      "step": 12788
    },
    {
      "epoch": 0.9226939865084233,
      "grad_norm": 3.8594486368544376,
      "learning_rate": 6.23094094014831e-08,
      "loss": 0.8908,
      "step": 12789
    },
    {
      "epoch": 0.9227661339778507,
      "grad_norm": 3.493463952871798,
      "learning_rate": 6.219371284582276e-08,
      "loss": 0.9648,
      "step": 12790
    },
    {
      "epoch": 0.9228382814472782,
      "grad_norm": 5.040182205115863,
      "learning_rate": 6.207812210659069e-08,
      "loss": 1.0274,
      "step": 12791
    },
    {
      "epoch": 0.9229104289167057,
      "grad_norm": 4.27469774036532,
      "learning_rate": 6.196263719009809e-08,
      "loss": 0.9659,
      "step": 12792
    },
    {
      "epoch": 0.9229825763861332,
      "grad_norm": 0.6744996415102651,
      "learning_rate": 6.184725810265145e-08,
      "loss": 0.8038,
      "step": 12793
    },
    {
      "epoch": 0.9230547238555608,
      "grad_norm": 4.509608712487824,
      "learning_rate": 6.173198485055153e-08,
      "loss": 0.9118,
      "step": 12794
    },
    {
      "epoch": 0.9231268713249883,
      "grad_norm": 11.751600440579958,
      "learning_rate": 6.161681744009239e-08,
      "loss": 0.9572,
      "step": 12795
    },
    {
      "epoch": 0.9231990187944158,
      "grad_norm": 2.7143666678381355,
      "learning_rate": 6.150175587756323e-08,
      "loss": 0.9197,
      "step": 12796
    },
    {
      "epoch": 0.9232711662638433,
      "grad_norm": 3.797628822376719,
      "learning_rate": 6.138680016924725e-08,
      "loss": 0.9459,
      "step": 12797
    },
    {
      "epoch": 0.9233433137332708,
      "grad_norm": 4.747809858851821,
      "learning_rate": 6.127195032142163e-08,
      "loss": 0.9921,
      "step": 12798
    },
    {
      "epoch": 0.9234154612026984,
      "grad_norm": 6.064510680584368,
      "learning_rate": 6.115720634035781e-08,
      "loss": 0.9117,
      "step": 12799
    },
    {
      "epoch": 0.9234876086721259,
      "grad_norm": 3.91364960767414,
      "learning_rate": 6.104256823232124e-08,
      "loss": 0.891,
      "step": 12800
    },
    {
      "epoch": 0.9235597561415534,
      "grad_norm": 5.078697195827802,
      "learning_rate": 6.092803600357288e-08,
      "loss": 0.9777,
      "step": 12801
    },
    {
      "epoch": 0.9236319036109808,
      "grad_norm": 5.611772540213192,
      "learning_rate": 6.081360966036597e-08,
      "loss": 0.9192,
      "step": 12802
    },
    {
      "epoch": 0.9237040510804083,
      "grad_norm": 5.5045170008566355,
      "learning_rate": 6.069928920894929e-08,
      "loss": 0.9061,
      "step": 12803
    },
    {
      "epoch": 0.9237761985498358,
      "grad_norm": 4.653183509761107,
      "learning_rate": 6.05850746555654e-08,
      "loss": 0.8651,
      "step": 12804
    },
    {
      "epoch": 0.9238483460192634,
      "grad_norm": 3.472025931002209,
      "learning_rate": 6.047096600645108e-08,
      "loss": 0.9741,
      "step": 12805
    },
    {
      "epoch": 0.9239204934886909,
      "grad_norm": 3.2928316171494294,
      "learning_rate": 6.035696326783713e-08,
      "loss": 0.9628,
      "step": 12806
    },
    {
      "epoch": 0.9239926409581184,
      "grad_norm": 7.712345570605576,
      "learning_rate": 6.024306644594923e-08,
      "loss": 1.07,
      "step": 12807
    },
    {
      "epoch": 0.9240647884275459,
      "grad_norm": 3.6284034799556806,
      "learning_rate": 6.012927554700709e-08,
      "loss": 0.9252,
      "step": 12808
    },
    {
      "epoch": 0.9241369358969734,
      "grad_norm": 4.164805848724327,
      "learning_rate": 6.001559057722394e-08,
      "loss": 0.8858,
      "step": 12809
    },
    {
      "epoch": 0.924209083366401,
      "grad_norm": 6.069629999035799,
      "learning_rate": 5.99020115428075e-08,
      "loss": 0.9919,
      "step": 12810
    },
    {
      "epoch": 0.9242812308358285,
      "grad_norm": 4.696377295484051,
      "learning_rate": 5.978853844996079e-08,
      "loss": 1.0018,
      "step": 12811
    },
    {
      "epoch": 0.924353378305256,
      "grad_norm": 4.6435588977044535,
      "learning_rate": 5.967517130487886e-08,
      "loss": 0.9049,
      "step": 12812
    },
    {
      "epoch": 0.9244255257746834,
      "grad_norm": 3.8290889849720107,
      "learning_rate": 5.956191011375322e-08,
      "loss": 0.8916,
      "step": 12813
    },
    {
      "epoch": 0.9244976732441109,
      "grad_norm": 6.306079885647653,
      "learning_rate": 5.9448754882768684e-08,
      "loss": 1.0169,
      "step": 12814
    },
    {
      "epoch": 0.9245698207135384,
      "grad_norm": 5.0009880996925595,
      "learning_rate": 5.933570561810386e-08,
      "loss": 0.8931,
      "step": 12815
    },
    {
      "epoch": 0.924641968182966,
      "grad_norm": 3.602955050401662,
      "learning_rate": 5.922276232593204e-08,
      "loss": 0.9605,
      "step": 12816
    },
    {
      "epoch": 0.9247141156523935,
      "grad_norm": 3.7645919459120765,
      "learning_rate": 5.910992501242051e-08,
      "loss": 0.9817,
      "step": 12817
    },
    {
      "epoch": 0.924786263121821,
      "grad_norm": 4.24018714347831,
      "learning_rate": 5.899719368373146e-08,
      "loss": 0.8898,
      "step": 12818
    },
    {
      "epoch": 0.9248584105912485,
      "grad_norm": 5.306580061245768,
      "learning_rate": 5.888456834601974e-08,
      "loss": 0.893,
      "step": 12819
    },
    {
      "epoch": 0.924930558060676,
      "grad_norm": 3.9114241854637,
      "learning_rate": 5.8772049005436195e-08,
      "loss": 1.0169,
      "step": 12820
    },
    {
      "epoch": 0.9250027055301036,
      "grad_norm": 4.708074151841523,
      "learning_rate": 5.8659635668125265e-08,
      "loss": 1.0063,
      "step": 12821
    },
    {
      "epoch": 0.9250748529995311,
      "grad_norm": 5.500407290550262,
      "learning_rate": 5.8547328340224687e-08,
      "loss": 0.9633,
      "step": 12822
    },
    {
      "epoch": 0.9251470004689586,
      "grad_norm": 3.4636838190645673,
      "learning_rate": 5.843512702786735e-08,
      "loss": 0.9145,
      "step": 12823
    },
    {
      "epoch": 0.9252191479383861,
      "grad_norm": 4.498299595321279,
      "learning_rate": 5.8323031737180337e-08,
      "loss": 1.0174,
      "step": 12824
    },
    {
      "epoch": 0.9252912954078135,
      "grad_norm": 8.564955755989098,
      "learning_rate": 5.8211042474284986e-08,
      "loss": 0.9179,
      "step": 12825
    },
    {
      "epoch": 0.925363442877241,
      "grad_norm": 2.639074558842574,
      "learning_rate": 5.809915924529618e-08,
      "loss": 0.9521,
      "step": 12826
    },
    {
      "epoch": 0.9254355903466686,
      "grad_norm": 7.887179931337126,
      "learning_rate": 5.79873820563237e-08,
      "loss": 0.8525,
      "step": 12827
    },
    {
      "epoch": 0.9255077378160961,
      "grad_norm": 6.658176770078377,
      "learning_rate": 5.787571091347132e-08,
      "loss": 0.8844,
      "step": 12828
    },
    {
      "epoch": 0.9255798852855236,
      "grad_norm": 3.8469557381035377,
      "learning_rate": 5.776414582283662e-08,
      "loss": 0.8564,
      "step": 12829
    },
    {
      "epoch": 0.9256520327549511,
      "grad_norm": 10.121534343097201,
      "learning_rate": 5.765268679051183e-08,
      "loss": 0.9587,
      "step": 12830
    },
    {
      "epoch": 0.9257241802243786,
      "grad_norm": 3.584115290524495,
      "learning_rate": 5.754133382258386e-08,
      "loss": 0.9396,
      "step": 12831
    },
    {
      "epoch": 0.9257963276938062,
      "grad_norm": 0.7947932333741218,
      "learning_rate": 5.743008692513274e-08,
      "loss": 0.8782,
      "step": 12832
    },
    {
      "epoch": 0.9258684751632337,
      "grad_norm": 6.488117801598694,
      "learning_rate": 5.73189461042336e-08,
      "loss": 0.9112,
      "step": 12833
    },
    {
      "epoch": 0.9259406226326612,
      "grad_norm": 5.601798974316842,
      "learning_rate": 5.720791136595515e-08,
      "loss": 0.9808,
      "step": 12834
    },
    {
      "epoch": 0.9260127701020887,
      "grad_norm": 6.7816325136932365,
      "learning_rate": 5.7096982716360763e-08,
      "loss": 0.9659,
      "step": 12835
    },
    {
      "epoch": 0.9260849175715162,
      "grad_norm": 5.038259989391489,
      "learning_rate": 5.698616016150737e-08,
      "loss": 0.9598,
      "step": 12836
    },
    {
      "epoch": 0.9261570650409436,
      "grad_norm": 6.29432857698956,
      "learning_rate": 5.6875443707447235e-08,
      "loss": 0.9812,
      "step": 12837
    },
    {
      "epoch": 0.9262292125103712,
      "grad_norm": 4.204320347638996,
      "learning_rate": 5.6764833360226194e-08,
      "loss": 0.9109,
      "step": 12838
    },
    {
      "epoch": 0.9263013599797987,
      "grad_norm": 4.7344555737780585,
      "learning_rate": 5.665432912588386e-08,
      "loss": 0.9898,
      "step": 12839
    },
    {
      "epoch": 0.9263735074492262,
      "grad_norm": 5.610965258944753,
      "learning_rate": 5.654393101045452e-08,
      "loss": 0.9567,
      "step": 12840
    },
    {
      "epoch": 0.9264456549186537,
      "grad_norm": 5.670954858214466,
      "learning_rate": 5.643363901996667e-08,
      "loss": 0.9466,
      "step": 12841
    },
    {
      "epoch": 0.9265178023880812,
      "grad_norm": 5.235416502319446,
      "learning_rate": 5.632345316044307e-08,
      "loss": 0.9319,
      "step": 12842
    },
    {
      "epoch": 0.9265899498575088,
      "grad_norm": 0.6679321635973572,
      "learning_rate": 5.6213373437900005e-08,
      "loss": 0.7745,
      "step": 12843
    },
    {
      "epoch": 0.9266620973269363,
      "grad_norm": 9.724918284722564,
      "learning_rate": 5.6103399858349335e-08,
      "loss": 0.9536,
      "step": 12844
    },
    {
      "epoch": 0.9267342447963638,
      "grad_norm": 3.0657076483268533,
      "learning_rate": 5.599353242779603e-08,
      "loss": 0.9619,
      "step": 12845
    },
    {
      "epoch": 0.9268063922657913,
      "grad_norm": 2.8812820796877023,
      "learning_rate": 5.58837711522393e-08,
      "loss": 0.8127,
      "step": 12846
    },
    {
      "epoch": 0.9268785397352188,
      "grad_norm": 4.538335272463691,
      "learning_rate": 5.577411603767301e-08,
      "loss": 1.0513,
      "step": 12847
    },
    {
      "epoch": 0.9269506872046464,
      "grad_norm": 6.1348797272300395,
      "learning_rate": 5.5664567090085044e-08,
      "loss": 0.8345,
      "step": 12848
    },
    {
      "epoch": 0.9270228346740738,
      "grad_norm": 15.117220809071831,
      "learning_rate": 5.5555124315457056e-08,
      "loss": 0.7942,
      "step": 12849
    },
    {
      "epoch": 0.9270949821435013,
      "grad_norm": 2.812956539823195,
      "learning_rate": 5.5445787719765824e-08,
      "loss": 0.8695,
      "step": 12850
    },
    {
      "epoch": 0.9271671296129288,
      "grad_norm": 0.6915102088489082,
      "learning_rate": 5.533655730898168e-08,
      "loss": 0.8176,
      "step": 12851
    },
    {
      "epoch": 0.9272392770823563,
      "grad_norm": 3.075134993900738,
      "learning_rate": 5.522743308906941e-08,
      "loss": 0.9023,
      "step": 12852
    },
    {
      "epoch": 0.9273114245517838,
      "grad_norm": 8.07809524973366,
      "learning_rate": 5.511841506598758e-08,
      "loss": 0.8791,
      "step": 12853
    },
    {
      "epoch": 0.9273835720212114,
      "grad_norm": 7.586957994486816,
      "learning_rate": 5.5009503245688985e-08,
      "loss": 0.9159,
      "step": 12854
    },
    {
      "epoch": 0.9274557194906389,
      "grad_norm": 3.469406065886812,
      "learning_rate": 5.4900697634121974e-08,
      "loss": 0.9944,
      "step": 12855
    },
    {
      "epoch": 0.9275278669600664,
      "grad_norm": 3.9643418468885474,
      "learning_rate": 5.479199823722691e-08,
      "loss": 1.0038,
      "step": 12856
    },
    {
      "epoch": 0.9276000144294939,
      "grad_norm": 4.137747286410949,
      "learning_rate": 5.468340506094016e-08,
      "loss": 1.0568,
      "step": 12857
    },
    {
      "epoch": 0.9276721618989214,
      "grad_norm": 4.6207414558891795,
      "learning_rate": 5.457491811119119e-08,
      "loss": 0.8427,
      "step": 12858
    },
    {
      "epoch": 0.927744309368349,
      "grad_norm": 0.6907953052346311,
      "learning_rate": 5.446653739390439e-08,
      "loss": 0.8312,
      "step": 12859
    },
    {
      "epoch": 0.9278164568377764,
      "grad_norm": 5.6401023728298565,
      "learning_rate": 5.4358262914997455e-08,
      "loss": 0.875,
      "step": 12860
    },
    {
      "epoch": 0.9278886043072039,
      "grad_norm": 8.625956606682594,
      "learning_rate": 5.4250094680383665e-08,
      "loss": 1.0039,
      "step": 12861
    },
    {
      "epoch": 0.9279607517766314,
      "grad_norm": 2.776151377005241,
      "learning_rate": 5.41420326959694e-08,
      "loss": 0.8732,
      "step": 12862
    },
    {
      "epoch": 0.9280328992460589,
      "grad_norm": 0.6497813398032474,
      "learning_rate": 5.403407696765505e-08,
      "loss": 0.8388,
      "step": 12863
    },
    {
      "epoch": 0.9281050467154864,
      "grad_norm": 4.322780132052562,
      "learning_rate": 5.392622750133613e-08,
      "loss": 0.9125,
      "step": 12864
    },
    {
      "epoch": 0.928177194184914,
      "grad_norm": 3.0541036296617197,
      "learning_rate": 5.3818484302902146e-08,
      "loss": 0.8279,
      "step": 12865
    },
    {
      "epoch": 0.9282493416543415,
      "grad_norm": 14.508270009950898,
      "learning_rate": 5.371084737823594e-08,
      "loss": 0.8711,
      "step": 12866
    },
    {
      "epoch": 0.928321489123769,
      "grad_norm": 2.772742471781203,
      "learning_rate": 5.3603316733215054e-08,
      "loss": 0.9535,
      "step": 12867
    },
    {
      "epoch": 0.9283936365931965,
      "grad_norm": 13.074543780505666,
      "learning_rate": 5.3495892373712104e-08,
      "loss": 0.8935,
      "step": 12868
    },
    {
      "epoch": 0.928465784062624,
      "grad_norm": 4.275729218690363,
      "learning_rate": 5.338857430559307e-08,
      "loss": 1.0684,
      "step": 12869
    },
    {
      "epoch": 0.9285379315320516,
      "grad_norm": 4.180128983328235,
      "learning_rate": 5.328136253471749e-08,
      "loss": 1.0132,
      "step": 12870
    },
    {
      "epoch": 0.9286100790014791,
      "grad_norm": 4.082144324091793,
      "learning_rate": 5.317425706694023e-08,
      "loss": 0.9648,
      "step": 12871
    },
    {
      "epoch": 0.9286822264709065,
      "grad_norm": 6.762543784967571,
      "learning_rate": 5.306725790811018e-08,
      "loss": 1.0025,
      "step": 12872
    },
    {
      "epoch": 0.928754373940334,
      "grad_norm": 4.471817990838678,
      "learning_rate": 5.2960365064069314e-08,
      "loss": 0.9781,
      "step": 12873
    },
    {
      "epoch": 0.9288265214097615,
      "grad_norm": 7.8267716324811545,
      "learning_rate": 5.285357854065542e-08,
      "loss": 1.0255,
      "step": 12874
    },
    {
      "epoch": 0.928898668879189,
      "grad_norm": 3.764189513816069,
      "learning_rate": 5.274689834369983e-08,
      "loss": 0.9047,
      "step": 12875
    },
    {
      "epoch": 0.9289708163486166,
      "grad_norm": 3.620168952582083,
      "learning_rate": 5.2640324479027444e-08,
      "loss": 0.9415,
      "step": 12876
    },
    {
      "epoch": 0.9290429638180441,
      "grad_norm": 3.82735251592912,
      "learning_rate": 5.253385695245804e-08,
      "loss": 0.8802,
      "step": 12877
    },
    {
      "epoch": 0.9291151112874716,
      "grad_norm": 9.323617737621227,
      "learning_rate": 5.242749576980521e-08,
      "loss": 0.9562,
      "step": 12878
    },
    {
      "epoch": 0.9291872587568991,
      "grad_norm": 3.285892943743624,
      "learning_rate": 5.232124093687762e-08,
      "loss": 0.8492,
      "step": 12879
    },
    {
      "epoch": 0.9292594062263266,
      "grad_norm": 3.498251614524105,
      "learning_rate": 5.2215092459476865e-08,
      "loss": 0.9008,
      "step": 12880
    },
    {
      "epoch": 0.9293315536957542,
      "grad_norm": 4.722147545816205,
      "learning_rate": 5.2109050343399184e-08,
      "loss": 0.9338,
      "step": 12881
    },
    {
      "epoch": 0.9294037011651817,
      "grad_norm": 4.145744412233753,
      "learning_rate": 5.200311459443596e-08,
      "loss": 0.9373,
      "step": 12882
    },
    {
      "epoch": 0.9294758486346092,
      "grad_norm": 0.7468846308514208,
      "learning_rate": 5.1897285218371e-08,
      "loss": 0.8735,
      "step": 12883
    },
    {
      "epoch": 0.9295479961040366,
      "grad_norm": 3.7796353762731774,
      "learning_rate": 5.179156222098346e-08,
      "loss": 0.9932,
      "step": 12884
    },
    {
      "epoch": 0.9296201435734641,
      "grad_norm": 4.740419865937098,
      "learning_rate": 5.168594560804695e-08,
      "loss": 0.9395,
      "step": 12885
    },
    {
      "epoch": 0.9296922910428916,
      "grad_norm": 2.393585167530839,
      "learning_rate": 5.1580435385328634e-08,
      "loss": 0.9967,
      "step": 12886
    },
    {
      "epoch": 0.9297644385123192,
      "grad_norm": 3.269536792016659,
      "learning_rate": 5.1475031558589895e-08,
      "loss": 0.9282,
      "step": 12887
    },
    {
      "epoch": 0.9298365859817467,
      "grad_norm": 3.411993940788515,
      "learning_rate": 5.1369734133586364e-08,
      "loss": 0.9764,
      "step": 12888
    },
    {
      "epoch": 0.9299087334511742,
      "grad_norm": 4.79959158749349,
      "learning_rate": 5.1264543116068315e-08,
      "loss": 0.9996,
      "step": 12889
    },
    {
      "epoch": 0.9299808809206017,
      "grad_norm": 3.830454837494493,
      "learning_rate": 5.115945851177939e-08,
      "loss": 0.8414,
      "step": 12890
    },
    {
      "epoch": 0.9300530283900292,
      "grad_norm": 3.9218737841600455,
      "learning_rate": 5.1054480326458094e-08,
      "loss": 0.9341,
      "step": 12891
    },
    {
      "epoch": 0.9301251758594568,
      "grad_norm": 4.565568493043648,
      "learning_rate": 5.0949608565836965e-08,
      "loss": 0.9224,
      "step": 12892
    },
    {
      "epoch": 0.9301973233288843,
      "grad_norm": 2.805831208446828,
      "learning_rate": 5.084484323564253e-08,
      "loss": 0.9283,
      "step": 12893
    },
    {
      "epoch": 0.9302694707983118,
      "grad_norm": 10.440764322181957,
      "learning_rate": 5.074018434159555e-08,
      "loss": 0.9359,
      "step": 12894
    },
    {
      "epoch": 0.9303416182677393,
      "grad_norm": 2.7936830687559326,
      "learning_rate": 5.0635631889411444e-08,
      "loss": 0.9126,
      "step": 12895
    },
    {
      "epoch": 0.9304137657371667,
      "grad_norm": 0.7888987531813881,
      "learning_rate": 5.053118588479921e-08,
      "loss": 0.8506,
      "step": 12896
    },
    {
      "epoch": 0.9304859132065942,
      "grad_norm": 0.678951017881322,
      "learning_rate": 5.042684633346184e-08,
      "loss": 0.9074,
      "step": 12897
    },
    {
      "epoch": 0.9305580606760218,
      "grad_norm": 3.6857571119902115,
      "learning_rate": 5.032261324109743e-08,
      "loss": 0.987,
      "step": 12898
    },
    {
      "epoch": 0.9306302081454493,
      "grad_norm": 3.5398529045880713,
      "learning_rate": 5.021848661339789e-08,
      "loss": 0.936,
      "step": 12899
    },
    {
      "epoch": 0.9307023556148768,
      "grad_norm": 4.184069238706155,
      "learning_rate": 5.011446645604889e-08,
      "loss": 0.9065,
      "step": 12900
    },
    {
      "epoch": 0.9307745030843043,
      "grad_norm": 3.6512796902126885,
      "learning_rate": 5.001055277473054e-08,
      "loss": 1.0227,
      "step": 12901
    },
    {
      "epoch": 0.9308466505537318,
      "grad_norm": 0.7463904784358812,
      "learning_rate": 4.990674557511698e-08,
      "loss": 0.8419,
      "step": 12902
    },
    {
      "epoch": 0.9309187980231594,
      "grad_norm": 3.9875326173219627,
      "learning_rate": 4.980304486287767e-08,
      "loss": 1.0037,
      "step": 12903
    },
    {
      "epoch": 0.9309909454925869,
      "grad_norm": 5.118102479624979,
      "learning_rate": 4.969945064367431e-08,
      "loss": 0.8786,
      "step": 12904
    },
    {
      "epoch": 0.9310630929620144,
      "grad_norm": 5.10600762416511,
      "learning_rate": 4.959596292316437e-08,
      "loss": 0.9219,
      "step": 12905
    },
    {
      "epoch": 0.9311352404314419,
      "grad_norm": 3.119276833232554,
      "learning_rate": 4.949258170699888e-08,
      "loss": 0.9754,
      "step": 12906
    },
    {
      "epoch": 0.9312073879008694,
      "grad_norm": 3.606521072645566,
      "learning_rate": 4.938930700082267e-08,
      "loss": 1.0023,
      "step": 12907
    },
    {
      "epoch": 0.9312795353702968,
      "grad_norm": 4.874415827253808,
      "learning_rate": 4.9286138810275436e-08,
      "loss": 1.0537,
      "step": 12908
    },
    {
      "epoch": 0.9313516828397244,
      "grad_norm": 27.450196823776373,
      "learning_rate": 4.918307714099113e-08,
      "loss": 1.0272,
      "step": 12909
    },
    {
      "epoch": 0.9314238303091519,
      "grad_norm": 3.496388888554842,
      "learning_rate": 4.908012199859723e-08,
      "loss": 0.9899,
      "step": 12910
    },
    {
      "epoch": 0.9314959777785794,
      "grad_norm": 4.834780158614134,
      "learning_rate": 4.8977273388715715e-08,
      "loss": 1.0277,
      "step": 12911
    },
    {
      "epoch": 0.9315681252480069,
      "grad_norm": 3.597183710102401,
      "learning_rate": 4.887453131696273e-08,
      "loss": 0.8916,
      "step": 12912
    },
    {
      "epoch": 0.9316402727174344,
      "grad_norm": 7.53891652020262,
      "learning_rate": 4.8771895788949136e-08,
      "loss": 0.8301,
      "step": 12913
    },
    {
      "epoch": 0.931712420186862,
      "grad_norm": 3.0631092010570073,
      "learning_rate": 4.866936681027889e-08,
      "loss": 0.8515,
      "step": 12914
    },
    {
      "epoch": 0.9317845676562895,
      "grad_norm": 2.6430027656497406,
      "learning_rate": 4.856694438655085e-08,
      "loss": 0.887,
      "step": 12915
    },
    {
      "epoch": 0.931856715125717,
      "grad_norm": 4.616417008866926,
      "learning_rate": 4.846462852335831e-08,
      "loss": 0.9683,
      "step": 12916
    },
    {
      "epoch": 0.9319288625951445,
      "grad_norm": 3.9110510047233915,
      "learning_rate": 4.836241922628792e-08,
      "loss": 0.9214,
      "step": 12917
    },
    {
      "epoch": 0.932001010064572,
      "grad_norm": 5.5738449131715955,
      "learning_rate": 4.8260316500920996e-08,
      "loss": 0.9997,
      "step": 12918
    },
    {
      "epoch": 0.9320731575339994,
      "grad_norm": 3.3162377340504756,
      "learning_rate": 4.8158320352833516e-08,
      "loss": 0.8905,
      "step": 12919
    },
    {
      "epoch": 0.932145305003427,
      "grad_norm": 4.331756500529873,
      "learning_rate": 4.805643078759436e-08,
      "loss": 1.0169,
      "step": 12920
    },
    {
      "epoch": 0.9322174524728545,
      "grad_norm": 5.914608807949535,
      "learning_rate": 4.79546478107673e-08,
      "loss": 0.9263,
      "step": 12921
    },
    {
      "epoch": 0.932289599942282,
      "grad_norm": 3.523133799792485,
      "learning_rate": 4.7852971427911224e-08,
      "loss": 0.9771,
      "step": 12922
    },
    {
      "epoch": 0.9323617474117095,
      "grad_norm": 3.322277824832465,
      "learning_rate": 4.7751401644577694e-08,
      "loss": 0.9716,
      "step": 12923
    },
    {
      "epoch": 0.932433894881137,
      "grad_norm": 4.030307866682661,
      "learning_rate": 4.764993846631316e-08,
      "loss": 0.8255,
      "step": 12924
    },
    {
      "epoch": 0.9325060423505646,
      "grad_norm": 3.3749816328891136,
      "learning_rate": 4.754858189865807e-08,
      "loss": 1.0244,
      "step": 12925
    },
    {
      "epoch": 0.9325781898199921,
      "grad_norm": 5.631880218770991,
      "learning_rate": 4.744733194714734e-08,
      "loss": 0.9951,
      "step": 12926
    },
    {
      "epoch": 0.9326503372894196,
      "grad_norm": 3.2894613064984797,
      "learning_rate": 4.7346188617309214e-08,
      "loss": 0.9335,
      "step": 12927
    },
    {
      "epoch": 0.9327224847588471,
      "grad_norm": 5.175374947201044,
      "learning_rate": 4.724515191466749e-08,
      "loss": 1.036,
      "step": 12928
    },
    {
      "epoch": 0.9327946322282746,
      "grad_norm": 3.5122623892677,
      "learning_rate": 4.714422184473932e-08,
      "loss": 0.9736,
      "step": 12929
    },
    {
      "epoch": 0.9328667796977022,
      "grad_norm": 2.573629355212968,
      "learning_rate": 4.704339841303606e-08,
      "loss": 0.9646,
      "step": 12930
    },
    {
      "epoch": 0.9329389271671296,
      "grad_norm": 17.681167601268324,
      "learning_rate": 4.69426816250631e-08,
      "loss": 0.8989,
      "step": 12931
    },
    {
      "epoch": 0.9330110746365571,
      "grad_norm": 2.9078223784297883,
      "learning_rate": 4.6842071486320026e-08,
      "loss": 1.0714,
      "step": 12932
    },
    {
      "epoch": 0.9330832221059846,
      "grad_norm": 3.0201191200359205,
      "learning_rate": 4.674156800230156e-08,
      "loss": 0.9716,
      "step": 12933
    },
    {
      "epoch": 0.9331553695754121,
      "grad_norm": 3.1889094808034004,
      "learning_rate": 4.6641171178495086e-08,
      "loss": 0.9853,
      "step": 12934
    },
    {
      "epoch": 0.9332275170448396,
      "grad_norm": 4.663530840196185,
      "learning_rate": 4.6540881020383336e-08,
      "loss": 0.874,
      "step": 12935
    },
    {
      "epoch": 0.9332996645142672,
      "grad_norm": 7.290273159021841,
      "learning_rate": 4.6440697533442594e-08,
      "loss": 0.9361,
      "step": 12936
    },
    {
      "epoch": 0.9333718119836947,
      "grad_norm": 5.783956687965576,
      "learning_rate": 4.6340620723143376e-08,
      "loss": 0.9169,
      "step": 12937
    },
    {
      "epoch": 0.9334439594531222,
      "grad_norm": 2.895786605201808,
      "learning_rate": 4.624065059495086e-08,
      "loss": 0.9281,
      "step": 12938
    },
    {
      "epoch": 0.9335161069225497,
      "grad_norm": 5.051771687861526,
      "learning_rate": 4.6140787154323346e-08,
      "loss": 1.0103,
      "step": 12939
    },
    {
      "epoch": 0.9335882543919772,
      "grad_norm": 0.6870067083830823,
      "learning_rate": 4.604103040671514e-08,
      "loss": 0.8103,
      "step": 12940
    },
    {
      "epoch": 0.9336604018614048,
      "grad_norm": 3.2736000621553267,
      "learning_rate": 4.594138035757256e-08,
      "loss": 0.8582,
      "step": 12941
    },
    {
      "epoch": 0.9337325493308323,
      "grad_norm": 5.071316800334551,
      "learning_rate": 4.584183701233768e-08,
      "loss": 0.9261,
      "step": 12942
    },
    {
      "epoch": 0.9338046968002597,
      "grad_norm": 0.9998064151785504,
      "learning_rate": 4.5742400376446167e-08,
      "loss": 0.8569,
      "step": 12943
    },
    {
      "epoch": 0.9338768442696872,
      "grad_norm": 3.5903244277871513,
      "learning_rate": 4.564307045532767e-08,
      "loss": 0.9038,
      "step": 12944
    },
    {
      "epoch": 0.9339489917391147,
      "grad_norm": 2.3998572903165507,
      "learning_rate": 4.554384725440585e-08,
      "loss": 1.0056,
      "step": 12945
    },
    {
      "epoch": 0.9340211392085422,
      "grad_norm": 3.62741968361757,
      "learning_rate": 4.544473077909994e-08,
      "loss": 0.8603,
      "step": 12946
    },
    {
      "epoch": 0.9340932866779698,
      "grad_norm": 2.9092314818945884,
      "learning_rate": 4.534572103482159e-08,
      "loss": 1.01,
      "step": 12947
    },
    {
      "epoch": 0.9341654341473973,
      "grad_norm": 7.092983120373151,
      "learning_rate": 4.5246818026977605e-08,
      "loss": 0.9645,
      "step": 12948
    },
    {
      "epoch": 0.9342375816168248,
      "grad_norm": 3.171613861007894,
      "learning_rate": 4.514802176096855e-08,
      "loss": 1.0123,
      "step": 12949
    },
    {
      "epoch": 0.9343097290862523,
      "grad_norm": 4.197879737458082,
      "learning_rate": 4.504933224218965e-08,
      "loss": 0.9045,
      "step": 12950
    },
    {
      "epoch": 0.9343818765556798,
      "grad_norm": 3.790060576505867,
      "learning_rate": 4.495074947602928e-08,
      "loss": 0.9774,
      "step": 12951
    },
    {
      "epoch": 0.9344540240251074,
      "grad_norm": 4.737324717960385,
      "learning_rate": 4.485227346787135e-08,
      "loss": 0.8807,
      "step": 12952
    },
    {
      "epoch": 0.9345261714945349,
      "grad_norm": 4.506930841870476,
      "learning_rate": 4.475390422309333e-08,
      "loss": 0.772,
      "step": 12953
    },
    {
      "epoch": 0.9345983189639624,
      "grad_norm": 5.528807192253697,
      "learning_rate": 4.4655641747066265e-08,
      "loss": 0.9731,
      "step": 12954
    },
    {
      "epoch": 0.9346704664333898,
      "grad_norm": 8.048036361885213,
      "learning_rate": 4.455748604515608e-08,
      "loss": 0.9896,
      "step": 12955
    },
    {
      "epoch": 0.9347426139028173,
      "grad_norm": 3.728719730695527,
      "learning_rate": 4.445943712272271e-08,
      "loss": 1.0064,
      "step": 12956
    },
    {
      "epoch": 0.9348147613722448,
      "grad_norm": 5.783887766592053,
      "learning_rate": 4.436149498512054e-08,
      "loss": 0.9957,
      "step": 12957
    },
    {
      "epoch": 0.9348869088416724,
      "grad_norm": 21.830413545473807,
      "learning_rate": 4.4263659637697515e-08,
      "loss": 0.929,
      "step": 12958
    },
    {
      "epoch": 0.9349590563110999,
      "grad_norm": 0.827878663474399,
      "learning_rate": 4.4165931085796246e-08,
      "loss": 0.8555,
      "step": 12959
    },
    {
      "epoch": 0.9350312037805274,
      "grad_norm": 3.7742565206044483,
      "learning_rate": 4.406830933475336e-08,
      "loss": 0.8731,
      "step": 12960
    },
    {
      "epoch": 0.9351033512499549,
      "grad_norm": 5.157823172358601,
      "learning_rate": 4.3970794389899256e-08,
      "loss": 0.9437,
      "step": 12961
    },
    {
      "epoch": 0.9351754987193824,
      "grad_norm": 4.3193665170675954,
      "learning_rate": 4.387338625655923e-08,
      "loss": 1.0465,
      "step": 12962
    },
    {
      "epoch": 0.93524764618881,
      "grad_norm": 3.7340111974253047,
      "learning_rate": 4.3776084940052135e-08,
      "loss": 0.8455,
      "step": 12963
    },
    {
      "epoch": 0.9353197936582375,
      "grad_norm": 0.835286252187028,
      "learning_rate": 4.367889044569129e-08,
      "loss": 0.8137,
      "step": 12964
    },
    {
      "epoch": 0.935391941127665,
      "grad_norm": 0.5948523277263464,
      "learning_rate": 4.35818027787842e-08,
      "loss": 0.7706,
      "step": 12965
    },
    {
      "epoch": 0.9354640885970924,
      "grad_norm": 2.5422670770439773,
      "learning_rate": 4.348482194463243e-08,
      "loss": 0.9089,
      "step": 12966
    },
    {
      "epoch": 0.9355362360665199,
      "grad_norm": 3.65954259004917,
      "learning_rate": 4.3387947948532175e-08,
      "loss": 0.9585,
      "step": 12967
    },
    {
      "epoch": 0.9356083835359474,
      "grad_norm": 4.138790083921224,
      "learning_rate": 4.329118079577254e-08,
      "loss": 0.8898,
      "step": 12968
    },
    {
      "epoch": 0.935680531005375,
      "grad_norm": 10.483260116766218,
      "learning_rate": 4.3194520491637963e-08,
      "loss": 1.0442,
      "step": 12969
    },
    {
      "epoch": 0.9357526784748025,
      "grad_norm": 3.4905373858570017,
      "learning_rate": 4.309796704140711e-08,
      "loss": 0.9911,
      "step": 12970
    },
    {
      "epoch": 0.93582482594423,
      "grad_norm": 3.911912887445839,
      "learning_rate": 4.300152045035177e-08,
      "loss": 0.969,
      "step": 12971
    },
    {
      "epoch": 0.9358969734136575,
      "grad_norm": 3.385076949852233,
      "learning_rate": 4.2905180723739054e-08,
      "loss": 1.0021,
      "step": 12972
    },
    {
      "epoch": 0.935969120883085,
      "grad_norm": 5.07194128457389,
      "learning_rate": 4.28089478668292e-08,
      "loss": 0.9475,
      "step": 12973
    },
    {
      "epoch": 0.9360412683525126,
      "grad_norm": 8.78913107612136,
      "learning_rate": 4.2712821884878015e-08,
      "loss": 0.9535,
      "step": 12974
    },
    {
      "epoch": 0.9361134158219401,
      "grad_norm": 3.443398703157427,
      "learning_rate": 4.2616802783133285e-08,
      "loss": 0.9337,
      "step": 12975
    },
    {
      "epoch": 0.9361855632913676,
      "grad_norm": 3.090544147417446,
      "learning_rate": 4.252089056683927e-08,
      "loss": 0.9421,
      "step": 12976
    },
    {
      "epoch": 0.9362577107607951,
      "grad_norm": 3.1145608108029377,
      "learning_rate": 4.242508524123334e-08,
      "loss": 1.0487,
      "step": 12977
    },
    {
      "epoch": 0.9363298582302225,
      "grad_norm": 3.74725941013603,
      "learning_rate": 4.232938681154663e-08,
      "loss": 0.9735,
      "step": 12978
    },
    {
      "epoch": 0.93640200569965,
      "grad_norm": 3.47384091264127,
      "learning_rate": 4.223379528300497e-08,
      "loss": 1.0166,
      "step": 12979
    },
    {
      "epoch": 0.9364741531690776,
      "grad_norm": 3.674010090273653,
      "learning_rate": 4.2138310660828625e-08,
      "loss": 0.978,
      "step": 12980
    },
    {
      "epoch": 0.9365463006385051,
      "grad_norm": 9.414593718803628,
      "learning_rate": 4.2042932950230756e-08,
      "loss": 1.0179,
      "step": 12981
    },
    {
      "epoch": 0.9366184481079326,
      "grad_norm": 0.6352560408348913,
      "learning_rate": 4.194766215642076e-08,
      "loss": 0.7738,
      "step": 12982
    },
    {
      "epoch": 0.9366905955773601,
      "grad_norm": 21.163570067407033,
      "learning_rate": 4.185249828460025e-08,
      "loss": 0.9994,
      "step": 12983
    },
    {
      "epoch": 0.9367627430467876,
      "grad_norm": 2.944937198921366,
      "learning_rate": 4.175744133996617e-08,
      "loss": 0.9429,
      "step": 12984
    },
    {
      "epoch": 0.9368348905162152,
      "grad_norm": 4.294904333254878,
      "learning_rate": 4.166249132770905e-08,
      "loss": 0.9746,
      "step": 12985
    },
    {
      "epoch": 0.9369070379856427,
      "grad_norm": 4.2092261295855025,
      "learning_rate": 4.1567648253013624e-08,
      "loss": 0.9421,
      "step": 12986
    },
    {
      "epoch": 0.9369791854550702,
      "grad_norm": 5.1125416872894505,
      "learning_rate": 4.1472912121059076e-08,
      "loss": 0.9478,
      "step": 12987
    },
    {
      "epoch": 0.9370513329244977,
      "grad_norm": 8.722736281400653,
      "learning_rate": 4.137828293701839e-08,
      "loss": 0.9951,
      "step": 12988
    },
    {
      "epoch": 0.9371234803939252,
      "grad_norm": 0.8048347782796924,
      "learning_rate": 4.128376070605943e-08,
      "loss": 0.8316,
      "step": 12989
    },
    {
      "epoch": 0.9371956278633526,
      "grad_norm": 2.449223519887949,
      "learning_rate": 4.118934543334318e-08,
      "loss": 0.913,
      "step": 12990
    },
    {
      "epoch": 0.9372677753327802,
      "grad_norm": 4.457758477248346,
      "learning_rate": 4.1095037124025734e-08,
      "loss": 0.99,
      "step": 12991
    },
    {
      "epoch": 0.9373399228022077,
      "grad_norm": 4.361892246879635,
      "learning_rate": 4.100083578325653e-08,
      "loss": 0.9991,
      "step": 12992
    },
    {
      "epoch": 0.9374120702716352,
      "grad_norm": 3.515791825468934,
      "learning_rate": 4.0906741416179445e-08,
      "loss": 0.9153,
      "step": 12993
    },
    {
      "epoch": 0.9374842177410627,
      "grad_norm": 13.265640528369886,
      "learning_rate": 4.0812754027933496e-08,
      "loss": 0.9563,
      "step": 12994
    },
    {
      "epoch": 0.9375563652104902,
      "grad_norm": 4.289462489531372,
      "learning_rate": 4.0718873623649896e-08,
      "loss": 0.8182,
      "step": 12995
    },
    {
      "epoch": 0.9376285126799178,
      "grad_norm": 3.4502416111473946,
      "learning_rate": 4.062510020845589e-08,
      "loss": 1.0009,
      "step": 12996
    },
    {
      "epoch": 0.9377006601493453,
      "grad_norm": 3.7979130836814488,
      "learning_rate": 4.053143378747182e-08,
      "loss": 0.9085,
      "step": 12997
    },
    {
      "epoch": 0.9377728076187728,
      "grad_norm": 5.725485248906788,
      "learning_rate": 4.0437874365812256e-08,
      "loss": 1.0107,
      "step": 12998
    },
    {
      "epoch": 0.9378449550882003,
      "grad_norm": 2.7851557363801245,
      "learning_rate": 4.034442194858623e-08,
      "loss": 0.873,
      "step": 12999
    },
    {
      "epoch": 0.9379171025576278,
      "grad_norm": 2.462214450449026,
      "learning_rate": 4.025107654089699e-08,
      "loss": 0.9695,
      "step": 13000
    },
    {
      "epoch": 0.9379892500270554,
      "grad_norm": 4.307458335422568,
      "learning_rate": 4.0157838147842013e-08,
      "loss": 1.0008,
      "step": 13001
    },
    {
      "epoch": 0.9380613974964828,
      "grad_norm": 7.699670848994867,
      "learning_rate": 4.0064706774512124e-08,
      "loss": 0.9476,
      "step": 13002
    },
    {
      "epoch": 0.9381335449659103,
      "grad_norm": 4.648641014452504,
      "learning_rate": 3.9971682425993244e-08,
      "loss": 0.8096,
      "step": 13003
    },
    {
      "epoch": 0.9382056924353378,
      "grad_norm": 4.136796440857444,
      "learning_rate": 3.9878765107365096e-08,
      "loss": 0.9492,
      "step": 13004
    },
    {
      "epoch": 0.9382778399047653,
      "grad_norm": 5.357216269580463,
      "learning_rate": 3.9785954823701175e-08,
      "loss": 0.9967,
      "step": 13005
    },
    {
      "epoch": 0.9383499873741928,
      "grad_norm": 3.363747665194662,
      "learning_rate": 3.9693251580070087e-08,
      "loss": 0.9971,
      "step": 13006
    },
    {
      "epoch": 0.9384221348436204,
      "grad_norm": 3.3210985397702757,
      "learning_rate": 3.9600655381533786e-08,
      "loss": 0.9669,
      "step": 13007
    },
    {
      "epoch": 0.9384942823130479,
      "grad_norm": 3.6932495036291724,
      "learning_rate": 3.950816623314823e-08,
      "loss": 0.8931,
      "step": 13008
    },
    {
      "epoch": 0.9385664297824754,
      "grad_norm": 4.267692986869578,
      "learning_rate": 3.941578413996449e-08,
      "loss": 0.9178,
      "step": 13009
    },
    {
      "epoch": 0.9386385772519029,
      "grad_norm": 2.7224395096489307,
      "learning_rate": 3.9323509107026756e-08,
      "loss": 0.9276,
      "step": 13010
    },
    {
      "epoch": 0.9387107247213304,
      "grad_norm": 3.92531492585638,
      "learning_rate": 3.92313411393741e-08,
      "loss": 1.0207,
      "step": 13011
    },
    {
      "epoch": 0.938782872190758,
      "grad_norm": 3.9623663317159745,
      "learning_rate": 3.913928024203939e-08,
      "loss": 0.9026,
      "step": 13012
    },
    {
      "epoch": 0.9388550196601854,
      "grad_norm": 6.94081407724855,
      "learning_rate": 3.904732642004971e-08,
      "loss": 1.0323,
      "step": 13013
    },
    {
      "epoch": 0.9389271671296129,
      "grad_norm": 4.629747377099191,
      "learning_rate": 3.8955479678426385e-08,
      "loss": 1.0362,
      "step": 13014
    },
    {
      "epoch": 0.9389993145990404,
      "grad_norm": 3.129519894130924,
      "learning_rate": 3.886374002218473e-08,
      "loss": 0.927,
      "step": 13015
    },
    {
      "epoch": 0.9390714620684679,
      "grad_norm": 5.04363342679867,
      "learning_rate": 3.8772107456334304e-08,
      "loss": 0.9913,
      "step": 13016
    },
    {
      "epoch": 0.9391436095378954,
      "grad_norm": 6.918611122949856,
      "learning_rate": 3.868058198587887e-08,
      "loss": 0.9815,
      "step": 13017
    },
    {
      "epoch": 0.939215757007323,
      "grad_norm": 4.7114296018759125,
      "learning_rate": 3.8589163615816436e-08,
      "loss": 0.9075,
      "step": 13018
    },
    {
      "epoch": 0.9392879044767505,
      "grad_norm": 3.6047022574080914,
      "learning_rate": 3.849785235113878e-08,
      "loss": 0.8548,
      "step": 13019
    },
    {
      "epoch": 0.939360051946178,
      "grad_norm": 6.36392536497918,
      "learning_rate": 3.840664819683215e-08,
      "loss": 0.8999,
      "step": 13020
    },
    {
      "epoch": 0.9394321994156055,
      "grad_norm": 5.091562187425668,
      "learning_rate": 3.831555115787721e-08,
      "loss": 0.9198,
      "step": 13021
    },
    {
      "epoch": 0.939504346885033,
      "grad_norm": 3.356196687583955,
      "learning_rate": 3.8224561239247773e-08,
      "loss": 0.965,
      "step": 13022
    },
    {
      "epoch": 0.9395764943544606,
      "grad_norm": 4.80353354886314,
      "learning_rate": 3.813367844591275e-08,
      "loss": 0.8894,
      "step": 13023
    },
    {
      "epoch": 0.9396486418238881,
      "grad_norm": 3.2799084055137224,
      "learning_rate": 3.8042902782835063e-08,
      "loss": 0.9701,
      "step": 13024
    },
    {
      "epoch": 0.9397207892933155,
      "grad_norm": 3.7995383885740064,
      "learning_rate": 3.7952234254971624e-08,
      "loss": 0.9816,
      "step": 13025
    },
    {
      "epoch": 0.939792936762743,
      "grad_norm": 4.255396838392075,
      "learning_rate": 3.7861672867273375e-08,
      "loss": 0.9867,
      "step": 13026
    },
    {
      "epoch": 0.9398650842321705,
      "grad_norm": 4.035586840983979,
      "learning_rate": 3.7771218624685466e-08,
      "loss": 0.9677,
      "step": 13027
    },
    {
      "epoch": 0.939937231701598,
      "grad_norm": 4.513075376057394,
      "learning_rate": 3.7680871532147715e-08,
      "loss": 0.9673,
      "step": 13028
    },
    {
      "epoch": 0.9400093791710256,
      "grad_norm": 3.441735034352862,
      "learning_rate": 3.7590631594592857e-08,
      "loss": 0.9175,
      "step": 13029
    },
    {
      "epoch": 0.9400815266404531,
      "grad_norm": 3.363483843754821,
      "learning_rate": 3.750049881694917e-08,
      "loss": 0.993,
      "step": 13030
    },
    {
      "epoch": 0.9401536741098806,
      "grad_norm": 2.6106279300365487,
      "learning_rate": 3.74104732041387e-08,
      "loss": 0.9711,
      "step": 13031
    },
    {
      "epoch": 0.9402258215793081,
      "grad_norm": 4.078591740232907,
      "learning_rate": 3.732055476107665e-08,
      "loss": 0.9525,
      "step": 13032
    },
    {
      "epoch": 0.9402979690487356,
      "grad_norm": 2.8502235927210022,
      "learning_rate": 3.723074349267352e-08,
      "loss": 0.9757,
      "step": 13033
    },
    {
      "epoch": 0.9403701165181632,
      "grad_norm": 3.4287166394409296,
      "learning_rate": 3.7141039403833616e-08,
      "loss": 0.9505,
      "step": 13034
    },
    {
      "epoch": 0.9404422639875907,
      "grad_norm": 5.715201522464699,
      "learning_rate": 3.705144249945524e-08,
      "loss": 1.0085,
      "step": 13035
    },
    {
      "epoch": 0.9405144114570182,
      "grad_norm": 6.546916078424975,
      "learning_rate": 3.696195278443115e-08,
      "loss": 0.9418,
      "step": 13036
    },
    {
      "epoch": 0.9405865589264456,
      "grad_norm": 5.80632396441469,
      "learning_rate": 3.6872570263647876e-08,
      "loss": 0.9691,
      "step": 13037
    },
    {
      "epoch": 0.9406587063958731,
      "grad_norm": 4.2292460667465095,
      "learning_rate": 3.67832949419864e-08,
      "loss": 0.9211,
      "step": 13038
    },
    {
      "epoch": 0.9407308538653006,
      "grad_norm": 4.773294449830762,
      "learning_rate": 3.669412682432127e-08,
      "loss": 0.9076,
      "step": 13039
    },
    {
      "epoch": 0.9408030013347282,
      "grad_norm": 3.6141772009974775,
      "learning_rate": 3.6605065915522145e-08,
      "loss": 1.0924,
      "step": 13040
    },
    {
      "epoch": 0.9408751488041557,
      "grad_norm": 3.956789512066298,
      "learning_rate": 3.651611222045225e-08,
      "loss": 0.9517,
      "step": 13041
    },
    {
      "epoch": 0.9409472962735832,
      "grad_norm": 3.700740384437332,
      "learning_rate": 3.642726574396882e-08,
      "loss": 0.9016,
      "step": 13042
    },
    {
      "epoch": 0.9410194437430107,
      "grad_norm": 5.114772360650035,
      "learning_rate": 3.633852649092328e-08,
      "loss": 1.0103,
      "step": 13043
    },
    {
      "epoch": 0.9410915912124382,
      "grad_norm": 9.4327946737139,
      "learning_rate": 3.624989446616178e-08,
      "loss": 0.9987,
      "step": 13044
    },
    {
      "epoch": 0.9411637386818658,
      "grad_norm": 0.8010298180334251,
      "learning_rate": 3.616136967452421e-08,
      "loss": 0.8498,
      "step": 13045
    },
    {
      "epoch": 0.9412358861512933,
      "grad_norm": 3.813177330009405,
      "learning_rate": 3.6072952120844046e-08,
      "loss": 0.9376,
      "step": 13046
    },
    {
      "epoch": 0.9413080336207208,
      "grad_norm": 9.035751540862726,
      "learning_rate": 3.598464180994964e-08,
      "loss": 0.8505,
      "step": 13047
    },
    {
      "epoch": 0.9413801810901483,
      "grad_norm": 4.820218127051763,
      "learning_rate": 3.589643874666382e-08,
      "loss": 0.8433,
      "step": 13048
    },
    {
      "epoch": 0.9414523285595757,
      "grad_norm": 2.552329278629917,
      "learning_rate": 3.580834293580226e-08,
      "loss": 0.9647,
      "step": 13049
    },
    {
      "epoch": 0.9415244760290032,
      "grad_norm": 0.7163011879463362,
      "learning_rate": 3.5720354382176023e-08,
      "loss": 0.8345,
      "step": 13050
    },
    {
      "epoch": 0.9415966234984308,
      "grad_norm": 7.502215503254715,
      "learning_rate": 3.563247309058992e-08,
      "loss": 0.9033,
      "step": 13051
    },
    {
      "epoch": 0.9416687709678583,
      "grad_norm": 6.150572661743198,
      "learning_rate": 3.554469906584256e-08,
      "loss": 0.9191,
      "step": 13052
    },
    {
      "epoch": 0.9417409184372858,
      "grad_norm": 0.7350140084339527,
      "learning_rate": 3.545703231272679e-08,
      "loss": 0.8385,
      "step": 13053
    },
    {
      "epoch": 0.9418130659067133,
      "grad_norm": 35.361187398314975,
      "learning_rate": 3.536947283603009e-08,
      "loss": 0.9046,
      "step": 13054
    },
    {
      "epoch": 0.9418852133761408,
      "grad_norm": 5.7958970388846245,
      "learning_rate": 3.5282020640533764e-08,
      "loss": 1.1614,
      "step": 13055
    },
    {
      "epoch": 0.9419573608455684,
      "grad_norm": 9.591671796234465,
      "learning_rate": 3.519467573101309e-08,
      "loss": 1.0436,
      "step": 13056
    },
    {
      "epoch": 0.9420295083149959,
      "grad_norm": 5.526269426844494,
      "learning_rate": 3.5107438112237595e-08,
      "loss": 1.0364,
      "step": 13057
    },
    {
      "epoch": 0.9421016557844234,
      "grad_norm": 2.595844755765012,
      "learning_rate": 3.502030778897147e-08,
      "loss": 0.9045,
      "step": 13058
    },
    {
      "epoch": 0.9421738032538509,
      "grad_norm": 5.2048740139389125,
      "learning_rate": 3.493328476597179e-08,
      "loss": 0.9604,
      "step": 13059
    },
    {
      "epoch": 0.9422459507232784,
      "grad_norm": 4.877647854564703,
      "learning_rate": 3.484636904799121e-08,
      "loss": 1.0077,
      "step": 13060
    },
    {
      "epoch": 0.9423180981927058,
      "grad_norm": 11.22900873661746,
      "learning_rate": 3.475956063977547e-08,
      "loss": 0.9324,
      "step": 13061
    },
    {
      "epoch": 0.9423902456621334,
      "grad_norm": 3.604340315735086,
      "learning_rate": 3.467285954606547e-08,
      "loss": 0.9312,
      "step": 13062
    },
    {
      "epoch": 0.9424623931315609,
      "grad_norm": 4.2771000483515,
      "learning_rate": 3.458626577159496e-08,
      "loss": 1.0544,
      "step": 13063
    },
    {
      "epoch": 0.9425345406009884,
      "grad_norm": 15.836116720125471,
      "learning_rate": 3.449977932109282e-08,
      "loss": 0.9886,
      "step": 13064
    },
    {
      "epoch": 0.9426066880704159,
      "grad_norm": 2.5109681807081525,
      "learning_rate": 3.441340019928174e-08,
      "loss": 1.0065,
      "step": 13065
    },
    {
      "epoch": 0.9426788355398434,
      "grad_norm": 0.6791812721585433,
      "learning_rate": 3.432712841087837e-08,
      "loss": 0.8297,
      "step": 13066
    },
    {
      "epoch": 0.942750983009271,
      "grad_norm": 2.809691404319361,
      "learning_rate": 3.424096396059384e-08,
      "loss": 0.8949,
      "step": 13067
    },
    {
      "epoch": 0.9428231304786985,
      "grad_norm": 4.034277437125175,
      "learning_rate": 3.415490685313349e-08,
      "loss": 0.9078,
      "step": 13068
    },
    {
      "epoch": 0.942895277948126,
      "grad_norm": 4.1154480359519905,
      "learning_rate": 3.406895709319624e-08,
      "loss": 0.9641,
      "step": 13069
    },
    {
      "epoch": 0.9429674254175535,
      "grad_norm": 3.898333530912575,
      "learning_rate": 3.398311468547566e-08,
      "loss": 0.9879,
      "step": 13070
    },
    {
      "epoch": 0.943039572886981,
      "grad_norm": 3.478447632336615,
      "learning_rate": 3.3897379634659104e-08,
      "loss": 1.0773,
      "step": 13071
    },
    {
      "epoch": 0.9431117203564084,
      "grad_norm": 3.2202587850607953,
      "learning_rate": 3.3811751945428624e-08,
      "loss": 0.9071,
      "step": 13072
    },
    {
      "epoch": 0.943183867825836,
      "grad_norm": 9.454224577707189,
      "learning_rate": 3.3726231622459576e-08,
      "loss": 0.918,
      "step": 13073
    },
    {
      "epoch": 0.9432560152952635,
      "grad_norm": 6.523190430856422,
      "learning_rate": 3.3640818670422235e-08,
      "loss": 0.9789,
      "step": 13074
    },
    {
      "epoch": 0.943328162764691,
      "grad_norm": 2.855495134432653,
      "learning_rate": 3.3555513093980635e-08,
      "loss": 0.8784,
      "step": 13075
    },
    {
      "epoch": 0.9434003102341185,
      "grad_norm": 6.161262446751001,
      "learning_rate": 3.347031489779284e-08,
      "loss": 0.9707,
      "step": 13076
    },
    {
      "epoch": 0.943472457703546,
      "grad_norm": 3.8575540177128764,
      "learning_rate": 3.3385224086511124e-08,
      "loss": 0.8674,
      "step": 13077
    },
    {
      "epoch": 0.9435446051729736,
      "grad_norm": 3.9628067571753176,
      "learning_rate": 3.3300240664782206e-08,
      "loss": 0.8782,
      "step": 13078
    },
    {
      "epoch": 0.9436167526424011,
      "grad_norm": 0.7621391505647827,
      "learning_rate": 3.3215364637247053e-08,
      "loss": 0.9465,
      "step": 13079
    },
    {
      "epoch": 0.9436889001118286,
      "grad_norm": 3.518110964596812,
      "learning_rate": 3.313059600853974e-08,
      "loss": 0.8718,
      "step": 13080
    },
    {
      "epoch": 0.9437610475812561,
      "grad_norm": 3.8603184585445014,
      "learning_rate": 3.304593478328943e-08,
      "loss": 0.9558,
      "step": 13081
    },
    {
      "epoch": 0.9438331950506836,
      "grad_norm": 5.115926199815436,
      "learning_rate": 3.2961380966119554e-08,
      "loss": 0.9055,
      "step": 13082
    },
    {
      "epoch": 0.9439053425201112,
      "grad_norm": 6.235514218830889,
      "learning_rate": 3.287693456164642e-08,
      "loss": 0.9326,
      "step": 13083
    },
    {
      "epoch": 0.9439774899895386,
      "grad_norm": 4.690652829671565,
      "learning_rate": 3.2792595574482105e-08,
      "loss": 0.9367,
      "step": 13084
    },
    {
      "epoch": 0.9440496374589661,
      "grad_norm": 2.905825840340761,
      "learning_rate": 3.2708364009232045e-08,
      "loss": 0.857,
      "step": 13085
    },
    {
      "epoch": 0.9441217849283936,
      "grad_norm": 4.9330882389598765,
      "learning_rate": 3.262423987049523e-08,
      "loss": 0.9894,
      "step": 13086
    },
    {
      "epoch": 0.9441939323978211,
      "grad_norm": 11.30102935423725,
      "learning_rate": 3.2540223162865755e-08,
      "loss": 0.9782,
      "step": 13087
    },
    {
      "epoch": 0.9442660798672486,
      "grad_norm": 0.7406874103245122,
      "learning_rate": 3.245631389093151e-08,
      "loss": 0.7836,
      "step": 13088
    },
    {
      "epoch": 0.9443382273366762,
      "grad_norm": 4.068291629221719,
      "learning_rate": 3.2372512059274384e-08,
      "loss": 0.9716,
      "step": 13089
    },
    {
      "epoch": 0.9444103748061037,
      "grad_norm": 4.4153037697479425,
      "learning_rate": 3.228881767247027e-08,
      "loss": 0.9964,
      "step": 13090
    },
    {
      "epoch": 0.9444825222755312,
      "grad_norm": 3.7261733645672015,
      "learning_rate": 3.220523073508974e-08,
      "loss": 0.9419,
      "step": 13091
    },
    {
      "epoch": 0.9445546697449587,
      "grad_norm": 6.934916522150029,
      "learning_rate": 3.2121751251697135e-08,
      "loss": 0.9414,
      "step": 13092
    },
    {
      "epoch": 0.9446268172143862,
      "grad_norm": 5.633422310632467,
      "learning_rate": 3.203837922685082e-08,
      "loss": 0.9714,
      "step": 13093
    },
    {
      "epoch": 0.9446989646838138,
      "grad_norm": 3.472249920204739,
      "learning_rate": 3.1955114665103367e-08,
      "loss": 0.9571,
      "step": 13094
    },
    {
      "epoch": 0.9447711121532413,
      "grad_norm": 3.947886739464748,
      "learning_rate": 3.187195757100181e-08,
      "loss": 0.9384,
      "step": 13095
    },
    {
      "epoch": 0.9448432596226687,
      "grad_norm": 8.334629975213685,
      "learning_rate": 3.178890794908651e-08,
      "loss": 0.9295,
      "step": 13096
    },
    {
      "epoch": 0.9449154070920962,
      "grad_norm": 4.294330967408264,
      "learning_rate": 3.1705965803893174e-08,
      "loss": 0.8229,
      "step": 13097
    },
    {
      "epoch": 0.9449875545615237,
      "grad_norm": 3.0482528309927432,
      "learning_rate": 3.162313113995063e-08,
      "loss": 0.9343,
      "step": 13098
    },
    {
      "epoch": 0.9450597020309512,
      "grad_norm": 0.6791457505506101,
      "learning_rate": 3.1540403961782144e-08,
      "loss": 0.8448,
      "step": 13099
    },
    {
      "epoch": 0.9451318495003788,
      "grad_norm": 3.424692605139513,
      "learning_rate": 3.145778427390544e-08,
      "loss": 0.9499,
      "step": 13100
    },
    {
      "epoch": 0.9452039969698063,
      "grad_norm": 3.7592918989298854,
      "learning_rate": 3.137527208083135e-08,
      "loss": 0.9562,
      "step": 13101
    },
    {
      "epoch": 0.9452761444392338,
      "grad_norm": 2.9405923353077528,
      "learning_rate": 3.129286738706649e-08,
      "loss": 0.9042,
      "step": 13102
    },
    {
      "epoch": 0.9453482919086613,
      "grad_norm": 4.393783139447059,
      "learning_rate": 3.1210570197110155e-08,
      "loss": 0.9114,
      "step": 13103
    },
    {
      "epoch": 0.9454204393780888,
      "grad_norm": 10.900670064226377,
      "learning_rate": 3.1128380515456296e-08,
      "loss": 0.9036,
      "step": 13104
    },
    {
      "epoch": 0.9454925868475164,
      "grad_norm": 7.125372358262863,
      "learning_rate": 3.104629834659311e-08,
      "loss": 0.914,
      "step": 13105
    },
    {
      "epoch": 0.9455647343169439,
      "grad_norm": 2.408975123212482,
      "learning_rate": 3.096432369500301e-08,
      "loss": 0.933,
      "step": 13106
    },
    {
      "epoch": 0.9456368817863714,
      "grad_norm": 4.189809816910534,
      "learning_rate": 3.088245656516153e-08,
      "loss": 0.9399,
      "step": 13107
    },
    {
      "epoch": 0.9457090292557988,
      "grad_norm": 2.8980126905038373,
      "learning_rate": 3.080069696153997e-08,
      "loss": 0.9358,
      "step": 13108
    },
    {
      "epoch": 0.9457811767252263,
      "grad_norm": 3.835110639346384,
      "learning_rate": 3.0719044888602776e-08,
      "loss": 1.037,
      "step": 13109
    },
    {
      "epoch": 0.9458533241946538,
      "grad_norm": 7.702747389879466,
      "learning_rate": 3.063750035080836e-08,
      "loss": 1.0007,
      "step": 13110
    },
    {
      "epoch": 0.9459254716640814,
      "grad_norm": 4.0382667215048675,
      "learning_rate": 3.055606335260963e-08,
      "loss": 0.9868,
      "step": 13111
    },
    {
      "epoch": 0.9459976191335089,
      "grad_norm": 3.582760269709309,
      "learning_rate": 3.04747338984539e-08,
      "loss": 1.0386,
      "step": 13112
    },
    {
      "epoch": 0.9460697666029364,
      "grad_norm": 3.6922514740014156,
      "learning_rate": 3.039351199278184e-08,
      "loss": 0.9063,
      "step": 13113
    },
    {
      "epoch": 0.9461419140723639,
      "grad_norm": 3.6716263057818237,
      "learning_rate": 3.0312397640028575e-08,
      "loss": 0.9566,
      "step": 13114
    },
    {
      "epoch": 0.9462140615417914,
      "grad_norm": 9.622390591521464,
      "learning_rate": 3.023139084462412e-08,
      "loss": 0.8958,
      "step": 13115
    },
    {
      "epoch": 0.946286209011219,
      "grad_norm": 4.779496214107606,
      "learning_rate": 3.015049161099159e-08,
      "loss": 0.982,
      "step": 13116
    },
    {
      "epoch": 0.9463583564806465,
      "grad_norm": 3.5650939866954654,
      "learning_rate": 3.006969994354835e-08,
      "loss": 0.9348,
      "step": 13117
    },
    {
      "epoch": 0.946430503950074,
      "grad_norm": 2.8285594627545625,
      "learning_rate": 2.9989015846706654e-08,
      "loss": 0.8895,
      "step": 13118
    },
    {
      "epoch": 0.9465026514195014,
      "grad_norm": 4.342856335819141,
      "learning_rate": 2.990843932487186e-08,
      "loss": 0.9567,
      "step": 13119
    },
    {
      "epoch": 0.9465747988889289,
      "grad_norm": 0.7502169692596392,
      "learning_rate": 2.982797038244422e-08,
      "loss": 0.8295,
      "step": 13120
    },
    {
      "epoch": 0.9466469463583564,
      "grad_norm": 3.7703244338338773,
      "learning_rate": 2.9747609023817564e-08,
      "loss": 0.9322,
      "step": 13121
    },
    {
      "epoch": 0.946719093827784,
      "grad_norm": 2.9025366245426643,
      "learning_rate": 2.96673552533806e-08,
      "loss": 0.9994,
      "step": 13122
    },
    {
      "epoch": 0.9467912412972115,
      "grad_norm": 4.202899005018645,
      "learning_rate": 2.9587209075515597e-08,
      "loss": 0.9274,
      "step": 13123
    },
    {
      "epoch": 0.946863388766639,
      "grad_norm": 0.675405805956596,
      "learning_rate": 2.950717049459861e-08,
      "loss": 0.8031,
      "step": 13124
    },
    {
      "epoch": 0.9469355362360665,
      "grad_norm": 4.250255577032136,
      "learning_rate": 2.942723951500059e-08,
      "loss": 0.9737,
      "step": 13125
    },
    {
      "epoch": 0.947007683705494,
      "grad_norm": 5.523825625065731,
      "learning_rate": 2.9347416141086266e-08,
      "loss": 1.0834,
      "step": 13126
    },
    {
      "epoch": 0.9470798311749216,
      "grad_norm": 3.20315075375508,
      "learning_rate": 2.926770037721438e-08,
      "loss": 1.0116,
      "step": 13127
    },
    {
      "epoch": 0.9471519786443491,
      "grad_norm": 5.8358436995683,
      "learning_rate": 2.9188092227738105e-08,
      "loss": 1.0045,
      "step": 13128
    },
    {
      "epoch": 0.9472241261137766,
      "grad_norm": 7.778967832923228,
      "learning_rate": 2.9108591697004415e-08,
      "loss": 0.9111,
      "step": 13129
    },
    {
      "epoch": 0.9472962735832041,
      "grad_norm": 3.400023202255965,
      "learning_rate": 2.9029198789354502e-08,
      "loss": 1.0017,
      "step": 13130
    },
    {
      "epoch": 0.9473684210526315,
      "grad_norm": 3.262803606544214,
      "learning_rate": 2.894991350912357e-08,
      "loss": 0.9265,
      "step": 13131
    },
    {
      "epoch": 0.947440568522059,
      "grad_norm": 3.1330436944675495,
      "learning_rate": 2.887073586064126e-08,
      "loss": 1.0365,
      "step": 13132
    },
    {
      "epoch": 0.9475127159914866,
      "grad_norm": 4.225688963926313,
      "learning_rate": 2.879166584823145e-08,
      "loss": 0.8245,
      "step": 13133
    },
    {
      "epoch": 0.9475848634609141,
      "grad_norm": 4.122058860258612,
      "learning_rate": 2.8712703476211576e-08,
      "loss": 0.9872,
      "step": 13134
    },
    {
      "epoch": 0.9476570109303416,
      "grad_norm": 8.506242591467888,
      "learning_rate": 2.8633848748893518e-08,
      "loss": 1.0125,
      "step": 13135
    },
    {
      "epoch": 0.9477291583997691,
      "grad_norm": 5.764965650285751,
      "learning_rate": 2.8555101670583172e-08,
      "loss": 0.9829,
      "step": 13136
    },
    {
      "epoch": 0.9478013058691966,
      "grad_norm": 6.922654222152959,
      "learning_rate": 2.8476462245580644e-08,
      "loss": 0.8521,
      "step": 13137
    },
    {
      "epoch": 0.9478734533386242,
      "grad_norm": 4.796018642067968,
      "learning_rate": 2.8397930478180064e-08,
      "loss": 0.9056,
      "step": 13138
    },
    {
      "epoch": 0.9479456008080517,
      "grad_norm": 2.946547676750666,
      "learning_rate": 2.831950637267022e-08,
      "loss": 0.8783,
      "step": 13139
    },
    {
      "epoch": 0.9480177482774792,
      "grad_norm": 3.674390734557586,
      "learning_rate": 2.824118993333302e-08,
      "loss": 0.9516,
      "step": 13140
    },
    {
      "epoch": 0.9480898957469067,
      "grad_norm": 3.9894604112033236,
      "learning_rate": 2.8162981164445265e-08,
      "loss": 0.9418,
      "step": 13141
    },
    {
      "epoch": 0.9481620432163342,
      "grad_norm": 9.326861271967298,
      "learning_rate": 2.808488007027754e-08,
      "loss": 0.8935,
      "step": 13142
    },
    {
      "epoch": 0.9482341906857616,
      "grad_norm": 9.717913527231755,
      "learning_rate": 2.8006886655094875e-08,
      "loss": 1.0403,
      "step": 13143
    },
    {
      "epoch": 0.9483063381551892,
      "grad_norm": 5.857338838920434,
      "learning_rate": 2.7929000923155866e-08,
      "loss": 1.0683,
      "step": 13144
    },
    {
      "epoch": 0.9483784856246167,
      "grad_norm": 4.395492082460468,
      "learning_rate": 2.7851222878713776e-08,
      "loss": 0.9969,
      "step": 13145
    },
    {
      "epoch": 0.9484506330940442,
      "grad_norm": 3.1668523098253405,
      "learning_rate": 2.7773552526015875e-08,
      "loss": 1.0645,
      "step": 13146
    },
    {
      "epoch": 0.9485227805634717,
      "grad_norm": 3.6742588826996347,
      "learning_rate": 2.7695989869303437e-08,
      "loss": 0.8593,
      "step": 13147
    },
    {
      "epoch": 0.9485949280328992,
      "grad_norm": 4.772710417020023,
      "learning_rate": 2.761853491281152e-08,
      "loss": 1.0035,
      "step": 13148
    },
    {
      "epoch": 0.9486670755023268,
      "grad_norm": 4.722222875457918,
      "learning_rate": 2.7541187660769848e-08,
      "loss": 0.9282,
      "step": 13149
    },
    {
      "epoch": 0.9487392229717543,
      "grad_norm": 4.1611308644166405,
      "learning_rate": 2.74639481174026e-08,
      "loss": 0.9156,
      "step": 13150
    },
    {
      "epoch": 0.9488113704411818,
      "grad_norm": 3.627098528402725,
      "learning_rate": 2.7386816286926627e-08,
      "loss": 1.0097,
      "step": 13151
    },
    {
      "epoch": 0.9488835179106093,
      "grad_norm": 3.479771331202103,
      "learning_rate": 2.7309792173554557e-08,
      "loss": 0.9323,
      "step": 13152
    },
    {
      "epoch": 0.9489556653800368,
      "grad_norm": 4.130173184733138,
      "learning_rate": 2.7232875781491916e-08,
      "loss": 1.112,
      "step": 13153
    },
    {
      "epoch": 0.9490278128494644,
      "grad_norm": 2.8601658462242563,
      "learning_rate": 2.715606711493912e-08,
      "loss": 0.9463,
      "step": 13154
    },
    {
      "epoch": 0.9490999603188918,
      "grad_norm": 3.4012687728010476,
      "learning_rate": 2.7079366178089923e-08,
      "loss": 1.012,
      "step": 13155
    },
    {
      "epoch": 0.9491721077883193,
      "grad_norm": 3.53670195619149,
      "learning_rate": 2.7002772975133426e-08,
      "loss": 0.9505,
      "step": 13156
    },
    {
      "epoch": 0.9492442552577468,
      "grad_norm": 7.634908273242077,
      "learning_rate": 2.692628751025161e-08,
      "loss": 0.8836,
      "step": 13157
    },
    {
      "epoch": 0.9493164027271743,
      "grad_norm": 2.8697026418365157,
      "learning_rate": 2.684990978762114e-08,
      "loss": 0.8441,
      "step": 13158
    },
    {
      "epoch": 0.9493885501966018,
      "grad_norm": 0.7598314428381954,
      "learning_rate": 2.6773639811412674e-08,
      "loss": 0.8247,
      "step": 13159
    },
    {
      "epoch": 0.9494606976660294,
      "grad_norm": 3.9955294183558174,
      "learning_rate": 2.6697477585791328e-08,
      "loss": 1.0038,
      "step": 13160
    },
    {
      "epoch": 0.9495328451354569,
      "grad_norm": 0.6463292330457222,
      "learning_rate": 2.6621423114915554e-08,
      "loss": 0.7534,
      "step": 13161
    },
    {
      "epoch": 0.9496049926048844,
      "grad_norm": 2.7610132592368846,
      "learning_rate": 2.6545476402938695e-08,
      "loss": 0.8724,
      "step": 13162
    },
    {
      "epoch": 0.9496771400743119,
      "grad_norm": 6.00336965671009,
      "learning_rate": 2.64696374540081e-08,
      "loss": 1.0096,
      "step": 13163
    },
    {
      "epoch": 0.9497492875437394,
      "grad_norm": 4.049467103537432,
      "learning_rate": 2.6393906272264677e-08,
      "loss": 0.8953,
      "step": 13164
    },
    {
      "epoch": 0.949821435013167,
      "grad_norm": 3.137119291392231,
      "learning_rate": 2.631828286184401e-08,
      "loss": 0.8748,
      "step": 13165
    },
    {
      "epoch": 0.9498935824825945,
      "grad_norm": 4.939074108444421,
      "learning_rate": 2.6242767226875463e-08,
      "loss": 0.9928,
      "step": 13166
    },
    {
      "epoch": 0.9499657299520219,
      "grad_norm": 4.258958238907928,
      "learning_rate": 2.6167359371483065e-08,
      "loss": 0.8174,
      "step": 13167
    },
    {
      "epoch": 0.9500378774214494,
      "grad_norm": 3.8518416669159374,
      "learning_rate": 2.6092059299783974e-08,
      "loss": 0.9588,
      "step": 13168
    },
    {
      "epoch": 0.9501100248908769,
      "grad_norm": 5.951458074325083,
      "learning_rate": 2.6016867015890677e-08,
      "loss": 0.9383,
      "step": 13169
    },
    {
      "epoch": 0.9501821723603044,
      "grad_norm": 3.0421218183154175,
      "learning_rate": 2.5941782523908772e-08,
      "loss": 0.9628,
      "step": 13170
    },
    {
      "epoch": 0.950254319829732,
      "grad_norm": 5.020454909545478,
      "learning_rate": 2.586680582793832e-08,
      "loss": 0.8585,
      "step": 13171
    },
    {
      "epoch": 0.9503264672991595,
      "grad_norm": 3.184429410070917,
      "learning_rate": 2.579193693207382e-08,
      "loss": 0.94,
      "step": 13172
    },
    {
      "epoch": 0.950398614768587,
      "grad_norm": 5.529372246759781,
      "learning_rate": 2.5717175840403337e-08,
      "loss": 1.0379,
      "step": 13173
    },
    {
      "epoch": 0.9504707622380145,
      "grad_norm": 3.908178601052179,
      "learning_rate": 2.564252255700894e-08,
      "loss": 0.8902,
      "step": 13174
    },
    {
      "epoch": 0.950542909707442,
      "grad_norm": 4.270584967609961,
      "learning_rate": 2.5567977085967806e-08,
      "loss": 0.867,
      "step": 13175
    },
    {
      "epoch": 0.9506150571768696,
      "grad_norm": 2.96297318425888,
      "learning_rate": 2.5493539431350465e-08,
      "loss": 0.8893,
      "step": 13176
    },
    {
      "epoch": 0.9506872046462971,
      "grad_norm": 4.568385051966638,
      "learning_rate": 2.5419209597221436e-08,
      "loss": 0.9671,
      "step": 13177
    },
    {
      "epoch": 0.9507593521157245,
      "grad_norm": 4.061230739971692,
      "learning_rate": 2.53449875876397e-08,
      "loss": 0.8932,
      "step": 13178
    },
    {
      "epoch": 0.950831499585152,
      "grad_norm": 4.379823260290515,
      "learning_rate": 2.5270873406658233e-08,
      "loss": 0.9815,
      "step": 13179
    },
    {
      "epoch": 0.9509036470545795,
      "grad_norm": 3.8520481507568816,
      "learning_rate": 2.5196867058324466e-08,
      "loss": 0.9384,
      "step": 13180
    },
    {
      "epoch": 0.950975794524007,
      "grad_norm": 2.87327357991951,
      "learning_rate": 2.512296854667917e-08,
      "loss": 0.8968,
      "step": 13181
    },
    {
      "epoch": 0.9510479419934346,
      "grad_norm": 4.493298945384967,
      "learning_rate": 2.5049177875757553e-08,
      "loss": 0.9174,
      "step": 13182
    },
    {
      "epoch": 0.9511200894628621,
      "grad_norm": 4.081916770972491,
      "learning_rate": 2.4975495049589735e-08,
      "loss": 0.9446,
      "step": 13183
    },
    {
      "epoch": 0.9511922369322896,
      "grad_norm": 4.814438268549593,
      "learning_rate": 2.490192007219871e-08,
      "loss": 0.9957,
      "step": 13184
    },
    {
      "epoch": 0.9512643844017171,
      "grad_norm": 5.845940979854433,
      "learning_rate": 2.4828452947602163e-08,
      "loss": 0.9129,
      "step": 13185
    },
    {
      "epoch": 0.9513365318711446,
      "grad_norm": 2.9234595260728233,
      "learning_rate": 2.4755093679811768e-08,
      "loss": 0.9476,
      "step": 13186
    },
    {
      "epoch": 0.9514086793405722,
      "grad_norm": 4.785095662881745,
      "learning_rate": 2.4681842272833873e-08,
      "loss": 1.0144,
      "step": 13187
    },
    {
      "epoch": 0.9514808268099997,
      "grad_norm": 3.129413539800769,
      "learning_rate": 2.4608698730668175e-08,
      "loss": 0.9041,
      "step": 13188
    },
    {
      "epoch": 0.9515529742794272,
      "grad_norm": 0.7740591614165708,
      "learning_rate": 2.4535663057308807e-08,
      "loss": 0.8978,
      "step": 13189
    },
    {
      "epoch": 0.9516251217488546,
      "grad_norm": 6.023474548463398,
      "learning_rate": 2.4462735256743915e-08,
      "loss": 0.9314,
      "step": 13190
    },
    {
      "epoch": 0.9516972692182821,
      "grad_norm": 3.5317574195548955,
      "learning_rate": 2.4389915332955868e-08,
      "loss": 0.9296,
      "step": 13191
    },
    {
      "epoch": 0.9517694166877096,
      "grad_norm": 5.31130411488147,
      "learning_rate": 2.431720328992104e-08,
      "loss": 0.9829,
      "step": 13192
    },
    {
      "epoch": 0.9518415641571372,
      "grad_norm": 3.9352805527909704,
      "learning_rate": 2.424459913161003e-08,
      "loss": 0.9656,
      "step": 13193
    },
    {
      "epoch": 0.9519137116265647,
      "grad_norm": 3.3787910926908276,
      "learning_rate": 2.4172102861987677e-08,
      "loss": 1.0626,
      "step": 13194
    },
    {
      "epoch": 0.9519858590959922,
      "grad_norm": 3.057284838762517,
      "learning_rate": 2.4099714485012133e-08,
      "loss": 0.9064,
      "step": 13195
    },
    {
      "epoch": 0.9520580065654197,
      "grad_norm": 4.552326722552676,
      "learning_rate": 2.4027434004636916e-08,
      "loss": 0.8763,
      "step": 13196
    },
    {
      "epoch": 0.9521301540348472,
      "grad_norm": 2.79556755322441,
      "learning_rate": 2.395526142480886e-08,
      "loss": 0.8791,
      "step": 13197
    },
    {
      "epoch": 0.9522023015042748,
      "grad_norm": 7.198344665551398,
      "learning_rate": 2.388319674946859e-08,
      "loss": 0.98,
      "step": 13198
    },
    {
      "epoch": 0.9522744489737023,
      "grad_norm": 0.7799421043795042,
      "learning_rate": 2.3811239982551635e-08,
      "loss": 0.8209,
      "step": 13199
    },
    {
      "epoch": 0.9523465964431298,
      "grad_norm": 3.2421927946116407,
      "learning_rate": 2.3739391127987506e-08,
      "loss": 1.0272,
      "step": 13200
    },
    {
      "epoch": 0.9524187439125573,
      "grad_norm": 4.5181446899373405,
      "learning_rate": 2.3667650189699297e-08,
      "loss": 1.0021,
      "step": 13201
    },
    {
      "epoch": 0.9524908913819847,
      "grad_norm": 3.197639494739498,
      "learning_rate": 2.359601717160431e-08,
      "loss": 1.0534,
      "step": 13202
    },
    {
      "epoch": 0.9525630388514122,
      "grad_norm": 2.899994054327822,
      "learning_rate": 2.3524492077614533e-08,
      "loss": 0.8915,
      "step": 13203
    },
    {
      "epoch": 0.9526351863208398,
      "grad_norm": 3.5193005146428917,
      "learning_rate": 2.345307491163573e-08,
      "loss": 0.9393,
      "step": 13204
    },
    {
      "epoch": 0.9527073337902673,
      "grad_norm": 2.529211282938837,
      "learning_rate": 2.3381765677567666e-08,
      "loss": 0.9558,
      "step": 13205
    },
    {
      "epoch": 0.9527794812596948,
      "grad_norm": 2.877231063935582,
      "learning_rate": 2.3310564379303898e-08,
      "loss": 0.9554,
      "step": 13206
    },
    {
      "epoch": 0.9528516287291223,
      "grad_norm": 7.610052998943471,
      "learning_rate": 2.3239471020733092e-08,
      "loss": 0.9859,
      "step": 13207
    },
    {
      "epoch": 0.9529237761985498,
      "grad_norm": 3.2355101431668616,
      "learning_rate": 2.316848560573681e-08,
      "loss": 1.0538,
      "step": 13208
    },
    {
      "epoch": 0.9529959236679774,
      "grad_norm": 4.513193076211571,
      "learning_rate": 2.3097608138191503e-08,
      "loss": 0.998,
      "step": 13209
    },
    {
      "epoch": 0.9530680711374049,
      "grad_norm": 12.04728727711366,
      "learning_rate": 2.3026838621967637e-08,
      "loss": 1.1212,
      "step": 13210
    },
    {
      "epoch": 0.9531402186068324,
      "grad_norm": 2.9121476905804298,
      "learning_rate": 2.295617706092967e-08,
      "loss": 1.0379,
      "step": 13211
    },
    {
      "epoch": 0.9532123660762599,
      "grad_norm": 3.2372668325935314,
      "learning_rate": 2.2885623458936076e-08,
      "loss": 0.9735,
      "step": 13212
    },
    {
      "epoch": 0.9532845135456874,
      "grad_norm": 5.306429817019547,
      "learning_rate": 2.281517781983955e-08,
      "loss": 0.9508,
      "step": 13213
    },
    {
      "epoch": 0.9533566610151148,
      "grad_norm": 3.0446193072491377,
      "learning_rate": 2.274484014748701e-08,
      "loss": 0.9158,
      "step": 13214
    },
    {
      "epoch": 0.9534288084845424,
      "grad_norm": 3.913311732122622,
      "learning_rate": 2.2674610445719167e-08,
      "loss": 0.9619,
      "step": 13215
    },
    {
      "epoch": 0.9535009559539699,
      "grad_norm": 8.12935627283271,
      "learning_rate": 2.260448871837073e-08,
      "loss": 0.943,
      "step": 13216
    },
    {
      "epoch": 0.9535731034233974,
      "grad_norm": 0.7542793655141223,
      "learning_rate": 2.2534474969271522e-08,
      "loss": 0.8615,
      "step": 13217
    },
    {
      "epoch": 0.9536452508928249,
      "grad_norm": 4.991932463959368,
      "learning_rate": 2.2464569202244043e-08,
      "loss": 1.0503,
      "step": 13218
    },
    {
      "epoch": 0.9537173983622524,
      "grad_norm": 5.143241425189769,
      "learning_rate": 2.2394771421105907e-08,
      "loss": 0.929,
      "step": 13219
    },
    {
      "epoch": 0.95378954583168,
      "grad_norm": 4.504325483252962,
      "learning_rate": 2.2325081629668506e-08,
      "loss": 0.8669,
      "step": 13220
    },
    {
      "epoch": 0.9538616933011075,
      "grad_norm": 8.333444467439291,
      "learning_rate": 2.2255499831737246e-08,
      "loss": 1.0029,
      "step": 13221
    },
    {
      "epoch": 0.953933840770535,
      "grad_norm": 3.9347534288565575,
      "learning_rate": 2.2186026031111748e-08,
      "loss": 0.9219,
      "step": 13222
    },
    {
      "epoch": 0.9540059882399625,
      "grad_norm": 3.3275292091606623,
      "learning_rate": 2.211666023158587e-08,
      "loss": 0.8943,
      "step": 13223
    },
    {
      "epoch": 0.95407813570939,
      "grad_norm": 2.524389223312274,
      "learning_rate": 2.2047402436947248e-08,
      "loss": 0.9591,
      "step": 13224
    },
    {
      "epoch": 0.9541502831788174,
      "grad_norm": 3.349736638890699,
      "learning_rate": 2.1978252650977746e-08,
      "loss": 0.8919,
      "step": 13225
    },
    {
      "epoch": 0.954222430648245,
      "grad_norm": 3.457478812174169,
      "learning_rate": 2.1909210877453678e-08,
      "loss": 1.0039,
      "step": 13226
    },
    {
      "epoch": 0.9542945781176725,
      "grad_norm": 0.7557876119134922,
      "learning_rate": 2.184027712014469e-08,
      "loss": 0.8766,
      "step": 13227
    },
    {
      "epoch": 0.9543667255871,
      "grad_norm": 4.094346097687806,
      "learning_rate": 2.1771451382815553e-08,
      "loss": 1.1221,
      "step": 13228
    },
    {
      "epoch": 0.9544388730565275,
      "grad_norm": 4.499612367676863,
      "learning_rate": 2.170273366922415e-08,
      "loss": 0.9969,
      "step": 13229
    },
    {
      "epoch": 0.954511020525955,
      "grad_norm": 3.303997571995099,
      "learning_rate": 2.163412398312303e-08,
      "loss": 1.0048,
      "step": 13230
    },
    {
      "epoch": 0.9545831679953826,
      "grad_norm": 5.162312036376054,
      "learning_rate": 2.156562232825898e-08,
      "loss": 0.973,
      "step": 13231
    },
    {
      "epoch": 0.9546553154648101,
      "grad_norm": 10.819564716376323,
      "learning_rate": 2.1497228708372116e-08,
      "loss": 0.8826,
      "step": 13232
    },
    {
      "epoch": 0.9547274629342376,
      "grad_norm": 4.477808381179298,
      "learning_rate": 2.1428943127197453e-08,
      "loss": 0.9397,
      "step": 13233
    },
    {
      "epoch": 0.9547996104036651,
      "grad_norm": 4.025994712223498,
      "learning_rate": 2.1360765588464003e-08,
      "loss": 1.0276,
      "step": 13234
    },
    {
      "epoch": 0.9548717578730926,
      "grad_norm": 6.60864628952324,
      "learning_rate": 2.1292696095894125e-08,
      "loss": 0.8326,
      "step": 13235
    },
    {
      "epoch": 0.9549439053425202,
      "grad_norm": 4.947154589407344,
      "learning_rate": 2.1224734653205512e-08,
      "loss": 0.8992,
      "step": 13236
    },
    {
      "epoch": 0.9550160528119476,
      "grad_norm": 2.82131694966098,
      "learning_rate": 2.115688126410875e-08,
      "loss": 0.981,
      "step": 13237
    },
    {
      "epoch": 0.9550882002813751,
      "grad_norm": 4.6410331161294,
      "learning_rate": 2.1089135932309542e-08,
      "loss": 1.0546,
      "step": 13238
    },
    {
      "epoch": 0.9551603477508026,
      "grad_norm": 3.7176155475650075,
      "learning_rate": 2.102149866150671e-08,
      "loss": 0.9823,
      "step": 13239
    },
    {
      "epoch": 0.9552324952202301,
      "grad_norm": 0.7859082772072071,
      "learning_rate": 2.0953969455393738e-08,
      "loss": 0.8854,
      "step": 13240
    },
    {
      "epoch": 0.9553046426896576,
      "grad_norm": 4.316841041450153,
      "learning_rate": 2.088654831765879e-08,
      "loss": 0.9642,
      "step": 13241
    },
    {
      "epoch": 0.9553767901590852,
      "grad_norm": 5.267027897256991,
      "learning_rate": 2.0819235251982705e-08,
      "loss": 1.1083,
      "step": 13242
    },
    {
      "epoch": 0.9554489376285127,
      "grad_norm": 14.060478370309921,
      "learning_rate": 2.075203026204142e-08,
      "loss": 0.9591,
      "step": 13243
    },
    {
      "epoch": 0.9555210850979402,
      "grad_norm": 3.7196650300929237,
      "learning_rate": 2.068493335150512e-08,
      "loss": 0.8415,
      "step": 13244
    },
    {
      "epoch": 0.9555932325673677,
      "grad_norm": 3.2955934465367474,
      "learning_rate": 2.0617944524037313e-08,
      "loss": 0.9231,
      "step": 13245
    },
    {
      "epoch": 0.9556653800367952,
      "grad_norm": 2.962853126565198,
      "learning_rate": 2.0551063783295965e-08,
      "loss": 0.9357,
      "step": 13246
    },
    {
      "epoch": 0.9557375275062228,
      "grad_norm": 4.385354132972847,
      "learning_rate": 2.0484291132933483e-08,
      "loss": 0.8716,
      "step": 13247
    },
    {
      "epoch": 0.9558096749756503,
      "grad_norm": 5.125746695854515,
      "learning_rate": 2.041762657659607e-08,
      "loss": 0.9771,
      "step": 13248
    },
    {
      "epoch": 0.9558818224450777,
      "grad_norm": 4.92137584274427,
      "learning_rate": 2.0351070117923918e-08,
      "loss": 1.0096,
      "step": 13249
    },
    {
      "epoch": 0.9559539699145052,
      "grad_norm": 0.7281790357193884,
      "learning_rate": 2.0284621760551235e-08,
      "loss": 0.8092,
      "step": 13250
    },
    {
      "epoch": 0.9560261173839327,
      "grad_norm": 4.284804379283328,
      "learning_rate": 2.0218281508106894e-08,
      "loss": 1.0025,
      "step": 13251
    },
    {
      "epoch": 0.9560982648533602,
      "grad_norm": 4.676566635624528,
      "learning_rate": 2.0152049364212888e-08,
      "loss": 1.0675,
      "step": 13252
    },
    {
      "epoch": 0.9561704123227878,
      "grad_norm": 18.630308073304906,
      "learning_rate": 2.0085925332486764e-08,
      "loss": 0.9696,
      "step": 13253
    },
    {
      "epoch": 0.9562425597922153,
      "grad_norm": 4.962637926774503,
      "learning_rate": 2.0019909416538528e-08,
      "loss": 0.9348,
      "step": 13254
    },
    {
      "epoch": 0.9563147072616428,
      "grad_norm": 4.524014925297166,
      "learning_rate": 1.9954001619973737e-08,
      "loss": 0.892,
      "step": 13255
    },
    {
      "epoch": 0.9563868547310703,
      "grad_norm": 3.732439249697917,
      "learning_rate": 1.988820194639085e-08,
      "loss": 1.021,
      "step": 13256
    },
    {
      "epoch": 0.9564590022004978,
      "grad_norm": 4.3081359896788785,
      "learning_rate": 1.982251039938321e-08,
      "loss": 1.0446,
      "step": 13257
    },
    {
      "epoch": 0.9565311496699254,
      "grad_norm": 3.68976213282954,
      "learning_rate": 1.9756926982537945e-08,
      "loss": 0.9267,
      "step": 13258
    },
    {
      "epoch": 0.9566032971393529,
      "grad_norm": 2.5843079379512726,
      "learning_rate": 1.96914516994362e-08,
      "loss": 0.9911,
      "step": 13259
    },
    {
      "epoch": 0.9566754446087804,
      "grad_norm": 8.421228614022358,
      "learning_rate": 1.9626084553653332e-08,
      "loss": 0.9658,
      "step": 13260
    },
    {
      "epoch": 0.9567475920782078,
      "grad_norm": 2.981954341833111,
      "learning_rate": 1.9560825548759152e-08,
      "loss": 0.8589,
      "step": 13261
    },
    {
      "epoch": 0.9568197395476353,
      "grad_norm": 3.0735614854388302,
      "learning_rate": 1.949567468831681e-08,
      "loss": 0.8978,
      "step": 13262
    },
    {
      "epoch": 0.9568918870170628,
      "grad_norm": 5.651712863515447,
      "learning_rate": 1.9430631975884127e-08,
      "loss": 0.8903,
      "step": 13263
    },
    {
      "epoch": 0.9569640344864904,
      "grad_norm": 4.419369260568173,
      "learning_rate": 1.9365697415012483e-08,
      "loss": 0.9141,
      "step": 13264
    },
    {
      "epoch": 0.9570361819559179,
      "grad_norm": 4.816479709433212,
      "learning_rate": 1.9300871009248598e-08,
      "loss": 0.9906,
      "step": 13265
    },
    {
      "epoch": 0.9571083294253454,
      "grad_norm": 3.4534797313642724,
      "learning_rate": 1.923615276213142e-08,
      "loss": 0.9645,
      "step": 13266
    },
    {
      "epoch": 0.9571804768947729,
      "grad_norm": 8.227712407762182,
      "learning_rate": 1.9171542677195672e-08,
      "loss": 0.9766,
      "step": 13267
    },
    {
      "epoch": 0.9572526243642004,
      "grad_norm": 5.4833079112351575,
      "learning_rate": 1.9107040757969206e-08,
      "loss": 0.9837,
      "step": 13268
    },
    {
      "epoch": 0.957324771833628,
      "grad_norm": 5.599240299192434,
      "learning_rate": 1.9042647007974088e-08,
      "loss": 0.9724,
      "step": 13269
    },
    {
      "epoch": 0.9573969193030555,
      "grad_norm": 4.056765455695512,
      "learning_rate": 1.897836143072662e-08,
      "loss": 1.0011,
      "step": 13270
    },
    {
      "epoch": 0.957469066772483,
      "grad_norm": 3.5165338994375634,
      "learning_rate": 1.8914184029737546e-08,
      "loss": 0.9451,
      "step": 13271
    },
    {
      "epoch": 0.9575412142419104,
      "grad_norm": 8.648516453879667,
      "learning_rate": 1.8850114808511176e-08,
      "loss": 0.9226,
      "step": 13272
    },
    {
      "epoch": 0.9576133617113379,
      "grad_norm": 6.769726785079949,
      "learning_rate": 1.8786153770546044e-08,
      "loss": 0.9521,
      "step": 13273
    },
    {
      "epoch": 0.9576855091807654,
      "grad_norm": 4.747912500348118,
      "learning_rate": 1.872230091933491e-08,
      "loss": 0.8479,
      "step": 13274
    },
    {
      "epoch": 0.957757656650193,
      "grad_norm": 12.618636841098926,
      "learning_rate": 1.8658556258364543e-08,
      "loss": 0.9502,
      "step": 13275
    },
    {
      "epoch": 0.9578298041196205,
      "grad_norm": 2.9538177807909376,
      "learning_rate": 1.8594919791115494e-08,
      "loss": 0.8435,
      "step": 13276
    },
    {
      "epoch": 0.957901951589048,
      "grad_norm": 4.435997937077283,
      "learning_rate": 1.853139152106298e-08,
      "loss": 0.8457,
      "step": 13277
    },
    {
      "epoch": 0.9579740990584755,
      "grad_norm": 2.428527555149541,
      "learning_rate": 1.8467971451676224e-08,
      "loss": 1.0421,
      "step": 13278
    },
    {
      "epoch": 0.958046246527903,
      "grad_norm": 3.1724178056462775,
      "learning_rate": 1.8404659586418014e-08,
      "loss": 0.9676,
      "step": 13279
    },
    {
      "epoch": 0.9581183939973306,
      "grad_norm": 7.008042755937136,
      "learning_rate": 1.8341455928745807e-08,
      "loss": 0.9138,
      "step": 13280
    },
    {
      "epoch": 0.9581905414667581,
      "grad_norm": 4.279576329447847,
      "learning_rate": 1.8278360482110844e-08,
      "loss": 0.8906,
      "step": 13281
    },
    {
      "epoch": 0.9582626889361856,
      "grad_norm": 4.112210647931823,
      "learning_rate": 1.8215373249958587e-08,
      "loss": 0.946,
      "step": 13282
    },
    {
      "epoch": 0.9583348364056131,
      "grad_norm": 3.465736798752545,
      "learning_rate": 1.8152494235728288e-08,
      "loss": 0.9347,
      "step": 13283
    },
    {
      "epoch": 0.9584069838750405,
      "grad_norm": 4.238292725718911,
      "learning_rate": 1.8089723442853865e-08,
      "loss": 0.9454,
      "step": 13284
    },
    {
      "epoch": 0.958479131344468,
      "grad_norm": 6.308870386252765,
      "learning_rate": 1.8027060874763022e-08,
      "loss": 0.9632,
      "step": 13285
    },
    {
      "epoch": 0.9585512788138956,
      "grad_norm": 3.377222812665993,
      "learning_rate": 1.796450653487702e-08,
      "loss": 0.9852,
      "step": 13286
    },
    {
      "epoch": 0.9586234262833231,
      "grad_norm": 2.947875994254582,
      "learning_rate": 1.7902060426612243e-08,
      "loss": 1.0471,
      "step": 13287
    },
    {
      "epoch": 0.9586955737527506,
      "grad_norm": 3.7629999531977294,
      "learning_rate": 1.7839722553378622e-08,
      "loss": 0.9776,
      "step": 13288
    },
    {
      "epoch": 0.9587677212221781,
      "grad_norm": 3.9290103518533397,
      "learning_rate": 1.7777492918579884e-08,
      "loss": 0.9449,
      "step": 13289
    },
    {
      "epoch": 0.9588398686916056,
      "grad_norm": 5.413804966029445,
      "learning_rate": 1.7715371525614198e-08,
      "loss": 0.9865,
      "step": 13290
    },
    {
      "epoch": 0.9589120161610332,
      "grad_norm": 3.7869002317427563,
      "learning_rate": 1.7653358377874184e-08,
      "loss": 0.7964,
      "step": 13291
    },
    {
      "epoch": 0.9589841636304607,
      "grad_norm": 5.4173684863722045,
      "learning_rate": 1.75914534787458e-08,
      "loss": 0.9525,
      "step": 13292
    },
    {
      "epoch": 0.9590563110998882,
      "grad_norm": 4.315668683449036,
      "learning_rate": 1.7529656831609452e-08,
      "loss": 1.0175,
      "step": 13293
    },
    {
      "epoch": 0.9591284585693157,
      "grad_norm": 4.062856570514549,
      "learning_rate": 1.7467968439839554e-08,
      "loss": 0.8884,
      "step": 13294
    },
    {
      "epoch": 0.9592006060387432,
      "grad_norm": 0.7029154359263657,
      "learning_rate": 1.7406388306804964e-08,
      "loss": 0.824,
      "step": 13295
    },
    {
      "epoch": 0.9592727535081707,
      "grad_norm": 4.1982693330783825,
      "learning_rate": 1.734491643586833e-08,
      "loss": 0.9977,
      "step": 13296
    },
    {
      "epoch": 0.9593449009775982,
      "grad_norm": 2.903390605754334,
      "learning_rate": 1.7283552830386074e-08,
      "loss": 0.9693,
      "step": 13297
    },
    {
      "epoch": 0.9594170484470257,
      "grad_norm": 4.857644119362889,
      "learning_rate": 1.7222297493709292e-08,
      "loss": 0.9781,
      "step": 13298
    },
    {
      "epoch": 0.9594891959164532,
      "grad_norm": 2.7584371318667795,
      "learning_rate": 1.7161150429182868e-08,
      "loss": 0.9738,
      "step": 13299
    },
    {
      "epoch": 0.9595613433858807,
      "grad_norm": 5.254659129374423,
      "learning_rate": 1.7100111640145686e-08,
      "loss": 1.0014,
      "step": 13300
    },
    {
      "epoch": 0.9596334908553082,
      "grad_norm": 3.861276135169409,
      "learning_rate": 1.7039181129931078e-08,
      "loss": 1.0094,
      "step": 13301
    },
    {
      "epoch": 0.9597056383247358,
      "grad_norm": 3.928276036043858,
      "learning_rate": 1.6978358901866164e-08,
      "loss": 0.9284,
      "step": 13302
    },
    {
      "epoch": 0.9597777857941633,
      "grad_norm": 2.2654439919611544,
      "learning_rate": 1.691764495927206e-08,
      "loss": 0.9347,
      "step": 13303
    },
    {
      "epoch": 0.9598499332635908,
      "grad_norm": 8.219325984599545,
      "learning_rate": 1.6857039305464336e-08,
      "loss": 0.9277,
      "step": 13304
    },
    {
      "epoch": 0.9599220807330183,
      "grad_norm": 4.584064471565357,
      "learning_rate": 1.679654194375235e-08,
      "loss": 0.9675,
      "step": 13305
    },
    {
      "epoch": 0.9599942282024458,
      "grad_norm": 3.7764179471957093,
      "learning_rate": 1.673615287743946e-08,
      "loss": 0.9067,
      "step": 13306
    },
    {
      "epoch": 0.9600663756718734,
      "grad_norm": 3.6026604358704866,
      "learning_rate": 1.6675872109823687e-08,
      "loss": 1.0188,
      "step": 13307
    },
    {
      "epoch": 0.9601385231413008,
      "grad_norm": 3.8086586032138614,
      "learning_rate": 1.66156996441964e-08,
      "loss": 1.019,
      "step": 13308
    },
    {
      "epoch": 0.9602106706107283,
      "grad_norm": 3.274249356271739,
      "learning_rate": 1.6555635483843645e-08,
      "loss": 0.9442,
      "step": 13309
    },
    {
      "epoch": 0.9602828180801558,
      "grad_norm": 2.893353962832035,
      "learning_rate": 1.6495679632045233e-08,
      "loss": 0.9543,
      "step": 13310
    },
    {
      "epoch": 0.9603549655495833,
      "grad_norm": 3.8916704236379345,
      "learning_rate": 1.6435832092074776e-08,
      "loss": 0.9481,
      "step": 13311
    },
    {
      "epoch": 0.9604271130190108,
      "grad_norm": 5.352170500361644,
      "learning_rate": 1.6376092867200987e-08,
      "loss": 0.9036,
      "step": 13312
    },
    {
      "epoch": 0.9604992604884384,
      "grad_norm": 3.4715663714365395,
      "learning_rate": 1.6316461960685257e-08,
      "loss": 0.9532,
      "step": 13313
    },
    {
      "epoch": 0.9605714079578659,
      "grad_norm": 0.847736583366186,
      "learning_rate": 1.625693937578432e-08,
      "loss": 0.8607,
      "step": 13314
    },
    {
      "epoch": 0.9606435554272934,
      "grad_norm": 3.319702810521313,
      "learning_rate": 1.6197525115748456e-08,
      "loss": 0.8381,
      "step": 13315
    },
    {
      "epoch": 0.9607157028967209,
      "grad_norm": 3.208291107164078,
      "learning_rate": 1.613821918382219e-08,
      "loss": 1.0137,
      "step": 13316
    },
    {
      "epoch": 0.9607878503661484,
      "grad_norm": 3.4030866020005033,
      "learning_rate": 1.6079021583243367e-08,
      "loss": 0.9325,
      "step": 13317
    },
    {
      "epoch": 0.960859997835576,
      "grad_norm": 3.5906143440639586,
      "learning_rate": 1.6019932317244966e-08,
      "loss": 0.9232,
      "step": 13318
    },
    {
      "epoch": 0.9609321453050035,
      "grad_norm": 4.106911736850724,
      "learning_rate": 1.5960951389053956e-08,
      "loss": 1.0343,
      "step": 13319
    },
    {
      "epoch": 0.9610042927744309,
      "grad_norm": 4.529744075594212,
      "learning_rate": 1.590207880189043e-08,
      "loss": 1.0183,
      "step": 13320
    },
    {
      "epoch": 0.9610764402438584,
      "grad_norm": 5.894282412221974,
      "learning_rate": 1.5843314558969588e-08,
      "loss": 0.9936,
      "step": 13321
    },
    {
      "epoch": 0.9611485877132859,
      "grad_norm": 4.532348710074271,
      "learning_rate": 1.5784658663500206e-08,
      "loss": 0.9757,
      "step": 13322
    },
    {
      "epoch": 0.9612207351827134,
      "grad_norm": 3.1041442733742057,
      "learning_rate": 1.5726111118685493e-08,
      "loss": 0.9985,
      "step": 13323
    },
    {
      "epoch": 0.961292882652141,
      "grad_norm": 3.392247668716222,
      "learning_rate": 1.5667671927722003e-08,
      "loss": 1.0144,
      "step": 13324
    },
    {
      "epoch": 0.9613650301215685,
      "grad_norm": 3.692056201048302,
      "learning_rate": 1.560934109380141e-08,
      "loss": 0.9325,
      "step": 13325
    },
    {
      "epoch": 0.961437177590996,
      "grad_norm": 3.162057198747267,
      "learning_rate": 1.5551118620108717e-08,
      "loss": 0.9313,
      "step": 13326
    },
    {
      "epoch": 0.9615093250604235,
      "grad_norm": 3.098560638981794,
      "learning_rate": 1.549300450982316e-08,
      "loss": 0.9465,
      "step": 13327
    },
    {
      "epoch": 0.961581472529851,
      "grad_norm": 4.058295318822349,
      "learning_rate": 1.5434998766118423e-08,
      "loss": 1.0926,
      "step": 13328
    },
    {
      "epoch": 0.9616536199992786,
      "grad_norm": 6.516432897304945,
      "learning_rate": 1.537710139216175e-08,
      "loss": 0.8804,
      "step": 13329
    },
    {
      "epoch": 0.9617257674687061,
      "grad_norm": 3.1268392109657492,
      "learning_rate": 1.5319312391114393e-08,
      "loss": 1.055,
      "step": 13330
    },
    {
      "epoch": 0.9617979149381335,
      "grad_norm": 6.438616313253716,
      "learning_rate": 1.526163176613271e-08,
      "loss": 1.0767,
      "step": 13331
    },
    {
      "epoch": 0.961870062407561,
      "grad_norm": 7.568672120554373,
      "learning_rate": 1.5204059520366185e-08,
      "loss": 0.9743,
      "step": 13332
    },
    {
      "epoch": 0.9619422098769885,
      "grad_norm": 19.13050433964203,
      "learning_rate": 1.5146595656958304e-08,
      "loss": 0.9241,
      "step": 13333
    },
    {
      "epoch": 0.962014357346416,
      "grad_norm": 4.399506081688873,
      "learning_rate": 1.5089240179047003e-08,
      "loss": 0.9049,
      "step": 13334
    },
    {
      "epoch": 0.9620865048158436,
      "grad_norm": 3.238925993895678,
      "learning_rate": 1.503199308976466e-08,
      "loss": 0.8594,
      "step": 13335
    },
    {
      "epoch": 0.9621586522852711,
      "grad_norm": 4.320363049616683,
      "learning_rate": 1.4974854392237003e-08,
      "loss": 0.9221,
      "step": 13336
    },
    {
      "epoch": 0.9622307997546986,
      "grad_norm": 0.7579136023967161,
      "learning_rate": 1.4917824089584198e-08,
      "loss": 0.8399,
      "step": 13337
    },
    {
      "epoch": 0.9623029472241261,
      "grad_norm": 4.157265675284353,
      "learning_rate": 1.4860902184920421e-08,
      "loss": 0.9595,
      "step": 13338
    },
    {
      "epoch": 0.9623750946935536,
      "grad_norm": 3.608178406899474,
      "learning_rate": 1.4804088681354298e-08,
      "loss": 1.0759,
      "step": 13339
    },
    {
      "epoch": 0.9624472421629812,
      "grad_norm": 2.92816920292893,
      "learning_rate": 1.4747383581987794e-08,
      "loss": 0.9483,
      "step": 13340
    },
    {
      "epoch": 0.9625193896324087,
      "grad_norm": 3.9387602227458776,
      "learning_rate": 1.4690786889917317e-08,
      "loss": 0.9548,
      "step": 13341
    },
    {
      "epoch": 0.9625915371018362,
      "grad_norm": 10.997901022663228,
      "learning_rate": 1.463429860823373e-08,
      "loss": 0.8431,
      "step": 13342
    },
    {
      "epoch": 0.9626636845712636,
      "grad_norm": 2.3509087651900304,
      "learning_rate": 1.4577918740021677e-08,
      "loss": 0.9233,
      "step": 13343
    },
    {
      "epoch": 0.9627358320406911,
      "grad_norm": 6.880240090638842,
      "learning_rate": 1.4521647288359585e-08,
      "loss": 0.9672,
      "step": 13344
    },
    {
      "epoch": 0.9628079795101186,
      "grad_norm": 4.594013647701068,
      "learning_rate": 1.4465484256320325e-08,
      "loss": 0.9507,
      "step": 13345
    },
    {
      "epoch": 0.9628801269795462,
      "grad_norm": 0.6781138256576247,
      "learning_rate": 1.4409429646971006e-08,
      "loss": 0.7875,
      "step": 13346
    },
    {
      "epoch": 0.9629522744489737,
      "grad_norm": 3.3941881996356504,
      "learning_rate": 1.4353483463372062e-08,
      "loss": 0.9723,
      "step": 13347
    },
    {
      "epoch": 0.9630244219184012,
      "grad_norm": 3.767706204892669,
      "learning_rate": 1.4297645708578832e-08,
      "loss": 0.9516,
      "step": 13348
    },
    {
      "epoch": 0.9630965693878287,
      "grad_norm": 3.460053410641545,
      "learning_rate": 1.4241916385640428e-08,
      "loss": 0.8922,
      "step": 13349
    },
    {
      "epoch": 0.9631687168572562,
      "grad_norm": 2.8658861746376116,
      "learning_rate": 1.4186295497599975e-08,
      "loss": 1.0352,
      "step": 13350
    },
    {
      "epoch": 0.9632408643266838,
      "grad_norm": 33.170566757645815,
      "learning_rate": 1.41307830474946e-08,
      "loss": 0.9451,
      "step": 13351
    },
    {
      "epoch": 0.9633130117961113,
      "grad_norm": 3.159044445037945,
      "learning_rate": 1.407537903835565e-08,
      "loss": 0.9155,
      "step": 13352
    },
    {
      "epoch": 0.9633851592655388,
      "grad_norm": 2.6023640815355202,
      "learning_rate": 1.4020083473208933e-08,
      "loss": 0.9968,
      "step": 13353
    },
    {
      "epoch": 0.9634573067349663,
      "grad_norm": 4.286195668932757,
      "learning_rate": 1.396489635507314e-08,
      "loss": 0.93,
      "step": 13354
    },
    {
      "epoch": 0.9635294542043937,
      "grad_norm": 3.8666169333274065,
      "learning_rate": 1.3909817686962533e-08,
      "loss": 0.8295,
      "step": 13355
    },
    {
      "epoch": 0.9636016016738213,
      "grad_norm": 3.941126530894375,
      "learning_rate": 1.38548474718847e-08,
      "loss": 0.9509,
      "step": 13356
    },
    {
      "epoch": 0.9636737491432488,
      "grad_norm": 2.889718073256925,
      "learning_rate": 1.3799985712841023e-08,
      "loss": 1.0121,
      "step": 13357
    },
    {
      "epoch": 0.9637458966126763,
      "grad_norm": 4.327904210157027,
      "learning_rate": 1.3745232412827323e-08,
      "loss": 1.0098,
      "step": 13358
    },
    {
      "epoch": 0.9638180440821038,
      "grad_norm": 4.513392546663035,
      "learning_rate": 1.3690587574833878e-08,
      "loss": 0.957,
      "step": 13359
    },
    {
      "epoch": 0.9638901915515313,
      "grad_norm": 5.738555797763316,
      "learning_rate": 1.36360512018443e-08,
      "loss": 0.7644,
      "step": 13360
    },
    {
      "epoch": 0.9639623390209588,
      "grad_norm": 0.7963506431992305,
      "learning_rate": 1.3581623296836652e-08,
      "loss": 0.8385,
      "step": 13361
    },
    {
      "epoch": 0.9640344864903864,
      "grad_norm": 3.5042612837915734,
      "learning_rate": 1.3527303862783002e-08,
      "loss": 0.9886,
      "step": 13362
    },
    {
      "epoch": 0.9641066339598139,
      "grad_norm": 3.2896781580768697,
      "learning_rate": 1.3473092902649862e-08,
      "loss": 1.0461,
      "step": 13363
    },
    {
      "epoch": 0.9641787814292414,
      "grad_norm": 2.2807195386373396,
      "learning_rate": 1.3418990419397092e-08,
      "loss": 0.8796,
      "step": 13364
    },
    {
      "epoch": 0.9642509288986689,
      "grad_norm": 5.219549608933436,
      "learning_rate": 1.3364996415979213e-08,
      "loss": 0.9448,
      "step": 13365
    },
    {
      "epoch": 0.9643230763680964,
      "grad_norm": 5.7485698082295125,
      "learning_rate": 1.3311110895344535e-08,
      "loss": 0.9041,
      "step": 13366
    },
    {
      "epoch": 0.9643952238375239,
      "grad_norm": 3.5259466266454798,
      "learning_rate": 1.3257333860435594e-08,
      "loss": 0.9567,
      "step": 13367
    },
    {
      "epoch": 0.9644673713069514,
      "grad_norm": 4.23312843808111,
      "learning_rate": 1.320366531418915e-08,
      "loss": 0.945,
      "step": 13368
    },
    {
      "epoch": 0.9645395187763789,
      "grad_norm": 10.229680832464034,
      "learning_rate": 1.3150105259535526e-08,
      "loss": 1.0785,
      "step": 13369
    },
    {
      "epoch": 0.9646116662458064,
      "grad_norm": 0.6809738308097874,
      "learning_rate": 1.3096653699399719e-08,
      "loss": 0.8019,
      "step": 13370
    },
    {
      "epoch": 0.9646838137152339,
      "grad_norm": 3.698786000253514,
      "learning_rate": 1.3043310636700278e-08,
      "loss": 0.8899,
      "step": 13371
    },
    {
      "epoch": 0.9647559611846614,
      "grad_norm": 2.9740137917892113,
      "learning_rate": 1.299007607435021e-08,
      "loss": 1.0008,
      "step": 13372
    },
    {
      "epoch": 0.964828108654089,
      "grad_norm": 2.8732362395637234,
      "learning_rate": 1.293695001525652e-08,
      "loss": 0.9501,
      "step": 13373
    },
    {
      "epoch": 0.9649002561235165,
      "grad_norm": 3.7872793504757527,
      "learning_rate": 1.2883932462320002e-08,
      "loss": 1.0087,
      "step": 13374
    },
    {
      "epoch": 0.964972403592944,
      "grad_norm": 9.715555236310722,
      "learning_rate": 1.2831023418435894e-08,
      "loss": 0.9625,
      "step": 13375
    },
    {
      "epoch": 0.9650445510623715,
      "grad_norm": 0.7349679054986259,
      "learning_rate": 1.277822288649344e-08,
      "loss": 0.8221,
      "step": 13376
    },
    {
      "epoch": 0.965116698531799,
      "grad_norm": 5.025574699674826,
      "learning_rate": 1.272553086937589e-08,
      "loss": 1.0351,
      "step": 13377
    },
    {
      "epoch": 0.9651888460012265,
      "grad_norm": 2.7186929981789354,
      "learning_rate": 1.2672947369960053e-08,
      "loss": 1.0037,
      "step": 13378
    },
    {
      "epoch": 0.965260993470654,
      "grad_norm": 3.7136507094086944,
      "learning_rate": 1.2620472391118075e-08,
      "loss": 1.0009,
      "step": 13379
    },
    {
      "epoch": 0.9653331409400815,
      "grad_norm": 6.580302188508507,
      "learning_rate": 1.2568105935715001e-08,
      "loss": 1.0317,
      "step": 13380
    },
    {
      "epoch": 0.965405288409509,
      "grad_norm": 3.4858400831370733,
      "learning_rate": 1.251584800661054e-08,
      "loss": 0.9844,
      "step": 13381
    },
    {
      "epoch": 0.9654774358789365,
      "grad_norm": 2.439035665497653,
      "learning_rate": 1.2463698606657968e-08,
      "loss": 1.0086,
      "step": 13382
    },
    {
      "epoch": 0.965549583348364,
      "grad_norm": 3.5023645860306214,
      "learning_rate": 1.241165773870545e-08,
      "loss": 0.992,
      "step": 13383
    },
    {
      "epoch": 0.9656217308177916,
      "grad_norm": 4.263422470731751,
      "learning_rate": 1.2359725405594268e-08,
      "loss": 0.9708,
      "step": 13384
    },
    {
      "epoch": 0.9656938782872191,
      "grad_norm": 3.5260557609212557,
      "learning_rate": 1.2307901610160599e-08,
      "loss": 0.9224,
      "step": 13385
    },
    {
      "epoch": 0.9657660257566466,
      "grad_norm": 5.568744197889531,
      "learning_rate": 1.2256186355234177e-08,
      "loss": 0.9165,
      "step": 13386
    },
    {
      "epoch": 0.9658381732260741,
      "grad_norm": 3.04067926623073,
      "learning_rate": 1.220457964363919e-08,
      "loss": 1.0246,
      "step": 13387
    },
    {
      "epoch": 0.9659103206955016,
      "grad_norm": 3.0472416583770214,
      "learning_rate": 1.2153081478193606e-08,
      "loss": 0.8927,
      "step": 13388
    },
    {
      "epoch": 0.9659824681649292,
      "grad_norm": 3.795128817175069,
      "learning_rate": 1.2101691861709173e-08,
      "loss": 0.9352,
      "step": 13389
    },
    {
      "epoch": 0.9660546156343566,
      "grad_norm": 2.89765956751967,
      "learning_rate": 1.205041079699276e-08,
      "loss": 1.0278,
      "step": 13390
    },
    {
      "epoch": 0.9661267631037841,
      "grad_norm": 4.144151100320204,
      "learning_rate": 1.1999238286843904e-08,
      "loss": 0.9899,
      "step": 13391
    },
    {
      "epoch": 0.9661989105732116,
      "grad_norm": 4.032838730133726,
      "learning_rate": 1.194817433405748e-08,
      "loss": 0.9143,
      "step": 13392
    },
    {
      "epoch": 0.9662710580426391,
      "grad_norm": 3.9177858228186753,
      "learning_rate": 1.1897218941421927e-08,
      "loss": 1.0005,
      "step": 13393
    },
    {
      "epoch": 0.9663432055120666,
      "grad_norm": 9.423533633675827,
      "learning_rate": 1.1846372111719239e-08,
      "loss": 0.8847,
      "step": 13394
    },
    {
      "epoch": 0.9664153529814942,
      "grad_norm": 3.064090082413063,
      "learning_rate": 1.1795633847726527e-08,
      "loss": 0.9906,
      "step": 13395
    },
    {
      "epoch": 0.9664875004509217,
      "grad_norm": 3.1496806754882223,
      "learning_rate": 1.1745004152213799e-08,
      "loss": 0.8761,
      "step": 13396
    },
    {
      "epoch": 0.9665596479203492,
      "grad_norm": 5.494919771506049,
      "learning_rate": 1.1694483027946622e-08,
      "loss": 0.9966,
      "step": 13397
    },
    {
      "epoch": 0.9666317953897767,
      "grad_norm": 7.367848954183977,
      "learning_rate": 1.164407047768301e-08,
      "loss": 0.9977,
      "step": 13398
    },
    {
      "epoch": 0.9667039428592042,
      "grad_norm": 4.874166955441872,
      "learning_rate": 1.1593766504176095e-08,
      "loss": 0.9983,
      "step": 13399
    },
    {
      "epoch": 0.9667760903286318,
      "grad_norm": 3.0921509588780163,
      "learning_rate": 1.154357111017279e-08,
      "loss": 0.8725,
      "step": 13400
    },
    {
      "epoch": 0.9668482377980593,
      "grad_norm": 4.665317999007179,
      "learning_rate": 1.1493484298414013e-08,
      "loss": 0.9788,
      "step": 13401
    },
    {
      "epoch": 0.9669203852674867,
      "grad_norm": 5.386811696136574,
      "learning_rate": 1.144350607163469e-08,
      "loss": 0.853,
      "step": 13402
    },
    {
      "epoch": 0.9669925327369142,
      "grad_norm": 5.438210978010822,
      "learning_rate": 1.139363643256419e-08,
      "loss": 0.8584,
      "step": 13403
    },
    {
      "epoch": 0.9670646802063417,
      "grad_norm": 8.772810929735504,
      "learning_rate": 1.1343875383925894e-08,
      "loss": 0.9724,
      "step": 13404
    },
    {
      "epoch": 0.9671368276757693,
      "grad_norm": 3.5822248000973667,
      "learning_rate": 1.1294222928436736e-08,
      "loss": 0.9523,
      "step": 13405
    },
    {
      "epoch": 0.9672089751451968,
      "grad_norm": 17.252113102288096,
      "learning_rate": 1.1244679068807884e-08,
      "loss": 0.9972,
      "step": 13406
    },
    {
      "epoch": 0.9672811226146243,
      "grad_norm": 2.9523984630932674,
      "learning_rate": 1.1195243807745392e-08,
      "loss": 1.0622,
      "step": 13407
    },
    {
      "epoch": 0.9673532700840518,
      "grad_norm": 3.934553588013599,
      "learning_rate": 1.1145917147947992e-08,
      "loss": 0.9933,
      "step": 13408
    },
    {
      "epoch": 0.9674254175534793,
      "grad_norm": 3.351974782061344,
      "learning_rate": 1.109669909210953e-08,
      "loss": 0.9894,
      "step": 13409
    },
    {
      "epoch": 0.9674975650229068,
      "grad_norm": 4.081704625838069,
      "learning_rate": 1.1047589642917853e-08,
      "loss": 1.0032,
      "step": 13410
    },
    {
      "epoch": 0.9675697124923344,
      "grad_norm": 6.628710877074494,
      "learning_rate": 1.0998588803054375e-08,
      "loss": 0.9618,
      "step": 13411
    },
    {
      "epoch": 0.9676418599617619,
      "grad_norm": 2.6983299282524515,
      "learning_rate": 1.0949696575194956e-08,
      "loss": 1.0056,
      "step": 13412
    },
    {
      "epoch": 0.9677140074311894,
      "grad_norm": 4.0668574512525755,
      "learning_rate": 1.0900912962009457e-08,
      "loss": 0.9931,
      "step": 13413
    },
    {
      "epoch": 0.9677861549006168,
      "grad_norm": 10.048380737429088,
      "learning_rate": 1.0852237966161748e-08,
      "loss": 0.8538,
      "step": 13414
    },
    {
      "epoch": 0.9678583023700443,
      "grad_norm": 5.357651146889612,
      "learning_rate": 1.0803671590309481e-08,
      "loss": 0.8511,
      "step": 13415
    },
    {
      "epoch": 0.9679304498394719,
      "grad_norm": 6.657685747352307,
      "learning_rate": 1.0755213837105204e-08,
      "loss": 1.0885,
      "step": 13416
    },
    {
      "epoch": 0.9680025973088994,
      "grad_norm": 4.1650339106430625,
      "learning_rate": 1.0706864709194797e-08,
      "loss": 0.9366,
      "step": 13417
    },
    {
      "epoch": 0.9680747447783269,
      "grad_norm": 3.6146907882035677,
      "learning_rate": 1.0658624209218148e-08,
      "loss": 0.984,
      "step": 13418
    },
    {
      "epoch": 0.9681468922477544,
      "grad_norm": 4.235306190492215,
      "learning_rate": 1.0610492339810039e-08,
      "loss": 0.9073,
      "step": 13419
    },
    {
      "epoch": 0.9682190397171819,
      "grad_norm": 3.874715363908187,
      "learning_rate": 1.0562469103598371e-08,
      "loss": 1.0075,
      "step": 13420
    },
    {
      "epoch": 0.9682911871866094,
      "grad_norm": 3.840274138599317,
      "learning_rate": 1.0514554503205486e-08,
      "loss": 1.0034,
      "step": 13421
    },
    {
      "epoch": 0.968363334656037,
      "grad_norm": 3.00349778671018,
      "learning_rate": 1.0466748541247961e-08,
      "loss": 0.9124,
      "step": 13422
    },
    {
      "epoch": 0.9684354821254645,
      "grad_norm": 6.6981663714226904,
      "learning_rate": 1.0419051220336373e-08,
      "loss": 0.9365,
      "step": 13423
    },
    {
      "epoch": 0.968507629594892,
      "grad_norm": 6.261712048694168,
      "learning_rate": 1.0371462543075304e-08,
      "loss": 1.018,
      "step": 13424
    },
    {
      "epoch": 0.9685797770643194,
      "grad_norm": 3.685198825560316,
      "learning_rate": 1.0323982512063124e-08,
      "loss": 0.991,
      "step": 13425
    },
    {
      "epoch": 0.9686519245337469,
      "grad_norm": 4.26399663788339,
      "learning_rate": 1.0276611129892865e-08,
      "loss": 1.0559,
      "step": 13426
    },
    {
      "epoch": 0.9687240720031745,
      "grad_norm": 3.64718155410095,
      "learning_rate": 1.0229348399151128e-08,
      "loss": 0.8889,
      "step": 13427
    },
    {
      "epoch": 0.968796219472602,
      "grad_norm": 3.143070919643431,
      "learning_rate": 1.0182194322418736e-08,
      "loss": 0.9811,
      "step": 13428
    },
    {
      "epoch": 0.9688683669420295,
      "grad_norm": 3.375011161503173,
      "learning_rate": 1.013514890227074e-08,
      "loss": 0.8857,
      "step": 13429
    },
    {
      "epoch": 0.968940514411457,
      "grad_norm": 3.649569807437297,
      "learning_rate": 1.0088212141276197e-08,
      "loss": 0.9142,
      "step": 13430
    },
    {
      "epoch": 0.9690126618808845,
      "grad_norm": 5.395602639862394,
      "learning_rate": 1.0041384041997725e-08,
      "loss": 1.0324,
      "step": 13431
    },
    {
      "epoch": 0.969084809350312,
      "grad_norm": 2.8984396386974085,
      "learning_rate": 9.994664606992831e-09,
      "loss": 0.9453,
      "step": 13432
    },
    {
      "epoch": 0.9691569568197396,
      "grad_norm": 3.583468915163404,
      "learning_rate": 9.948053838812365e-09,
      "loss": 0.9298,
      "step": 13433
    },
    {
      "epoch": 0.9692291042891671,
      "grad_norm": 7.905829980119769,
      "learning_rate": 9.901551740002068e-09,
      "loss": 1.0816,
      "step": 13434
    },
    {
      "epoch": 0.9693012517585946,
      "grad_norm": 3.427696675585063,
      "learning_rate": 9.855158313100798e-09,
      "loss": 1.0112,
      "step": 13435
    },
    {
      "epoch": 0.9693733992280221,
      "grad_norm": 6.209853973883212,
      "learning_rate": 9.808873560642083e-09,
      "loss": 0.8833,
      "step": 13436
    },
    {
      "epoch": 0.9694455466974495,
      "grad_norm": 10.239199647024229,
      "learning_rate": 9.762697485153237e-09,
      "loss": 0.976,
      "step": 13437
    },
    {
      "epoch": 0.969517694166877,
      "grad_norm": 5.318073714727761,
      "learning_rate": 9.71663008915602e-09,
      "loss": 0.9917,
      "step": 13438
    },
    {
      "epoch": 0.9695898416363046,
      "grad_norm": 3.660836237029144,
      "learning_rate": 9.67067137516575e-09,
      "loss": 0.9645,
      "step": 13439
    },
    {
      "epoch": 0.9696619891057321,
      "grad_norm": 3.701321577200052,
      "learning_rate": 9.624821345692202e-09,
      "loss": 0.9114,
      "step": 13440
    },
    {
      "epoch": 0.9697341365751596,
      "grad_norm": 4.798387916380807,
      "learning_rate": 9.579080003239148e-09,
      "loss": 0.9589,
      "step": 13441
    },
    {
      "epoch": 0.9698062840445871,
      "grad_norm": 6.325142466030008,
      "learning_rate": 9.533447350304147e-09,
      "loss": 0.9823,
      "step": 13442
    },
    {
      "epoch": 0.9698784315140146,
      "grad_norm": 3.2243473309259376,
      "learning_rate": 9.487923389378982e-09,
      "loss": 0.8899,
      "step": 13443
    },
    {
      "epoch": 0.9699505789834422,
      "grad_norm": 2.6973099977681576,
      "learning_rate": 9.442508122949889e-09,
      "loss": 0.8863,
      "step": 13444
    },
    {
      "epoch": 0.9700227264528697,
      "grad_norm": 4.558634753725187,
      "learning_rate": 9.397201553496658e-09,
      "loss": 0.9828,
      "step": 13445
    },
    {
      "epoch": 0.9700948739222972,
      "grad_norm": 4.262690502395625,
      "learning_rate": 9.352003683493093e-09,
      "loss": 0.8429,
      "step": 13446
    },
    {
      "epoch": 0.9701670213917247,
      "grad_norm": 6.348695828333811,
      "learning_rate": 9.30691451540766e-09,
      "loss": 0.8693,
      "step": 13447
    },
    {
      "epoch": 0.9702391688611522,
      "grad_norm": 4.257239290902601,
      "learning_rate": 9.26193405170239e-09,
      "loss": 0.8678,
      "step": 13448
    },
    {
      "epoch": 0.9703113163305797,
      "grad_norm": 4.908850691277229,
      "learning_rate": 9.217062294833321e-09,
      "loss": 0.9393,
      "step": 13449
    },
    {
      "epoch": 0.9703834638000072,
      "grad_norm": 4.128644546977626,
      "learning_rate": 9.172299247250714e-09,
      "loss": 0.9297,
      "step": 13450
    },
    {
      "epoch": 0.9704556112694347,
      "grad_norm": 5.850700284336797,
      "learning_rate": 9.127644911399278e-09,
      "loss": 0.9192,
      "step": 13451
    },
    {
      "epoch": 0.9705277587388622,
      "grad_norm": 1.9200636228311823,
      "learning_rate": 9.08309928971729e-09,
      "loss": 0.9229,
      "step": 13452
    },
    {
      "epoch": 0.9705999062082897,
      "grad_norm": 2.7114336510696386,
      "learning_rate": 9.038662384636797e-09,
      "loss": 0.926,
      "step": 13453
    },
    {
      "epoch": 0.9706720536777173,
      "grad_norm": 2.5322189360362883,
      "learning_rate": 8.994334198584974e-09,
      "loss": 0.8898,
      "step": 13454
    },
    {
      "epoch": 0.9707442011471448,
      "grad_norm": 3.7545588439158544,
      "learning_rate": 8.950114733981884e-09,
      "loss": 0.9048,
      "step": 13455
    },
    {
      "epoch": 0.9708163486165723,
      "grad_norm": 3.9070863362505843,
      "learning_rate": 8.906003993242484e-09,
      "loss": 0.8548,
      "step": 13456
    },
    {
      "epoch": 0.9708884960859998,
      "grad_norm": 4.132938952755899,
      "learning_rate": 8.862001978775512e-09,
      "loss": 1.0409,
      "step": 13457
    },
    {
      "epoch": 0.9709606435554273,
      "grad_norm": 4.298270591898183,
      "learning_rate": 8.818108692983495e-09,
      "loss": 1.0063,
      "step": 13458
    },
    {
      "epoch": 0.9710327910248548,
      "grad_norm": 4.556781686495295,
      "learning_rate": 8.774324138263622e-09,
      "loss": 1.0011,
      "step": 13459
    },
    {
      "epoch": 0.9711049384942824,
      "grad_norm": 4.254519359727936,
      "learning_rate": 8.73064831700665e-09,
      "loss": 0.8945,
      "step": 13460
    },
    {
      "epoch": 0.9711770859637098,
      "grad_norm": 5.718794692234859,
      "learning_rate": 8.68708123159756e-09,
      "loss": 0.8918,
      "step": 13461
    },
    {
      "epoch": 0.9712492334331373,
      "grad_norm": 7.177807716462036,
      "learning_rate": 8.643622884415335e-09,
      "loss": 0.9745,
      "step": 13462
    },
    {
      "epoch": 0.9713213809025648,
      "grad_norm": 4.339645971809743,
      "learning_rate": 8.60027327783297e-09,
      "loss": 0.9162,
      "step": 13463
    },
    {
      "epoch": 0.9713935283719923,
      "grad_norm": 3.491734281141693,
      "learning_rate": 8.557032414217902e-09,
      "loss": 0.8855,
      "step": 13464
    },
    {
      "epoch": 0.9714656758414199,
      "grad_norm": 2.8564680154012323,
      "learning_rate": 8.51390029593113e-09,
      "loss": 0.9025,
      "step": 13465
    },
    {
      "epoch": 0.9715378233108474,
      "grad_norm": 7.458256328618567,
      "learning_rate": 8.470876925328107e-09,
      "loss": 0.9672,
      "step": 13466
    },
    {
      "epoch": 0.9716099707802749,
      "grad_norm": 6.868879888693131,
      "learning_rate": 8.427962304758063e-09,
      "loss": 0.9546,
      "step": 13467
    },
    {
      "epoch": 0.9716821182497024,
      "grad_norm": 4.703293299911269,
      "learning_rate": 8.385156436564677e-09,
      "loss": 0.9143,
      "step": 13468
    },
    {
      "epoch": 0.9717542657191299,
      "grad_norm": 4.417423231475726,
      "learning_rate": 8.342459323084972e-09,
      "loss": 1.0103,
      "step": 13469
    },
    {
      "epoch": 0.9718264131885574,
      "grad_norm": 6.479943657147389,
      "learning_rate": 8.299870966650635e-09,
      "loss": 0.9764,
      "step": 13470
    },
    {
      "epoch": 0.971898560657985,
      "grad_norm": 1.9700889878095116,
      "learning_rate": 8.257391369587585e-09,
      "loss": 0.825,
      "step": 13471
    },
    {
      "epoch": 0.9719707081274125,
      "grad_norm": 4.273767992033663,
      "learning_rate": 8.215020534215078e-09,
      "loss": 0.8743,
      "step": 13472
    },
    {
      "epoch": 0.9720428555968399,
      "grad_norm": 4.741115090592729,
      "learning_rate": 8.172758462846818e-09,
      "loss": 0.9562,
      "step": 13473
    },
    {
      "epoch": 0.9721150030662674,
      "grad_norm": 3.950361047009931,
      "learning_rate": 8.130605157790737e-09,
      "loss": 0.8538,
      "step": 13474
    },
    {
      "epoch": 0.9721871505356949,
      "grad_norm": 3.5981077890413538,
      "learning_rate": 8.088560621348995e-09,
      "loss": 0.9672,
      "step": 13475
    },
    {
      "epoch": 0.9722592980051225,
      "grad_norm": 10.037181776045061,
      "learning_rate": 8.046624855816864e-09,
      "loss": 0.8792,
      "step": 13476
    },
    {
      "epoch": 0.97233144547455,
      "grad_norm": 4.916389435964117,
      "learning_rate": 8.004797863484514e-09,
      "loss": 0.9461,
      "step": 13477
    },
    {
      "epoch": 0.9724035929439775,
      "grad_norm": 3.0447373936858315,
      "learning_rate": 7.963079646636339e-09,
      "loss": 0.8897,
      "step": 13478
    },
    {
      "epoch": 0.972475740413405,
      "grad_norm": 4.6710371714252,
      "learning_rate": 7.921470207549852e-09,
      "loss": 1.0359,
      "step": 13479
    },
    {
      "epoch": 0.9725478878828325,
      "grad_norm": 7.278398602176045,
      "learning_rate": 7.879969548497677e-09,
      "loss": 1.061,
      "step": 13480
    },
    {
      "epoch": 0.97262003535226,
      "grad_norm": 3.392266504605592,
      "learning_rate": 7.838577671745783e-09,
      "loss": 0.9959,
      "step": 13481
    },
    {
      "epoch": 0.9726921828216876,
      "grad_norm": 5.159775332209905,
      "learning_rate": 7.797294579554137e-09,
      "loss": 0.9158,
      "step": 13482
    },
    {
      "epoch": 0.9727643302911151,
      "grad_norm": 2.4292900520863143,
      "learning_rate": 7.756120274177602e-09,
      "loss": 0.9803,
      "step": 13483
    },
    {
      "epoch": 0.9728364777605425,
      "grad_norm": 7.740077403861892,
      "learning_rate": 7.715054757864158e-09,
      "loss": 0.8647,
      "step": 13484
    },
    {
      "epoch": 0.97290862522997,
      "grad_norm": 4.3431330180921,
      "learning_rate": 7.674098032856679e-09,
      "loss": 0.9553,
      "step": 13485
    },
    {
      "epoch": 0.9729807726993975,
      "grad_norm": 3.3163309077498986,
      "learning_rate": 7.633250101391154e-09,
      "loss": 0.9701,
      "step": 13486
    },
    {
      "epoch": 0.973052920168825,
      "grad_norm": 3.7320378224213373,
      "learning_rate": 7.59251096569824e-09,
      "loss": 0.9148,
      "step": 13487
    },
    {
      "epoch": 0.9731250676382526,
      "grad_norm": 4.117897623917386,
      "learning_rate": 7.551880628002827e-09,
      "loss": 0.9535,
      "step": 13488
    },
    {
      "epoch": 0.9731972151076801,
      "grad_norm": 8.428133816938983,
      "learning_rate": 7.511359090523361e-09,
      "loss": 0.9771,
      "step": 13489
    },
    {
      "epoch": 0.9732693625771076,
      "grad_norm": 9.05605407882018,
      "learning_rate": 7.470946355472518e-09,
      "loss": 0.9849,
      "step": 13490
    },
    {
      "epoch": 0.9733415100465351,
      "grad_norm": 5.952900078797269,
      "learning_rate": 7.430642425057421e-09,
      "loss": 0.8967,
      "step": 13491
    },
    {
      "epoch": 0.9734136575159626,
      "grad_norm": 3.1039696103899153,
      "learning_rate": 7.390447301478753e-09,
      "loss": 0.947,
      "step": 13492
    },
    {
      "epoch": 0.9734858049853902,
      "grad_norm": 3.7766304486640427,
      "learning_rate": 7.350360986931204e-09,
      "loss": 0.9224,
      "step": 13493
    },
    {
      "epoch": 0.9735579524548177,
      "grad_norm": 3.254267165373853,
      "learning_rate": 7.31038348360391e-09,
      "loss": 0.9711,
      "step": 13494
    },
    {
      "epoch": 0.9736300999242452,
      "grad_norm": 3.5493895905858626,
      "learning_rate": 7.270514793679794e-09,
      "loss": 0.9972,
      "step": 13495
    },
    {
      "epoch": 0.9737022473936726,
      "grad_norm": 3.876888307274601,
      "learning_rate": 7.230754919336224e-09,
      "loss": 0.9839,
      "step": 13496
    },
    {
      "epoch": 0.9737743948631001,
      "grad_norm": 3.10181124287894,
      "learning_rate": 7.19110386274413e-09,
      "loss": 0.9801,
      "step": 13497
    },
    {
      "epoch": 0.9738465423325277,
      "grad_norm": 7.957610119085569,
      "learning_rate": 7.151561626068669e-09,
      "loss": 0.9478,
      "step": 13498
    },
    {
      "epoch": 0.9739186898019552,
      "grad_norm": 3.8982536563067174,
      "learning_rate": 7.112128211469226e-09,
      "loss": 0.9331,
      "step": 13499
    },
    {
      "epoch": 0.9739908372713827,
      "grad_norm": 7.619913624752031,
      "learning_rate": 7.072803621098966e-09,
      "loss": 1.0484,
      "step": 13500
    },
    {
      "epoch": 0.9740629847408102,
      "grad_norm": 3.5436114787052655,
      "learning_rate": 7.033587857105283e-09,
      "loss": 0.8736,
      "step": 13501
    },
    {
      "epoch": 0.9741351322102377,
      "grad_norm": 3.041757051004637,
      "learning_rate": 6.994480921629798e-09,
      "loss": 1.0079,
      "step": 13502
    },
    {
      "epoch": 0.9742072796796653,
      "grad_norm": 3.3038323199970705,
      "learning_rate": 6.95548281680769e-09,
      "loss": 0.7959,
      "step": 13503
    },
    {
      "epoch": 0.9742794271490928,
      "grad_norm": 3.2067985496709,
      "learning_rate": 6.916593544768812e-09,
      "loss": 0.8309,
      "step": 13504
    },
    {
      "epoch": 0.9743515746185203,
      "grad_norm": 4.891228684287025,
      "learning_rate": 6.8778131076365765e-09,
      "loss": 0.9312,
      "step": 13505
    },
    {
      "epoch": 0.9744237220879478,
      "grad_norm": 2.805188657566727,
      "learning_rate": 6.839141507528623e-09,
      "loss": 0.8844,
      "step": 13506
    },
    {
      "epoch": 0.9744958695573753,
      "grad_norm": 3.618412477766034,
      "learning_rate": 6.800578746556817e-09,
      "loss": 0.9748,
      "step": 13507
    },
    {
      "epoch": 0.9745680170268027,
      "grad_norm": 6.200553303302041,
      "learning_rate": 6.7621248268268095e-09,
      "loss": 0.9138,
      "step": 13508
    },
    {
      "epoch": 0.9746401644962303,
      "grad_norm": 3.741071753724608,
      "learning_rate": 6.723779750438474e-09,
      "loss": 1.0455,
      "step": 13509
    },
    {
      "epoch": 0.9747123119656578,
      "grad_norm": 2.702502228544518,
      "learning_rate": 6.6855435194854705e-09,
      "loss": 0.9051,
      "step": 13510
    },
    {
      "epoch": 0.9747844594350853,
      "grad_norm": 3.064342333879977,
      "learning_rate": 6.647416136056127e-09,
      "loss": 0.796,
      "step": 13511
    },
    {
      "epoch": 0.9748566069045128,
      "grad_norm": 3.745194853334192,
      "learning_rate": 6.609397602232336e-09,
      "loss": 0.8339,
      "step": 13512
    },
    {
      "epoch": 0.9749287543739403,
      "grad_norm": 3.977927702440954,
      "learning_rate": 6.571487920089769e-09,
      "loss": 0.9609,
      "step": 13513
    },
    {
      "epoch": 0.9750009018433679,
      "grad_norm": 2.6090326141681732,
      "learning_rate": 6.533687091698991e-09,
      "loss": 0.9475,
      "step": 13514
    },
    {
      "epoch": 0.9750730493127954,
      "grad_norm": 4.118662271448097,
      "learning_rate": 6.495995119124131e-09,
      "loss": 0.9932,
      "step": 13515
    },
    {
      "epoch": 0.9751451967822229,
      "grad_norm": 3.1464918007458156,
      "learning_rate": 6.458412004423097e-09,
      "loss": 0.8845,
      "step": 13516
    },
    {
      "epoch": 0.9752173442516504,
      "grad_norm": 4.54440838478993,
      "learning_rate": 6.42093774964847e-09,
      "loss": 0.9138,
      "step": 13517
    },
    {
      "epoch": 0.9752894917210779,
      "grad_norm": 3.8948905878529136,
      "learning_rate": 6.383572356846389e-09,
      "loss": 0.8916,
      "step": 13518
    },
    {
      "epoch": 0.9753616391905054,
      "grad_norm": 3.5198814565031125,
      "learning_rate": 6.346315828057225e-09,
      "loss": 1.0828,
      "step": 13519
    },
    {
      "epoch": 0.9754337866599329,
      "grad_norm": 3.734503452533858,
      "learning_rate": 6.30916816531557e-09,
      "loss": 0.8545,
      "step": 13520
    },
    {
      "epoch": 0.9755059341293604,
      "grad_norm": 2.8851383410098617,
      "learning_rate": 6.272129370650025e-09,
      "loss": 0.967,
      "step": 13521
    },
    {
      "epoch": 0.9755780815987879,
      "grad_norm": 4.210601396895123,
      "learning_rate": 6.2351994460827505e-09,
      "loss": 0.9935,
      "step": 13522
    },
    {
      "epoch": 0.9756502290682154,
      "grad_norm": 7.525288988818594,
      "learning_rate": 6.198378393630576e-09,
      "loss": 0.9865,
      "step": 13523
    },
    {
      "epoch": 0.9757223765376429,
      "grad_norm": 0.6518813910312748,
      "learning_rate": 6.16166621530434e-09,
      "loss": 0.7819,
      "step": 13524
    },
    {
      "epoch": 0.9757945240070705,
      "grad_norm": 4.461882536240943,
      "learning_rate": 6.125062913108437e-09,
      "loss": 0.8744,
      "step": 13525
    },
    {
      "epoch": 0.975866671476498,
      "grad_norm": 4.372468488436829,
      "learning_rate": 6.088568489041712e-09,
      "loss": 0.8033,
      "step": 13526
    },
    {
      "epoch": 0.9759388189459255,
      "grad_norm": 3.0555200247191903,
      "learning_rate": 6.052182945097239e-09,
      "loss": 0.9451,
      "step": 13527
    },
    {
      "epoch": 0.976010966415353,
      "grad_norm": 2.9920718018795167,
      "learning_rate": 6.015906283261429e-09,
      "loss": 0.9231,
      "step": 13528
    },
    {
      "epoch": 0.9760831138847805,
      "grad_norm": 4.3014153923378995,
      "learning_rate": 5.979738505515808e-09,
      "loss": 0.9654,
      "step": 13529
    },
    {
      "epoch": 0.976155261354208,
      "grad_norm": 3.72804189303194,
      "learning_rate": 5.943679613834796e-09,
      "loss": 0.9654,
      "step": 13530
    },
    {
      "epoch": 0.9762274088236355,
      "grad_norm": 4.913173188731566,
      "learning_rate": 5.907729610187928e-09,
      "loss": 0.9259,
      "step": 13531
    },
    {
      "epoch": 0.976299556293063,
      "grad_norm": 3.500409374819008,
      "learning_rate": 5.8718884965378575e-09,
      "loss": 0.9044,
      "step": 13532
    },
    {
      "epoch": 0.9763717037624905,
      "grad_norm": 6.617975217374511,
      "learning_rate": 5.83615627484213e-09,
      "loss": 0.7218,
      "step": 13533
    },
    {
      "epoch": 0.976443851231918,
      "grad_norm": 3.106006580539863,
      "learning_rate": 5.800532947051628e-09,
      "loss": 0.9122,
      "step": 13534
    },
    {
      "epoch": 0.9765159987013455,
      "grad_norm": 3.714747478790013,
      "learning_rate": 5.765018515111686e-09,
      "loss": 0.9604,
      "step": 13535
    },
    {
      "epoch": 0.976588146170773,
      "grad_norm": 4.835991336849458,
      "learning_rate": 5.729612980961862e-09,
      "loss": 0.9988,
      "step": 13536
    },
    {
      "epoch": 0.9766602936402006,
      "grad_norm": 9.044569917642143,
      "learning_rate": 5.694316346535277e-09,
      "loss": 0.8971,
      "step": 13537
    },
    {
      "epoch": 0.9767324411096281,
      "grad_norm": 3.8632341002083597,
      "learning_rate": 5.659128613759501e-09,
      "loss": 0.8299,
      "step": 13538
    },
    {
      "epoch": 0.9768045885790556,
      "grad_norm": 5.330057548799144,
      "learning_rate": 5.624049784555884e-09,
      "loss": 0.9473,
      "step": 13539
    },
    {
      "epoch": 0.9768767360484831,
      "grad_norm": 4.252695463017764,
      "learning_rate": 5.589079860839785e-09,
      "loss": 0.9669,
      "step": 13540
    },
    {
      "epoch": 0.9769488835179106,
      "grad_norm": 0.8172845815610283,
      "learning_rate": 5.554218844521008e-09,
      "loss": 0.9048,
      "step": 13541
    },
    {
      "epoch": 0.9770210309873382,
      "grad_norm": 4.792030298560749,
      "learning_rate": 5.519466737503364e-09,
      "loss": 0.9268,
      "step": 13542
    },
    {
      "epoch": 0.9770931784567656,
      "grad_norm": 2.978963369266586,
      "learning_rate": 5.484823541684225e-09,
      "loss": 0.9395,
      "step": 13543
    },
    {
      "epoch": 0.9771653259261931,
      "grad_norm": 3.238117503776186,
      "learning_rate": 5.450289258955409e-09,
      "loss": 0.8908,
      "step": 13544
    },
    {
      "epoch": 0.9772374733956206,
      "grad_norm": 3.9231910473203335,
      "learning_rate": 5.415863891202743e-09,
      "loss": 1.0068,
      "step": 13545
    },
    {
      "epoch": 0.9773096208650481,
      "grad_norm": 3.1867440019052675,
      "learning_rate": 5.381547440306056e-09,
      "loss": 0.8889,
      "step": 13546
    },
    {
      "epoch": 0.9773817683344757,
      "grad_norm": 3.7405655756590694,
      "learning_rate": 5.347339908139182e-09,
      "loss": 0.9428,
      "step": 13547
    },
    {
      "epoch": 0.9774539158039032,
      "grad_norm": 2.7991695262382903,
      "learning_rate": 5.313241296570181e-09,
      "loss": 0.9473,
      "step": 13548
    },
    {
      "epoch": 0.9775260632733307,
      "grad_norm": 2.747763417800269,
      "learning_rate": 5.279251607461122e-09,
      "loss": 0.9926,
      "step": 13549
    },
    {
      "epoch": 0.9775982107427582,
      "grad_norm": 0.7129434226355146,
      "learning_rate": 5.245370842667851e-09,
      "loss": 0.8101,
      "step": 13550
    },
    {
      "epoch": 0.9776703582121857,
      "grad_norm": 4.186689839349589,
      "learning_rate": 5.211599004040446e-09,
      "loss": 0.901,
      "step": 13551
    },
    {
      "epoch": 0.9777425056816132,
      "grad_norm": 9.152531724888894,
      "learning_rate": 5.177936093423207e-09,
      "loss": 0.9486,
      "step": 13552
    },
    {
      "epoch": 0.9778146531510408,
      "grad_norm": 3.378008172964324,
      "learning_rate": 5.144382112654222e-09,
      "loss": 0.929,
      "step": 13553
    },
    {
      "epoch": 0.9778868006204683,
      "grad_norm": 6.020300536907397,
      "learning_rate": 5.1109370635660235e-09,
      "loss": 1.0492,
      "step": 13554
    },
    {
      "epoch": 0.9779589480898957,
      "grad_norm": 3.965834380169275,
      "learning_rate": 5.077600947984484e-09,
      "loss": 0.9401,
      "step": 13555
    },
    {
      "epoch": 0.9780310955593232,
      "grad_norm": 4.21152108701037,
      "learning_rate": 5.04437376773037e-09,
      "loss": 0.9713,
      "step": 13556
    },
    {
      "epoch": 0.9781032430287507,
      "grad_norm": 9.892220463364167,
      "learning_rate": 5.011255524618007e-09,
      "loss": 1.0222,
      "step": 13557
    },
    {
      "epoch": 0.9781753904981783,
      "grad_norm": 4.7086034151556895,
      "learning_rate": 4.978246220455506e-09,
      "loss": 0.8805,
      "step": 13558
    },
    {
      "epoch": 0.9782475379676058,
      "grad_norm": 3.5545313538850594,
      "learning_rate": 4.945345857046091e-09,
      "loss": 0.8974,
      "step": 13559
    },
    {
      "epoch": 0.9783196854370333,
      "grad_norm": 4.26210743279387,
      "learning_rate": 4.9125544361854346e-09,
      "loss": 0.9276,
      "step": 13560
    },
    {
      "epoch": 0.9783918329064608,
      "grad_norm": 0.8152183395877796,
      "learning_rate": 4.879871959664772e-09,
      "loss": 0.8443,
      "step": 13561
    },
    {
      "epoch": 0.9784639803758883,
      "grad_norm": 4.02191168732053,
      "learning_rate": 4.847298429268676e-09,
      "loss": 0.9422,
      "step": 13562
    },
    {
      "epoch": 0.9785361278453159,
      "grad_norm": 4.475487811390211,
      "learning_rate": 4.814833846775946e-09,
      "loss": 0.977,
      "step": 13563
    },
    {
      "epoch": 0.9786082753147434,
      "grad_norm": 3.4312878049008124,
      "learning_rate": 4.7824782139589405e-09,
      "loss": 1.0061,
      "step": 13564
    },
    {
      "epoch": 0.9786804227841709,
      "grad_norm": 2.856698039033573,
      "learning_rate": 4.750231532584914e-09,
      "loss": 1.0276,
      "step": 13565
    },
    {
      "epoch": 0.9787525702535984,
      "grad_norm": 3.009561718687029,
      "learning_rate": 4.71809380441468e-09,
      "loss": 0.9429,
      "step": 13566
    },
    {
      "epoch": 0.9788247177230258,
      "grad_norm": 17.395685191484986,
      "learning_rate": 4.686065031202835e-09,
      "loss": 0.9594,
      "step": 13567
    },
    {
      "epoch": 0.9788968651924533,
      "grad_norm": 2.8893810998124865,
      "learning_rate": 4.654145214698646e-09,
      "loss": 0.9983,
      "step": 13568
    },
    {
      "epoch": 0.9789690126618809,
      "grad_norm": 5.336957931687265,
      "learning_rate": 4.6223343566451636e-09,
      "loss": 0.9893,
      "step": 13569
    },
    {
      "epoch": 0.9790411601313084,
      "grad_norm": 5.202167961380437,
      "learning_rate": 4.590632458779442e-09,
      "loss": 0.9982,
      "step": 13570
    },
    {
      "epoch": 0.9791133076007359,
      "grad_norm": 4.502028961535878,
      "learning_rate": 4.55903952283232e-09,
      "loss": 0.8558,
      "step": 13571
    },
    {
      "epoch": 0.9791854550701634,
      "grad_norm": 0.896775906767385,
      "learning_rate": 4.527555550529305e-09,
      "loss": 0.9548,
      "step": 13572
    },
    {
      "epoch": 0.9792576025395909,
      "grad_norm": 6.28044829305351,
      "learning_rate": 4.4961805435894676e-09,
      "loss": 0.934,
      "step": 13573
    },
    {
      "epoch": 0.9793297500090185,
      "grad_norm": 5.001727187339128,
      "learning_rate": 4.464914503726103e-09,
      "loss": 0.8869,
      "step": 13574
    },
    {
      "epoch": 0.979401897478446,
      "grad_norm": 2.7693035424891788,
      "learning_rate": 4.433757432646512e-09,
      "loss": 0.8959,
      "step": 13575
    },
    {
      "epoch": 0.9794740449478735,
      "grad_norm": 3.6892109150742955,
      "learning_rate": 4.402709332052224e-09,
      "loss": 1.0097,
      "step": 13576
    },
    {
      "epoch": 0.979546192417301,
      "grad_norm": 3.8643956442992105,
      "learning_rate": 4.3717702036383255e-09,
      "loss": 1.0118,
      "step": 13577
    },
    {
      "epoch": 0.9796183398867284,
      "grad_norm": 5.33730029552208,
      "learning_rate": 4.340940049094577e-09,
      "loss": 0.8393,
      "step": 13578
    },
    {
      "epoch": 0.9796904873561559,
      "grad_norm": 3.871311154693913,
      "learning_rate": 4.3102188701042985e-09,
      "loss": 0.9359,
      "step": 13579
    },
    {
      "epoch": 0.9797626348255835,
      "grad_norm": 3.6694980151218775,
      "learning_rate": 4.279606668345481e-09,
      "loss": 0.8557,
      "step": 13580
    },
    {
      "epoch": 0.979834782295011,
      "grad_norm": 9.345618427973054,
      "learning_rate": 4.249103445489232e-09,
      "loss": 0.9026,
      "step": 13581
    },
    {
      "epoch": 0.9799069297644385,
      "grad_norm": 0.7210802368352683,
      "learning_rate": 4.218709203201332e-09,
      "loss": 0.8525,
      "step": 13582
    },
    {
      "epoch": 0.979979077233866,
      "grad_norm": 3.367528628175932,
      "learning_rate": 4.188423943141783e-09,
      "loss": 0.9997,
      "step": 13583
    },
    {
      "epoch": 0.9800512247032935,
      "grad_norm": 3.515345040936371,
      "learning_rate": 4.1582476669641545e-09,
      "loss": 0.9276,
      "step": 13584
    },
    {
      "epoch": 0.980123372172721,
      "grad_norm": 6.492844971785228,
      "learning_rate": 4.128180376316015e-09,
      "loss": 0.9367,
      "step": 13585
    },
    {
      "epoch": 0.9801955196421486,
      "grad_norm": 4.60154529820904,
      "learning_rate": 4.09822207283983e-09,
      "loss": 0.9799,
      "step": 13586
    },
    {
      "epoch": 0.9802676671115761,
      "grad_norm": 4.037139375720048,
      "learning_rate": 4.068372758170957e-09,
      "loss": 0.9951,
      "step": 13587
    },
    {
      "epoch": 0.9803398145810036,
      "grad_norm": 8.198638095804483,
      "learning_rate": 4.038632433939648e-09,
      "loss": 0.9846,
      "step": 13588
    },
    {
      "epoch": 0.9804119620504311,
      "grad_norm": 7.645152492162466,
      "learning_rate": 4.009001101769716e-09,
      "loss": 0.8851,
      "step": 13589
    },
    {
      "epoch": 0.9804841095198585,
      "grad_norm": 4.085546062676203,
      "learning_rate": 3.97947876327942e-09,
      "loss": 0.9761,
      "step": 13590
    },
    {
      "epoch": 0.9805562569892861,
      "grad_norm": 3.6540463319376872,
      "learning_rate": 3.950065420080806e-09,
      "loss": 0.9726,
      "step": 13591
    },
    {
      "epoch": 0.9806284044587136,
      "grad_norm": 4.4148240234279905,
      "learning_rate": 3.920761073779921e-09,
      "loss": 0.7128,
      "step": 13592
    },
    {
      "epoch": 0.9807005519281411,
      "grad_norm": 4.538261513610467,
      "learning_rate": 3.891565725977042e-09,
      "loss": 1.0066,
      "step": 13593
    },
    {
      "epoch": 0.9807726993975686,
      "grad_norm": 5.033327988543841,
      "learning_rate": 3.862479378266448e-09,
      "loss": 0.9845,
      "step": 13594
    },
    {
      "epoch": 0.9808448468669961,
      "grad_norm": 2.4804079063342424,
      "learning_rate": 3.8335020322364244e-09,
      "loss": 0.9607,
      "step": 13595
    },
    {
      "epoch": 0.9809169943364237,
      "grad_norm": 3.7810709256731356,
      "learning_rate": 3.804633689469261e-09,
      "loss": 0.906,
      "step": 13596
    },
    {
      "epoch": 0.9809891418058512,
      "grad_norm": 4.480683299135126,
      "learning_rate": 3.775874351541475e-09,
      "loss": 0.8917,
      "step": 13597
    },
    {
      "epoch": 0.9810612892752787,
      "grad_norm": 5.335307630055104,
      "learning_rate": 3.747224020023365e-09,
      "loss": 0.9218,
      "step": 13598
    },
    {
      "epoch": 0.9811334367447062,
      "grad_norm": 3.9686856001787802,
      "learning_rate": 3.7186826964794584e-09,
      "loss": 0.8525,
      "step": 13599
    },
    {
      "epoch": 0.9812055842141337,
      "grad_norm": 4.593325640741633,
      "learning_rate": 3.6902503824682852e-09,
      "loss": 0.8652,
      "step": 13600
    },
    {
      "epoch": 0.9812777316835612,
      "grad_norm": 0.7855226411500619,
      "learning_rate": 3.661927079542382e-09,
      "loss": 0.8583,
      "step": 13601
    },
    {
      "epoch": 0.9813498791529887,
      "grad_norm": 4.60828347940824,
      "learning_rate": 3.633712789248511e-09,
      "loss": 0.9411,
      "step": 13602
    },
    {
      "epoch": 0.9814220266224162,
      "grad_norm": 5.555191553694812,
      "learning_rate": 3.60560751312744e-09,
      "loss": 0.9232,
      "step": 13603
    },
    {
      "epoch": 0.9814941740918437,
      "grad_norm": 6.7875280896021835,
      "learning_rate": 3.577611252713719e-09,
      "loss": 1.0622,
      "step": 13604
    },
    {
      "epoch": 0.9815663215612712,
      "grad_norm": 3.1170377384626717,
      "learning_rate": 3.5497240095359036e-09,
      "loss": 0.8855,
      "step": 13605
    },
    {
      "epoch": 0.9816384690306987,
      "grad_norm": 5.222943294514582,
      "learning_rate": 3.5219457851172197e-09,
      "loss": 0.8462,
      "step": 13606
    },
    {
      "epoch": 0.9817106165001263,
      "grad_norm": 2.8319888197252525,
      "learning_rate": 3.4942765809744536e-09,
      "loss": 0.9394,
      "step": 13607
    },
    {
      "epoch": 0.9817827639695538,
      "grad_norm": 3.1522039786625013,
      "learning_rate": 3.466716398618175e-09,
      "loss": 0.9286,
      "step": 13608
    },
    {
      "epoch": 0.9818549114389813,
      "grad_norm": 2.8286729144871665,
      "learning_rate": 3.439265239553846e-09,
      "loss": 0.8245,
      "step": 13609
    },
    {
      "epoch": 0.9819270589084088,
      "grad_norm": 0.749067720499256,
      "learning_rate": 3.411923105280268e-09,
      "loss": 0.8441,
      "step": 13610
    },
    {
      "epoch": 0.9819992063778363,
      "grad_norm": 2.7500228880883455,
      "learning_rate": 3.3846899972902467e-09,
      "loss": 0.955,
      "step": 13611
    },
    {
      "epoch": 0.9820713538472639,
      "grad_norm": 3.9340750921594663,
      "learning_rate": 3.357565917071037e-09,
      "loss": 0.9501,
      "step": 13612
    },
    {
      "epoch": 0.9821435013166914,
      "grad_norm": 81.65559521838506,
      "learning_rate": 3.3305508661041205e-09,
      "loss": 0.9886,
      "step": 13613
    },
    {
      "epoch": 0.9822156487861188,
      "grad_norm": 3.266497189803048,
      "learning_rate": 3.303644845864095e-09,
      "loss": 0.8999,
      "step": 13614
    },
    {
      "epoch": 0.9822877962555463,
      "grad_norm": 13.47222416583874,
      "learning_rate": 3.2768478578206747e-09,
      "loss": 1.0773,
      "step": 13615
    },
    {
      "epoch": 0.9823599437249738,
      "grad_norm": 5.521410450071095,
      "learning_rate": 3.250159903437133e-09,
      "loss": 1.0452,
      "step": 13616
    },
    {
      "epoch": 0.9824320911944013,
      "grad_norm": 4.604957040923931,
      "learning_rate": 3.2235809841703042e-09,
      "loss": 0.9452,
      "step": 13617
    },
    {
      "epoch": 0.9825042386638289,
      "grad_norm": 7.158029126892773,
      "learning_rate": 3.1971111014721386e-09,
      "loss": 1.056,
      "step": 13618
    },
    {
      "epoch": 0.9825763861332564,
      "grad_norm": 0.7393677804808866,
      "learning_rate": 3.170750256787702e-09,
      "loss": 0.8276,
      "step": 13619
    },
    {
      "epoch": 0.9826485336026839,
      "grad_norm": 3.386784360397486,
      "learning_rate": 3.144498451556732e-09,
      "loss": 0.8374,
      "step": 13620
    },
    {
      "epoch": 0.9827206810721114,
      "grad_norm": 4.206428902663063,
      "learning_rate": 3.1183556872125262e-09,
      "loss": 0.9339,
      "step": 13621
    },
    {
      "epoch": 0.9827928285415389,
      "grad_norm": 4.339447964836814,
      "learning_rate": 3.0923219651826095e-09,
      "loss": 1.0176,
      "step": 13622
    },
    {
      "epoch": 0.9828649760109665,
      "grad_norm": 2.81916620941414,
      "learning_rate": 3.0663972868887332e-09,
      "loss": 0.978,
      "step": 13623
    },
    {
      "epoch": 0.982937123480394,
      "grad_norm": 4.863818721869073,
      "learning_rate": 3.0405816537466546e-09,
      "loss": 0.9363,
      "step": 13624
    },
    {
      "epoch": 0.9830092709498215,
      "grad_norm": 3.01790268026065,
      "learning_rate": 3.0148750671656896e-09,
      "loss": 0.999,
      "step": 13625
    },
    {
      "epoch": 0.9830814184192489,
      "grad_norm": 5.040542077779553,
      "learning_rate": 2.9892775285500494e-09,
      "loss": 1.026,
      "step": 13626
    },
    {
      "epoch": 0.9831535658886764,
      "grad_norm": 4.638346616886119,
      "learning_rate": 2.963789039297282e-09,
      "loss": 0.9577,
      "step": 13627
    },
    {
      "epoch": 0.9832257133581039,
      "grad_norm": 4.897463761059754,
      "learning_rate": 2.9384096007989414e-09,
      "loss": 0.9978,
      "step": 13628
    },
    {
      "epoch": 0.9832978608275315,
      "grad_norm": 3.0071637135103764,
      "learning_rate": 2.9131392144414736e-09,
      "loss": 0.9406,
      "step": 13629
    },
    {
      "epoch": 0.983370008296959,
      "grad_norm": 3.26360539677395,
      "learning_rate": 2.887977881604442e-09,
      "loss": 0.9458,
      "step": 13630
    },
    {
      "epoch": 0.9834421557663865,
      "grad_norm": 3.409252837988124,
      "learning_rate": 2.8629256036618587e-09,
      "loss": 0.9568,
      "step": 13631
    },
    {
      "epoch": 0.983514303235814,
      "grad_norm": 3.0854391347482193,
      "learning_rate": 2.8379823819815186e-09,
      "loss": 0.871,
      "step": 13632
    },
    {
      "epoch": 0.9835864507052415,
      "grad_norm": 4.009782273619845,
      "learning_rate": 2.813148217925887e-09,
      "loss": 0.9151,
      "step": 13633
    },
    {
      "epoch": 0.983658598174669,
      "grad_norm": 4.473430609822671,
      "learning_rate": 2.7884231128509907e-09,
      "loss": 0.9639,
      "step": 13634
    },
    {
      "epoch": 0.9837307456440966,
      "grad_norm": 20.623022366124513,
      "learning_rate": 2.7638070681068604e-09,
      "loss": 0.9769,
      "step": 13635
    },
    {
      "epoch": 0.9838028931135241,
      "grad_norm": 2.616536620324288,
      "learning_rate": 2.7393000850377547e-09,
      "loss": 0.8731,
      "step": 13636
    },
    {
      "epoch": 0.9838750405829515,
      "grad_norm": 11.521538508783234,
      "learning_rate": 2.7149021649817138e-09,
      "loss": 1.0043,
      "step": 13637
    },
    {
      "epoch": 0.983947188052379,
      "grad_norm": 3.176366773167977,
      "learning_rate": 2.690613309271228e-09,
      "loss": 0.9443,
      "step": 13638
    },
    {
      "epoch": 0.9840193355218065,
      "grad_norm": 5.284660366842044,
      "learning_rate": 2.6664335192325694e-09,
      "loss": 0.8822,
      "step": 13639
    },
    {
      "epoch": 0.9840914829912341,
      "grad_norm": 4.05765444007418,
      "learning_rate": 2.642362796186237e-09,
      "loss": 0.9005,
      "step": 13640
    },
    {
      "epoch": 0.9841636304606616,
      "grad_norm": 2.8765546492967284,
      "learning_rate": 2.6184011414462915e-09,
      "loss": 1.0205,
      "step": 13641
    },
    {
      "epoch": 0.9842357779300891,
      "grad_norm": 0.6216747277745406,
      "learning_rate": 2.594548556321685e-09,
      "loss": 0.8,
      "step": 13642
    },
    {
      "epoch": 0.9843079253995166,
      "grad_norm": 2.3989803929928972,
      "learning_rate": 2.5708050421142658e-09,
      "loss": 0.9775,
      "step": 13643
    },
    {
      "epoch": 0.9843800728689441,
      "grad_norm": 2.7175612207445985,
      "learning_rate": 2.5471706001212177e-09,
      "loss": 0.8608,
      "step": 13644
    },
    {
      "epoch": 0.9844522203383717,
      "grad_norm": 2.8899545252829024,
      "learning_rate": 2.5236452316328427e-09,
      "loss": 0.9736,
      "step": 13645
    },
    {
      "epoch": 0.9845243678077992,
      "grad_norm": 4.157600814321885,
      "learning_rate": 2.5002289379336685e-09,
      "loss": 0.9325,
      "step": 13646
    },
    {
      "epoch": 0.9845965152772267,
      "grad_norm": 4.6447072168227015,
      "learning_rate": 2.4769217203026716e-09,
      "loss": 0.9423,
      "step": 13647
    },
    {
      "epoch": 0.9846686627466542,
      "grad_norm": 3.2443832867082425,
      "learning_rate": 2.4537235800123902e-09,
      "loss": 0.892,
      "step": 13648
    },
    {
      "epoch": 0.9847408102160816,
      "grad_norm": 3.8059181557303194,
      "learning_rate": 2.4306345183293663e-09,
      "loss": 0.8287,
      "step": 13649
    },
    {
      "epoch": 0.9848129576855091,
      "grad_norm": 6.60715427471338,
      "learning_rate": 2.407654536514814e-09,
      "loss": 0.9893,
      "step": 13650
    },
    {
      "epoch": 0.9848851051549367,
      "grad_norm": 3.588144059800871,
      "learning_rate": 2.384783635823506e-09,
      "loss": 1.0585,
      "step": 13651
    },
    {
      "epoch": 0.9849572526243642,
      "grad_norm": 2.7122712386041026,
      "learning_rate": 2.3620218175040006e-09,
      "loss": 1.0157,
      "step": 13652
    },
    {
      "epoch": 0.9850294000937917,
      "grad_norm": 2.765187460561602,
      "learning_rate": 2.339369082799747e-09,
      "loss": 0.9849,
      "step": 13653
    },
    {
      "epoch": 0.9851015475632192,
      "grad_norm": 4.813400778751419,
      "learning_rate": 2.3168254329473114e-09,
      "loss": 1.0464,
      "step": 13654
    },
    {
      "epoch": 0.9851736950326467,
      "grad_norm": 6.359669821463777,
      "learning_rate": 2.2943908691779313e-09,
      "loss": 0.9494,
      "step": 13655
    },
    {
      "epoch": 0.9852458425020743,
      "grad_norm": 3.549344316562518,
      "learning_rate": 2.272065392716405e-09,
      "loss": 0.9241,
      "step": 13656
    },
    {
      "epoch": 0.9853179899715018,
      "grad_norm": 4.169343902055622,
      "learning_rate": 2.2498490047824226e-09,
      "loss": 0.9204,
      "step": 13657
    },
    {
      "epoch": 0.9853901374409293,
      "grad_norm": 3.97423624822605,
      "learning_rate": 2.2277417065885706e-09,
      "loss": 0.9874,
      "step": 13658
    },
    {
      "epoch": 0.9854622849103568,
      "grad_norm": 3.9300909423012467,
      "learning_rate": 2.2057434993421054e-09,
      "loss": 0.8978,
      "step": 13659
    },
    {
      "epoch": 0.9855344323797843,
      "grad_norm": 4.9221191345702255,
      "learning_rate": 2.18385438424451e-09,
      "loss": 0.9515,
      "step": 13660
    },
    {
      "epoch": 0.9856065798492117,
      "grad_norm": 3.8967971561491206,
      "learning_rate": 2.162074362491051e-09,
      "loss": 1.0344,
      "step": 13661
    },
    {
      "epoch": 0.9856787273186393,
      "grad_norm": 2.2949798446800775,
      "learning_rate": 2.1404034352709987e-09,
      "loss": 0.9901,
      "step": 13662
    },
    {
      "epoch": 0.9857508747880668,
      "grad_norm": 2.9317726954345065,
      "learning_rate": 2.118841603767407e-09,
      "loss": 0.9521,
      "step": 13663
    },
    {
      "epoch": 0.9858230222574943,
      "grad_norm": 3.7537732532452974,
      "learning_rate": 2.097388869158223e-09,
      "loss": 0.9284,
      "step": 13664
    },
    {
      "epoch": 0.9858951697269218,
      "grad_norm": 3.7374073948439417,
      "learning_rate": 2.0760452326147317e-09,
      "loss": 0.8964,
      "step": 13665
    },
    {
      "epoch": 0.9859673171963493,
      "grad_norm": 4.574158287788918,
      "learning_rate": 2.0548106953022226e-09,
      "loss": 0.8864,
      "step": 13666
    },
    {
      "epoch": 0.9860394646657769,
      "grad_norm": 8.466196949261894,
      "learning_rate": 2.0336852583802135e-09,
      "loss": 0.7955,
      "step": 13667
    },
    {
      "epoch": 0.9861116121352044,
      "grad_norm": 3.9490337686956827,
      "learning_rate": 2.0126689230026693e-09,
      "loss": 0.9513,
      "step": 13668
    },
    {
      "epoch": 0.9861837596046319,
      "grad_norm": 6.369755401466505,
      "learning_rate": 1.9917616903168954e-09,
      "loss": 0.8936,
      "step": 13669
    },
    {
      "epoch": 0.9862559070740594,
      "grad_norm": 4.142624857925344,
      "learning_rate": 1.9709635614646446e-09,
      "loss": 0.9234,
      "step": 13670
    },
    {
      "epoch": 0.9863280545434869,
      "grad_norm": 0.7067285806670903,
      "learning_rate": 1.9502745375814534e-09,
      "loss": 0.8184,
      "step": 13671
    },
    {
      "epoch": 0.9864002020129145,
      "grad_norm": 3.6484060020639144,
      "learning_rate": 1.929694619797528e-09,
      "loss": 1.0688,
      "step": 13672
    },
    {
      "epoch": 0.9864723494823419,
      "grad_norm": 3.146102607436082,
      "learning_rate": 1.9092238092361934e-09,
      "loss": 0.9578,
      "step": 13673
    },
    {
      "epoch": 0.9865444969517694,
      "grad_norm": 0.6847200978056799,
      "learning_rate": 1.888862107015443e-09,
      "loss": 0.7782,
      "step": 13674
    },
    {
      "epoch": 0.9866166444211969,
      "grad_norm": 0.6766756126167929,
      "learning_rate": 1.868609514247277e-09,
      "loss": 0.7973,
      "step": 13675
    },
    {
      "epoch": 0.9866887918906244,
      "grad_norm": 2.6890065605733273,
      "learning_rate": 1.8484660320374768e-09,
      "loss": 0.9076,
      "step": 13676
    },
    {
      "epoch": 0.9867609393600519,
      "grad_norm": 4.009366513107249,
      "learning_rate": 1.8284316614860517e-09,
      "loss": 0.9392,
      "step": 13677
    },
    {
      "epoch": 0.9868330868294795,
      "grad_norm": 5.817608956831782,
      "learning_rate": 1.8085064036872377e-09,
      "loss": 0.9178,
      "step": 13678
    },
    {
      "epoch": 0.986905234298907,
      "grad_norm": 2.8187027244240457,
      "learning_rate": 1.788690259728387e-09,
      "loss": 0.967,
      "step": 13679
    },
    {
      "epoch": 0.9869773817683345,
      "grad_norm": 4.3200760788752905,
      "learning_rate": 1.7689832306924113e-09,
      "loss": 0.9423,
      "step": 13680
    },
    {
      "epoch": 0.987049529237762,
      "grad_norm": 4.117862190108313,
      "learning_rate": 1.7493853176548945e-09,
      "loss": 0.8716,
      "step": 13681
    },
    {
      "epoch": 0.9871216767071895,
      "grad_norm": 4.751836321391234,
      "learning_rate": 1.7298965216860917e-09,
      "loss": 1.0454,
      "step": 13682
    },
    {
      "epoch": 0.987193824176617,
      "grad_norm": 4.853225791459299,
      "learning_rate": 1.7105168438504846e-09,
      "loss": 0.8642,
      "step": 13683
    },
    {
      "epoch": 0.9872659716460445,
      "grad_norm": 3.242865287069533,
      "learning_rate": 1.691246285206116e-09,
      "loss": 0.8174,
      "step": 13684
    },
    {
      "epoch": 0.987338119115472,
      "grad_norm": 6.224694252931663,
      "learning_rate": 1.672084846805033e-09,
      "loss": 0.994,
      "step": 13685
    },
    {
      "epoch": 0.9874102665848995,
      "grad_norm": 3.1254046368889754,
      "learning_rate": 1.6530325296941762e-09,
      "loss": 0.9821,
      "step": 13686
    },
    {
      "epoch": 0.987482414054327,
      "grad_norm": 3.259104666832613,
      "learning_rate": 1.63408933491338e-09,
      "loss": 0.8968,
      "step": 13687
    },
    {
      "epoch": 0.9875545615237545,
      "grad_norm": 4.309383427180847,
      "learning_rate": 1.6152552634973726e-09,
      "loss": 0.953,
      "step": 13688
    },
    {
      "epoch": 0.9876267089931821,
      "grad_norm": 2.891820711838641,
      "learning_rate": 1.5965303164744425e-09,
      "loss": 0.9107,
      "step": 13689
    },
    {
      "epoch": 0.9876988564626096,
      "grad_norm": 5.4873028626365095,
      "learning_rate": 1.5779144948671052e-09,
      "loss": 0.9273,
      "step": 13690
    },
    {
      "epoch": 0.9877710039320371,
      "grad_norm": 3.553095534444872,
      "learning_rate": 1.559407799692103e-09,
      "loss": 0.9638,
      "step": 13691
    },
    {
      "epoch": 0.9878431514014646,
      "grad_norm": 2.8765490132212013,
      "learning_rate": 1.5410102319597386e-09,
      "loss": 0.9787,
      "step": 13692
    },
    {
      "epoch": 0.9879152988708921,
      "grad_norm": 9.267104135336766,
      "learning_rate": 1.5227217926749859e-09,
      "loss": 0.9127,
      "step": 13693
    },
    {
      "epoch": 0.9879874463403197,
      "grad_norm": 2.5145749572091867,
      "learning_rate": 1.5045424828359355e-09,
      "loss": 0.9766,
      "step": 13694
    },
    {
      "epoch": 0.9880595938097472,
      "grad_norm": 8.88412393435924,
      "learning_rate": 1.4864723034360148e-09,
      "loss": 0.8918,
      "step": 13695
    },
    {
      "epoch": 0.9881317412791746,
      "grad_norm": 2.6420099362309957,
      "learning_rate": 1.4685112554613245e-09,
      "loss": 0.9462,
      "step": 13696
    },
    {
      "epoch": 0.9882038887486021,
      "grad_norm": 3.390846491316488,
      "learning_rate": 1.4506593398930788e-09,
      "loss": 1.033,
      "step": 13697
    },
    {
      "epoch": 0.9882760362180296,
      "grad_norm": 4.554774183753815,
      "learning_rate": 1.4329165577058322e-09,
      "loss": 0.9629,
      "step": 13698
    },
    {
      "epoch": 0.9883481836874571,
      "grad_norm": 7.266525639323332,
      "learning_rate": 1.415282909868587e-09,
      "loss": 0.9862,
      "step": 13699
    },
    {
      "epoch": 0.9884203311568847,
      "grad_norm": 3.5878041048784852,
      "learning_rate": 1.3977583973443508e-09,
      "loss": 0.9227,
      "step": 13700
    },
    {
      "epoch": 0.9884924786263122,
      "grad_norm": 5.619483955828112,
      "learning_rate": 1.3803430210899137e-09,
      "loss": 0.9113,
      "step": 13701
    },
    {
      "epoch": 0.9885646260957397,
      "grad_norm": 5.0116959152831315,
      "learning_rate": 1.3630367820562926e-09,
      "loss": 0.9645,
      "step": 13702
    },
    {
      "epoch": 0.9886367735651672,
      "grad_norm": 6.9558953070546625,
      "learning_rate": 1.3458396811885099e-09,
      "loss": 0.9961,
      "step": 13703
    },
    {
      "epoch": 0.9887089210345947,
      "grad_norm": 3.6300580482990896,
      "learning_rate": 1.3287517194255915e-09,
      "loss": 0.9873,
      "step": 13704
    },
    {
      "epoch": 0.9887810685040223,
      "grad_norm": 4.620272101444009,
      "learning_rate": 1.3117728977007913e-09,
      "loss": 0.9026,
      "step": 13705
    },
    {
      "epoch": 0.9888532159734498,
      "grad_norm": 5.502909670965148,
      "learning_rate": 1.2949032169411457e-09,
      "loss": 0.9612,
      "step": 13706
    },
    {
      "epoch": 0.9889253634428773,
      "grad_norm": 4.5837315935442025,
      "learning_rate": 1.2781426780679173e-09,
      "loss": 0.8603,
      "step": 13707
    },
    {
      "epoch": 0.9889975109123047,
      "grad_norm": 4.362121154599816,
      "learning_rate": 1.261491281996374e-09,
      "loss": 0.8432,
      "step": 13708
    },
    {
      "epoch": 0.9890696583817322,
      "grad_norm": 4.742954248724434,
      "learning_rate": 1.244949029635567e-09,
      "loss": 1.0663,
      "step": 13709
    },
    {
      "epoch": 0.9891418058511597,
      "grad_norm": 4.122706382494364,
      "learning_rate": 1.2285159218889952e-09,
      "loss": 0.9117,
      "step": 13710
    },
    {
      "epoch": 0.9892139533205873,
      "grad_norm": 7.1010261223971565,
      "learning_rate": 1.212191959653941e-09,
      "loss": 0.9723,
      "step": 13711
    },
    {
      "epoch": 0.9892861007900148,
      "grad_norm": 5.836793073370655,
      "learning_rate": 1.1959771438219135e-09,
      "loss": 0.9532,
      "step": 13712
    },
    {
      "epoch": 0.9893582482594423,
      "grad_norm": 3.7570517041464035,
      "learning_rate": 1.1798714752782047e-09,
      "loss": 0.9039,
      "step": 13713
    },
    {
      "epoch": 0.9894303957288698,
      "grad_norm": 4.531574316586142,
      "learning_rate": 1.163874954902333e-09,
      "loss": 1.0005,
      "step": 13714
    },
    {
      "epoch": 0.9895025431982973,
      "grad_norm": 2.760773449758881,
      "learning_rate": 1.1479875835678222e-09,
      "loss": 1.0631,
      "step": 13715
    },
    {
      "epoch": 0.9895746906677249,
      "grad_norm": 4.2610207326080705,
      "learning_rate": 1.1322093621422e-09,
      "loss": 0.8743,
      "step": 13716
    },
    {
      "epoch": 0.9896468381371524,
      "grad_norm": 4.250328948810398,
      "learning_rate": 1.1165402914872224e-09,
      "loss": 0.9837,
      "step": 13717
    },
    {
      "epoch": 0.9897189856065799,
      "grad_norm": 3.271042122625932,
      "learning_rate": 1.1009803724582044e-09,
      "loss": 0.8447,
      "step": 13718
    },
    {
      "epoch": 0.9897911330760074,
      "grad_norm": 4.18228143078734,
      "learning_rate": 1.0855296059049113e-09,
      "loss": 0.9624,
      "step": 13719
    },
    {
      "epoch": 0.9898632805454348,
      "grad_norm": 3.246540722790011,
      "learning_rate": 1.0701879926713342e-09,
      "loss": 1.0317,
      "step": 13720
    },
    {
      "epoch": 0.9899354280148623,
      "grad_norm": 3.7397704152921256,
      "learning_rate": 1.0549555335948035e-09,
      "loss": 0.9139,
      "step": 13721
    },
    {
      "epoch": 0.9900075754842899,
      "grad_norm": 2.8456384445935208,
      "learning_rate": 1.0398322295073204e-09,
      "loss": 1.0075,
      "step": 13722
    },
    {
      "epoch": 0.9900797229537174,
      "grad_norm": 3.747825627955633,
      "learning_rate": 1.0248180812348906e-09,
      "loss": 0.9737,
      "step": 13723
    },
    {
      "epoch": 0.9901518704231449,
      "grad_norm": 6.8203701091137985,
      "learning_rate": 1.009913089597081e-09,
      "loss": 0.9367,
      "step": 13724
    },
    {
      "epoch": 0.9902240178925724,
      "grad_norm": 3.405326648118107,
      "learning_rate": 9.951172554079068e-10,
      "loss": 0.8736,
      "step": 13725
    },
    {
      "epoch": 0.9902961653619999,
      "grad_norm": 18.94432262233571,
      "learning_rate": 9.804305794753886e-10,
      "loss": 0.9795,
      "step": 13726
    },
    {
      "epoch": 0.9903683128314275,
      "grad_norm": 3.522023663772205,
      "learning_rate": 9.658530626013295e-10,
      "loss": 0.9181,
      "step": 13727
    },
    {
      "epoch": 0.990440460300855,
      "grad_norm": 3.8175792941169036,
      "learning_rate": 9.513847055817593e-10,
      "loss": 0.9205,
      "step": 13728
    },
    {
      "epoch": 0.9905126077702825,
      "grad_norm": 5.571588014410195,
      "learning_rate": 9.370255092071566e-10,
      "loss": 0.9304,
      "step": 13729
    },
    {
      "epoch": 0.99058475523971,
      "grad_norm": 2.494358086123623,
      "learning_rate": 9.227754742611171e-10,
      "loss": 0.9291,
      "step": 13730
    },
    {
      "epoch": 0.9906569027091375,
      "grad_norm": 3.72799469559419,
      "learning_rate": 9.086346015219071e-10,
      "loss": 0.8793,
      "step": 13731
    },
    {
      "epoch": 0.9907290501785649,
      "grad_norm": 4.828100111813887,
      "learning_rate": 8.946028917620197e-10,
      "loss": 1.0968,
      "step": 13732
    },
    {
      "epoch": 0.9908011976479925,
      "grad_norm": 4.6461243588196375,
      "learning_rate": 8.806803457472866e-10,
      "loss": 0.8002,
      "step": 13733
    },
    {
      "epoch": 0.99087334511742,
      "grad_norm": 2.283944406905082,
      "learning_rate": 8.668669642379889e-10,
      "loss": 1.0409,
      "step": 13734
    },
    {
      "epoch": 0.9909454925868475,
      "grad_norm": 4.793825652757731,
      "learning_rate": 8.531627479886339e-10,
      "loss": 0.9269,
      "step": 13735
    },
    {
      "epoch": 0.991017640056275,
      "grad_norm": 3.9593029373945403,
      "learning_rate": 8.395676977475119e-10,
      "loss": 1.0196,
      "step": 13736
    },
    {
      "epoch": 0.9910897875257025,
      "grad_norm": 3.8649851683496146,
      "learning_rate": 8.260818142569181e-10,
      "loss": 0.9522,
      "step": 13737
    },
    {
      "epoch": 0.9911619349951301,
      "grad_norm": 11.76642515746937,
      "learning_rate": 8.127050982533745e-10,
      "loss": 0.9596,
      "step": 13738
    },
    {
      "epoch": 0.9912340824645576,
      "grad_norm": 0.7103715677200251,
      "learning_rate": 7.994375504674078e-10,
      "loss": 0.8073,
      "step": 13739
    },
    {
      "epoch": 0.9913062299339851,
      "grad_norm": 5.218357094233726,
      "learning_rate": 7.862791716231054e-10,
      "loss": 1.1011,
      "step": 13740
    },
    {
      "epoch": 0.9913783774034126,
      "grad_norm": 2.815214161425152,
      "learning_rate": 7.732299624394478e-10,
      "loss": 0.899,
      "step": 13741
    },
    {
      "epoch": 0.9914505248728401,
      "grad_norm": 2.662824863440503,
      "learning_rate": 7.602899236287541e-10,
      "loss": 0.9922,
      "step": 13742
    },
    {
      "epoch": 0.9915226723422675,
      "grad_norm": 4.172787227264116,
      "learning_rate": 7.474590558975702e-10,
      "loss": 0.9444,
      "step": 13743
    },
    {
      "epoch": 0.9915948198116951,
      "grad_norm": 3.8014390930891944,
      "learning_rate": 7.347373599468909e-10,
      "loss": 0.9094,
      "step": 13744
    },
    {
      "epoch": 0.9916669672811226,
      "grad_norm": 9.533227734103148,
      "learning_rate": 7.221248364710497e-10,
      "loss": 0.8463,
      "step": 13745
    },
    {
      "epoch": 0.9917391147505501,
      "grad_norm": 3.132911737634903,
      "learning_rate": 7.09621486158829e-10,
      "loss": 0.9714,
      "step": 13746
    },
    {
      "epoch": 0.9918112622199776,
      "grad_norm": 6.56998764849798,
      "learning_rate": 6.972273096930159e-10,
      "loss": 0.9942,
      "step": 13747
    },
    {
      "epoch": 0.9918834096894051,
      "grad_norm": 3.6379838130144124,
      "learning_rate": 6.849423077506245e-10,
      "loss": 0.8914,
      "step": 13748
    },
    {
      "epoch": 0.9919555571588327,
      "grad_norm": 4.396031861846669,
      "learning_rate": 6.727664810022293e-10,
      "loss": 1.0225,
      "step": 13749
    },
    {
      "epoch": 0.9920277046282602,
      "grad_norm": 4.5627922069045255,
      "learning_rate": 6.606998301128541e-10,
      "loss": 0.9713,
      "step": 13750
    },
    {
      "epoch": 0.9920998520976877,
      "grad_norm": 3.3962059674657112,
      "learning_rate": 6.487423557415272e-10,
      "loss": 1.0104,
      "step": 13751
    },
    {
      "epoch": 0.9921719995671152,
      "grad_norm": 6.529726863832803,
      "learning_rate": 6.368940585408377e-10,
      "loss": 0.9194,
      "step": 13752
    },
    {
      "epoch": 0.9922441470365427,
      "grad_norm": 3.5758490507879377,
      "learning_rate": 6.251549391578237e-10,
      "loss": 0.9886,
      "step": 13753
    },
    {
      "epoch": 0.9923162945059703,
      "grad_norm": 5.002726574389269,
      "learning_rate": 6.135249982339719e-10,
      "loss": 0.9315,
      "step": 13754
    },
    {
      "epoch": 0.9923884419753977,
      "grad_norm": 7.682971892333635,
      "learning_rate": 6.020042364038858e-10,
      "loss": 0.9223,
      "step": 13755
    },
    {
      "epoch": 0.9924605894448252,
      "grad_norm": 3.5655234373373776,
      "learning_rate": 5.905926542968398e-10,
      "loss": 0.9497,
      "step": 13756
    },
    {
      "epoch": 0.9925327369142527,
      "grad_norm": 3.7619076183278355,
      "learning_rate": 5.792902525358912e-10,
      "loss": 0.9815,
      "step": 13757
    },
    {
      "epoch": 0.9926048843836802,
      "grad_norm": 3.7620928014881927,
      "learning_rate": 5.680970317385458e-10,
      "loss": 0.9896,
      "step": 13758
    },
    {
      "epoch": 0.9926770318531077,
      "grad_norm": 3.6647710379448104,
      "learning_rate": 5.570129925156486e-10,
      "loss": 0.8824,
      "step": 13759
    },
    {
      "epoch": 0.9927491793225353,
      "grad_norm": 3.4061571861011086,
      "learning_rate": 5.46038135472493e-10,
      "loss": 0.9136,
      "step": 13760
    },
    {
      "epoch": 0.9928213267919628,
      "grad_norm": 4.925999635821076,
      "learning_rate": 5.351724612085995e-10,
      "loss": 0.8833,
      "step": 13761
    },
    {
      "epoch": 0.9928934742613903,
      "grad_norm": 3.807065749173538,
      "learning_rate": 5.244159703170492e-10,
      "loss": 0.9312,
      "step": 13762
    },
    {
      "epoch": 0.9929656217308178,
      "grad_norm": 3.078543232809846,
      "learning_rate": 5.137686633853722e-10,
      "loss": 0.8823,
      "step": 13763
    },
    {
      "epoch": 0.9930377692002453,
      "grad_norm": 3.6336319850237846,
      "learning_rate": 5.032305409951032e-10,
      "loss": 1.0168,
      "step": 13764
    },
    {
      "epoch": 0.9931099166696729,
      "grad_norm": 0.7010651191049675,
      "learning_rate": 4.928016037213378e-10,
      "loss": 0.8256,
      "step": 13765
    },
    {
      "epoch": 0.9931820641391004,
      "grad_norm": 0.7744668603435435,
      "learning_rate": 4.824818521338425e-10,
      "loss": 0.8473,
      "step": 13766
    },
    {
      "epoch": 0.9932542116085278,
      "grad_norm": 5.109759861628124,
      "learning_rate": 4.722712867959445e-10,
      "loss": 0.9843,
      "step": 13767
    },
    {
      "epoch": 0.9933263590779553,
      "grad_norm": 4.2140417882787276,
      "learning_rate": 4.6216990826541955e-10,
      "loss": 0.9665,
      "step": 13768
    },
    {
      "epoch": 0.9933985065473828,
      "grad_norm": 0.7958900498024283,
      "learning_rate": 4.521777170938268e-10,
      "loss": 0.8512,
      "step": 13769
    },
    {
      "epoch": 0.9934706540168103,
      "grad_norm": 3.700160517948045,
      "learning_rate": 4.422947138265076e-10,
      "loss": 0.9439,
      "step": 13770
    },
    {
      "epoch": 0.9935428014862379,
      "grad_norm": 4.343142899275854,
      "learning_rate": 4.325208990034745e-10,
      "loss": 0.8785,
      "step": 13771
    },
    {
      "epoch": 0.9936149489556654,
      "grad_norm": 7.959490733509048,
      "learning_rate": 4.228562731585228e-10,
      "loss": 0.9829,
      "step": 13772
    },
    {
      "epoch": 0.9936870964250929,
      "grad_norm": 3.1683445718142194,
      "learning_rate": 4.133008368190083e-10,
      "loss": 1.0838,
      "step": 13773
    },
    {
      "epoch": 0.9937592438945204,
      "grad_norm": 2.620851781281849,
      "learning_rate": 4.0385459050695794e-10,
      "loss": 0.8821,
      "step": 13774
    },
    {
      "epoch": 0.9938313913639479,
      "grad_norm": 3.836039803999517,
      "learning_rate": 3.945175347381813e-10,
      "loss": 0.947,
      "step": 13775
    },
    {
      "epoch": 0.9939035388333755,
      "grad_norm": 4.238484433133734,
      "learning_rate": 3.852896700224928e-10,
      "loss": 0.8951,
      "step": 13776
    },
    {
      "epoch": 0.993975686302803,
      "grad_norm": 2.730226781599137,
      "learning_rate": 3.761709968637117e-10,
      "loss": 0.9524,
      "step": 13777
    },
    {
      "epoch": 0.9940478337722305,
      "grad_norm": 0.6602359926716039,
      "learning_rate": 3.6716151575988397e-10,
      "loss": 0.8265,
      "step": 13778
    },
    {
      "epoch": 0.9941199812416579,
      "grad_norm": 135.08047369993932,
      "learning_rate": 3.5826122720328256e-10,
      "loss": 0.9257,
      "step": 13779
    },
    {
      "epoch": 0.9941921287110854,
      "grad_norm": 3.1585586715739353,
      "learning_rate": 3.4947013167929694e-10,
      "loss": 1.02,
      "step": 13780
    },
    {
      "epoch": 0.9942642761805129,
      "grad_norm": 3.2867860142031793,
      "learning_rate": 3.407882296684317e-10,
      "loss": 0.8425,
      "step": 13781
    },
    {
      "epoch": 0.9943364236499405,
      "grad_norm": 3.521810963953107,
      "learning_rate": 3.3221552164452994e-10,
      "loss": 0.9486,
      "step": 13782
    },
    {
      "epoch": 0.994408571119368,
      "grad_norm": 3.482455695774846,
      "learning_rate": 3.2375200807588376e-10,
      "loss": 1.0286,
      "step": 13783
    },
    {
      "epoch": 0.9944807185887955,
      "grad_norm": 3.2242831475842655,
      "learning_rate": 3.1539768942479004e-10,
      "loss": 0.8908,
      "step": 13784
    },
    {
      "epoch": 0.994552866058223,
      "grad_norm": 6.788750082713818,
      "learning_rate": 3.0715256614688434e-10,
      "loss": 0.884,
      "step": 13785
    },
    {
      "epoch": 0.9946250135276505,
      "grad_norm": 3.9471260963479735,
      "learning_rate": 2.990166386931392e-10,
      "loss": 0.887,
      "step": 13786
    },
    {
      "epoch": 0.9946971609970781,
      "grad_norm": 3.5167333593568033,
      "learning_rate": 2.909899075071998e-10,
      "loss": 0.8714,
      "step": 13787
    },
    {
      "epoch": 0.9947693084665056,
      "grad_norm": 4.205208301079597,
      "learning_rate": 2.8307237302760413e-10,
      "loss": 1.0069,
      "step": 13788
    },
    {
      "epoch": 0.9948414559359331,
      "grad_norm": 3.8181581878343724,
      "learning_rate": 2.7526403568689515e-10,
      "loss": 1.0208,
      "step": 13789
    },
    {
      "epoch": 0.9949136034053605,
      "grad_norm": 4.799782732178931,
      "learning_rate": 2.675648959111765e-10,
      "loss": 0.9596,
      "step": 13790
    },
    {
      "epoch": 0.994985750874788,
      "grad_norm": 4.455906477818234,
      "learning_rate": 2.599749541210006e-10,
      "loss": 0.9194,
      "step": 13791
    },
    {
      "epoch": 0.9950578983442155,
      "grad_norm": 0.7510881874254645,
      "learning_rate": 2.524942107309247e-10,
      "loss": 0.8844,
      "step": 13792
    },
    {
      "epoch": 0.9951300458136431,
      "grad_norm": 4.994730079544327,
      "learning_rate": 2.451226661492889e-10,
      "loss": 0.9158,
      "step": 13793
    },
    {
      "epoch": 0.9952021932830706,
      "grad_norm": 3.8132577361992075,
      "learning_rate": 2.37860320778438e-10,
      "loss": 0.9482,
      "step": 13794
    },
    {
      "epoch": 0.9952743407524981,
      "grad_norm": 4.631611814792954,
      "learning_rate": 2.3070717501538773e-10,
      "loss": 0.9163,
      "step": 13795
    },
    {
      "epoch": 0.9953464882219256,
      "grad_norm": 4.587368431595316,
      "learning_rate": 2.2366322925049252e-10,
      "loss": 1.0463,
      "step": 13796
    },
    {
      "epoch": 0.9954186356913531,
      "grad_norm": 2.716007032837562,
      "learning_rate": 2.1672848386833364e-10,
      "loss": 0.9784,
      "step": 13797
    },
    {
      "epoch": 0.9954907831607807,
      "grad_norm": 3.3889530120230638,
      "learning_rate": 2.0990293924794123e-10,
      "loss": 0.8449,
      "step": 13798
    },
    {
      "epoch": 0.9955629306302082,
      "grad_norm": 5.696014068050371,
      "learning_rate": 2.0318659576146202e-10,
      "loss": 0.9714,
      "step": 13799
    },
    {
      "epoch": 0.9956350780996357,
      "grad_norm": 2.6579473121536044,
      "learning_rate": 1.965794537761578e-10,
      "loss": 0.9556,
      "step": 13800
    },
    {
      "epoch": 0.9957072255690632,
      "grad_norm": 3.9384362454931847,
      "learning_rate": 1.9008151365262903e-10,
      "loss": 0.9043,
      "step": 13801
    },
    {
      "epoch": 0.9957793730384906,
      "grad_norm": 2.640010875188218,
      "learning_rate": 1.8369277574570295e-10,
      "loss": 0.9402,
      "step": 13802
    },
    {
      "epoch": 0.9958515205079181,
      "grad_norm": 3.4277925231000927,
      "learning_rate": 1.7741324040421167e-10,
      "loss": 0.9155,
      "step": 13803
    },
    {
      "epoch": 0.9959236679773457,
      "grad_norm": 3.6445214372607047,
      "learning_rate": 1.7124290797099206e-10,
      "loss": 0.9735,
      "step": 13804
    },
    {
      "epoch": 0.9959958154467732,
      "grad_norm": 4.077421281102661,
      "learning_rate": 1.6518177878332983e-10,
      "loss": 1.0492,
      "step": 13805
    },
    {
      "epoch": 0.9960679629162007,
      "grad_norm": 11.575064436071997,
      "learning_rate": 1.5922985317184945e-10,
      "loss": 0.8548,
      "step": 13806
    },
    {
      "epoch": 0.9961401103856282,
      "grad_norm": 4.752664120091311,
      "learning_rate": 1.5338713146140215e-10,
      "loss": 0.9463,
      "step": 13807
    },
    {
      "epoch": 0.9962122578550557,
      "grad_norm": 3.2275276864871376,
      "learning_rate": 1.4765361397173214e-10,
      "loss": 0.8974,
      "step": 13808
    },
    {
      "epoch": 0.9962844053244833,
      "grad_norm": 4.42232184966135,
      "learning_rate": 1.4202930101525624e-10,
      "loss": 0.9763,
      "step": 13809
    },
    {
      "epoch": 0.9963565527939108,
      "grad_norm": 0.7001402978286569,
      "learning_rate": 1.3651419289928413e-10,
      "loss": 0.8211,
      "step": 13810
    },
    {
      "epoch": 0.9964287002633383,
      "grad_norm": 4.171041455381349,
      "learning_rate": 1.3110828992513034e-10,
      "loss": 0.9191,
      "step": 13811
    },
    {
      "epoch": 0.9965008477327658,
      "grad_norm": 2.8347091326446914,
      "learning_rate": 1.2581159238789217e-10,
      "loss": 0.8805,
      "step": 13812
    },
    {
      "epoch": 0.9965729952021933,
      "grad_norm": 3.9236633357320514,
      "learning_rate": 1.2062410057689376e-10,
      "loss": 0.8835,
      "step": 13813
    },
    {
      "epoch": 0.9966451426716207,
      "grad_norm": 4.439227990056951,
      "learning_rate": 1.1554581477524195e-10,
      "loss": 1.0234,
      "step": 13814
    },
    {
      "epoch": 0.9967172901410483,
      "grad_norm": 4.880466428572663,
      "learning_rate": 1.1057673526027044e-10,
      "loss": 0.9964,
      "step": 13815
    },
    {
      "epoch": 0.9967894376104758,
      "grad_norm": 0.7796921647721609,
      "learning_rate": 1.0571686230331778e-10,
      "loss": 0.885,
      "step": 13816
    },
    {
      "epoch": 0.9968615850799033,
      "grad_norm": 3.624570886929911,
      "learning_rate": 1.0096619616994927e-10,
      "loss": 0.8953,
      "step": 13817
    },
    {
      "epoch": 0.9969337325493308,
      "grad_norm": 3.072329569622211,
      "learning_rate": 9.632473711929101e-11,
      "loss": 0.9652,
      "step": 13818
    },
    {
      "epoch": 0.9970058800187583,
      "grad_norm": 3.1759214851225597,
      "learning_rate": 9.179248540513995e-11,
      "loss": 1.0064,
      "step": 13819
    },
    {
      "epoch": 0.9970780274881859,
      "grad_norm": 3.1016764199501385,
      "learning_rate": 8.736944127463176e-11,
      "loss": 1.0654,
      "step": 13820
    },
    {
      "epoch": 0.9971501749576134,
      "grad_norm": 4.00053974324774,
      "learning_rate": 8.305560496957297e-11,
      "loss": 0.9586,
      "step": 13821
    },
    {
      "epoch": 0.9972223224270409,
      "grad_norm": 4.025932886283426,
      "learning_rate": 7.885097672533092e-11,
      "loss": 0.9308,
      "step": 13822
    },
    {
      "epoch": 0.9972944698964684,
      "grad_norm": 3.8829736196450493,
      "learning_rate": 7.475555677149969e-11,
      "loss": 0.8817,
      "step": 13823
    },
    {
      "epoch": 0.9973666173658959,
      "grad_norm": 3.6809611131043645,
      "learning_rate": 7.076934533190027e-11,
      "loss": 0.8915,
      "step": 13824
    },
    {
      "epoch": 0.9974387648353235,
      "grad_norm": 4.6748900619880684,
      "learning_rate": 6.689234262413634e-11,
      "loss": 1.0085,
      "step": 13825
    },
    {
      "epoch": 0.9975109123047509,
      "grad_norm": 4.286920064513259,
      "learning_rate": 6.312454885981644e-11,
      "loss": 0.9663,
      "step": 13826
    },
    {
      "epoch": 0.9975830597741784,
      "grad_norm": 5.471191907386467,
      "learning_rate": 5.94659642447759e-11,
      "loss": 0.9075,
      "step": 13827
    },
    {
      "epoch": 0.9976552072436059,
      "grad_norm": 13.35064819133911,
      "learning_rate": 5.5916588978854874e-11,
      "loss": 0.9671,
      "step": 13828
    },
    {
      "epoch": 0.9977273547130334,
      "grad_norm": 9.390597376727259,
      "learning_rate": 5.247642325567625e-11,
      "loss": 1.0771,
      "step": 13829
    },
    {
      "epoch": 0.9977995021824609,
      "grad_norm": 3.850061460722773,
      "learning_rate": 4.914546726331181e-11,
      "loss": 0.9722,
      "step": 13830
    },
    {
      "epoch": 0.9978716496518885,
      "grad_norm": 4.573212679910944,
      "learning_rate": 4.592372118339405e-11,
      "loss": 0.9526,
      "step": 13831
    },
    {
      "epoch": 0.997943797121316,
      "grad_norm": 4.361753409697394,
      "learning_rate": 4.2811185192226374e-11,
      "loss": 1.0243,
      "step": 13832
    },
    {
      "epoch": 0.9980159445907435,
      "grad_norm": 4.345368886373974,
      "learning_rate": 3.980785945945087e-11,
      "loss": 0.8735,
      "step": 13833
    },
    {
      "epoch": 0.998088092060171,
      "grad_norm": 3.6222922636903983,
      "learning_rate": 3.691374414893644e-11,
      "loss": 1.0211,
      "step": 13834
    },
    {
      "epoch": 0.9981602395295985,
      "grad_norm": 4.05699394939104,
      "learning_rate": 3.412883941922295e-11,
      "loss": 0.9038,
      "step": 13835
    },
    {
      "epoch": 0.9982323869990261,
      "grad_norm": 3.180693446191415,
      "learning_rate": 3.1453145421966865e-11,
      "loss": 0.9571,
      "step": 13836
    },
    {
      "epoch": 0.9983045344684535,
      "grad_norm": 6.041029832973107,
      "learning_rate": 2.8886662303717614e-11,
      "loss": 0.9494,
      "step": 13837
    },
    {
      "epoch": 0.998376681937881,
      "grad_norm": 4.899463211552113,
      "learning_rate": 2.642939020391921e-11,
      "loss": 1.078,
      "step": 13838
    },
    {
      "epoch": 0.9984488294073085,
      "grad_norm": 5.030493258792602,
      "learning_rate": 2.4081329257352733e-11,
      "loss": 0.8729,
      "step": 13839
    },
    {
      "epoch": 0.998520976876736,
      "grad_norm": 4.05754044846959,
      "learning_rate": 2.184247959213792e-11,
      "loss": 0.9218,
      "step": 13840
    },
    {
      "epoch": 0.9985931243461635,
      "grad_norm": 3.5853910673928078,
      "learning_rate": 1.97128413303993e-11,
      "loss": 0.9633,
      "step": 13841
    },
    {
      "epoch": 0.9986652718155911,
      "grad_norm": 4.873803603255916,
      "learning_rate": 1.76924145882662e-11,
      "loss": 0.9313,
      "step": 13842
    },
    {
      "epoch": 0.9987374192850186,
      "grad_norm": 3.3787737340795636,
      "learning_rate": 1.5781199476538887e-11,
      "loss": 0.9192,
      "step": 13843
    },
    {
      "epoch": 0.9988095667544461,
      "grad_norm": 2.8812537799404945,
      "learning_rate": 1.3979196099134227e-11,
      "loss": 0.9096,
      "step": 13844
    },
    {
      "epoch": 0.9988817142238736,
      "grad_norm": 5.3899670185502595,
      "learning_rate": 1.2286404554862073e-11,
      "loss": 0.9342,
      "step": 13845
    },
    {
      "epoch": 0.9989538616933011,
      "grad_norm": 5.913006183013961,
      "learning_rate": 1.0702824935648891e-11,
      "loss": 0.9665,
      "step": 13846
    },
    {
      "epoch": 0.9990260091627287,
      "grad_norm": 3.9963868034651195,
      "learning_rate": 9.22845732831412e-12,
      "loss": 1.0504,
      "step": 13847
    },
    {
      "epoch": 0.9990981566321562,
      "grad_norm": 4.390366081667301,
      "learning_rate": 7.863301813459954e-12,
      "loss": 0.9258,
      "step": 13848
    },
    {
      "epoch": 0.9991703041015836,
      "grad_norm": 3.469276457005385,
      "learning_rate": 6.607358465249291e-12,
      "loss": 0.8958,
      "step": 13849
    },
    {
      "epoch": 0.9992424515710111,
      "grad_norm": 13.849718309973854,
      "learning_rate": 5.460627352738001e-12,
      "loss": 0.9029,
      "step": 13850
    },
    {
      "epoch": 0.9993145990404386,
      "grad_norm": 2.974793717100693,
      "learning_rate": 4.423108538098574e-12,
      "loss": 0.9947,
      "step": 13851
    },
    {
      "epoch": 0.9993867465098661,
      "grad_norm": 6.734469030303517,
      "learning_rate": 3.4948020781744304e-12,
      "loss": 1.0097,
      "step": 13852
    },
    {
      "epoch": 0.9994588939792937,
      "grad_norm": 2.9266504440495233,
      "learning_rate": 2.6757080238137832e-12,
      "loss": 0.9875,
      "step": 13853
    },
    {
      "epoch": 0.9995310414487212,
      "grad_norm": 2.920405217102599,
      "learning_rate": 1.9658264194255538e-12,
      "loss": 0.9409,
      "step": 13854
    },
    {
      "epoch": 0.9996031889181487,
      "grad_norm": 7.796654585358304,
      "learning_rate": 1.3651573040895925e-12,
      "loss": 1.003,
      "step": 13855
    },
    {
      "epoch": 0.9996753363875762,
      "grad_norm": 0.7006087951079244,
      "learning_rate": 8.737007104464567e-13,
      "loss": 0.8214,
      "step": 13856
    },
    {
      "epoch": 0.9997474838570037,
      "grad_norm": 5.208101292845416,
      "learning_rate": 4.914566651414986e-13,
      "loss": 0.9876,
      "step": 13857
    },
    {
      "epoch": 0.9998196313264313,
      "grad_norm": 3.597117695380081,
      "learning_rate": 2.1842518949100052e-13,
      "loss": 0.9981,
      "step": 13858
    },
    {
      "epoch": 0.9998917787958588,
      "grad_norm": 3.631817293747522,
      "learning_rate": 5.460629814990625e-14,
      "loss": 0.9851,
      "step": 13859
    },
    {
      "epoch": 0.9999639262652863,
      "grad_norm": 13.084135836794998,
      "learning_rate": 0.0,
      "loss": 0.8209,
      "step": 13860
    },
    {
      "epoch": 0.9999639262652863,
      "step": 13860,
      "total_flos": 9.616305352836383e+17,
      "train_loss": 0.1470793764484565,
      "train_runtime": 31984.6409,
      "train_samples_per_second": 20.801,
      "train_steps_per_second": 0.433
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 13860,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 9.616305352836383e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}