{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 20.984308131241086,
  "eval_steps": 500,
  "global_step": 11025,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0019020446980504042,
      "grad_norm": 0.9932524561882019,
      "learning_rate": 2e-05,
      "loss": 1.3348,
      "step": 1
    },
    {
      "epoch": 0.0038040893961008085,
      "grad_norm": 0.9241018295288086,
      "learning_rate": 4e-05,
      "loss": 1.3131,
      "step": 2
    },
    {
      "epoch": 0.005706134094151213,
      "grad_norm": 1.1556137800216675,
      "learning_rate": 6e-05,
      "loss": 1.5644,
      "step": 3
    },
    {
      "epoch": 0.007608178792201617,
      "grad_norm": 0.8612737059593201,
      "learning_rate": 8e-05,
      "loss": 1.2192,
      "step": 4
    },
    {
      "epoch": 0.009510223490252021,
      "grad_norm": 0.8998388648033142,
      "learning_rate": 0.0001,
      "loss": 1.3651,
      "step": 5
    },
    {
      "epoch": 0.011412268188302425,
      "grad_norm": 0.7211980819702148,
      "learning_rate": 9.999364877738964e-05,
      "loss": 1.2525,
      "step": 6
    },
    {
      "epoch": 0.01331431288635283,
      "grad_norm": 0.44894707202911377,
      "learning_rate": 9.998729755477931e-05,
      "loss": 1.1999,
      "step": 7
    },
    {
      "epoch": 0.015216357584403234,
      "grad_norm": 0.4338511824607849,
      "learning_rate": 9.998094633216895e-05,
      "loss": 1.0147,
      "step": 8
    },
    {
      "epoch": 0.017118402282453638,
      "grad_norm": 0.5658989548683167,
      "learning_rate": 9.99745951095586e-05,
      "loss": 1.1997,
      "step": 9
    },
    {
      "epoch": 0.019020446980504042,
      "grad_norm": 0.4467356503009796,
      "learning_rate": 9.996824388694824e-05,
      "loss": 1.0424,
      "step": 10
    },
    {
      "epoch": 0.020922491678554447,
      "grad_norm": 0.3743385374546051,
      "learning_rate": 9.996189266433789e-05,
      "loss": 1.0902,
      "step": 11
    },
    {
      "epoch": 0.02282453637660485,
      "grad_norm": 0.30667275190353394,
      "learning_rate": 9.995554144172754e-05,
      "loss": 0.8736,
      "step": 12
    },
    {
      "epoch": 0.024726581074655255,
      "grad_norm": 0.48634254932403564,
      "learning_rate": 9.994919021911718e-05,
      "loss": 0.977,
      "step": 13
    },
    {
      "epoch": 0.02662862577270566,
      "grad_norm": 0.4229658246040344,
      "learning_rate": 9.994283899650683e-05,
      "loss": 0.9673,
      "step": 14
    },
    {
      "epoch": 0.028530670470756064,
      "grad_norm": 0.39269882440567017,
      "learning_rate": 9.993648777389648e-05,
      "loss": 1.0001,
      "step": 15
    },
    {
      "epoch": 0.030432715168806468,
      "grad_norm": 0.38597363233566284,
      "learning_rate": 9.993013655128612e-05,
      "loss": 0.9705,
      "step": 16
    },
    {
      "epoch": 0.03233475986685687,
      "grad_norm": 0.40809136629104614,
      "learning_rate": 9.992378532867577e-05,
      "loss": 0.9246,
      "step": 17
    },
    {
      "epoch": 0.034236804564907276,
      "grad_norm": 0.4431133270263672,
      "learning_rate": 9.991743410606542e-05,
      "loss": 1.0409,
      "step": 18
    },
    {
      "epoch": 0.03613884926295768,
      "grad_norm": 0.5659255981445312,
      "learning_rate": 9.991108288345506e-05,
      "loss": 1.1118,
      "step": 19
    },
    {
      "epoch": 0.038040893961008085,
      "grad_norm": 0.4943106472492218,
      "learning_rate": 9.990473166084471e-05,
      "loss": 0.9213,
      "step": 20
    },
    {
      "epoch": 0.039942938659058486,
      "grad_norm": 0.48820945620536804,
      "learning_rate": 9.989838043823437e-05,
      "loss": 0.9108,
      "step": 21
    },
    {
      "epoch": 0.04184498335710889,
      "grad_norm": 0.4464576542377472,
      "learning_rate": 9.989202921562402e-05,
      "loss": 0.8959,
      "step": 22
    },
    {
      "epoch": 0.043747028055159294,
      "grad_norm": 0.3870016038417816,
      "learning_rate": 9.988567799301366e-05,
      "loss": 0.8013,
      "step": 23
    },
    {
      "epoch": 0.0456490727532097,
      "grad_norm": 0.42381179332733154,
      "learning_rate": 9.987932677040331e-05,
      "loss": 0.8584,
      "step": 24
    },
    {
      "epoch": 0.0475511174512601,
      "grad_norm": 0.37170907855033875,
      "learning_rate": 9.987297554779296e-05,
      "loss": 0.7849,
      "step": 25
    },
    {
      "epoch": 0.04945316214931051,
      "grad_norm": 0.4516700506210327,
      "learning_rate": 9.98666243251826e-05,
      "loss": 0.8902,
      "step": 26
    },
    {
      "epoch": 0.05135520684736091,
      "grad_norm": 0.3525027334690094,
      "learning_rate": 9.986027310257225e-05,
      "loss": 0.6029,
      "step": 27
    },
    {
      "epoch": 0.05325725154541132,
      "grad_norm": 0.437707781791687,
      "learning_rate": 9.98539218799619e-05,
      "loss": 0.7387,
      "step": 28
    },
    {
      "epoch": 0.05515929624346172,
      "grad_norm": 0.45205071568489075,
      "learning_rate": 9.984757065735154e-05,
      "loss": 0.7468,
      "step": 29
    },
    {
      "epoch": 0.05706134094151213,
      "grad_norm": 0.3709086775779724,
      "learning_rate": 9.984121943474119e-05,
      "loss": 0.7365,
      "step": 30
    },
    {
      "epoch": 0.05896338563956253,
      "grad_norm": 0.4089844822883606,
      "learning_rate": 9.983486821213084e-05,
      "loss": 0.6563,
      "step": 31
    },
    {
      "epoch": 0.060865430337612936,
      "grad_norm": 0.45955532789230347,
      "learning_rate": 9.982851698952048e-05,
      "loss": 0.8021,
      "step": 32
    },
    {
      "epoch": 0.06276747503566334,
      "grad_norm": 0.5240988731384277,
      "learning_rate": 9.982216576691013e-05,
      "loss": 0.6933,
      "step": 33
    },
    {
      "epoch": 0.06466951973371374,
      "grad_norm": 0.4703526496887207,
      "learning_rate": 9.981581454429977e-05,
      "loss": 0.7339,
      "step": 34
    },
    {
      "epoch": 0.06657156443176415,
      "grad_norm": 0.5659805536270142,
      "learning_rate": 9.980946332168944e-05,
      "loss": 0.8139,
      "step": 35
    },
    {
      "epoch": 0.06847360912981455,
      "grad_norm": 0.39259326457977295,
      "learning_rate": 9.980311209907908e-05,
      "loss": 0.5838,
      "step": 36
    },
    {
      "epoch": 0.07037565382786495,
      "grad_norm": 0.4165003001689911,
      "learning_rate": 9.979676087646871e-05,
      "loss": 0.674,
      "step": 37
    },
    {
      "epoch": 0.07227769852591535,
      "grad_norm": 0.4533802568912506,
      "learning_rate": 9.979040965385838e-05,
      "loss": 0.6974,
      "step": 38
    },
    {
      "epoch": 0.07417974322396577,
      "grad_norm": 0.5213814973831177,
      "learning_rate": 9.978405843124802e-05,
      "loss": 0.7896,
      "step": 39
    },
    {
      "epoch": 0.07608178792201617,
      "grad_norm": 0.3241259753704071,
      "learning_rate": 9.977770720863767e-05,
      "loss": 0.5895,
      "step": 40
    },
    {
      "epoch": 0.07798383262006657,
      "grad_norm": 0.34446167945861816,
      "learning_rate": 9.977135598602731e-05,
      "loss": 0.6222,
      "step": 41
    },
    {
      "epoch": 0.07988587731811697,
      "grad_norm": 0.49035167694091797,
      "learning_rate": 9.976500476341696e-05,
      "loss": 0.6978,
      "step": 42
    },
    {
      "epoch": 0.08178792201616739,
      "grad_norm": 0.4795296788215637,
      "learning_rate": 9.975865354080661e-05,
      "loss": 0.7368,
      "step": 43
    },
    {
      "epoch": 0.08368996671421779,
      "grad_norm": 0.44959381222724915,
      "learning_rate": 9.975230231819625e-05,
      "loss": 0.57,
      "step": 44
    },
    {
      "epoch": 0.08559201141226819,
      "grad_norm": 0.4577605426311493,
      "learning_rate": 9.974595109558592e-05,
      "loss": 0.691,
      "step": 45
    },
    {
      "epoch": 0.08749405611031859,
      "grad_norm": 0.41654840111732483,
      "learning_rate": 9.973959987297555e-05,
      "loss": 0.6346,
      "step": 46
    },
    {
      "epoch": 0.089396100808369,
      "grad_norm": 0.6599829196929932,
      "learning_rate": 9.973324865036519e-05,
      "loss": 0.6358,
      "step": 47
    },
    {
      "epoch": 0.0912981455064194,
      "grad_norm": 0.38539162278175354,
      "learning_rate": 9.972689742775484e-05,
      "loss": 0.5723,
      "step": 48
    },
    {
      "epoch": 0.0932001902044698,
      "grad_norm": 0.4626316428184509,
      "learning_rate": 9.97205462051445e-05,
      "loss": 0.6845,
      "step": 49
    },
    {
      "epoch": 0.0951022349025202,
      "grad_norm": 0.348387211561203,
      "learning_rate": 9.971419498253413e-05,
      "loss": 0.4857,
      "step": 50
    },
    {
      "epoch": 0.09700427960057062,
      "grad_norm": 0.4964020252227783,
      "learning_rate": 9.970784375992379e-05,
      "loss": 0.7141,
      "step": 51
    },
    {
      "epoch": 0.09890632429862102,
      "grad_norm": 0.4282241463661194,
      "learning_rate": 9.970149253731344e-05,
      "loss": 0.6619,
      "step": 52
    },
    {
      "epoch": 0.10080836899667142,
      "grad_norm": 0.35991716384887695,
      "learning_rate": 9.969514131470309e-05,
      "loss": 0.4727,
      "step": 53
    },
    {
      "epoch": 0.10271041369472182,
      "grad_norm": 0.3936012387275696,
      "learning_rate": 9.968879009209273e-05,
      "loss": 0.5644,
      "step": 54
    },
    {
      "epoch": 0.10461245839277224,
      "grad_norm": 0.39267924427986145,
      "learning_rate": 9.968243886948238e-05,
      "loss": 0.5126,
      "step": 55
    },
    {
      "epoch": 0.10651450309082264,
      "grad_norm": 0.4119136333465576,
      "learning_rate": 9.967608764687203e-05,
      "loss": 0.471,
      "step": 56
    },
    {
      "epoch": 0.10841654778887304,
      "grad_norm": 0.5160384178161621,
      "learning_rate": 9.966973642426167e-05,
      "loss": 0.6555,
      "step": 57
    },
    {
      "epoch": 0.11031859248692344,
      "grad_norm": 0.4742174744606018,
      "learning_rate": 9.966338520165132e-05,
      "loss": 0.6093,
      "step": 58
    },
    {
      "epoch": 0.11222063718497385,
      "grad_norm": 0.3615169823169708,
      "learning_rate": 9.965703397904097e-05,
      "loss": 0.5527,
      "step": 59
    },
    {
      "epoch": 0.11412268188302425,
      "grad_norm": 0.5700575113296509,
      "learning_rate": 9.965068275643061e-05,
      "loss": 0.5713,
      "step": 60
    },
    {
      "epoch": 0.11602472658107466,
      "grad_norm": 0.4825727939605713,
      "learning_rate": 9.964433153382026e-05,
      "loss": 0.5142,
      "step": 61
    },
    {
      "epoch": 0.11792677127912506,
      "grad_norm": 0.392088919878006,
      "learning_rate": 9.963798031120992e-05,
      "loss": 0.513,
      "step": 62
    },
    {
      "epoch": 0.11982881597717546,
      "grad_norm": 0.35883110761642456,
      "learning_rate": 9.963162908859957e-05,
      "loss": 0.501,
      "step": 63
    },
    {
      "epoch": 0.12173086067522587,
      "grad_norm": 0.39946749806404114,
      "learning_rate": 9.96252778659892e-05,
      "loss": 0.5532,
      "step": 64
    },
    {
      "epoch": 0.12363290537327627,
      "grad_norm": 0.4191288352012634,
      "learning_rate": 9.961892664337886e-05,
      "loss": 0.5258,
      "step": 65
    },
    {
      "epoch": 0.12553495007132667,
      "grad_norm": 0.3662487268447876,
      "learning_rate": 9.961257542076851e-05,
      "loss": 0.5121,
      "step": 66
    },
    {
      "epoch": 0.1274369947693771,
      "grad_norm": 0.5582164525985718,
      "learning_rate": 9.960622419815815e-05,
      "loss": 0.6494,
      "step": 67
    },
    {
      "epoch": 0.12933903946742747,
      "grad_norm": 0.485128790140152,
      "learning_rate": 9.959987297554779e-05,
      "loss": 0.6022,
      "step": 68
    },
    {
      "epoch": 0.1312410841654779,
      "grad_norm": 0.3816944360733032,
      "learning_rate": 9.959352175293745e-05,
      "loss": 0.4851,
      "step": 69
    },
    {
      "epoch": 0.1331431288635283,
      "grad_norm": 0.3637336194515228,
      "learning_rate": 9.958717053032709e-05,
      "loss": 0.4344,
      "step": 70
    },
    {
      "epoch": 0.1350451735615787,
      "grad_norm": 0.4418705105781555,
      "learning_rate": 9.958081930771674e-05,
      "loss": 0.6008,
      "step": 71
    },
    {
      "epoch": 0.1369472182596291,
      "grad_norm": 0.44138631224632263,
      "learning_rate": 9.95744680851064e-05,
      "loss": 0.5319,
      "step": 72
    },
    {
      "epoch": 0.1388492629576795,
      "grad_norm": 0.37523001432418823,
      "learning_rate": 9.956811686249603e-05,
      "loss": 0.657,
      "step": 73
    },
    {
      "epoch": 0.1407513076557299,
      "grad_norm": 0.4489665627479553,
      "learning_rate": 9.956176563988568e-05,
      "loss": 0.5526,
      "step": 74
    },
    {
      "epoch": 0.14265335235378032,
      "grad_norm": 0.39318791031837463,
      "learning_rate": 9.955541441727532e-05,
      "loss": 0.6046,
      "step": 75
    },
    {
      "epoch": 0.1445553970518307,
      "grad_norm": 0.4817538261413574,
      "learning_rate": 9.954906319466499e-05,
      "loss": 0.5149,
      "step": 76
    },
    {
      "epoch": 0.14645744174988112,
      "grad_norm": 0.4451163411140442,
      "learning_rate": 9.954271197205463e-05,
      "loss": 0.4892,
      "step": 77
    },
    {
      "epoch": 0.14835948644793154,
      "grad_norm": 0.29836660623550415,
      "learning_rate": 9.953636074944426e-05,
      "loss": 0.4005,
      "step": 78
    },
    {
      "epoch": 0.15026153114598192,
      "grad_norm": 0.3185100555419922,
      "learning_rate": 9.953000952683393e-05,
      "loss": 0.4168,
      "step": 79
    },
    {
      "epoch": 0.15216357584403234,
      "grad_norm": 0.26550424098968506,
      "learning_rate": 9.952365830422357e-05,
      "loss": 0.39,
      "step": 80
    },
    {
      "epoch": 0.15406562054208273,
      "grad_norm": 0.4328240156173706,
      "learning_rate": 9.951730708161322e-05,
      "loss": 0.5041,
      "step": 81
    },
    {
      "epoch": 0.15596766524013314,
      "grad_norm": 0.5178936123847961,
      "learning_rate": 9.951095585900286e-05,
      "loss": 0.6017,
      "step": 82
    },
    {
      "epoch": 0.15786970993818356,
      "grad_norm": 0.45657551288604736,
      "learning_rate": 9.950460463639251e-05,
      "loss": 0.5734,
      "step": 83
    },
    {
      "epoch": 0.15977175463623394,
      "grad_norm": 0.5482913851737976,
      "learning_rate": 9.949825341378216e-05,
      "loss": 0.6015,
      "step": 84
    },
    {
      "epoch": 0.16167379933428436,
      "grad_norm": 0.39362308382987976,
      "learning_rate": 9.94919021911718e-05,
      "loss": 0.5712,
      "step": 85
    },
    {
      "epoch": 0.16357584403233477,
      "grad_norm": 0.4381113350391388,
      "learning_rate": 9.948555096856145e-05,
      "loss": 0.5194,
      "step": 86
    },
    {
      "epoch": 0.16547788873038516,
      "grad_norm": 0.5021312236785889,
      "learning_rate": 9.94791997459511e-05,
      "loss": 0.5279,
      "step": 87
    },
    {
      "epoch": 0.16737993342843557,
      "grad_norm": 0.4364267587661743,
      "learning_rate": 9.947284852334074e-05,
      "loss": 0.5892,
      "step": 88
    },
    {
      "epoch": 0.16928197812648596,
      "grad_norm": 0.37873050570487976,
      "learning_rate": 9.94664973007304e-05,
      "loss": 0.5328,
      "step": 89
    },
    {
      "epoch": 0.17118402282453637,
      "grad_norm": 0.4768919050693512,
      "learning_rate": 9.946014607812005e-05,
      "loss": 0.4889,
      "step": 90
    },
    {
      "epoch": 0.1730860675225868,
      "grad_norm": 0.3834541440010071,
      "learning_rate": 9.945379485550968e-05,
      "loss": 0.4642,
      "step": 91
    },
    {
      "epoch": 0.17498811222063718,
      "grad_norm": 0.48581764101982117,
      "learning_rate": 9.944744363289934e-05,
      "loss": 0.4741,
      "step": 92
    },
    {
      "epoch": 0.1768901569186876,
      "grad_norm": 0.39364808797836304,
      "learning_rate": 9.944109241028899e-05,
      "loss": 0.5684,
      "step": 93
    },
    {
      "epoch": 0.178792201616738,
      "grad_norm": 0.4657204747200012,
      "learning_rate": 9.943474118767864e-05,
      "loss": 0.609,
      "step": 94
    },
    {
      "epoch": 0.1806942463147884,
      "grad_norm": 0.40989887714385986,
      "learning_rate": 9.942838996506828e-05,
      "loss": 0.4319,
      "step": 95
    },
    {
      "epoch": 0.1825962910128388,
      "grad_norm": 0.43797624111175537,
      "learning_rate": 9.942203874245793e-05,
      "loss": 0.4997,
      "step": 96
    },
    {
      "epoch": 0.1844983357108892,
      "grad_norm": 0.3887675106525421,
      "learning_rate": 9.941568751984758e-05,
      "loss": 0.5548,
      "step": 97
    },
    {
      "epoch": 0.1864003804089396,
      "grad_norm": 0.39017003774642944,
      "learning_rate": 9.940933629723722e-05,
      "loss": 0.5113,
      "step": 98
    },
    {
      "epoch": 0.18830242510699002,
      "grad_norm": 0.41409194469451904,
      "learning_rate": 9.940298507462687e-05,
      "loss": 0.5496,
      "step": 99
    },
    {
      "epoch": 0.1902044698050404,
      "grad_norm": 0.34578803181648254,
      "learning_rate": 9.939663385201652e-05,
      "loss": 0.4048,
      "step": 100
    },
    {
      "epoch": 0.19210651450309082,
      "grad_norm": 0.32233092188835144,
      "learning_rate": 9.939028262940616e-05,
      "loss": 0.4442,
      "step": 101
    },
    {
      "epoch": 0.19400855920114124,
      "grad_norm": 0.45841965079307556,
      "learning_rate": 9.938393140679581e-05,
      "loss": 0.5646,
      "step": 102
    },
    {
      "epoch": 0.19591060389919163,
      "grad_norm": 0.3825596272945404,
      "learning_rate": 9.937758018418547e-05,
      "loss": 0.4583,
      "step": 103
    },
    {
      "epoch": 0.19781264859724204,
      "grad_norm": 0.44690102338790894,
      "learning_rate": 9.93712289615751e-05,
      "loss": 0.5799,
      "step": 104
    },
    {
      "epoch": 0.19971469329529243,
      "grad_norm": 0.4881773591041565,
      "learning_rate": 9.936487773896476e-05,
      "loss": 0.4094,
      "step": 105
    },
    {
      "epoch": 0.20161673799334284,
      "grad_norm": 0.4745669960975647,
      "learning_rate": 9.93585265163544e-05,
      "loss": 0.6068,
      "step": 106
    },
    {
      "epoch": 0.20351878269139326,
      "grad_norm": 0.5497081279754639,
      "learning_rate": 9.935217529374406e-05,
      "loss": 0.4654,
      "step": 107
    },
    {
      "epoch": 0.20542082738944364,
      "grad_norm": 0.3564707636833191,
      "learning_rate": 9.93458240711337e-05,
      "loss": 0.5678,
      "step": 108
    },
    {
      "epoch": 0.20732287208749406,
      "grad_norm": 0.446321964263916,
      "learning_rate": 9.933947284852334e-05,
      "loss": 0.4503,
      "step": 109
    },
    {
      "epoch": 0.20922491678554447,
      "grad_norm": 0.4253140389919281,
      "learning_rate": 9.9333121625913e-05,
      "loss": 0.538,
      "step": 110
    },
    {
      "epoch": 0.21112696148359486,
      "grad_norm": 0.4123047888278961,
      "learning_rate": 9.932677040330264e-05,
      "loss": 0.4359,
      "step": 111
    },
    {
      "epoch": 0.21302900618164528,
      "grad_norm": 0.3887772262096405,
      "learning_rate": 9.932041918069229e-05,
      "loss": 0.5534,
      "step": 112
    },
    {
      "epoch": 0.21493105087969566,
      "grad_norm": 0.38153669238090515,
      "learning_rate": 9.931406795808193e-05,
      "loss": 0.4296,
      "step": 113
    },
    {
      "epoch": 0.21683309557774608,
      "grad_norm": 0.43017521500587463,
      "learning_rate": 9.930771673547158e-05,
      "loss": 0.5899,
      "step": 114
    },
    {
      "epoch": 0.2187351402757965,
      "grad_norm": 0.40156394243240356,
      "learning_rate": 9.930136551286123e-05,
      "loss": 0.3917,
      "step": 115
    },
    {
      "epoch": 0.22063718497384688,
      "grad_norm": 0.3576590120792389,
      "learning_rate": 9.929501429025087e-05,
      "loss": 0.3908,
      "step": 116
    },
    {
      "epoch": 0.2225392296718973,
      "grad_norm": 0.33245769143104553,
      "learning_rate": 9.928866306764054e-05,
      "loss": 0.4043,
      "step": 117
    },
    {
      "epoch": 0.2244412743699477,
      "grad_norm": 0.43169739842414856,
      "learning_rate": 9.928231184503018e-05,
      "loss": 0.5569,
      "step": 118
    },
    {
      "epoch": 0.2263433190679981,
      "grad_norm": 0.4004412293434143,
      "learning_rate": 9.927596062241981e-05,
      "loss": 0.4931,
      "step": 119
    },
    {
      "epoch": 0.2282453637660485,
      "grad_norm": 0.3550797998905182,
      "learning_rate": 9.926960939980947e-05,
      "loss": 0.4505,
      "step": 120
    },
    {
      "epoch": 0.2301474084640989,
      "grad_norm": 0.3701287508010864,
      "learning_rate": 9.926325817719912e-05,
      "loss": 0.4967,
      "step": 121
    },
    {
      "epoch": 0.2320494531621493,
      "grad_norm": 0.4120308756828308,
      "learning_rate": 9.925690695458876e-05,
      "loss": 0.4408,
      "step": 122
    },
    {
      "epoch": 0.23395149786019973,
      "grad_norm": 0.4737403392791748,
      "learning_rate": 9.925055573197841e-05,
      "loss": 0.7221,
      "step": 123
    },
    {
      "epoch": 0.2358535425582501,
      "grad_norm": 0.37103158235549927,
      "learning_rate": 9.924420450936806e-05,
      "loss": 0.4419,
      "step": 124
    },
    {
      "epoch": 0.23775558725630053,
      "grad_norm": 0.48644623160362244,
      "learning_rate": 9.923785328675771e-05,
      "loss": 0.5006,
      "step": 125
    },
    {
      "epoch": 0.2396576319543509,
      "grad_norm": 0.3381918966770172,
      "learning_rate": 9.923150206414735e-05,
      "loss": 0.4786,
      "step": 126
    },
    {
      "epoch": 0.24155967665240133,
      "grad_norm": 0.4500490128993988,
      "learning_rate": 9.9225150841537e-05,
      "loss": 0.4984,
      "step": 127
    },
    {
      "epoch": 0.24346172135045174,
      "grad_norm": 0.5506143569946289,
      "learning_rate": 9.921879961892665e-05,
      "loss": 0.4857,
      "step": 128
    },
    {
      "epoch": 0.24536376604850213,
      "grad_norm": 0.4111080467700958,
      "learning_rate": 9.921244839631629e-05,
      "loss": 0.4464,
      "step": 129
    },
    {
      "epoch": 0.24726581074655254,
      "grad_norm": 0.52936851978302,
      "learning_rate": 9.920609717370594e-05,
      "loss": 0.5664,
      "step": 130
    },
    {
      "epoch": 0.24916785544460296,
      "grad_norm": 0.465009480714798,
      "learning_rate": 9.91997459510956e-05,
      "loss": 0.4318,
      "step": 131
    },
    {
      "epoch": 0.25106990014265335,
      "grad_norm": 0.3044665455818176,
      "learning_rate": 9.919339472848523e-05,
      "loss": 0.4284,
      "step": 132
    },
    {
      "epoch": 0.25297194484070373,
      "grad_norm": 0.4849638342857361,
      "learning_rate": 9.918704350587488e-05,
      "loss": 0.5956,
      "step": 133
    },
    {
      "epoch": 0.2548739895387542,
      "grad_norm": 0.4701893925666809,
      "learning_rate": 9.918069228326454e-05,
      "loss": 0.4541,
      "step": 134
    },
    {
      "epoch": 0.25677603423680456,
      "grad_norm": 0.42524924874305725,
      "learning_rate": 9.917434106065419e-05,
      "loss": 0.4991,
      "step": 135
    },
    {
      "epoch": 0.25867807893485495,
      "grad_norm": 0.46284592151641846,
      "learning_rate": 9.916798983804383e-05,
      "loss": 0.453,
      "step": 136
    },
    {
      "epoch": 0.2605801236329054,
      "grad_norm": 0.40281572937965393,
      "learning_rate": 9.916163861543348e-05,
      "loss": 0.4771,
      "step": 137
    },
    {
      "epoch": 0.2624821683309558,
      "grad_norm": 0.425214558839798,
      "learning_rate": 9.915528739282313e-05,
      "loss": 0.4665,
      "step": 138
    },
    {
      "epoch": 0.26438421302900617,
      "grad_norm": 0.4181045889854431,
      "learning_rate": 9.914893617021277e-05,
      "loss": 0.5014,
      "step": 139
    },
    {
      "epoch": 0.2662862577270566,
      "grad_norm": 0.4024779498577118,
      "learning_rate": 9.914258494760241e-05,
      "loss": 0.5905,
      "step": 140
    },
    {
      "epoch": 0.268188302425107,
      "grad_norm": 0.3768770694732666,
      "learning_rate": 9.913623372499207e-05,
      "loss": 0.408,
      "step": 141
    },
    {
      "epoch": 0.2700903471231574,
      "grad_norm": 0.4033905267715454,
      "learning_rate": 9.912988250238171e-05,
      "loss": 0.4511,
      "step": 142
    },
    {
      "epoch": 0.2719923918212078,
      "grad_norm": 0.32505708932876587,
      "learning_rate": 9.912353127977136e-05,
      "loss": 0.4395,
      "step": 143
    },
    {
      "epoch": 0.2738944365192582,
      "grad_norm": 0.3487790822982788,
      "learning_rate": 9.9117180057161e-05,
      "loss": 0.3601,
      "step": 144
    },
    {
      "epoch": 0.2757964812173086,
      "grad_norm": 0.30558326840400696,
      "learning_rate": 9.911082883455065e-05,
      "loss": 0.4607,
      "step": 145
    },
    {
      "epoch": 0.277698525915359,
      "grad_norm": 0.3752080500125885,
      "learning_rate": 9.91044776119403e-05,
      "loss": 0.3957,
      "step": 146
    },
    {
      "epoch": 0.2796005706134094,
      "grad_norm": 0.3506644368171692,
      "learning_rate": 9.909812638932994e-05,
      "loss": 0.366,
      "step": 147
    },
    {
      "epoch": 0.2815026153114598,
      "grad_norm": 0.43430307507514954,
      "learning_rate": 9.909177516671961e-05,
      "loss": 0.4542,
      "step": 148
    },
    {
      "epoch": 0.2834046600095102,
      "grad_norm": 0.41930171847343445,
      "learning_rate": 9.908542394410925e-05,
      "loss": 0.709,
      "step": 149
    },
    {
      "epoch": 0.28530670470756064,
      "grad_norm": 0.3717108964920044,
      "learning_rate": 9.907907272149888e-05,
      "loss": 0.4701,
      "step": 150
    },
    {
      "epoch": 0.28720874940561103,
      "grad_norm": 0.4177984595298767,
      "learning_rate": 9.907272149888854e-05,
      "loss": 0.6189,
      "step": 151
    },
    {
      "epoch": 0.2891107941036614,
      "grad_norm": 0.37706881761550903,
      "learning_rate": 9.906637027627819e-05,
      "loss": 0.4546,
      "step": 152
    },
    {
      "epoch": 0.29101283880171186,
      "grad_norm": 0.4210599660873413,
      "learning_rate": 9.906001905366784e-05,
      "loss": 0.4716,
      "step": 153
    },
    {
      "epoch": 0.29291488349976225,
      "grad_norm": 0.3707990050315857,
      "learning_rate": 9.905366783105748e-05,
      "loss": 0.4644,
      "step": 154
    },
    {
      "epoch": 0.29481692819781263,
      "grad_norm": 0.36913537979125977,
      "learning_rate": 9.904731660844713e-05,
      "loss": 0.4605,
      "step": 155
    },
    {
      "epoch": 0.2967189728958631,
      "grad_norm": 0.41291072964668274,
      "learning_rate": 9.904096538583678e-05,
      "loss": 0.4294,
      "step": 156
    },
    {
      "epoch": 0.29862101759391346,
      "grad_norm": 0.30809640884399414,
      "learning_rate": 9.903461416322642e-05,
      "loss": 0.4369,
      "step": 157
    },
    {
      "epoch": 0.30052306229196385,
      "grad_norm": 0.4266267716884613,
      "learning_rate": 9.902826294061607e-05,
      "loss": 0.456,
      "step": 158
    },
    {
      "epoch": 0.3024251069900143,
      "grad_norm": 0.37408629059791565,
      "learning_rate": 9.902191171800572e-05,
      "loss": 0.4359,
      "step": 159
    },
    {
      "epoch": 0.3043271516880647,
      "grad_norm": 0.40199100971221924,
      "learning_rate": 9.901556049539536e-05,
      "loss": 0.4433,
      "step": 160
    },
    {
      "epoch": 0.30622919638611507,
      "grad_norm": 0.3430602252483368,
      "learning_rate": 9.900920927278501e-05,
      "loss": 0.4317,
      "step": 161
    },
    {
      "epoch": 0.30813124108416545,
      "grad_norm": 0.5091786980628967,
      "learning_rate": 9.900285805017467e-05,
      "loss": 0.5824,
      "step": 162
    },
    {
      "epoch": 0.3100332857822159,
      "grad_norm": 0.34287527203559875,
      "learning_rate": 9.89965068275643e-05,
      "loss": 0.4025,
      "step": 163
    },
    {
      "epoch": 0.3119353304802663,
      "grad_norm": 0.4919246733188629,
      "learning_rate": 9.899015560495396e-05,
      "loss": 0.5612,
      "step": 164
    },
    {
      "epoch": 0.31383737517831667,
      "grad_norm": 0.35404297709465027,
      "learning_rate": 9.898380438234361e-05,
      "loss": 0.4731,
      "step": 165
    },
    {
      "epoch": 0.3157394198763671,
      "grad_norm": 0.3590085506439209,
      "learning_rate": 9.897745315973326e-05,
      "loss": 0.4365,
      "step": 166
    },
    {
      "epoch": 0.3176414645744175,
      "grad_norm": 0.4132196605205536,
      "learning_rate": 9.89711019371229e-05,
      "loss": 0.3485,
      "step": 167
    },
    {
      "epoch": 0.3195435092724679,
      "grad_norm": 0.46459728479385376,
      "learning_rate": 9.896475071451255e-05,
      "loss": 0.4327,
      "step": 168
    },
    {
      "epoch": 0.3214455539705183,
      "grad_norm": 0.435651957988739,
      "learning_rate": 9.89583994919022e-05,
      "loss": 0.4684,
      "step": 169
    },
    {
      "epoch": 0.3233475986685687,
      "grad_norm": 0.38278958201408386,
      "learning_rate": 9.895204826929184e-05,
      "loss": 0.4265,
      "step": 170
    },
    {
      "epoch": 0.3252496433666191,
      "grad_norm": 0.31499558687210083,
      "learning_rate": 9.894569704668149e-05,
      "loss": 0.4099,
      "step": 171
    },
    {
      "epoch": 0.32715168806466954,
      "grad_norm": 0.40141284465789795,
      "learning_rate": 9.893934582407114e-05,
      "loss": 0.4461,
      "step": 172
    },
    {
      "epoch": 0.32905373276271993,
      "grad_norm": 0.42945384979248047,
      "learning_rate": 9.893299460146078e-05,
      "loss": 0.4379,
      "step": 173
    },
    {
      "epoch": 0.3309557774607703,
      "grad_norm": 0.5186269283294678,
      "learning_rate": 9.892664337885043e-05,
      "loss": 0.5134,
      "step": 174
    },
    {
      "epoch": 0.33285782215882076,
      "grad_norm": 0.3771612048149109,
      "learning_rate": 9.892029215624009e-05,
      "loss": 0.4617,
      "step": 175
    },
    {
      "epoch": 0.33475986685687115,
      "grad_norm": 0.48396849632263184,
      "learning_rate": 9.891394093362972e-05,
      "loss": 0.4944,
      "step": 176
    },
    {
      "epoch": 0.33666191155492153,
      "grad_norm": 0.5303121209144592,
      "learning_rate": 9.890758971101938e-05,
      "loss": 0.4049,
      "step": 177
    },
    {
      "epoch": 0.3385639562529719,
      "grad_norm": 0.33063024282455444,
      "learning_rate": 9.890123848840901e-05,
      "loss": 0.401,
      "step": 178
    },
    {
      "epoch": 0.34046600095102236,
      "grad_norm": 0.3764759302139282,
      "learning_rate": 9.889488726579868e-05,
      "loss": 0.4222,
      "step": 179
    },
    {
      "epoch": 0.34236804564907275,
      "grad_norm": 0.27206951379776,
      "learning_rate": 9.888853604318832e-05,
      "loss": 0.3206,
      "step": 180
    },
    {
      "epoch": 0.34427009034712314,
      "grad_norm": 0.3893122971057892,
      "learning_rate": 9.888218482057796e-05,
      "loss": 0.3558,
      "step": 181
    },
    {
      "epoch": 0.3461721350451736,
      "grad_norm": 0.42340540885925293,
      "learning_rate": 9.887583359796762e-05,
      "loss": 0.3948,
      "step": 182
    },
    {
      "epoch": 0.34807417974322397,
      "grad_norm": 0.4103796184062958,
      "learning_rate": 9.886948237535726e-05,
      "loss": 0.4769,
      "step": 183
    },
    {
      "epoch": 0.34997622444127435,
      "grad_norm": 0.39225244522094727,
      "learning_rate": 9.886313115274691e-05,
      "loss": 0.441,
      "step": 184
    },
    {
      "epoch": 0.3518782691393248,
      "grad_norm": 0.3774043023586273,
      "learning_rate": 9.885677993013655e-05,
      "loss": 0.3018,
      "step": 185
    },
    {
      "epoch": 0.3537803138373752,
      "grad_norm": 0.4012366235256195,
      "learning_rate": 9.88504287075262e-05,
      "loss": 0.4217,
      "step": 186
    },
    {
      "epoch": 0.35568235853542557,
      "grad_norm": 0.37299972772598267,
      "learning_rate": 9.884407748491585e-05,
      "loss": 0.4518,
      "step": 187
    },
    {
      "epoch": 0.357584403233476,
      "grad_norm": 0.34713125228881836,
      "learning_rate": 9.883772626230549e-05,
      "loss": 0.3882,
      "step": 188
    },
    {
      "epoch": 0.3594864479315264,
      "grad_norm": 0.4148958623409271,
      "learning_rate": 9.883137503969516e-05,
      "loss": 0.4979,
      "step": 189
    },
    {
      "epoch": 0.3613884926295768,
      "grad_norm": 0.3979155421257019,
      "learning_rate": 9.88250238170848e-05,
      "loss": 0.3854,
      "step": 190
    },
    {
      "epoch": 0.36329053732762717,
      "grad_norm": 0.42723751068115234,
      "learning_rate": 9.881867259447443e-05,
      "loss": 0.4325,
      "step": 191
    },
    {
      "epoch": 0.3651925820256776,
      "grad_norm": 0.4195951521396637,
      "learning_rate": 9.881232137186409e-05,
      "loss": 0.3917,
      "step": 192
    },
    {
      "epoch": 0.367094626723728,
      "grad_norm": 0.43937554955482483,
      "learning_rate": 9.880597014925374e-05,
      "loss": 0.3907,
      "step": 193
    },
    {
      "epoch": 0.3689966714217784,
      "grad_norm": 0.3176072835922241,
      "learning_rate": 9.879961892664338e-05,
      "loss": 0.3581,
      "step": 194
    },
    {
      "epoch": 0.37089871611982883,
      "grad_norm": 0.39909854531288147,
      "learning_rate": 9.879326770403303e-05,
      "loss": 0.5881,
      "step": 195
    },
    {
      "epoch": 0.3728007608178792,
      "grad_norm": 0.35058659315109253,
      "learning_rate": 9.878691648142268e-05,
      "loss": 0.4753,
      "step": 196
    },
    {
      "epoch": 0.3747028055159296,
      "grad_norm": 0.3353765904903412,
      "learning_rate": 9.878056525881233e-05,
      "loss": 0.4014,
      "step": 197
    },
    {
      "epoch": 0.37660485021398005,
      "grad_norm": 0.4102007746696472,
      "learning_rate": 9.877421403620197e-05,
      "loss": 0.4841,
      "step": 198
    },
    {
      "epoch": 0.37850689491203043,
      "grad_norm": 0.45450812578201294,
      "learning_rate": 9.876786281359162e-05,
      "loss": 0.4655,
      "step": 199
    },
    {
      "epoch": 0.3804089396100808,
      "grad_norm": 0.32525572180747986,
      "learning_rate": 9.876151159098127e-05,
      "loss": 0.3869,
      "step": 200
    },
    {
      "epoch": 0.38231098430813126,
      "grad_norm": 0.4488207697868347,
      "learning_rate": 9.875516036837091e-05,
      "loss": 0.4743,
      "step": 201
    },
    {
      "epoch": 0.38421302900618165,
      "grad_norm": 0.432962030172348,
      "learning_rate": 9.874880914576056e-05,
      "loss": 0.4171,
      "step": 202
    },
    {
      "epoch": 0.38611507370423204,
      "grad_norm": 0.4264095723628998,
      "learning_rate": 9.874245792315022e-05,
      "loss": 0.4344,
      "step": 203
    },
    {
      "epoch": 0.3880171184022825,
      "grad_norm": 0.43752139806747437,
      "learning_rate": 9.873610670053985e-05,
      "loss": 0.5248,
      "step": 204
    },
    {
      "epoch": 0.38991916310033287,
      "grad_norm": 0.42547503113746643,
      "learning_rate": 9.87297554779295e-05,
      "loss": 0.4011,
      "step": 205
    },
    {
      "epoch": 0.39182120779838325,
      "grad_norm": 0.34600159525871277,
      "learning_rate": 9.872340425531916e-05,
      "loss": 0.3444,
      "step": 206
    },
    {
      "epoch": 0.39372325249643364,
      "grad_norm": 0.3614776134490967,
      "learning_rate": 9.871705303270881e-05,
      "loss": 0.4784,
      "step": 207
    },
    {
      "epoch": 0.3956252971944841,
      "grad_norm": 0.47591882944107056,
      "learning_rate": 9.871070181009845e-05,
      "loss": 0.5159,
      "step": 208
    },
    {
      "epoch": 0.39752734189253447,
      "grad_norm": 0.3321515917778015,
      "learning_rate": 9.870435058748809e-05,
      "loss": 0.4382,
      "step": 209
    },
    {
      "epoch": 0.39942938659058486,
      "grad_norm": 0.45849499106407166,
      "learning_rate": 9.869799936487775e-05,
      "loss": 0.4269,
      "step": 210
    },
    {
      "epoch": 0.4013314312886353,
      "grad_norm": 0.3666900098323822,
      "learning_rate": 9.869164814226739e-05,
      "loss": 0.4077,
      "step": 211
    },
    {
      "epoch": 0.4032334759866857,
      "grad_norm": 0.3387741446495056,
      "learning_rate": 9.868529691965703e-05,
      "loss": 0.4485,
      "step": 212
    },
    {
      "epoch": 0.4051355206847361,
      "grad_norm": 0.3360239267349243,
      "learning_rate": 9.86789456970467e-05,
      "loss": 0.4042,
      "step": 213
    },
    {
      "epoch": 0.4070375653827865,
      "grad_norm": 0.40923500061035156,
      "learning_rate": 9.867259447443633e-05,
      "loss": 0.5001,
      "step": 214
    },
    {
      "epoch": 0.4089396100808369,
      "grad_norm": 0.3974573314189911,
      "learning_rate": 9.866624325182598e-05,
      "loss": 0.4984,
      "step": 215
    },
    {
      "epoch": 0.4108416547788873,
      "grad_norm": 0.4095960557460785,
      "learning_rate": 9.865989202921562e-05,
      "loss": 0.3837,
      "step": 216
    },
    {
      "epoch": 0.41274369947693773,
      "grad_norm": 0.3334168493747711,
      "learning_rate": 9.865354080660527e-05,
      "loss": 0.3935,
      "step": 217
    },
    {
      "epoch": 0.4146457441749881,
      "grad_norm": 0.5007266998291016,
      "learning_rate": 9.864718958399493e-05,
      "loss": 0.4443,
      "step": 218
    },
    {
      "epoch": 0.4165477888730385,
      "grad_norm": 0.35881495475769043,
      "learning_rate": 9.864083836138456e-05,
      "loss": 0.3835,
      "step": 219
    },
    {
      "epoch": 0.41844983357108895,
      "grad_norm": 0.3785092830657959,
      "learning_rate": 9.863448713877423e-05,
      "loss": 0.3884,
      "step": 220
    },
    {
      "epoch": 0.42035187826913933,
      "grad_norm": 0.41435107588768005,
      "learning_rate": 9.862813591616387e-05,
      "loss": 0.4116,
      "step": 221
    },
    {
      "epoch": 0.4222539229671897,
      "grad_norm": 0.41338756680488586,
      "learning_rate": 9.86217846935535e-05,
      "loss": 0.5235,
      "step": 222
    },
    {
      "epoch": 0.4241559676652401,
      "grad_norm": 0.4335710406303406,
      "learning_rate": 9.861543347094316e-05,
      "loss": 0.516,
      "step": 223
    },
    {
      "epoch": 0.42605801236329055,
      "grad_norm": 0.37374967336654663,
      "learning_rate": 9.860908224833281e-05,
      "loss": 0.4663,
      "step": 224
    },
    {
      "epoch": 0.42796005706134094,
      "grad_norm": 0.3213825821876526,
      "learning_rate": 9.860273102572246e-05,
      "loss": 0.3636,
      "step": 225
    },
    {
      "epoch": 0.4298621017593913,
      "grad_norm": 0.41535523533821106,
      "learning_rate": 9.85963798031121e-05,
      "loss": 0.3677,
      "step": 226
    },
    {
      "epoch": 0.43176414645744177,
      "grad_norm": 0.3543884754180908,
      "learning_rate": 9.859002858050175e-05,
      "loss": 0.376,
      "step": 227
    },
    {
      "epoch": 0.43366619115549215,
      "grad_norm": 0.4012312889099121,
      "learning_rate": 9.85836773578914e-05,
      "loss": 0.4886,
      "step": 228
    },
    {
      "epoch": 0.43556823585354254,
      "grad_norm": 0.3928169310092926,
      "learning_rate": 9.857732613528104e-05,
      "loss": 0.3741,
      "step": 229
    },
    {
      "epoch": 0.437470280551593,
      "grad_norm": 0.4982980191707611,
      "learning_rate": 9.85709749126707e-05,
      "loss": 0.5704,
      "step": 230
    },
    {
      "epoch": 0.43937232524964337,
      "grad_norm": 0.356545090675354,
      "learning_rate": 9.856462369006035e-05,
      "loss": 0.3618,
      "step": 231
    },
    {
      "epoch": 0.44127436994769376,
      "grad_norm": 0.5087487697601318,
      "learning_rate": 9.855827246744998e-05,
      "loss": 0.4733,
      "step": 232
    },
    {
      "epoch": 0.4431764146457442,
      "grad_norm": 0.3566097021102905,
      "learning_rate": 9.855192124483964e-05,
      "loss": 0.3771,
      "step": 233
    },
    {
      "epoch": 0.4450784593437946,
      "grad_norm": 0.3210541605949402,
      "learning_rate": 9.854557002222929e-05,
      "loss": 0.4341,
      "step": 234
    },
    {
      "epoch": 0.446980504041845,
      "grad_norm": 0.25422924757003784,
      "learning_rate": 9.853921879961893e-05,
      "loss": 0.3987,
      "step": 235
    },
    {
      "epoch": 0.4488825487398954,
      "grad_norm": 0.39164894819259644,
      "learning_rate": 9.853286757700858e-05,
      "loss": 0.4149,
      "step": 236
    },
    {
      "epoch": 0.4507845934379458,
      "grad_norm": 0.37471455335617065,
      "learning_rate": 9.852651635439823e-05,
      "loss": 0.4471,
      "step": 237
    },
    {
      "epoch": 0.4526866381359962,
      "grad_norm": 0.37678262591362,
      "learning_rate": 9.852016513178788e-05,
      "loss": 0.3943,
      "step": 238
    },
    {
      "epoch": 0.4545886828340466,
      "grad_norm": 0.4653976857662201,
      "learning_rate": 9.851381390917752e-05,
      "loss": 0.4848,
      "step": 239
    },
    {
      "epoch": 0.456490727532097,
      "grad_norm": 0.46764564514160156,
      "learning_rate": 9.850746268656717e-05,
      "loss": 0.4624,
      "step": 240
    },
    {
      "epoch": 0.4583927722301474,
      "grad_norm": 0.3803463876247406,
      "learning_rate": 9.850111146395682e-05,
      "loss": 0.442,
      "step": 241
    },
    {
      "epoch": 0.4602948169281978,
      "grad_norm": 0.33662229776382446,
      "learning_rate": 9.849476024134646e-05,
      "loss": 0.4564,
      "step": 242
    },
    {
      "epoch": 0.46219686162624823,
      "grad_norm": 0.42181041836738586,
      "learning_rate": 9.848840901873611e-05,
      "loss": 0.4702,
      "step": 243
    },
    {
      "epoch": 0.4640989063242986,
      "grad_norm": 0.40373390913009644,
      "learning_rate": 9.848205779612576e-05,
      "loss": 0.3745,
      "step": 244
    },
    {
      "epoch": 0.466000951022349,
      "grad_norm": 0.36634379625320435,
      "learning_rate": 9.84757065735154e-05,
      "loss": 0.428,
      "step": 245
    },
    {
      "epoch": 0.46790299572039945,
      "grad_norm": 0.35369235277175903,
      "learning_rate": 9.846935535090506e-05,
      "loss": 0.3986,
      "step": 246
    },
    {
      "epoch": 0.46980504041844984,
      "grad_norm": 0.4154004454612732,
      "learning_rate": 9.846300412829471e-05,
      "loss": 0.3512,
      "step": 247
    },
    {
      "epoch": 0.4717070851165002,
      "grad_norm": 0.3689868450164795,
      "learning_rate": 9.845665290568435e-05,
      "loss": 0.3708,
      "step": 248
    },
    {
      "epoch": 0.47360912981455067,
      "grad_norm": 0.38414841890335083,
      "learning_rate": 9.8450301683074e-05,
      "loss": 0.3401,
      "step": 249
    },
    {
      "epoch": 0.47551117451260105,
      "grad_norm": 0.39936143159866333,
      "learning_rate": 9.844395046046364e-05,
      "loss": 0.4328,
      "step": 250
    },
    {
      "epoch": 0.47741321921065144,
      "grad_norm": 0.30578187108039856,
      "learning_rate": 9.84375992378533e-05,
      "loss": 0.3694,
      "step": 251
    },
    {
      "epoch": 0.4793152639087018,
      "grad_norm": 0.39497658610343933,
      "learning_rate": 9.843124801524294e-05,
      "loss": 0.3945,
      "step": 252
    },
    {
      "epoch": 0.48121730860675227,
      "grad_norm": 0.44466689229011536,
      "learning_rate": 9.842489679263258e-05,
      "loss": 0.4485,
      "step": 253
    },
    {
      "epoch": 0.48311935330480266,
      "grad_norm": 0.3614617586135864,
      "learning_rate": 9.841854557002223e-05,
      "loss": 0.3701,
      "step": 254
    },
    {
      "epoch": 0.48502139800285304,
      "grad_norm": 0.3102608621120453,
      "learning_rate": 9.841219434741188e-05,
      "loss": 0.3677,
      "step": 255
    },
    {
      "epoch": 0.4869234427009035,
      "grad_norm": 0.36049678921699524,
      "learning_rate": 9.840584312480153e-05,
      "loss": 0.411,
      "step": 256
    },
    {
      "epoch": 0.4888254873989539,
      "grad_norm": 0.4025668501853943,
      "learning_rate": 9.839949190219117e-05,
      "loss": 0.433,
      "step": 257
    },
    {
      "epoch": 0.49072753209700426,
      "grad_norm": 0.4131562113761902,
      "learning_rate": 9.839314067958082e-05,
      "loss": 0.4818,
      "step": 258
    },
    {
      "epoch": 0.4926295767950547,
      "grad_norm": 0.481468141078949,
      "learning_rate": 9.838678945697047e-05,
      "loss": 0.5226,
      "step": 259
    },
    {
      "epoch": 0.4945316214931051,
      "grad_norm": 0.2845190167427063,
      "learning_rate": 9.838043823436011e-05,
      "loss": 0.3323,
      "step": 260
    },
    {
      "epoch": 0.4964336661911555,
      "grad_norm": 0.40381497144699097,
      "learning_rate": 9.837408701174976e-05,
      "loss": 0.4025,
      "step": 261
    },
    {
      "epoch": 0.4983357108892059,
      "grad_norm": 0.4109043478965759,
      "learning_rate": 9.836773578913942e-05,
      "loss": 0.4429,
      "step": 262
    },
    {
      "epoch": 0.5002377555872562,
      "grad_norm": 0.4256783425807953,
      "learning_rate": 9.836138456652906e-05,
      "loss": 0.3994,
      "step": 263
    },
    {
      "epoch": 0.5021398002853067,
      "grad_norm": 0.35044407844543457,
      "learning_rate": 9.835503334391871e-05,
      "loss": 0.4431,
      "step": 264
    },
    {
      "epoch": 0.5040418449833571,
      "grad_norm": 0.4456939697265625,
      "learning_rate": 9.834868212130836e-05,
      "loss": 0.5424,
      "step": 265
    },
    {
      "epoch": 0.5059438896814075,
      "grad_norm": 0.36340197920799255,
      "learning_rate": 9.8342330898698e-05,
      "loss": 0.4199,
      "step": 266
    },
    {
      "epoch": 0.5078459343794579,
      "grad_norm": 0.4018803536891937,
      "learning_rate": 9.833597967608765e-05,
      "loss": 0.4132,
      "step": 267
    },
    {
      "epoch": 0.5097479790775084,
      "grad_norm": 0.3372616469860077,
      "learning_rate": 9.83296284534773e-05,
      "loss": 0.3239,
      "step": 268
    },
    {
      "epoch": 0.5116500237755587,
      "grad_norm": 0.4497722387313843,
      "learning_rate": 9.832327723086695e-05,
      "loss": 0.4019,
      "step": 269
    },
    {
      "epoch": 0.5135520684736091,
      "grad_norm": 0.422269344329834,
      "learning_rate": 9.831692600825659e-05,
      "loss": 0.45,
      "step": 270
    },
    {
      "epoch": 0.5154541131716596,
      "grad_norm": 0.4167305529117584,
      "learning_rate": 9.831057478564624e-05,
      "loss": 0.4172,
      "step": 271
    },
    {
      "epoch": 0.5173561578697099,
      "grad_norm": 0.4340919554233551,
      "learning_rate": 9.83042235630359e-05,
      "loss": 0.5042,
      "step": 272
    },
    {
      "epoch": 0.5192582025677603,
      "grad_norm": 0.4179072380065918,
      "learning_rate": 9.829787234042553e-05,
      "loss": 0.3499,
      "step": 273
    },
    {
      "epoch": 0.5211602472658108,
      "grad_norm": 0.39216554164886475,
      "learning_rate": 9.829152111781518e-05,
      "loss": 0.4729,
      "step": 274
    },
    {
      "epoch": 0.5230622919638611,
      "grad_norm": 0.4485825002193451,
      "learning_rate": 9.828516989520484e-05,
      "loss": 0.4449,
      "step": 275
    },
    {
      "epoch": 0.5249643366619116,
      "grad_norm": 0.3843270242214203,
      "learning_rate": 9.827881867259447e-05,
      "loss": 0.5416,
      "step": 276
    },
    {
      "epoch": 0.526866381359962,
      "grad_norm": 0.30829140543937683,
      "learning_rate": 9.827246744998413e-05,
      "loss": 0.4004,
      "step": 277
    },
    {
      "epoch": 0.5287684260580123,
      "grad_norm": 0.2905525863170624,
      "learning_rate": 9.826611622737378e-05,
      "loss": 0.3574,
      "step": 278
    },
    {
      "epoch": 0.5306704707560628,
      "grad_norm": 0.3848637342453003,
      "learning_rate": 9.825976500476343e-05,
      "loss": 0.4021,
      "step": 279
    },
    {
      "epoch": 0.5325725154541132,
      "grad_norm": 0.32691988348960876,
      "learning_rate": 9.825341378215307e-05,
      "loss": 0.4317,
      "step": 280
    },
    {
      "epoch": 0.5344745601521635,
      "grad_norm": 0.3506065011024475,
      "learning_rate": 9.824706255954271e-05,
      "loss": 0.329,
      "step": 281
    },
    {
      "epoch": 0.536376604850214,
      "grad_norm": 0.3102387487888336,
      "learning_rate": 9.824071133693237e-05,
      "loss": 0.3695,
      "step": 282
    },
    {
      "epoch": 0.5382786495482644,
      "grad_norm": 0.45750680565834045,
      "learning_rate": 9.823436011432201e-05,
      "loss": 0.4232,
      "step": 283
    },
    {
      "epoch": 0.5401806942463148,
      "grad_norm": 0.297134131193161,
      "learning_rate": 9.822800889171165e-05,
      "loss": 0.4137,
      "step": 284
    },
    {
      "epoch": 0.5420827389443652,
      "grad_norm": 0.3696708679199219,
      "learning_rate": 9.822165766910131e-05,
      "loss": 0.4598,
      "step": 285
    },
    {
      "epoch": 0.5439847836424156,
      "grad_norm": 0.31236112117767334,
      "learning_rate": 9.821530644649095e-05,
      "loss": 0.314,
      "step": 286
    },
    {
      "epoch": 0.545886828340466,
      "grad_norm": 0.3596087694168091,
      "learning_rate": 9.82089552238806e-05,
      "loss": 0.4164,
      "step": 287
    },
    {
      "epoch": 0.5477888730385164,
      "grad_norm": 0.33347079157829285,
      "learning_rate": 9.820260400127024e-05,
      "loss": 0.3915,
      "step": 288
    },
    {
      "epoch": 0.5496909177365669,
      "grad_norm": 0.37818920612335205,
      "learning_rate": 9.81962527786599e-05,
      "loss": 0.3994,
      "step": 289
    },
    {
      "epoch": 0.5515929624346172,
      "grad_norm": 0.3968106806278229,
      "learning_rate": 9.818990155604955e-05,
      "loss": 0.3611,
      "step": 290
    },
    {
      "epoch": 0.5534950071326676,
      "grad_norm": 0.34991270303726196,
      "learning_rate": 9.818355033343918e-05,
      "loss": 0.3703,
      "step": 291
    },
    {
      "epoch": 0.555397051830718,
      "grad_norm": 0.4046263098716736,
      "learning_rate": 9.817719911082885e-05,
      "loss": 0.3302,
      "step": 292
    },
    {
      "epoch": 0.5572990965287684,
      "grad_norm": 0.35804587602615356,
      "learning_rate": 9.817084788821849e-05,
      "loss": 0.373,
      "step": 293
    },
    {
      "epoch": 0.5592011412268189,
      "grad_norm": 0.3538301885128021,
      "learning_rate": 9.816449666560813e-05,
      "loss": 0.3482,
      "step": 294
    },
    {
      "epoch": 0.5611031859248692,
      "grad_norm": 0.36835455894470215,
      "learning_rate": 9.815814544299778e-05,
      "loss": 0.3393,
      "step": 295
    },
    {
      "epoch": 0.5630052306229196,
      "grad_norm": 0.48919835686683655,
      "learning_rate": 9.815179422038743e-05,
      "loss": 0.4213,
      "step": 296
    },
    {
      "epoch": 0.5649072753209701,
      "grad_norm": 0.3472330570220947,
      "learning_rate": 9.814544299777708e-05,
      "loss": 0.3996,
      "step": 297
    },
    {
      "epoch": 0.5668093200190204,
      "grad_norm": 0.428611159324646,
      "learning_rate": 9.813909177516672e-05,
      "loss": 0.4524,
      "step": 298
    },
    {
      "epoch": 0.5687113647170708,
      "grad_norm": 0.4176979959011078,
      "learning_rate": 9.813274055255637e-05,
      "loss": 0.3787,
      "step": 299
    },
    {
      "epoch": 0.5706134094151213,
      "grad_norm": 0.41548797488212585,
      "learning_rate": 9.812638932994602e-05,
      "loss": 0.4758,
      "step": 300
    },
    {
      "epoch": 0.5725154541131716,
      "grad_norm": 0.3926902413368225,
      "learning_rate": 9.812003810733566e-05,
      "loss": 0.434,
      "step": 301
    },
    {
      "epoch": 0.5744174988112221,
      "grad_norm": 0.392846018075943,
      "learning_rate": 9.811368688472531e-05,
      "loss": 0.3928,
      "step": 302
    },
    {
      "epoch": 0.5763195435092725,
      "grad_norm": 0.36347585916519165,
      "learning_rate": 9.810733566211497e-05,
      "loss": 0.4264,
      "step": 303
    },
    {
      "epoch": 0.5782215882073228,
      "grad_norm": 0.4314410090446472,
      "learning_rate": 9.81009844395046e-05,
      "loss": 0.4199,
      "step": 304
    },
    {
      "epoch": 0.5801236329053733,
      "grad_norm": 0.337494820356369,
      "learning_rate": 9.809463321689426e-05,
      "loss": 0.4181,
      "step": 305
    },
    {
      "epoch": 0.5820256776034237,
      "grad_norm": 0.27786335349082947,
      "learning_rate": 9.808828199428391e-05,
      "loss": 0.3,
      "step": 306
    },
    {
      "epoch": 0.583927722301474,
      "grad_norm": 0.37235599756240845,
      "learning_rate": 9.808193077167355e-05,
      "loss": 0.3927,
      "step": 307
    },
    {
      "epoch": 0.5858297669995245,
      "grad_norm": 0.37353670597076416,
      "learning_rate": 9.80755795490632e-05,
      "loss": 0.4146,
      "step": 308
    },
    {
      "epoch": 0.5877318116975749,
      "grad_norm": 0.3919946551322937,
      "learning_rate": 9.806922832645285e-05,
      "loss": 0.5055,
      "step": 309
    },
    {
      "epoch": 0.5896338563956253,
      "grad_norm": 0.45411062240600586,
      "learning_rate": 9.80628771038425e-05,
      "loss": 0.5347,
      "step": 310
    },
    {
      "epoch": 0.5915359010936757,
      "grad_norm": 0.4087005853652954,
      "learning_rate": 9.805652588123214e-05,
      "loss": 0.3732,
      "step": 311
    },
    {
      "epoch": 0.5934379457917262,
      "grad_norm": 0.313297837972641,
      "learning_rate": 9.805017465862178e-05,
      "loss": 0.3093,
      "step": 312
    },
    {
      "epoch": 0.5953399904897765,
      "grad_norm": 0.40149226784706116,
      "learning_rate": 9.804382343601144e-05,
      "loss": 0.4404,
      "step": 313
    },
    {
      "epoch": 0.5972420351878269,
      "grad_norm": 0.34245574474334717,
      "learning_rate": 9.803747221340108e-05,
      "loss": 0.4036,
      "step": 314
    },
    {
      "epoch": 0.5991440798858774,
      "grad_norm": 0.38059449195861816,
      "learning_rate": 9.803112099079073e-05,
      "loss": 0.3763,
      "step": 315
    },
    {
      "epoch": 0.6010461245839277,
      "grad_norm": 0.4539381265640259,
      "learning_rate": 9.802476976818039e-05,
      "loss": 0.4551,
      "step": 316
    },
    {
      "epoch": 0.6029481692819781,
      "grad_norm": 0.4077235460281372,
      "learning_rate": 9.801841854557002e-05,
      "loss": 0.4641,
      "step": 317
    },
    {
      "epoch": 0.6048502139800286,
      "grad_norm": 0.3426643908023834,
      "learning_rate": 9.801206732295968e-05,
      "loss": 0.3684,
      "step": 318
    },
    {
      "epoch": 0.6067522586780789,
      "grad_norm": 0.3042270839214325,
      "learning_rate": 9.800571610034931e-05,
      "loss": 0.373,
      "step": 319
    },
    {
      "epoch": 0.6086543033761294,
      "grad_norm": 0.4373973309993744,
      "learning_rate": 9.799936487773897e-05,
      "loss": 0.5442,
      "step": 320
    },
    {
      "epoch": 0.6105563480741797,
      "grad_norm": 0.385797917842865,
      "learning_rate": 9.799301365512862e-05,
      "loss": 0.4218,
      "step": 321
    },
    {
      "epoch": 0.6124583927722301,
      "grad_norm": 0.33210891485214233,
      "learning_rate": 9.798666243251826e-05,
      "loss": 0.3062,
      "step": 322
    },
    {
      "epoch": 0.6143604374702806,
      "grad_norm": 0.3997063636779785,
      "learning_rate": 9.798031120990792e-05,
      "loss": 0.4104,
      "step": 323
    },
    {
      "epoch": 0.6162624821683309,
      "grad_norm": 0.4837460219860077,
      "learning_rate": 9.797395998729756e-05,
      "loss": 0.5271,
      "step": 324
    },
    {
      "epoch": 0.6181645268663813,
      "grad_norm": 0.36420971155166626,
      "learning_rate": 9.79676087646872e-05,
      "loss": 0.4033,
      "step": 325
    },
    {
      "epoch": 0.6200665715644318,
      "grad_norm": 0.33610865473747253,
      "learning_rate": 9.796125754207685e-05,
      "loss": 0.3992,
      "step": 326
    },
    {
      "epoch": 0.6219686162624821,
      "grad_norm": 0.28999099135398865,
      "learning_rate": 9.79549063194665e-05,
      "loss": 0.3675,
      "step": 327
    },
    {
      "epoch": 0.6238706609605326,
      "grad_norm": 0.359401673078537,
      "learning_rate": 9.794855509685615e-05,
      "loss": 0.4363,
      "step": 328
    },
    {
      "epoch": 0.625772705658583,
      "grad_norm": 0.3948569595813751,
      "learning_rate": 9.794220387424579e-05,
      "loss": 0.3698,
      "step": 329
    },
    {
      "epoch": 0.6276747503566333,
      "grad_norm": 0.3753513991832733,
      "learning_rate": 9.793585265163544e-05,
      "loss": 0.4397,
      "step": 330
    },
    {
      "epoch": 0.6295767950546838,
      "grad_norm": 0.32612451910972595,
      "learning_rate": 9.79295014290251e-05,
      "loss": 0.3846,
      "step": 331
    },
    {
      "epoch": 0.6314788397527342,
      "grad_norm": 0.40796539187431335,
      "learning_rate": 9.792315020641473e-05,
      "loss": 0.371,
      "step": 332
    },
    {
      "epoch": 0.6333808844507846,
      "grad_norm": 0.4358294904232025,
      "learning_rate": 9.791679898380439e-05,
      "loss": 0.4052,
      "step": 333
    },
    {
      "epoch": 0.635282929148835,
      "grad_norm": 0.39615437388420105,
      "learning_rate": 9.791044776119404e-05,
      "loss": 0.3686,
      "step": 334
    },
    {
      "epoch": 0.6371849738468854,
      "grad_norm": 0.32977715134620667,
      "learning_rate": 9.790409653858368e-05,
      "loss": 0.4404,
      "step": 335
    },
    {
      "epoch": 0.6390870185449358,
      "grad_norm": 0.38361093401908875,
      "learning_rate": 9.789774531597333e-05,
      "loss": 0.3709,
      "step": 336
    },
    {
      "epoch": 0.6409890632429862,
      "grad_norm": 0.40280988812446594,
      "learning_rate": 9.789139409336298e-05,
      "loss": 0.3322,
      "step": 337
    },
    {
      "epoch": 0.6428911079410367,
      "grad_norm": 0.3682766854763031,
      "learning_rate": 9.788504287075262e-05,
      "loss": 0.4144,
      "step": 338
    },
    {
      "epoch": 0.644793152639087,
      "grad_norm": 0.39864271879196167,
      "learning_rate": 9.787869164814227e-05,
      "loss": 0.4404,
      "step": 339
    },
    {
      "epoch": 0.6466951973371374,
      "grad_norm": 0.3244321048259735,
      "learning_rate": 9.787234042553192e-05,
      "loss": 0.3541,
      "step": 340
    },
    {
      "epoch": 0.6485972420351879,
      "grad_norm": 0.323403924703598,
      "learning_rate": 9.786598920292157e-05,
      "loss": 0.3374,
      "step": 341
    },
    {
      "epoch": 0.6504992867332382,
      "grad_norm": 0.3881044387817383,
      "learning_rate": 9.785963798031121e-05,
      "loss": 0.4415,
      "step": 342
    },
    {
      "epoch": 0.6524013314312886,
      "grad_norm": 0.35189467668533325,
      "learning_rate": 9.785328675770086e-05,
      "loss": 0.401,
      "step": 343
    },
    {
      "epoch": 0.6543033761293391,
      "grad_norm": 0.3553767800331116,
      "learning_rate": 9.784693553509052e-05,
      "loss": 0.456,
      "step": 344
    },
    {
      "epoch": 0.6562054208273894,
      "grad_norm": 0.3302605152130127,
      "learning_rate": 9.784058431248015e-05,
      "loss": 0.472,
      "step": 345
    },
    {
      "epoch": 0.6581074655254399,
      "grad_norm": 0.4526873826980591,
      "learning_rate": 9.78342330898698e-05,
      "loss": 0.3908,
      "step": 346
    },
    {
      "epoch": 0.6600095102234903,
      "grad_norm": 0.3232348561286926,
      "learning_rate": 9.782788186725946e-05,
      "loss": 0.3421,
      "step": 347
    },
    {
      "epoch": 0.6619115549215406,
      "grad_norm": 0.38508203625679016,
      "learning_rate": 9.78215306446491e-05,
      "loss": 0.4093,
      "step": 348
    },
    {
      "epoch": 0.6638135996195911,
      "grad_norm": 0.3187748193740845,
      "learning_rate": 9.781517942203875e-05,
      "loss": 0.4319,
      "step": 349
    },
    {
      "epoch": 0.6657156443176415,
      "grad_norm": 0.2614807188510895,
      "learning_rate": 9.78088281994284e-05,
      "loss": 0.314,
      "step": 350
    },
    {
      "epoch": 0.6676176890156919,
      "grad_norm": 0.40218180418014526,
      "learning_rate": 9.780247697681805e-05,
      "loss": 0.4404,
      "step": 351
    },
    {
      "epoch": 0.6695197337137423,
      "grad_norm": 0.4016517996788025,
      "learning_rate": 9.779612575420769e-05,
      "loss": 0.5063,
      "step": 352
    },
    {
      "epoch": 0.6714217784117926,
      "grad_norm": 0.3333278000354767,
      "learning_rate": 9.778977453159733e-05,
      "loss": 0.2966,
      "step": 353
    },
    {
      "epoch": 0.6733238231098431,
      "grad_norm": 0.4535547196865082,
      "learning_rate": 9.778342330898699e-05,
      "loss": 0.4077,
      "step": 354
    },
    {
      "epoch": 0.6752258678078935,
      "grad_norm": 0.4180653393268585,
      "learning_rate": 9.777707208637663e-05,
      "loss": 0.4554,
      "step": 355
    },
    {
      "epoch": 0.6771279125059438,
      "grad_norm": 0.43454670906066895,
      "learning_rate": 9.777072086376627e-05,
      "loss": 0.4403,
      "step": 356
    },
    {
      "epoch": 0.6790299572039943,
      "grad_norm": 0.45290321111679077,
      "learning_rate": 9.776436964115594e-05,
      "loss": 0.4037,
      "step": 357
    },
    {
      "epoch": 0.6809320019020447,
      "grad_norm": 0.34165212512016296,
      "learning_rate": 9.775801841854557e-05,
      "loss": 0.3044,
      "step": 358
    },
    {
      "epoch": 0.6828340466000951,
      "grad_norm": 0.435138463973999,
      "learning_rate": 9.775166719593523e-05,
      "loss": 0.4293,
      "step": 359
    },
    {
      "epoch": 0.6847360912981455,
      "grad_norm": 0.36061882972717285,
      "learning_rate": 9.774531597332486e-05,
      "loss": 0.4052,
      "step": 360
    },
    {
      "epoch": 0.6866381359961959,
      "grad_norm": 0.4023354947566986,
      "learning_rate": 9.773896475071452e-05,
      "loss": 0.4232,
      "step": 361
    },
    {
      "epoch": 0.6885401806942463,
      "grad_norm": 0.39200109243392944,
      "learning_rate": 9.773261352810417e-05,
      "loss": 0.3882,
      "step": 362
    },
    {
      "epoch": 0.6904422253922967,
      "grad_norm": 0.34504035115242004,
      "learning_rate": 9.77262623054938e-05,
      "loss": 0.4063,
      "step": 363
    },
    {
      "epoch": 0.6923442700903472,
      "grad_norm": 0.31081900000572205,
      "learning_rate": 9.771991108288346e-05,
      "loss": 0.251,
      "step": 364
    },
    {
      "epoch": 0.6942463147883975,
      "grad_norm": 0.3800300061702728,
      "learning_rate": 9.771355986027311e-05,
      "loss": 0.3722,
      "step": 365
    },
    {
      "epoch": 0.6961483594864479,
      "grad_norm": 0.3476494550704956,
      "learning_rate": 9.770720863766275e-05,
      "loss": 0.382,
      "step": 366
    },
    {
      "epoch": 0.6980504041844984,
      "grad_norm": 0.38069918751716614,
      "learning_rate": 9.77008574150524e-05,
      "loss": 0.4329,
      "step": 367
    },
    {
      "epoch": 0.6999524488825487,
      "grad_norm": 0.4034759998321533,
      "learning_rate": 9.769450619244205e-05,
      "loss": 0.4112,
      "step": 368
    },
    {
      "epoch": 0.7018544935805991,
      "grad_norm": 0.4232093393802643,
      "learning_rate": 9.76881549698317e-05,
      "loss": 0.4524,
      "step": 369
    },
    {
      "epoch": 0.7037565382786496,
      "grad_norm": 0.40627321600914,
      "learning_rate": 9.768180374722134e-05,
      "loss": 0.388,
      "step": 370
    },
    {
      "epoch": 0.7056585829766999,
      "grad_norm": 0.41021519899368286,
      "learning_rate": 9.767545252461099e-05,
      "loss": 0.3741,
      "step": 371
    },
    {
      "epoch": 0.7075606276747504,
      "grad_norm": 0.3615809679031372,
      "learning_rate": 9.766910130200065e-05,
      "loss": 0.4432,
      "step": 372
    },
    {
      "epoch": 0.7094626723728008,
      "grad_norm": 0.3088645935058594,
      "learning_rate": 9.766275007939028e-05,
      "loss": 0.3343,
      "step": 373
    },
    {
      "epoch": 0.7113647170708511,
      "grad_norm": 0.380659818649292,
      "learning_rate": 9.765639885677994e-05,
      "loss": 0.4092,
      "step": 374
    },
    {
      "epoch": 0.7132667617689016,
      "grad_norm": 0.28462380170822144,
      "learning_rate": 9.765004763416959e-05,
      "loss": 0.31,
      "step": 375
    },
    {
      "epoch": 0.715168806466952,
      "grad_norm": 0.3215513229370117,
      "learning_rate": 9.764369641155923e-05,
      "loss": 0.4115,
      "step": 376
    },
    {
      "epoch": 0.7170708511650024,
      "grad_norm": 0.397651731967926,
      "learning_rate": 9.763734518894888e-05,
      "loss": 0.4369,
      "step": 377
    },
    {
      "epoch": 0.7189728958630528,
      "grad_norm": 0.31436121463775635,
      "learning_rate": 9.763099396633853e-05,
      "loss": 0.4339,
      "step": 378
    },
    {
      "epoch": 0.7208749405611032,
      "grad_norm": 0.4024806320667267,
      "learning_rate": 9.762464274372817e-05,
      "loss": 0.4252,
      "step": 379
    },
    {
      "epoch": 0.7227769852591536,
      "grad_norm": 0.37994107604026794,
      "learning_rate": 9.761829152111782e-05,
      "loss": 0.3483,
      "step": 380
    },
    {
      "epoch": 0.724679029957204,
      "grad_norm": 0.44616061449050903,
      "learning_rate": 9.761194029850747e-05,
      "loss": 0.3809,
      "step": 381
    },
    {
      "epoch": 0.7265810746552543,
      "grad_norm": 0.3396744728088379,
      "learning_rate": 9.760558907589712e-05,
      "loss": 0.3382,
      "step": 382
    },
    {
      "epoch": 0.7284831193533048,
      "grad_norm": 0.334839791059494,
      "learning_rate": 9.759923785328676e-05,
      "loss": 0.3465,
      "step": 383
    },
    {
      "epoch": 0.7303851640513552,
      "grad_norm": 0.417478084564209,
      "learning_rate": 9.75928866306764e-05,
      "loss": 0.3191,
      "step": 384
    },
    {
      "epoch": 0.7322872087494056,
      "grad_norm": 0.30790823698043823,
      "learning_rate": 9.758653540806606e-05,
      "loss": 0.3139,
      "step": 385
    },
    {
      "epoch": 0.734189253447456,
      "grad_norm": 0.4008057415485382,
      "learning_rate": 9.75801841854557e-05,
      "loss": 0.419,
      "step": 386
    },
    {
      "epoch": 0.7360912981455064,
      "grad_norm": 0.42966723442077637,
      "learning_rate": 9.757383296284535e-05,
      "loss": 0.3634,
      "step": 387
    },
    {
      "epoch": 0.7379933428435568,
      "grad_norm": 0.33789002895355225,
      "learning_rate": 9.7567481740235e-05,
      "loss": 0.3966,
      "step": 388
    },
    {
      "epoch": 0.7398953875416072,
      "grad_norm": 0.35244229435920715,
      "learning_rate": 9.756113051762464e-05,
      "loss": 0.3991,
      "step": 389
    },
    {
      "epoch": 0.7417974322396577,
      "grad_norm": 0.3581864833831787,
      "learning_rate": 9.75547792950143e-05,
      "loss": 0.347,
      "step": 390
    },
    {
      "epoch": 0.743699476937708,
      "grad_norm": 0.30788975954055786,
      "learning_rate": 9.754842807240394e-05,
      "loss": 0.3485,
      "step": 391
    },
    {
      "epoch": 0.7456015216357584,
      "grad_norm": 0.5155593156814575,
      "learning_rate": 9.754207684979359e-05,
      "loss": 0.4793,
      "step": 392
    },
    {
      "epoch": 0.7475035663338089,
      "grad_norm": 0.4183029532432556,
      "learning_rate": 9.753572562718324e-05,
      "loss": 0.4064,
      "step": 393
    },
    {
      "epoch": 0.7494056110318592,
      "grad_norm": 0.36132046580314636,
      "learning_rate": 9.752937440457288e-05,
      "loss": 0.3539,
      "step": 394
    },
    {
      "epoch": 0.7513076557299097,
      "grad_norm": 0.4269217252731323,
      "learning_rate": 9.752302318196254e-05,
      "loss": 0.4358,
      "step": 395
    },
    {
      "epoch": 0.7532097004279601,
      "grad_norm": 0.38872459530830383,
      "learning_rate": 9.751667195935218e-05,
      "loss": 0.3238,
      "step": 396
    },
    {
      "epoch": 0.7551117451260104,
      "grad_norm": 0.4668743312358856,
      "learning_rate": 9.751032073674182e-05,
      "loss": 0.4218,
      "step": 397
    },
    {
      "epoch": 0.7570137898240609,
      "grad_norm": 0.3817143738269806,
      "learning_rate": 9.750396951413147e-05,
      "loss": 0.4332,
      "step": 398
    },
    {
      "epoch": 0.7589158345221113,
      "grad_norm": 0.4089401960372925,
      "learning_rate": 9.749761829152112e-05,
      "loss": 0.319,
      "step": 399
    },
    {
      "epoch": 0.7608178792201616,
      "grad_norm": 0.36516866087913513,
      "learning_rate": 9.749126706891077e-05,
      "loss": 0.3858,
      "step": 400
    },
    {
      "epoch": 0.7627199239182121,
      "grad_norm": 0.3843027949333191,
      "learning_rate": 9.748491584630041e-05,
      "loss": 0.4682,
      "step": 401
    },
    {
      "epoch": 0.7646219686162625,
      "grad_norm": 0.36987295746803284,
      "learning_rate": 9.747856462369006e-05,
      "loss": 0.3328,
      "step": 402
    },
    {
      "epoch": 0.7665240133143129,
      "grad_norm": 0.4972301721572876,
      "learning_rate": 9.747221340107972e-05,
      "loss": 0.3939,
      "step": 403
    },
    {
      "epoch": 0.7684260580123633,
      "grad_norm": 0.4319972097873688,
      "learning_rate": 9.746586217846935e-05,
      "loss": 0.3918,
      "step": 404
    },
    {
      "epoch": 0.7703281027104137,
      "grad_norm": 0.364364892244339,
      "learning_rate": 9.7459510955859e-05,
      "loss": 0.3871,
      "step": 405
    },
    {
      "epoch": 0.7722301474084641,
      "grad_norm": 0.43767908215522766,
      "learning_rate": 9.745315973324866e-05,
      "loss": 0.3973,
      "step": 406
    },
    {
      "epoch": 0.7741321921065145,
      "grad_norm": 0.44734928011894226,
      "learning_rate": 9.74468085106383e-05,
      "loss": 0.3884,
      "step": 407
    },
    {
      "epoch": 0.776034236804565,
      "grad_norm": 0.3817954957485199,
      "learning_rate": 9.744045728802795e-05,
      "loss": 0.3647,
      "step": 408
    },
    {
      "epoch": 0.7779362815026153,
      "grad_norm": 0.3619462251663208,
      "learning_rate": 9.74341060654176e-05,
      "loss": 0.4994,
      "step": 409
    },
    {
      "epoch": 0.7798383262006657,
      "grad_norm": 0.38225993514060974,
      "learning_rate": 9.742775484280724e-05,
      "loss": 0.4116,
      "step": 410
    },
    {
      "epoch": 0.7817403708987162,
      "grad_norm": 0.39784252643585205,
      "learning_rate": 9.742140362019689e-05,
      "loss": 0.3729,
      "step": 411
    },
    {
      "epoch": 0.7836424155967665,
      "grad_norm": 0.3188072443008423,
      "learning_rate": 9.741505239758654e-05,
      "loss": 0.3767,
      "step": 412
    },
    {
      "epoch": 0.785544460294817,
      "grad_norm": 0.4509223401546478,
      "learning_rate": 9.74087011749762e-05,
      "loss": 0.4595,
      "step": 413
    },
    {
      "epoch": 0.7874465049928673,
      "grad_norm": 0.40249937772750854,
      "learning_rate": 9.740234995236583e-05,
      "loss": 0.3761,
      "step": 414
    },
    {
      "epoch": 0.7893485496909177,
      "grad_norm": 0.3387410044670105,
      "learning_rate": 9.739599872975547e-05,
      "loss": 0.401,
      "step": 415
    },
    {
      "epoch": 0.7912505943889682,
      "grad_norm": 0.47670629620552063,
      "learning_rate": 9.738964750714514e-05,
      "loss": 0.3656,
      "step": 416
    },
    {
      "epoch": 0.7931526390870185,
      "grad_norm": 0.37239211797714233,
      "learning_rate": 9.738329628453477e-05,
      "loss": 0.4885,
      "step": 417
    },
    {
      "epoch": 0.7950546837850689,
      "grad_norm": 0.3347351849079132,
      "learning_rate": 9.737694506192443e-05,
      "loss": 0.291,
      "step": 418
    },
    {
      "epoch": 0.7969567284831194,
      "grad_norm": 0.3727717399597168,
      "learning_rate": 9.737059383931408e-05,
      "loss": 0.3506,
      "step": 419
    },
    {
      "epoch": 0.7988587731811697,
      "grad_norm": 0.3866841793060303,
      "learning_rate": 9.736424261670372e-05,
      "loss": 0.4355,
      "step": 420
    },
    {
      "epoch": 0.8007608178792202,
      "grad_norm": 0.39670372009277344,
      "learning_rate": 9.735789139409337e-05,
      "loss": 0.4041,
      "step": 421
    },
    {
      "epoch": 0.8026628625772706,
      "grad_norm": 0.35946765542030334,
      "learning_rate": 9.7351540171483e-05,
      "loss": 0.3378,
      "step": 422
    },
    {
      "epoch": 0.8045649072753209,
      "grad_norm": 0.24180381000041962,
      "learning_rate": 9.734518894887267e-05,
      "loss": 0.3133,
      "step": 423
    },
    {
      "epoch": 0.8064669519733714,
      "grad_norm": 0.4238085150718689,
      "learning_rate": 9.733883772626231e-05,
      "loss": 0.3968,
      "step": 424
    },
    {
      "epoch": 0.8083689966714218,
      "grad_norm": 0.35451412200927734,
      "learning_rate": 9.733248650365195e-05,
      "loss": 0.3456,
      "step": 425
    },
    {
      "epoch": 0.8102710413694721,
      "grad_norm": 0.49277418851852417,
      "learning_rate": 9.732613528104161e-05,
      "loss": 0.3916,
      "step": 426
    },
    {
      "epoch": 0.8121730860675226,
      "grad_norm": 0.34536874294281006,
      "learning_rate": 9.731978405843125e-05,
      "loss": 0.537,
      "step": 427
    },
    {
      "epoch": 0.814075130765573,
      "grad_norm": 0.3002311885356903,
      "learning_rate": 9.731343283582089e-05,
      "loss": 0.3842,
      "step": 428
    },
    {
      "epoch": 0.8159771754636234,
      "grad_norm": 0.29766812920570374,
      "learning_rate": 9.730708161321054e-05,
      "loss": 0.2979,
      "step": 429
    },
    {
      "epoch": 0.8178792201616738,
      "grad_norm": 0.34347230195999146,
      "learning_rate": 9.73007303906002e-05,
      "loss": 0.3996,
      "step": 430
    },
    {
      "epoch": 0.8197812648597242,
      "grad_norm": 0.42430102825164795,
      "learning_rate": 9.729437916798985e-05,
      "loss": 0.4677,
      "step": 431
    },
    {
      "epoch": 0.8216833095577746,
      "grad_norm": 0.3375668227672577,
      "learning_rate": 9.728802794537948e-05,
      "loss": 0.4257,
      "step": 432
    },
    {
      "epoch": 0.823585354255825,
      "grad_norm": 0.3718586266040802,
      "learning_rate": 9.728167672276914e-05,
      "loss": 0.3555,
      "step": 433
    },
    {
      "epoch": 0.8254873989538755,
      "grad_norm": 0.4310496151447296,
      "learning_rate": 9.727532550015879e-05,
      "loss": 0.4026,
      "step": 434
    },
    {
      "epoch": 0.8273894436519258,
      "grad_norm": 0.43832001090049744,
      "learning_rate": 9.726897427754843e-05,
      "loss": 0.4421,
      "step": 435
    },
    {
      "epoch": 0.8292914883499762,
      "grad_norm": 0.42209911346435547,
      "learning_rate": 9.726262305493808e-05,
      "loss": 0.397,
      "step": 436
    },
    {
      "epoch": 0.8311935330480267,
      "grad_norm": 0.4297396242618561,
      "learning_rate": 9.725627183232773e-05,
      "loss": 0.4244,
      "step": 437
    },
    {
      "epoch": 0.833095577746077,
      "grad_norm": 0.40587079524993896,
      "learning_rate": 9.724992060971737e-05,
      "loss": 0.3753,
      "step": 438
    },
    {
      "epoch": 0.8349976224441275,
      "grad_norm": 0.4127040505409241,
      "learning_rate": 9.724356938710702e-05,
      "loss": 0.3926,
      "step": 439
    },
    {
      "epoch": 0.8368996671421779,
      "grad_norm": 0.3734678030014038,
      "learning_rate": 9.723721816449667e-05,
      "loss": 0.3338,
      "step": 440
    },
    {
      "epoch": 0.8388017118402282,
      "grad_norm": 0.38152286410331726,
      "learning_rate": 9.723086694188632e-05,
      "loss": 0.3893,
      "step": 441
    },
    {
      "epoch": 0.8407037565382787,
      "grad_norm": 0.4234791398048401,
      "learning_rate": 9.722451571927596e-05,
      "loss": 0.3104,
      "step": 442
    },
    {
      "epoch": 0.842605801236329,
      "grad_norm": 0.49204525351524353,
      "learning_rate": 9.721816449666561e-05,
      "loss": 0.3698,
      "step": 443
    },
    {
      "epoch": 0.8445078459343794,
      "grad_norm": 0.40980932116508484,
      "learning_rate": 9.721181327405527e-05,
      "loss": 0.3901,
      "step": 444
    },
    {
      "epoch": 0.8464098906324299,
      "grad_norm": 0.3330426514148712,
      "learning_rate": 9.72054620514449e-05,
      "loss": 0.3118,
      "step": 445
    },
    {
      "epoch": 0.8483119353304802,
      "grad_norm": 0.3042624890804291,
      "learning_rate": 9.719911082883456e-05,
      "loss": 0.3003,
      "step": 446
    },
    {
      "epoch": 0.8502139800285307,
      "grad_norm": 0.34576475620269775,
      "learning_rate": 9.719275960622421e-05,
      "loss": 0.3332,
      "step": 447
    },
    {
      "epoch": 0.8521160247265811,
      "grad_norm": 0.2980082035064697,
      "learning_rate": 9.718640838361385e-05,
      "loss": 0.3285,
      "step": 448
    },
    {
      "epoch": 0.8540180694246314,
      "grad_norm": 0.31439459323883057,
      "learning_rate": 9.71800571610035e-05,
      "loss": 0.3178,
      "step": 449
    },
    {
      "epoch": 0.8559201141226819,
      "grad_norm": 0.37447845935821533,
      "learning_rate": 9.717370593839315e-05,
      "loss": 0.3861,
      "step": 450
    },
    {
      "epoch": 0.8578221588207323,
      "grad_norm": 0.4261024594306946,
      "learning_rate": 9.716735471578279e-05,
      "loss": 0.4377,
      "step": 451
    },
    {
      "epoch": 0.8597242035187826,
      "grad_norm": 0.3328630328178406,
      "learning_rate": 9.716100349317244e-05,
      "loss": 0.2791,
      "step": 452
    },
    {
      "epoch": 0.8616262482168331,
      "grad_norm": 0.41943463683128357,
      "learning_rate": 9.715465227056209e-05,
      "loss": 0.4693,
      "step": 453
    },
    {
      "epoch": 0.8635282929148835,
      "grad_norm": 0.4295640289783478,
      "learning_rate": 9.714830104795174e-05,
      "loss": 0.4105,
      "step": 454
    },
    {
      "epoch": 0.8654303376129339,
      "grad_norm": 0.3548508882522583,
      "learning_rate": 9.714194982534138e-05,
      "loss": 0.3024,
      "step": 455
    },
    {
      "epoch": 0.8673323823109843,
      "grad_norm": 0.5577777624130249,
      "learning_rate": 9.713559860273102e-05,
      "loss": 0.3961,
      "step": 456
    },
    {
      "epoch": 0.8692344270090347,
      "grad_norm": 0.4119040071964264,
      "learning_rate": 9.712924738012069e-05,
      "loss": 0.3143,
      "step": 457
    },
    {
      "epoch": 0.8711364717070851,
      "grad_norm": 0.40272560715675354,
      "learning_rate": 9.712289615751032e-05,
      "loss": 0.3452,
      "step": 458
    },
    {
      "epoch": 0.8730385164051355,
      "grad_norm": 0.456386536359787,
      "learning_rate": 9.711654493489998e-05,
      "loss": 0.403,
      "step": 459
    },
    {
      "epoch": 0.874940561103186,
      "grad_norm": 0.3982544541358948,
      "learning_rate": 9.711019371228963e-05,
      "loss": 0.4498,
      "step": 460
    },
    {
      "epoch": 0.8768426058012363,
      "grad_norm": 0.29361623525619507,
      "learning_rate": 9.710384248967927e-05,
      "loss": 0.3724,
      "step": 461
    },
    {
      "epoch": 0.8787446504992867,
      "grad_norm": 0.3854773938655853,
      "learning_rate": 9.709749126706892e-05,
      "loss": 0.4162,
      "step": 462
    },
    {
      "epoch": 0.8806466951973372,
      "grad_norm": 0.3760225474834442,
      "learning_rate": 9.709114004445856e-05,
      "loss": 0.4335,
      "step": 463
    },
    {
      "epoch": 0.8825487398953875,
      "grad_norm": 0.4936290383338928,
      "learning_rate": 9.708478882184821e-05,
      "loss": 0.3522,
      "step": 464
    },
    {
      "epoch": 0.884450784593438,
      "grad_norm": 0.3584468364715576,
      "learning_rate": 9.707843759923786e-05,
      "loss": 0.552,
      "step": 465
    },
    {
      "epoch": 0.8863528292914884,
      "grad_norm": 0.3523949086666107,
      "learning_rate": 9.70720863766275e-05,
      "loss": 0.3498,
      "step": 466
    },
    {
      "epoch": 0.8882548739895387,
      "grad_norm": 0.42082804441452026,
      "learning_rate": 9.706573515401716e-05,
      "loss": 0.4863,
      "step": 467
    },
    {
      "epoch": 0.8901569186875892,
      "grad_norm": 0.4284763038158417,
      "learning_rate": 9.70593839314068e-05,
      "loss": 0.4737,
      "step": 468
    },
    {
      "epoch": 0.8920589633856396,
      "grad_norm": 0.3609261214733124,
      "learning_rate": 9.705303270879644e-05,
      "loss": 0.3208,
      "step": 469
    },
    {
      "epoch": 0.89396100808369,
      "grad_norm": 0.31832849979400635,
      "learning_rate": 9.704668148618609e-05,
      "loss": 0.2545,
      "step": 470
    },
    {
      "epoch": 0.8958630527817404,
      "grad_norm": 0.38202738761901855,
      "learning_rate": 9.704033026357574e-05,
      "loss": 0.3952,
      "step": 471
    },
    {
      "epoch": 0.8977650974797908,
      "grad_norm": 0.347649484872818,
      "learning_rate": 9.70339790409654e-05,
      "loss": 0.3776,
      "step": 472
    },
    {
      "epoch": 0.8996671421778412,
      "grad_norm": 0.41626760363578796,
      "learning_rate": 9.702762781835503e-05,
      "loss": 0.4152,
      "step": 473
    },
    {
      "epoch": 0.9015691868758916,
      "grad_norm": 0.4042579233646393,
      "learning_rate": 9.702127659574469e-05,
      "loss": 0.3813,
      "step": 474
    },
    {
      "epoch": 0.9034712315739419,
      "grad_norm": 0.38196825981140137,
      "learning_rate": 9.701492537313434e-05,
      "loss": 0.4398,
      "step": 475
    },
    {
      "epoch": 0.9053732762719924,
      "grad_norm": 0.3867753744125366,
      "learning_rate": 9.700857415052398e-05,
      "loss": 0.4995,
      "step": 476
    },
    {
      "epoch": 0.9072753209700428,
      "grad_norm": 0.34228166937828064,
      "learning_rate": 9.700222292791363e-05,
      "loss": 0.284,
      "step": 477
    },
    {
      "epoch": 0.9091773656680932,
      "grad_norm": 0.3962937593460083,
      "learning_rate": 9.699587170530328e-05,
      "loss": 0.3501,
      "step": 478
    },
    {
      "epoch": 0.9110794103661436,
      "grad_norm": 0.3665268123149872,
      "learning_rate": 9.698952048269292e-05,
      "loss": 0.2737,
      "step": 479
    },
    {
      "epoch": 0.912981455064194,
      "grad_norm": 0.3775653839111328,
      "learning_rate": 9.698316926008257e-05,
      "loss": 0.3173,
      "step": 480
    },
    {
      "epoch": 0.9148834997622444,
      "grad_norm": 0.3584369421005249,
      "learning_rate": 9.697681803747222e-05,
      "loss": 0.3055,
      "step": 481
    },
    {
      "epoch": 0.9167855444602948,
      "grad_norm": 0.3510100245475769,
      "learning_rate": 9.697046681486186e-05,
      "loss": 0.3278,
      "step": 482
    },
    {
      "epoch": 0.9186875891583453,
      "grad_norm": 0.33394765853881836,
      "learning_rate": 9.696411559225151e-05,
      "loss": 0.2954,
      "step": 483
    },
    {
      "epoch": 0.9205896338563956,
      "grad_norm": 0.437014102935791,
      "learning_rate": 9.695776436964116e-05,
      "loss": 0.3797,
      "step": 484
    },
    {
      "epoch": 0.922491678554446,
      "grad_norm": 0.37421244382858276,
      "learning_rate": 9.695141314703082e-05,
      "loss": 0.3521,
      "step": 485
    },
    {
      "epoch": 0.9243937232524965,
      "grad_norm": 0.37696099281311035,
      "learning_rate": 9.694506192442045e-05,
      "loss": 0.3455,
      "step": 486
    },
    {
      "epoch": 0.9262957679505468,
      "grad_norm": 0.5452500581741333,
      "learning_rate": 9.693871070181009e-05,
      "loss": 0.3624,
      "step": 487
    },
    {
      "epoch": 0.9281978126485972,
      "grad_norm": 0.4049624502658844,
      "learning_rate": 9.693235947919976e-05,
      "loss": 0.4017,
      "step": 488
    },
    {
      "epoch": 0.9300998573466477,
      "grad_norm": 0.32757866382598877,
      "learning_rate": 9.69260082565894e-05,
      "loss": 0.3536,
      "step": 489
    },
    {
      "epoch": 0.932001902044698,
      "grad_norm": 0.298367977142334,
      "learning_rate": 9.691965703397905e-05,
      "loss": 0.3374,
      "step": 490
    },
    {
      "epoch": 0.9339039467427485,
      "grad_norm": 0.22035005688667297,
      "learning_rate": 9.69133058113687e-05,
      "loss": 0.2855,
      "step": 491
    },
    {
      "epoch": 0.9358059914407989,
      "grad_norm": 0.43000441789627075,
      "learning_rate": 9.690695458875834e-05,
      "loss": 0.4544,
      "step": 492
    },
    {
      "epoch": 0.9377080361388492,
      "grad_norm": 0.28024253249168396,
      "learning_rate": 9.690060336614799e-05,
      "loss": 0.308,
      "step": 493
    },
    {
      "epoch": 0.9396100808368997,
      "grad_norm": 0.53145432472229,
      "learning_rate": 9.689425214353763e-05,
      "loss": 0.4569,
      "step": 494
    },
    {
      "epoch": 0.9415121255349501,
      "grad_norm": 0.4006127715110779,
      "learning_rate": 9.688790092092729e-05,
      "loss": 0.419,
      "step": 495
    },
    {
      "epoch": 0.9434141702330004,
      "grad_norm": 0.4057261645793915,
      "learning_rate": 9.688154969831693e-05,
      "loss": 0.3553,
      "step": 496
    },
    {
      "epoch": 0.9453162149310509,
      "grad_norm": 0.40803465247154236,
      "learning_rate": 9.687519847570657e-05,
      "loss": 0.3735,
      "step": 497
    },
    {
      "epoch": 0.9472182596291013,
      "grad_norm": 0.34222155809402466,
      "learning_rate": 9.686884725309623e-05,
      "loss": 0.367,
      "step": 498
    },
    {
      "epoch": 0.9491203043271517,
      "grad_norm": 0.40403544902801514,
      "learning_rate": 9.686249603048587e-05,
      "loss": 0.416,
      "step": 499
    },
    {
      "epoch": 0.9510223490252021,
      "grad_norm": 0.33636951446533203,
      "learning_rate": 9.685614480787551e-05,
      "loss": 0.3423,
      "step": 500
    },
    {
      "epoch": 0.9529243937232525,
      "grad_norm": 0.3394258916378021,
      "learning_rate": 9.684979358526516e-05,
      "loss": 0.3282,
      "step": 501
    },
    {
      "epoch": 0.9548264384213029,
      "grad_norm": 0.3682473599910736,
      "learning_rate": 9.684344236265482e-05,
      "loss": 0.406,
      "step": 502
    },
    {
      "epoch": 0.9567284831193533,
      "grad_norm": 0.35073623061180115,
      "learning_rate": 9.683709114004447e-05,
      "loss": 0.376,
      "step": 503
    },
    {
      "epoch": 0.9586305278174037,
      "grad_norm": 0.36000022292137146,
      "learning_rate": 9.68307399174341e-05,
      "loss": 0.3969,
      "step": 504
    },
    {
      "epoch": 0.9605325725154541,
      "grad_norm": 0.361158162355423,
      "learning_rate": 9.682438869482376e-05,
      "loss": 0.347,
      "step": 505
    },
    {
      "epoch": 0.9624346172135045,
      "grad_norm": 0.3075178265571594,
      "learning_rate": 9.681803747221341e-05,
      "loss": 0.4362,
      "step": 506
    },
    {
      "epoch": 0.9643366619115549,
      "grad_norm": 0.30084747076034546,
      "learning_rate": 9.681168624960305e-05,
      "loss": 0.3563,
      "step": 507
    },
    {
      "epoch": 0.9662387066096053,
      "grad_norm": 0.3221014440059662,
      "learning_rate": 9.68053350269927e-05,
      "loss": 0.3366,
      "step": 508
    },
    {
      "epoch": 0.9681407513076558,
      "grad_norm": 0.36464688181877136,
      "learning_rate": 9.679898380438235e-05,
      "loss": 0.3992,
      "step": 509
    },
    {
      "epoch": 0.9700427960057061,
      "grad_norm": 0.32443803548812866,
      "learning_rate": 9.679263258177199e-05,
      "loss": 0.3293,
      "step": 510
    },
    {
      "epoch": 0.9719448407037565,
      "grad_norm": 0.3689454197883606,
      "learning_rate": 9.678628135916164e-05,
      "loss": 0.3546,
      "step": 511
    },
    {
      "epoch": 0.973846885401807,
      "grad_norm": 0.3754975199699402,
      "learning_rate": 9.677993013655129e-05,
      "loss": 0.3856,
      "step": 512
    },
    {
      "epoch": 0.9757489300998573,
      "grad_norm": 0.3642953634262085,
      "learning_rate": 9.677357891394094e-05,
      "loss": 0.4326,
      "step": 513
    },
    {
      "epoch": 0.9776509747979077,
      "grad_norm": 0.43278223276138306,
      "learning_rate": 9.676722769133058e-05,
      "loss": 0.3964,
      "step": 514
    },
    {
      "epoch": 0.9795530194959582,
      "grad_norm": 0.43771886825561523,
      "learning_rate": 9.676087646872023e-05,
      "loss": 0.3861,
      "step": 515
    },
    {
      "epoch": 0.9814550641940085,
      "grad_norm": 0.34908977150917053,
      "learning_rate": 9.675452524610989e-05,
      "loss": 0.3981,
      "step": 516
    },
    {
      "epoch": 0.983357108892059,
      "grad_norm": 0.35733312368392944,
      "learning_rate": 9.674817402349953e-05,
      "loss": 0.3636,
      "step": 517
    },
    {
      "epoch": 0.9852591535901094,
      "grad_norm": 0.3636298179626465,
      "learning_rate": 9.674182280088918e-05,
      "loss": 0.4336,
      "step": 518
    },
    {
      "epoch": 0.9871611982881597,
      "grad_norm": 0.32771605253219604,
      "learning_rate": 9.673547157827883e-05,
      "loss": 0.3481,
      "step": 519
    },
    {
      "epoch": 0.9890632429862102,
      "grad_norm": 0.40213117003440857,
      "learning_rate": 9.672912035566847e-05,
      "loss": 0.3707,
      "step": 520
    },
    {
      "epoch": 0.9909652876842606,
      "grad_norm": 0.3386654257774353,
      "learning_rate": 9.672276913305812e-05,
      "loss": 0.3384,
      "step": 521
    },
    {
      "epoch": 0.992867332382311,
      "grad_norm": 0.3965696096420288,
      "learning_rate": 9.671641791044777e-05,
      "loss": 0.3595,
      "step": 522
    },
    {
      "epoch": 0.9947693770803614,
      "grad_norm": 0.38238459825515747,
      "learning_rate": 9.671006668783741e-05,
      "loss": 0.3714,
      "step": 523
    },
    {
      "epoch": 0.9966714217784118,
      "grad_norm": 0.3248405456542969,
      "learning_rate": 9.670371546522706e-05,
      "loss": 0.394,
      "step": 524
    },
    {
      "epoch": 0.9985734664764622,
      "grad_norm": 0.3902266323566437,
      "learning_rate": 9.66973642426167e-05,
      "loss": 0.4115,
      "step": 525
    },
    {
      "epoch": 1.0004755111745125,
      "grad_norm": 0.4164808392524719,
      "learning_rate": 9.669101302000636e-05,
      "loss": 0.2972,
      "step": 526
    },
    {
      "epoch": 1.002377555872563,
      "grad_norm": 0.33123117685317993,
      "learning_rate": 9.6684661797396e-05,
      "loss": 0.3211,
      "step": 527
    },
    {
      "epoch": 1.0042796005706134,
      "grad_norm": 0.322803258895874,
      "learning_rate": 9.667831057478564e-05,
      "loss": 0.3424,
      "step": 528
    },
    {
      "epoch": 1.0061816452686638,
      "grad_norm": 0.29135918617248535,
      "learning_rate": 9.66719593521753e-05,
      "loss": 0.2882,
      "step": 529
    },
    {
      "epoch": 1.0080836899667143,
      "grad_norm": 0.3367983400821686,
      "learning_rate": 9.666560812956494e-05,
      "loss": 0.2776,
      "step": 530
    },
    {
      "epoch": 1.0099857346647647,
      "grad_norm": 0.304070383310318,
      "learning_rate": 9.66592569069546e-05,
      "loss": 0.249,
      "step": 531
    },
    {
      "epoch": 1.011887779362815,
      "grad_norm": 0.3832727372646332,
      "learning_rate": 9.665290568434423e-05,
      "loss": 0.3118,
      "step": 532
    },
    {
      "epoch": 1.0137898240608654,
      "grad_norm": 0.3365418612957001,
      "learning_rate": 9.664655446173389e-05,
      "loss": 0.197,
      "step": 533
    },
    {
      "epoch": 1.0156918687589158,
      "grad_norm": 0.4367881119251251,
      "learning_rate": 9.664020323912354e-05,
      "loss": 0.3121,
      "step": 534
    },
    {
      "epoch": 1.0175939134569663,
      "grad_norm": 0.43158653378486633,
      "learning_rate": 9.663385201651318e-05,
      "loss": 0.3543,
      "step": 535
    },
    {
      "epoch": 1.0194959581550167,
      "grad_norm": 0.43556904792785645,
      "learning_rate": 9.662750079390283e-05,
      "loss": 0.3121,
      "step": 536
    },
    {
      "epoch": 1.0213980028530671,
      "grad_norm": 0.31828534603118896,
      "learning_rate": 9.662114957129248e-05,
      "loss": 0.24,
      "step": 537
    },
    {
      "epoch": 1.0233000475511174,
      "grad_norm": 0.3935330808162689,
      "learning_rate": 9.661479834868212e-05,
      "loss": 0.2548,
      "step": 538
    },
    {
      "epoch": 1.0252020922491678,
      "grad_norm": 0.3288602828979492,
      "learning_rate": 9.660844712607177e-05,
      "loss": 0.2219,
      "step": 539
    },
    {
      "epoch": 1.0271041369472182,
      "grad_norm": 0.36314669251441956,
      "learning_rate": 9.660209590346142e-05,
      "loss": 0.2817,
      "step": 540
    },
    {
      "epoch": 1.0290061816452687,
      "grad_norm": 0.3528159558773041,
      "learning_rate": 9.659574468085106e-05,
      "loss": 0.2989,
      "step": 541
    },
    {
      "epoch": 1.0309082263433191,
      "grad_norm": 0.3235621750354767,
      "learning_rate": 9.658939345824071e-05,
      "loss": 0.2443,
      "step": 542
    },
    {
      "epoch": 1.0328102710413696,
      "grad_norm": 0.3819037675857544,
      "learning_rate": 9.658304223563036e-05,
      "loss": 0.3494,
      "step": 543
    },
    {
      "epoch": 1.0347123157394198,
      "grad_norm": 0.3885079324245453,
      "learning_rate": 9.657669101302002e-05,
      "loss": 0.3033,
      "step": 544
    },
    {
      "epoch": 1.0366143604374702,
      "grad_norm": 0.3339099884033203,
      "learning_rate": 9.657033979040965e-05,
      "loss": 0.2673,
      "step": 545
    },
    {
      "epoch": 1.0385164051355207,
      "grad_norm": 0.37009695172309875,
      "learning_rate": 9.65639885677993e-05,
      "loss": 0.3715,
      "step": 546
    },
    {
      "epoch": 1.0404184498335711,
      "grad_norm": 0.3462003171443939,
      "learning_rate": 9.655763734518896e-05,
      "loss": 0.2664,
      "step": 547
    },
    {
      "epoch": 1.0423204945316216,
      "grad_norm": 0.3916226327419281,
      "learning_rate": 9.65512861225786e-05,
      "loss": 0.3804,
      "step": 548
    },
    {
      "epoch": 1.044222539229672,
      "grad_norm": 0.3801763951778412,
      "learning_rate": 9.654493489996825e-05,
      "loss": 0.2672,
      "step": 549
    },
    {
      "epoch": 1.0461245839277222,
      "grad_norm": 0.37406545877456665,
      "learning_rate": 9.65385836773579e-05,
      "loss": 0.6203,
      "step": 550
    },
    {
      "epoch": 1.0480266286257727,
      "grad_norm": 0.43677276372909546,
      "learning_rate": 9.653223245474754e-05,
      "loss": 0.3866,
      "step": 551
    },
    {
      "epoch": 1.0499286733238231,
      "grad_norm": 0.26939406991004944,
      "learning_rate": 9.652588123213719e-05,
      "loss": 0.2169,
      "step": 552
    },
    {
      "epoch": 1.0518307180218736,
      "grad_norm": 0.41554608941078186,
      "learning_rate": 9.651953000952684e-05,
      "loss": 0.3705,
      "step": 553
    },
    {
      "epoch": 1.053732762719924,
      "grad_norm": 0.3090009391307831,
      "learning_rate": 9.651317878691648e-05,
      "loss": 0.2471,
      "step": 554
    },
    {
      "epoch": 1.0556348074179742,
      "grad_norm": 0.36705514788627625,
      "learning_rate": 9.650682756430613e-05,
      "loss": 0.2764,
      "step": 555
    },
    {
      "epoch": 1.0575368521160247,
      "grad_norm": 0.39900127053260803,
      "learning_rate": 9.650047634169578e-05,
      "loss": 0.2836,
      "step": 556
    },
    {
      "epoch": 1.059438896814075,
      "grad_norm": 0.31405431032180786,
      "learning_rate": 9.649412511908544e-05,
      "loss": 0.2464,
      "step": 557
    },
    {
      "epoch": 1.0613409415121255,
      "grad_norm": 0.39795488119125366,
      "learning_rate": 9.648777389647507e-05,
      "loss": 0.283,
      "step": 558
    },
    {
      "epoch": 1.063242986210176,
      "grad_norm": 0.36270254850387573,
      "learning_rate": 9.648142267386471e-05,
      "loss": 0.26,
      "step": 559
    },
    {
      "epoch": 1.0651450309082264,
      "grad_norm": 0.42650437355041504,
      "learning_rate": 9.647507145125438e-05,
      "loss": 0.2693,
      "step": 560
    },
    {
      "epoch": 1.0670470756062767,
      "grad_norm": 0.3075532019138336,
      "learning_rate": 9.646872022864402e-05,
      "loss": 0.2941,
      "step": 561
    },
    {
      "epoch": 1.068949120304327,
      "grad_norm": 0.4509059190750122,
      "learning_rate": 9.646236900603367e-05,
      "loss": 0.3525,
      "step": 562
    },
    {
      "epoch": 1.0708511650023775,
      "grad_norm": 0.3420471251010895,
      "learning_rate": 9.645601778342332e-05,
      "loss": 0.2601,
      "step": 563
    },
    {
      "epoch": 1.072753209700428,
      "grad_norm": 0.422493577003479,
      "learning_rate": 9.644966656081296e-05,
      "loss": 0.3441,
      "step": 564
    },
    {
      "epoch": 1.0746552543984784,
      "grad_norm": 0.3960445821285248,
      "learning_rate": 9.644331533820261e-05,
      "loss": 0.3049,
      "step": 565
    },
    {
      "epoch": 1.0765572990965289,
      "grad_norm": 0.32367074489593506,
      "learning_rate": 9.643696411559225e-05,
      "loss": 0.2694,
      "step": 566
    },
    {
      "epoch": 1.078459343794579,
      "grad_norm": 0.3480624258518219,
      "learning_rate": 9.643061289298191e-05,
      "loss": 0.2667,
      "step": 567
    },
    {
      "epoch": 1.0803613884926295,
      "grad_norm": 0.37603023648262024,
      "learning_rate": 9.642426167037155e-05,
      "loss": 0.2875,
      "step": 568
    },
    {
      "epoch": 1.08226343319068,
      "grad_norm": 0.391438752412796,
      "learning_rate": 9.641791044776119e-05,
      "loss": 0.2844,
      "step": 569
    },
    {
      "epoch": 1.0841654778887304,
      "grad_norm": 0.42726075649261475,
      "learning_rate": 9.641155922515086e-05,
      "loss": 0.3092,
      "step": 570
    },
    {
      "epoch": 1.0860675225867809,
      "grad_norm": 0.4007676839828491,
      "learning_rate": 9.64052080025405e-05,
      "loss": 0.2405,
      "step": 571
    },
    {
      "epoch": 1.0879695672848313,
      "grad_norm": 0.401592493057251,
      "learning_rate": 9.639885677993013e-05,
      "loss": 0.297,
      "step": 572
    },
    {
      "epoch": 1.0898716119828815,
      "grad_norm": 0.3883298635482788,
      "learning_rate": 9.639250555731978e-05,
      "loss": 0.3201,
      "step": 573
    },
    {
      "epoch": 1.091773656680932,
      "grad_norm": 0.41852253675460815,
      "learning_rate": 9.638615433470944e-05,
      "loss": 0.259,
      "step": 574
    },
    {
      "epoch": 1.0936757013789824,
      "grad_norm": 0.4559331238269806,
      "learning_rate": 9.637980311209909e-05,
      "loss": 0.3204,
      "step": 575
    },
    {
      "epoch": 1.0955777460770328,
      "grad_norm": 0.4163438379764557,
      "learning_rate": 9.637345188948873e-05,
      "loss": 0.267,
      "step": 576
    },
    {
      "epoch": 1.0974797907750833,
      "grad_norm": 0.38813936710357666,
      "learning_rate": 9.636710066687838e-05,
      "loss": 0.2653,
      "step": 577
    },
    {
      "epoch": 1.0993818354731335,
      "grad_norm": 0.373047798871994,
      "learning_rate": 9.636074944426803e-05,
      "loss": 0.2995,
      "step": 578
    },
    {
      "epoch": 1.101283880171184,
      "grad_norm": 0.39488789439201355,
      "learning_rate": 9.635439822165767e-05,
      "loss": 0.2972,
      "step": 579
    },
    {
      "epoch": 1.1031859248692344,
      "grad_norm": 0.37775856256484985,
      "learning_rate": 9.634804699904732e-05,
      "loss": 0.2833,
      "step": 580
    },
    {
      "epoch": 1.1050879695672848,
      "grad_norm": 0.3843298554420471,
      "learning_rate": 9.634169577643697e-05,
      "loss": 0.3413,
      "step": 581
    },
    {
      "epoch": 1.1069900142653353,
      "grad_norm": 0.3834189176559448,
      "learning_rate": 9.633534455382661e-05,
      "loss": 0.2792,
      "step": 582
    },
    {
      "epoch": 1.1088920589633857,
      "grad_norm": 0.37232789397239685,
      "learning_rate": 9.632899333121626e-05,
      "loss": 0.2724,
      "step": 583
    },
    {
      "epoch": 1.1107941036614362,
      "grad_norm": 0.2608899772167206,
      "learning_rate": 9.632264210860591e-05,
      "loss": 0.1966,
      "step": 584
    },
    {
      "epoch": 1.1126961483594864,
      "grad_norm": 0.2676723301410675,
      "learning_rate": 9.631629088599557e-05,
      "loss": 0.2149,
      "step": 585
    },
    {
      "epoch": 1.1145981930575368,
      "grad_norm": 0.40126022696495056,
      "learning_rate": 9.63099396633852e-05,
      "loss": 0.2937,
      "step": 586
    },
    {
      "epoch": 1.1165002377555873,
      "grad_norm": 0.3493163287639618,
      "learning_rate": 9.630358844077486e-05,
      "loss": 0.2461,
      "step": 587
    },
    {
      "epoch": 1.1184022824536377,
      "grad_norm": 0.39294591546058655,
      "learning_rate": 9.629723721816451e-05,
      "loss": 0.2922,
      "step": 588
    },
    {
      "epoch": 1.1203043271516882,
      "grad_norm": 0.3855053186416626,
      "learning_rate": 9.629088599555415e-05,
      "loss": 0.2541,
      "step": 589
    },
    {
      "epoch": 1.1222063718497384,
      "grad_norm": 0.3388477861881256,
      "learning_rate": 9.628453477294378e-05,
      "loss": 0.2234,
      "step": 590
    },
    {
      "epoch": 1.1241084165477888,
      "grad_norm": 0.3856431841850281,
      "learning_rate": 9.627818355033345e-05,
      "loss": 0.2836,
      "step": 591
    },
    {
      "epoch": 1.1260104612458393,
      "grad_norm": 0.39824768900871277,
      "learning_rate": 9.627183232772309e-05,
      "loss": 0.2562,
      "step": 592
    },
    {
      "epoch": 1.1279125059438897,
      "grad_norm": 0.44484448432922363,
      "learning_rate": 9.626548110511274e-05,
      "loss": 0.2685,
      "step": 593
    },
    {
      "epoch": 1.1298145506419401,
      "grad_norm": 0.4581182599067688,
      "learning_rate": 9.625912988250239e-05,
      "loss": 0.3208,
      "step": 594
    },
    {
      "epoch": 1.1317165953399906,
      "grad_norm": 0.3560565412044525,
      "learning_rate": 9.625277865989203e-05,
      "loss": 0.2834,
      "step": 595
    },
    {
      "epoch": 1.1336186400380408,
      "grad_norm": 0.4423635005950928,
      "learning_rate": 9.624642743728168e-05,
      "loss": 0.3154,
      "step": 596
    },
    {
      "epoch": 1.1355206847360912,
      "grad_norm": 0.3797377943992615,
      "learning_rate": 9.624007621467132e-05,
      "loss": 0.28,
      "step": 597
    },
    {
      "epoch": 1.1374227294341417,
      "grad_norm": 0.29780030250549316,
      "learning_rate": 9.623372499206099e-05,
      "loss": 0.2209,
      "step": 598
    },
    {
      "epoch": 1.1393247741321921,
      "grad_norm": 0.3372732996940613,
      "learning_rate": 9.622737376945062e-05,
      "loss": 0.2502,
      "step": 599
    },
    {
      "epoch": 1.1412268188302426,
      "grad_norm": 0.36365967988967896,
      "learning_rate": 9.622102254684026e-05,
      "loss": 0.2804,
      "step": 600
    },
    {
      "epoch": 1.1431288635282928,
      "grad_norm": 0.40790894627571106,
      "learning_rate": 9.621467132422993e-05,
      "loss": 0.3633,
      "step": 601
    },
    {
      "epoch": 1.1450309082263432,
      "grad_norm": 0.35693496465682983,
      "learning_rate": 9.620832010161957e-05,
      "loss": 0.3193,
      "step": 602
    },
    {
      "epoch": 1.1469329529243937,
      "grad_norm": 0.3701719045639038,
      "learning_rate": 9.620196887900922e-05,
      "loss": 0.2937,
      "step": 603
    },
    {
      "epoch": 1.1488349976224441,
      "grad_norm": 0.4299123287200928,
      "learning_rate": 9.619561765639886e-05,
      "loss": 0.2732,
      "step": 604
    },
    {
      "epoch": 1.1507370423204946,
      "grad_norm": 0.4082129895687103,
      "learning_rate": 9.618926643378851e-05,
      "loss": 0.2867,
      "step": 605
    },
    {
      "epoch": 1.152639087018545,
      "grad_norm": 0.49353981018066406,
      "learning_rate": 9.618291521117816e-05,
      "loss": 0.266,
      "step": 606
    },
    {
      "epoch": 1.1545411317165954,
      "grad_norm": 0.3889831006526947,
      "learning_rate": 9.61765639885678e-05,
      "loss": 0.2732,
      "step": 607
    },
    {
      "epoch": 1.1564431764146457,
      "grad_norm": 0.3464524745941162,
      "learning_rate": 9.617021276595745e-05,
      "loss": 0.2616,
      "step": 608
    },
    {
      "epoch": 1.158345221112696,
      "grad_norm": 0.3498656153678894,
      "learning_rate": 9.61638615433471e-05,
      "loss": 0.2538,
      "step": 609
    },
    {
      "epoch": 1.1602472658107466,
      "grad_norm": 0.31552717089653015,
      "learning_rate": 9.615751032073674e-05,
      "loss": 0.2283,
      "step": 610
    },
    {
      "epoch": 1.162149310508797,
      "grad_norm": 0.3225223422050476,
      "learning_rate": 9.615115909812639e-05,
      "loss": 0.2428,
      "step": 611
    },
    {
      "epoch": 1.1640513552068474,
      "grad_norm": 0.3108568489551544,
      "learning_rate": 9.614480787551604e-05,
      "loss": 0.2207,
      "step": 612
    },
    {
      "epoch": 1.1659533999048977,
      "grad_norm": 0.42909371852874756,
      "learning_rate": 9.613845665290568e-05,
      "loss": 0.3285,
      "step": 613
    },
    {
      "epoch": 1.167855444602948,
      "grad_norm": 0.3831368088722229,
      "learning_rate": 9.613210543029533e-05,
      "loss": 0.2425,
      "step": 614
    },
    {
      "epoch": 1.1697574893009985,
      "grad_norm": 0.3891592025756836,
      "learning_rate": 9.612575420768499e-05,
      "loss": 0.2849,
      "step": 615
    },
    {
      "epoch": 1.171659533999049,
      "grad_norm": 0.5383257865905762,
      "learning_rate": 9.611940298507464e-05,
      "loss": 0.3444,
      "step": 616
    },
    {
      "epoch": 1.1735615786970994,
      "grad_norm": 0.4203440845012665,
      "learning_rate": 9.611305176246428e-05,
      "loss": 0.3198,
      "step": 617
    },
    {
      "epoch": 1.1754636233951499,
      "grad_norm": 0.42422881722450256,
      "learning_rate": 9.610670053985393e-05,
      "loss": 0.3873,
      "step": 618
    },
    {
      "epoch": 1.1773656680932003,
      "grad_norm": 0.34799742698669434,
      "learning_rate": 9.610034931724358e-05,
      "loss": 0.2645,
      "step": 619
    },
    {
      "epoch": 1.1792677127912505,
      "grad_norm": 0.37579119205474854,
      "learning_rate": 9.609399809463322e-05,
      "loss": 0.3379,
      "step": 620
    },
    {
      "epoch": 1.181169757489301,
      "grad_norm": 0.3958894610404968,
      "learning_rate": 9.608764687202287e-05,
      "loss": 0.2792,
      "step": 621
    },
    {
      "epoch": 1.1830718021873514,
      "grad_norm": 0.30366870760917664,
      "learning_rate": 9.608129564941252e-05,
      "loss": 0.1871,
      "step": 622
    },
    {
      "epoch": 1.1849738468854019,
      "grad_norm": 0.39878007769584656,
      "learning_rate": 9.607494442680216e-05,
      "loss": 0.2675,
      "step": 623
    },
    {
      "epoch": 1.1868758915834523,
      "grad_norm": 0.35332080721855164,
      "learning_rate": 9.606859320419181e-05,
      "loss": 0.2856,
      "step": 624
    },
    {
      "epoch": 1.1887779362815025,
      "grad_norm": 0.3391731381416321,
      "learning_rate": 9.606224198158146e-05,
      "loss": 0.254,
      "step": 625
    },
    {
      "epoch": 1.190679980979553,
      "grad_norm": 0.39363861083984375,
      "learning_rate": 9.60558907589711e-05,
      "loss": 0.2447,
      "step": 626
    },
    {
      "epoch": 1.1925820256776034,
      "grad_norm": 0.4773564040660858,
      "learning_rate": 9.604953953636075e-05,
      "loss": 0.3447,
      "step": 627
    },
    {
      "epoch": 1.1944840703756539,
      "grad_norm": 0.34327152371406555,
      "learning_rate": 9.60431883137504e-05,
      "loss": 0.2353,
      "step": 628
    },
    {
      "epoch": 1.1963861150737043,
      "grad_norm": 0.37386631965637207,
      "learning_rate": 9.603683709114006e-05,
      "loss": 0.2792,
      "step": 629
    },
    {
      "epoch": 1.1982881597717547,
      "grad_norm": 0.4061308801174164,
      "learning_rate": 9.60304858685297e-05,
      "loss": 0.3216,
      "step": 630
    },
    {
      "epoch": 1.200190204469805,
      "grad_norm": 0.3440467417240143,
      "learning_rate": 9.602413464591933e-05,
      "loss": 0.2653,
      "step": 631
    },
    {
      "epoch": 1.2020922491678554,
      "grad_norm": 0.36648881435394287,
      "learning_rate": 9.6017783423309e-05,
      "loss": 0.2471,
      "step": 632
    },
    {
      "epoch": 1.2039942938659058,
      "grad_norm": 0.3737157881259918,
      "learning_rate": 9.601143220069864e-05,
      "loss": 0.3255,
      "step": 633
    },
    {
      "epoch": 1.2058963385639563,
      "grad_norm": 0.3840744197368622,
      "learning_rate": 9.600508097808829e-05,
      "loss": 0.2457,
      "step": 634
    },
    {
      "epoch": 1.2077983832620067,
      "grad_norm": 0.34374961256980896,
      "learning_rate": 9.599872975547793e-05,
      "loss": 0.2705,
      "step": 635
    },
    {
      "epoch": 1.209700427960057,
      "grad_norm": 0.3460882306098938,
      "learning_rate": 9.599237853286758e-05,
      "loss": 0.2308,
      "step": 636
    },
    {
      "epoch": 1.2116024726581074,
      "grad_norm": 0.33316507935523987,
      "learning_rate": 9.598602731025723e-05,
      "loss": 0.2562,
      "step": 637
    },
    {
      "epoch": 1.2135045173561578,
      "grad_norm": 0.3132528066635132,
      "learning_rate": 9.597967608764687e-05,
      "loss": 0.2331,
      "step": 638
    },
    {
      "epoch": 1.2154065620542083,
      "grad_norm": 0.3329333961009979,
      "learning_rate": 9.597332486503653e-05,
      "loss": 0.2224,
      "step": 639
    },
    {
      "epoch": 1.2173086067522587,
      "grad_norm": 0.35949432849884033,
      "learning_rate": 9.596697364242617e-05,
      "loss": 0.2337,
      "step": 640
    },
    {
      "epoch": 1.2192106514503092,
      "grad_norm": 0.33591121435165405,
      "learning_rate": 9.596062241981581e-05,
      "loss": 0.2441,
      "step": 641
    },
    {
      "epoch": 1.2211126961483596,
      "grad_norm": 0.38212794065475464,
      "learning_rate": 9.595427119720546e-05,
      "loss": 0.2569,
      "step": 642
    },
    {
      "epoch": 1.2230147408464098,
      "grad_norm": 0.4124354124069214,
      "learning_rate": 9.594791997459512e-05,
      "loss": 0.3143,
      "step": 643
    },
    {
      "epoch": 1.2249167855444603,
      "grad_norm": 0.4712159037590027,
      "learning_rate": 9.594156875198475e-05,
      "loss": 0.3153,
      "step": 644
    },
    {
      "epoch": 1.2268188302425107,
      "grad_norm": 0.3652181923389435,
      "learning_rate": 9.59352175293744e-05,
      "loss": 0.2448,
      "step": 645
    },
    {
      "epoch": 1.2287208749405611,
      "grad_norm": 0.40058213472366333,
      "learning_rate": 9.592886630676406e-05,
      "loss": 0.304,
      "step": 646
    },
    {
      "epoch": 1.2306229196386116,
      "grad_norm": 0.4105280041694641,
      "learning_rate": 9.592251508415371e-05,
      "loss": 0.251,
      "step": 647
    },
    {
      "epoch": 1.2325249643366618,
      "grad_norm": 0.3609527349472046,
      "learning_rate": 9.591616386154335e-05,
      "loss": 0.2311,
      "step": 648
    },
    {
      "epoch": 1.2344270090347123,
      "grad_norm": 0.3686671257019043,
      "learning_rate": 9.5909812638933e-05,
      "loss": 0.2214,
      "step": 649
    },
    {
      "epoch": 1.2363290537327627,
      "grad_norm": 0.27986517548561096,
      "learning_rate": 9.590346141632265e-05,
      "loss": 0.2531,
      "step": 650
    },
    {
      "epoch": 1.2382310984308131,
      "grad_norm": 0.4477519690990448,
      "learning_rate": 9.589711019371229e-05,
      "loss": 0.3039,
      "step": 651
    },
    {
      "epoch": 1.2401331431288636,
      "grad_norm": 0.33017873764038086,
      "learning_rate": 9.589075897110194e-05,
      "loss": 0.205,
      "step": 652
    },
    {
      "epoch": 1.242035187826914,
      "grad_norm": 0.31245800852775574,
      "learning_rate": 9.588440774849159e-05,
      "loss": 0.2493,
      "step": 653
    },
    {
      "epoch": 1.2439372325249642,
      "grad_norm": 0.33620285987854004,
      "learning_rate": 9.587805652588123e-05,
      "loss": 0.2629,
      "step": 654
    },
    {
      "epoch": 1.2458392772230147,
      "grad_norm": 0.34820401668548584,
      "learning_rate": 9.587170530327088e-05,
      "loss": 0.2446,
      "step": 655
    },
    {
      "epoch": 1.2477413219210651,
      "grad_norm": 0.4110179543495178,
      "learning_rate": 9.586535408066053e-05,
      "loss": 0.3345,
      "step": 656
    },
    {
      "epoch": 1.2496433666191156,
      "grad_norm": 0.3637439012527466,
      "learning_rate": 9.585900285805019e-05,
      "loss": 0.2052,
      "step": 657
    },
    {
      "epoch": 1.251545411317166,
      "grad_norm": 0.39023682475090027,
      "learning_rate": 9.585265163543982e-05,
      "loss": 0.2841,
      "step": 658
    },
    {
      "epoch": 1.2534474560152162,
      "grad_norm": 0.3623685836791992,
      "learning_rate": 9.584630041282948e-05,
      "loss": 0.2286,
      "step": 659
    },
    {
      "epoch": 1.2553495007132667,
      "grad_norm": 0.38151344656944275,
      "learning_rate": 9.583994919021913e-05,
      "loss": 0.2357,
      "step": 660
    },
    {
      "epoch": 1.2572515454113171,
      "grad_norm": 0.38236725330352783,
      "learning_rate": 9.583359796760877e-05,
      "loss": 0.2966,
      "step": 661
    },
    {
      "epoch": 1.2591535901093676,
      "grad_norm": 0.38568076491355896,
      "learning_rate": 9.58272467449984e-05,
      "loss": 0.3018,
      "step": 662
    },
    {
      "epoch": 1.261055634807418,
      "grad_norm": 0.3488738238811493,
      "learning_rate": 9.582089552238807e-05,
      "loss": 0.354,
      "step": 663
    },
    {
      "epoch": 1.2629576795054684,
      "grad_norm": 0.352860689163208,
      "learning_rate": 9.581454429977771e-05,
      "loss": 0.2143,
      "step": 664
    },
    {
      "epoch": 1.2648597242035189,
      "grad_norm": 0.3734944760799408,
      "learning_rate": 9.580819307716736e-05,
      "loss": 0.3486,
      "step": 665
    },
    {
      "epoch": 1.266761768901569,
      "grad_norm": 0.4024759531021118,
      "learning_rate": 9.580184185455701e-05,
      "loss": 0.2922,
      "step": 666
    },
    {
      "epoch": 1.2686638135996195,
      "grad_norm": 0.37389662861824036,
      "learning_rate": 9.579549063194665e-05,
      "loss": 0.2545,
      "step": 667
    },
    {
      "epoch": 1.27056585829767,
      "grad_norm": 0.42338186502456665,
      "learning_rate": 9.57891394093363e-05,
      "loss": 0.2961,
      "step": 668
    },
    {
      "epoch": 1.2724679029957204,
      "grad_norm": 0.3795355260372162,
      "learning_rate": 9.578278818672594e-05,
      "loss": 0.2777,
      "step": 669
    },
    {
      "epoch": 1.2743699476937709,
      "grad_norm": 0.3439030945301056,
      "learning_rate": 9.57764369641156e-05,
      "loss": 0.2179,
      "step": 670
    },
    {
      "epoch": 1.276271992391821,
      "grad_norm": 0.39637741446495056,
      "learning_rate": 9.577008574150524e-05,
      "loss": 0.2701,
      "step": 671
    },
    {
      "epoch": 1.2781740370898715,
      "grad_norm": 0.3348701298236847,
      "learning_rate": 9.576373451889488e-05,
      "loss": 0.2632,
      "step": 672
    },
    {
      "epoch": 1.280076081787922,
      "grad_norm": 0.3696272671222687,
      "learning_rate": 9.575738329628455e-05,
      "loss": 0.2228,
      "step": 673
    },
    {
      "epoch": 1.2819781264859724,
      "grad_norm": 0.3261694610118866,
      "learning_rate": 9.575103207367419e-05,
      "loss": 0.2589,
      "step": 674
    },
    {
      "epoch": 1.2838801711840229,
      "grad_norm": 0.39266085624694824,
      "learning_rate": 9.574468085106384e-05,
      "loss": 0.2893,
      "step": 675
    },
    {
      "epoch": 1.2857822158820733,
      "grad_norm": 0.4356357157230377,
      "learning_rate": 9.573832962845348e-05,
      "loss": 0.3249,
      "step": 676
    },
    {
      "epoch": 1.2876842605801238,
      "grad_norm": 0.38992395997047424,
      "learning_rate": 9.573197840584313e-05,
      "loss": 0.2697,
      "step": 677
    },
    {
      "epoch": 1.289586305278174,
      "grad_norm": 0.35415610671043396,
      "learning_rate": 9.572562718323278e-05,
      "loss": 0.2538,
      "step": 678
    },
    {
      "epoch": 1.2914883499762244,
      "grad_norm": 0.38410142064094543,
      "learning_rate": 9.571927596062242e-05,
      "loss": 0.2325,
      "step": 679
    },
    {
      "epoch": 1.2933903946742749,
      "grad_norm": 0.36036771535873413,
      "learning_rate": 9.571292473801207e-05,
      "loss": 0.242,
      "step": 680
    },
    {
      "epoch": 1.2952924393723253,
      "grad_norm": 0.3901429772377014,
      "learning_rate": 9.570657351540172e-05,
      "loss": 0.3141,
      "step": 681
    },
    {
      "epoch": 1.2971944840703755,
      "grad_norm": 0.3684573769569397,
      "learning_rate": 9.570022229279136e-05,
      "loss": 0.2725,
      "step": 682
    },
    {
      "epoch": 1.299096528768426,
      "grad_norm": 0.44199153780937195,
      "learning_rate": 9.569387107018101e-05,
      "loss": 0.2938,
      "step": 683
    },
    {
      "epoch": 1.3009985734664764,
      "grad_norm": 0.4435335695743561,
      "learning_rate": 9.568751984757066e-05,
      "loss": 0.3454,
      "step": 684
    },
    {
      "epoch": 1.3029006181645268,
      "grad_norm": 0.3713487386703491,
      "learning_rate": 9.56811686249603e-05,
      "loss": 0.25,
      "step": 685
    },
    {
      "epoch": 1.3048026628625773,
      "grad_norm": 0.394452840089798,
      "learning_rate": 9.567481740234995e-05,
      "loss": 0.3062,
      "step": 686
    },
    {
      "epoch": 1.3067047075606277,
      "grad_norm": 0.47593292593955994,
      "learning_rate": 9.56684661797396e-05,
      "loss": 0.3131,
      "step": 687
    },
    {
      "epoch": 1.3086067522586782,
      "grad_norm": 0.39060479402542114,
      "learning_rate": 9.566211495712926e-05,
      "loss": 0.3267,
      "step": 688
    },
    {
      "epoch": 1.3105087969567286,
      "grad_norm": 0.40931451320648193,
      "learning_rate": 9.56557637345189e-05,
      "loss": 0.2979,
      "step": 689
    },
    {
      "epoch": 1.3124108416547788,
      "grad_norm": 0.3557567000389099,
      "learning_rate": 9.564941251190855e-05,
      "loss": 0.213,
      "step": 690
    },
    {
      "epoch": 1.3143128863528293,
      "grad_norm": 0.43843701481819153,
      "learning_rate": 9.56430612892982e-05,
      "loss": 0.2835,
      "step": 691
    },
    {
      "epoch": 1.3162149310508797,
      "grad_norm": 0.33530867099761963,
      "learning_rate": 9.563671006668784e-05,
      "loss": 0.2392,
      "step": 692
    },
    {
      "epoch": 1.3181169757489302,
      "grad_norm": 0.35071656107902527,
      "learning_rate": 9.563035884407749e-05,
      "loss": 0.1916,
      "step": 693
    },
    {
      "epoch": 1.3200190204469804,
      "grad_norm": 0.3808371126651764,
      "learning_rate": 9.562400762146714e-05,
      "loss": 0.2426,
      "step": 694
    },
    {
      "epoch": 1.3219210651450308,
      "grad_norm": 0.46641990542411804,
      "learning_rate": 9.561765639885678e-05,
      "loss": 0.3399,
      "step": 695
    },
    {
      "epoch": 1.3238231098430813,
      "grad_norm": 0.4153888523578644,
      "learning_rate": 9.561130517624643e-05,
      "loss": 0.4152,
      "step": 696
    },
    {
      "epoch": 1.3257251545411317,
      "grad_norm": 0.4004898965358734,
      "learning_rate": 9.560495395363608e-05,
      "loss": 0.3637,
      "step": 697
    },
    {
      "epoch": 1.3276271992391822,
      "grad_norm": 0.421058714389801,
      "learning_rate": 9.559860273102572e-05,
      "loss": 0.2625,
      "step": 698
    },
    {
      "epoch": 1.3295292439372326,
      "grad_norm": 0.39722004532814026,
      "learning_rate": 9.559225150841537e-05,
      "loss": 0.3563,
      "step": 699
    },
    {
      "epoch": 1.331431288635283,
      "grad_norm": 0.3793489634990692,
      "learning_rate": 9.558590028580501e-05,
      "loss": 0.2306,
      "step": 700
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 0.43592244386672974,
      "learning_rate": 9.557954906319468e-05,
      "loss": 0.4354,
      "step": 701
    },
    {
      "epoch": 1.3352353780313837,
      "grad_norm": 0.30159738659858704,
      "learning_rate": 9.557319784058432e-05,
      "loss": 0.2062,
      "step": 702
    },
    {
      "epoch": 1.3371374227294341,
      "grad_norm": 0.34011465311050415,
      "learning_rate": 9.556684661797395e-05,
      "loss": 0.2363,
      "step": 703
    },
    {
      "epoch": 1.3390394674274846,
      "grad_norm": 0.41224443912506104,
      "learning_rate": 9.556049539536362e-05,
      "loss": 0.2913,
      "step": 704
    },
    {
      "epoch": 1.340941512125535,
      "grad_norm": 0.4105536937713623,
      "learning_rate": 9.555414417275326e-05,
      "loss": 0.2459,
      "step": 705
    },
    {
      "epoch": 1.3428435568235852,
      "grad_norm": 0.3158798813819885,
      "learning_rate": 9.554779295014291e-05,
      "loss": 0.1921,
      "step": 706
    },
    {
      "epoch": 1.3447456015216357,
      "grad_norm": 0.4023972451686859,
      "learning_rate": 9.554144172753255e-05,
      "loss": 0.2406,
      "step": 707
    },
    {
      "epoch": 1.3466476462196861,
      "grad_norm": 0.4204084277153015,
      "learning_rate": 9.55350905049222e-05,
      "loss": 0.2977,
      "step": 708
    },
    {
      "epoch": 1.3485496909177366,
      "grad_norm": 0.4853519797325134,
      "learning_rate": 9.552873928231185e-05,
      "loss": 0.3871,
      "step": 709
    },
    {
      "epoch": 1.350451735615787,
      "grad_norm": 0.3755006194114685,
      "learning_rate": 9.552238805970149e-05,
      "loss": 0.2399,
      "step": 710
    },
    {
      "epoch": 1.3523537803138375,
      "grad_norm": 0.37587347626686096,
      "learning_rate": 9.551603683709116e-05,
      "loss": 0.3029,
      "step": 711
    },
    {
      "epoch": 1.354255825011888,
      "grad_norm": 0.4257625937461853,
      "learning_rate": 9.55096856144808e-05,
      "loss": 0.2541,
      "step": 712
    },
    {
      "epoch": 1.3561578697099381,
      "grad_norm": 0.29570913314819336,
      "learning_rate": 9.550333439187043e-05,
      "loss": 0.1668,
      "step": 713
    },
    {
      "epoch": 1.3580599144079886,
      "grad_norm": 0.5089273452758789,
      "learning_rate": 9.549698316926008e-05,
      "loss": 0.4006,
      "step": 714
    },
    {
      "epoch": 1.359961959106039,
      "grad_norm": 0.43584999442100525,
      "learning_rate": 9.549063194664974e-05,
      "loss": 0.2996,
      "step": 715
    },
    {
      "epoch": 1.3618640038040895,
      "grad_norm": 0.4071057140827179,
      "learning_rate": 9.548428072403937e-05,
      "loss": 0.308,
      "step": 716
    },
    {
      "epoch": 1.3637660485021397,
      "grad_norm": 0.37772196531295776,
      "learning_rate": 9.547792950142903e-05,
      "loss": 0.2235,
      "step": 717
    },
    {
      "epoch": 1.3656680932001901,
      "grad_norm": 0.44488438963890076,
      "learning_rate": 9.547157827881868e-05,
      "loss": 0.2748,
      "step": 718
    },
    {
      "epoch": 1.3675701378982406,
      "grad_norm": 0.3227798640727997,
      "learning_rate": 9.546522705620833e-05,
      "loss": 0.2609,
      "step": 719
    },
    {
      "epoch": 1.369472182596291,
      "grad_norm": 0.3742448389530182,
      "learning_rate": 9.545887583359797e-05,
      "loss": 0.2417,
      "step": 720
    },
    {
      "epoch": 1.3713742272943414,
      "grad_norm": 0.3582020699977875,
      "learning_rate": 9.545252461098762e-05,
      "loss": 0.2688,
      "step": 721
    },
    {
      "epoch": 1.3732762719923919,
      "grad_norm": 0.3762567341327667,
      "learning_rate": 9.544617338837727e-05,
      "loss": 0.2939,
      "step": 722
    },
    {
      "epoch": 1.3751783166904423,
      "grad_norm": 0.38103973865509033,
      "learning_rate": 9.543982216576691e-05,
      "loss": 0.3335,
      "step": 723
    },
    {
      "epoch": 1.3770803613884925,
      "grad_norm": 0.3109844923019409,
      "learning_rate": 9.543347094315656e-05,
      "loss": 0.2094,
      "step": 724
    },
    {
      "epoch": 1.378982406086543,
      "grad_norm": 0.3642789125442505,
      "learning_rate": 9.542711972054621e-05,
      "loss": 0.2879,
      "step": 725
    },
    {
      "epoch": 1.3808844507845934,
      "grad_norm": 0.3879150152206421,
      "learning_rate": 9.542076849793585e-05,
      "loss": 0.2567,
      "step": 726
    },
    {
      "epoch": 1.3827864954826439,
      "grad_norm": 0.3364320993423462,
      "learning_rate": 9.54144172753255e-05,
      "loss": 0.2773,
      "step": 727
    },
    {
      "epoch": 1.3846885401806943,
      "grad_norm": 0.5071269273757935,
      "learning_rate": 9.540806605271516e-05,
      "loss": 0.2916,
      "step": 728
    },
    {
      "epoch": 1.3865905848787445,
      "grad_norm": 0.425793319940567,
      "learning_rate": 9.540171483010481e-05,
      "loss": 0.2948,
      "step": 729
    },
    {
      "epoch": 1.388492629576795,
      "grad_norm": 0.38478776812553406,
      "learning_rate": 9.539536360749445e-05,
      "loss": 0.2493,
      "step": 730
    },
    {
      "epoch": 1.3903946742748454,
      "grad_norm": 0.4016847014427185,
      "learning_rate": 9.53890123848841e-05,
      "loss": 0.3038,
      "step": 731
    },
    {
      "epoch": 1.3922967189728959,
      "grad_norm": 0.2799355983734131,
      "learning_rate": 9.538266116227375e-05,
      "loss": 0.2964,
      "step": 732
    },
    {
      "epoch": 1.3941987636709463,
      "grad_norm": 0.3720659613609314,
      "learning_rate": 9.537630993966339e-05,
      "loss": 0.2528,
      "step": 733
    },
    {
      "epoch": 1.3961008083689967,
      "grad_norm": 0.2954385578632355,
      "learning_rate": 9.536995871705303e-05,
      "loss": 0.2119,
      "step": 734
    },
    {
      "epoch": 1.3980028530670472,
      "grad_norm": 0.35636264085769653,
      "learning_rate": 9.536360749444269e-05,
      "loss": 0.3042,
      "step": 735
    },
    {
      "epoch": 1.3999048977650974,
      "grad_norm": 0.3219160735607147,
      "learning_rate": 9.535725627183233e-05,
      "loss": 0.2977,
      "step": 736
    },
    {
      "epoch": 1.4018069424631479,
      "grad_norm": 0.32340940833091736,
      "learning_rate": 9.535090504922198e-05,
      "loss": 0.2295,
      "step": 737
    },
    {
      "epoch": 1.4037089871611983,
      "grad_norm": 0.3884155750274658,
      "learning_rate": 9.534455382661163e-05,
      "loss": 0.2367,
      "step": 738
    },
    {
      "epoch": 1.4056110318592487,
      "grad_norm": 0.3708769381046295,
      "learning_rate": 9.533820260400127e-05,
      "loss": 0.2807,
      "step": 739
    },
    {
      "epoch": 1.407513076557299,
      "grad_norm": 0.3377797603607178,
      "learning_rate": 9.533185138139092e-05,
      "loss": 0.2459,
      "step": 740
    },
    {
      "epoch": 1.4094151212553494,
      "grad_norm": 0.542662501335144,
      "learning_rate": 9.532550015878056e-05,
      "loss": 0.3883,
      "step": 741
    },
    {
      "epoch": 1.4113171659533998,
      "grad_norm": 0.36908188462257385,
      "learning_rate": 9.531914893617023e-05,
      "loss": 0.2239,
      "step": 742
    },
    {
      "epoch": 1.4132192106514503,
      "grad_norm": 0.2898438572883606,
      "learning_rate": 9.531279771355987e-05,
      "loss": 0.1929,
      "step": 743
    },
    {
      "epoch": 1.4151212553495007,
      "grad_norm": 0.361965537071228,
      "learning_rate": 9.53064464909495e-05,
      "loss": 0.2758,
      "step": 744
    },
    {
      "epoch": 1.4170233000475512,
      "grad_norm": 0.42736831307411194,
      "learning_rate": 9.530009526833916e-05,
      "loss": 0.3103,
      "step": 745
    },
    {
      "epoch": 1.4189253447456016,
      "grad_norm": 0.3411954641342163,
      "learning_rate": 9.529374404572881e-05,
      "loss": 0.2498,
      "step": 746
    },
    {
      "epoch": 1.420827389443652,
      "grad_norm": 0.3671089708805084,
      "learning_rate": 9.528739282311846e-05,
      "loss": 0.2961,
      "step": 747
    },
    {
      "epoch": 1.4227294341417023,
      "grad_norm": 0.35021135210990906,
      "learning_rate": 9.52810416005081e-05,
      "loss": 0.2422,
      "step": 748
    },
    {
      "epoch": 1.4246314788397527,
      "grad_norm": 0.3203287422657013,
      "learning_rate": 9.527469037789775e-05,
      "loss": 0.2377,
      "step": 749
    },
    {
      "epoch": 1.4265335235378032,
      "grad_norm": 0.32512807846069336,
      "learning_rate": 9.52683391552874e-05,
      "loss": 0.2533,
      "step": 750
    },
    {
      "epoch": 1.4284355682358536,
      "grad_norm": 0.39963454008102417,
      "learning_rate": 9.526198793267704e-05,
      "loss": 0.3191,
      "step": 751
    },
    {
      "epoch": 1.4303376129339038,
      "grad_norm": 0.3722153306007385,
      "learning_rate": 9.525563671006669e-05,
      "loss": 0.2134,
      "step": 752
    },
    {
      "epoch": 1.4322396576319543,
      "grad_norm": 0.3429708182811737,
      "learning_rate": 9.524928548745634e-05,
      "loss": 0.2221,
      "step": 753
    },
    {
      "epoch": 1.4341417023300047,
      "grad_norm": 0.4014436602592468,
      "learning_rate": 9.524293426484598e-05,
      "loss": 0.2638,
      "step": 754
    },
    {
      "epoch": 1.4360437470280552,
      "grad_norm": 0.38329729437828064,
      "learning_rate": 9.523658304223563e-05,
      "loss": 0.25,
      "step": 755
    },
    {
      "epoch": 1.4379457917261056,
      "grad_norm": 0.37710002064704895,
      "learning_rate": 9.523023181962529e-05,
      "loss": 0.2623,
      "step": 756
    },
    {
      "epoch": 1.439847836424156,
      "grad_norm": 0.4223197102546692,
      "learning_rate": 9.522388059701492e-05,
      "loss": 0.408,
      "step": 757
    },
    {
      "epoch": 1.4417498811222065,
      "grad_norm": 0.45707425475120544,
      "learning_rate": 9.521752937440458e-05,
      "loss": 0.3491,
      "step": 758
    },
    {
      "epoch": 1.4436519258202567,
      "grad_norm": 0.39775991439819336,
      "learning_rate": 9.521117815179423e-05,
      "loss": 0.2498,
      "step": 759
    },
    {
      "epoch": 1.4455539705183071,
      "grad_norm": 0.3113288879394531,
      "learning_rate": 9.520482692918388e-05,
      "loss": 0.2191,
      "step": 760
    },
    {
      "epoch": 1.4474560152163576,
      "grad_norm": 0.35126394033432007,
      "learning_rate": 9.519847570657352e-05,
      "loss": 0.2689,
      "step": 761
    },
    {
      "epoch": 1.449358059914408,
      "grad_norm": 0.42121708393096924,
      "learning_rate": 9.519212448396317e-05,
      "loss": 0.2859,
      "step": 762
    },
    {
      "epoch": 1.4512601046124585,
      "grad_norm": 0.37913796305656433,
      "learning_rate": 9.518577326135282e-05,
      "loss": 0.2676,
      "step": 763
    },
    {
      "epoch": 1.4531621493105087,
      "grad_norm": 0.3767364025115967,
      "learning_rate": 9.517942203874246e-05,
      "loss": 0.2298,
      "step": 764
    },
    {
      "epoch": 1.4550641940085591,
      "grad_norm": 0.3317908048629761,
      "learning_rate": 9.517307081613211e-05,
      "loss": 0.2439,
      "step": 765
    },
    {
      "epoch": 1.4569662387066096,
      "grad_norm": 0.28014522790908813,
      "learning_rate": 9.516671959352176e-05,
      "loss": 0.207,
      "step": 766
    },
    {
      "epoch": 1.45886828340466,
      "grad_norm": 0.4119054675102234,
      "learning_rate": 9.51603683709114e-05,
      "loss": 0.2969,
      "step": 767
    },
    {
      "epoch": 1.4607703281027105,
      "grad_norm": 0.3351030647754669,
      "learning_rate": 9.515401714830105e-05,
      "loss": 0.2925,
      "step": 768
    },
    {
      "epoch": 1.462672372800761,
      "grad_norm": 0.5204692482948303,
      "learning_rate": 9.51476659256907e-05,
      "loss": 0.3546,
      "step": 769
    },
    {
      "epoch": 1.4645744174988113,
      "grad_norm": 0.42994043231010437,
      "learning_rate": 9.514131470308034e-05,
      "loss": 0.3284,
      "step": 770
    },
    {
      "epoch": 1.4664764621968616,
      "grad_norm": 0.3580436408519745,
      "learning_rate": 9.513496348047e-05,
      "loss": 0.2639,
      "step": 771
    },
    {
      "epoch": 1.468378506894912,
      "grad_norm": 0.37151291966438293,
      "learning_rate": 9.512861225785963e-05,
      "loss": 0.2556,
      "step": 772
    },
    {
      "epoch": 1.4702805515929624,
      "grad_norm": 0.33122384548187256,
      "learning_rate": 9.51222610352493e-05,
      "loss": 0.2565,
      "step": 773
    },
    {
      "epoch": 1.472182596291013,
      "grad_norm": 0.3718935251235962,
      "learning_rate": 9.511590981263894e-05,
      "loss": 0.2348,
      "step": 774
    },
    {
      "epoch": 1.474084640989063,
      "grad_norm": 0.3752667009830475,
      "learning_rate": 9.510955859002858e-05,
      "loss": 0.2933,
      "step": 775
    },
    {
      "epoch": 1.4759866856871136,
      "grad_norm": 0.44539371132850647,
      "learning_rate": 9.510320736741824e-05,
      "loss": 0.2699,
      "step": 776
    },
    {
      "epoch": 1.477888730385164,
      "grad_norm": 0.5468220114707947,
      "learning_rate": 9.509685614480788e-05,
      "loss": 0.4141,
      "step": 777
    },
    {
      "epoch": 1.4797907750832144,
      "grad_norm": 0.5036222338676453,
      "learning_rate": 9.509050492219753e-05,
      "loss": 0.3463,
      "step": 778
    },
    {
      "epoch": 1.4816928197812649,
      "grad_norm": 0.3742172420024872,
      "learning_rate": 9.508415369958717e-05,
      "loss": 0.3104,
      "step": 779
    },
    {
      "epoch": 1.4835948644793153,
      "grad_norm": 0.38696351647377014,
      "learning_rate": 9.507780247697682e-05,
      "loss": 0.2406,
      "step": 780
    },
    {
      "epoch": 1.4854969091773658,
      "grad_norm": 0.43431171774864197,
      "learning_rate": 9.507145125436647e-05,
      "loss": 0.307,
      "step": 781
    },
    {
      "epoch": 1.4873989538754162,
      "grad_norm": 0.3814404606819153,
      "learning_rate": 9.506510003175611e-05,
      "loss": 0.2681,
      "step": 782
    },
    {
      "epoch": 1.4893009985734664,
      "grad_norm": 0.350359708070755,
      "learning_rate": 9.505874880914578e-05,
      "loss": 0.2408,
      "step": 783
    },
    {
      "epoch": 1.4912030432715169,
      "grad_norm": 0.4443821609020233,
      "learning_rate": 9.505239758653541e-05,
      "loss": 0.3358,
      "step": 784
    },
    {
      "epoch": 1.4931050879695673,
      "grad_norm": 0.2963017225265503,
      "learning_rate": 9.504604636392505e-05,
      "loss": 0.2085,
      "step": 785
    },
    {
      "epoch": 1.4950071326676178,
      "grad_norm": 0.4765385389328003,
      "learning_rate": 9.50396951413147e-05,
      "loss": 0.396,
      "step": 786
    },
    {
      "epoch": 1.496909177365668,
      "grad_norm": 0.3389003574848175,
      "learning_rate": 9.503334391870436e-05,
      "loss": 0.327,
      "step": 787
    },
    {
      "epoch": 1.4988112220637184,
      "grad_norm": 0.42218640446662903,
      "learning_rate": 9.5026992696094e-05,
      "loss": 0.3078,
      "step": 788
    },
    {
      "epoch": 1.5007132667617689,
      "grad_norm": 0.4693278670310974,
      "learning_rate": 9.502064147348365e-05,
      "loss": 0.2853,
      "step": 789
    },
    {
      "epoch": 1.5026153114598193,
      "grad_norm": 0.3891851305961609,
      "learning_rate": 9.50142902508733e-05,
      "loss": 0.2493,
      "step": 790
    },
    {
      "epoch": 1.5045173561578697,
      "grad_norm": 0.3862535357475281,
      "learning_rate": 9.500793902826295e-05,
      "loss": 0.2673,
      "step": 791
    },
    {
      "epoch": 1.5064194008559202,
      "grad_norm": 0.34803205728530884,
      "learning_rate": 9.500158780565259e-05,
      "loss": 0.2814,
      "step": 792
    },
    {
      "epoch": 1.5083214455539706,
      "grad_norm": 0.3963899314403534,
      "learning_rate": 9.499523658304224e-05,
      "loss": 0.3018,
      "step": 793
    },
    {
      "epoch": 1.510223490252021,
      "grad_norm": 0.4004577398300171,
      "learning_rate": 9.498888536043189e-05,
      "loss": 0.313,
      "step": 794
    },
    {
      "epoch": 1.5121255349500713,
      "grad_norm": 0.32212579250335693,
      "learning_rate": 9.498253413782153e-05,
      "loss": 0.2081,
      "step": 795
    },
    {
      "epoch": 1.5140275796481217,
      "grad_norm": 0.32745805382728577,
      "learning_rate": 9.497618291521118e-05,
      "loss": 0.231,
      "step": 796
    },
    {
      "epoch": 1.5159296243461722,
      "grad_norm": 0.40773364901542664,
      "learning_rate": 9.496983169260083e-05,
      "loss": 0.2804,
      "step": 797
    },
    {
      "epoch": 1.5178316690442224,
      "grad_norm": 0.3848927319049835,
      "learning_rate": 9.496348046999047e-05,
      "loss": 0.288,
      "step": 798
    },
    {
      "epoch": 1.5197337137422728,
      "grad_norm": 0.317124605178833,
      "learning_rate": 9.495712924738012e-05,
      "loss": 0.2202,
      "step": 799
    },
    {
      "epoch": 1.5216357584403233,
      "grad_norm": 0.3564606010913849,
      "learning_rate": 9.495077802476978e-05,
      "loss": 0.2594,
      "step": 800
    },
    {
      "epoch": 1.5235378031383737,
      "grad_norm": 0.3151964545249939,
      "learning_rate": 9.494442680215943e-05,
      "loss": 0.2138,
      "step": 801
    },
    {
      "epoch": 1.5254398478364242,
      "grad_norm": 0.4009242057800293,
      "learning_rate": 9.493807557954907e-05,
      "loss": 0.3157,
      "step": 802
    },
    {
      "epoch": 1.5273418925344746,
      "grad_norm": 0.36916011571884155,
      "learning_rate": 9.49317243569387e-05,
      "loss": 0.2478,
      "step": 803
    },
    {
      "epoch": 1.529243937232525,
      "grad_norm": 0.372277170419693,
      "learning_rate": 9.492537313432837e-05,
      "loss": 0.2912,
      "step": 804
    },
    {
      "epoch": 1.5311459819305755,
      "grad_norm": 0.42100057005882263,
      "learning_rate": 9.491902191171801e-05,
      "loss": 0.2938,
      "step": 805
    },
    {
      "epoch": 1.533048026628626,
      "grad_norm": 0.3528178334236145,
      "learning_rate": 9.491267068910765e-05,
      "loss": 0.2519,
      "step": 806
    },
    {
      "epoch": 1.5349500713266762,
      "grad_norm": 0.3655840754508972,
      "learning_rate": 9.490631946649731e-05,
      "loss": 0.2685,
      "step": 807
    },
    {
      "epoch": 1.5368521160247266,
      "grad_norm": 0.34080174565315247,
      "learning_rate": 9.489996824388695e-05,
      "loss": 0.2339,
      "step": 808
    },
    {
      "epoch": 1.5387541607227768,
      "grad_norm": 0.3532484173774719,
      "learning_rate": 9.48936170212766e-05,
      "loss": 0.2448,
      "step": 809
    },
    {
      "epoch": 1.5406562054208273,
      "grad_norm": 0.33115965127944946,
      "learning_rate": 9.488726579866624e-05,
      "loss": 0.2549,
      "step": 810
    },
    {
      "epoch": 1.5425582501188777,
      "grad_norm": 0.40624433755874634,
      "learning_rate": 9.488091457605589e-05,
      "loss": 0.2847,
      "step": 811
    },
    {
      "epoch": 1.5444602948169281,
      "grad_norm": 0.35374221205711365,
      "learning_rate": 9.487456335344554e-05,
      "loss": 0.2704,
      "step": 812
    },
    {
      "epoch": 1.5463623395149786,
      "grad_norm": 0.3859337568283081,
      "learning_rate": 9.486821213083518e-05,
      "loss": 0.2969,
      "step": 813
    },
    {
      "epoch": 1.548264384213029,
      "grad_norm": 0.37984946370124817,
      "learning_rate": 9.486186090822485e-05,
      "loss": 0.2908,
      "step": 814
    },
    {
      "epoch": 1.5501664289110795,
      "grad_norm": 0.34984755516052246,
      "learning_rate": 9.485550968561449e-05,
      "loss": 0.2247,
      "step": 815
    },
    {
      "epoch": 1.55206847360913,
      "grad_norm": 0.32592761516571045,
      "learning_rate": 9.484915846300412e-05,
      "loss": 0.1985,
      "step": 816
    },
    {
      "epoch": 1.5539705183071804,
      "grad_norm": 0.4273107945919037,
      "learning_rate": 9.484280724039378e-05,
      "loss": 0.2875,
      "step": 817
    },
    {
      "epoch": 1.5558725630052306,
      "grad_norm": 0.35476601123809814,
      "learning_rate": 9.483645601778343e-05,
      "loss": 0.2721,
      "step": 818
    },
    {
      "epoch": 1.557774607703281,
      "grad_norm": 0.30542057752609253,
      "learning_rate": 9.483010479517308e-05,
      "loss": 0.1966,
      "step": 819
    },
    {
      "epoch": 1.5596766524013315,
      "grad_norm": 0.44310665130615234,
      "learning_rate": 9.482375357256272e-05,
      "loss": 0.2533,
      "step": 820
    },
    {
      "epoch": 1.5615786970993817,
      "grad_norm": 0.39837488532066345,
      "learning_rate": 9.481740234995237e-05,
      "loss": 0.3045,
      "step": 821
    },
    {
      "epoch": 1.5634807417974321,
      "grad_norm": 0.33650925755500793,
      "learning_rate": 9.481105112734202e-05,
      "loss": 0.3626,
      "step": 822
    },
    {
      "epoch": 1.5653827864954826,
      "grad_norm": 0.39762622117996216,
      "learning_rate": 9.480469990473166e-05,
      "loss": 0.2862,
      "step": 823
    },
    {
      "epoch": 1.567284831193533,
      "grad_norm": 0.36138975620269775,
      "learning_rate": 9.479834868212131e-05,
      "loss": 0.2434,
      "step": 824
    },
    {
      "epoch": 1.5691868758915835,
      "grad_norm": 0.37878358364105225,
      "learning_rate": 9.479199745951096e-05,
      "loss": 0.2421,
      "step": 825
    },
    {
      "epoch": 1.571088920589634,
      "grad_norm": 0.4009093642234802,
      "learning_rate": 9.47856462369006e-05,
      "loss": 0.2561,
      "step": 826
    },
    {
      "epoch": 1.5729909652876843,
      "grad_norm": 0.3085389733314514,
      "learning_rate": 9.477929501429025e-05,
      "loss": 0.2293,
      "step": 827
    },
    {
      "epoch": 1.5748930099857348,
      "grad_norm": 0.48082223534584045,
      "learning_rate": 9.47729437916799e-05,
      "loss": 0.3193,
      "step": 828
    },
    {
      "epoch": 1.5767950546837852,
      "grad_norm": 0.42938464879989624,
      "learning_rate": 9.476659256906954e-05,
      "loss": 0.3319,
      "step": 829
    },
    {
      "epoch": 1.5786970993818354,
      "grad_norm": 0.32788941264152527,
      "learning_rate": 9.47602413464592e-05,
      "loss": 0.2432,
      "step": 830
    },
    {
      "epoch": 1.5805991440798859,
      "grad_norm": 0.38157737255096436,
      "learning_rate": 9.475389012384885e-05,
      "loss": 0.3165,
      "step": 831
    },
    {
      "epoch": 1.5825011887779363,
      "grad_norm": 0.38666632771492004,
      "learning_rate": 9.47475389012385e-05,
      "loss": 0.2554,
      "step": 832
    },
    {
      "epoch": 1.5844032334759865,
      "grad_norm": 0.3475115895271301,
      "learning_rate": 9.474118767862814e-05,
      "loss": 0.2679,
      "step": 833
    },
    {
      "epoch": 1.586305278174037,
      "grad_norm": 0.35684680938720703,
      "learning_rate": 9.473483645601779e-05,
      "loss": 0.2574,
      "step": 834
    },
    {
      "epoch": 1.5882073228720874,
      "grad_norm": 0.5205959677696228,
      "learning_rate": 9.472848523340744e-05,
      "loss": 0.3646,
      "step": 835
    },
    {
      "epoch": 1.5901093675701379,
      "grad_norm": 0.37549740076065063,
      "learning_rate": 9.472213401079708e-05,
      "loss": 0.2741,
      "step": 836
    },
    {
      "epoch": 1.5920114122681883,
      "grad_norm": 0.5251928567886353,
      "learning_rate": 9.471578278818673e-05,
      "loss": 0.3799,
      "step": 837
    },
    {
      "epoch": 1.5939134569662388,
      "grad_norm": 0.42622271180152893,
      "learning_rate": 9.470943156557638e-05,
      "loss": 0.2991,
      "step": 838
    },
    {
      "epoch": 1.5958155016642892,
      "grad_norm": 0.3737063407897949,
      "learning_rate": 9.470308034296602e-05,
      "loss": 0.288,
      "step": 839
    },
    {
      "epoch": 1.5977175463623396,
      "grad_norm": 0.4851538836956024,
      "learning_rate": 9.469672912035567e-05,
      "loss": 0.3293,
      "step": 840
    },
    {
      "epoch": 1.5996195910603899,
      "grad_norm": 0.3662918508052826,
      "learning_rate": 9.469037789774533e-05,
      "loss": 0.2338,
      "step": 841
    },
    {
      "epoch": 1.6015216357584403,
      "grad_norm": 0.3263486325740814,
      "learning_rate": 9.468402667513496e-05,
      "loss": 0.2228,
      "step": 842
    },
    {
      "epoch": 1.6034236804564908,
      "grad_norm": 0.4000779092311859,
      "learning_rate": 9.467767545252462e-05,
      "loss": 0.2635,
      "step": 843
    },
    {
      "epoch": 1.605325725154541,
      "grad_norm": 0.4274492859840393,
      "learning_rate": 9.467132422991425e-05,
      "loss": 0.3063,
      "step": 844
    },
    {
      "epoch": 1.6072277698525914,
      "grad_norm": 0.4486158490180969,
      "learning_rate": 9.466497300730392e-05,
      "loss": 0.3039,
      "step": 845
    },
    {
      "epoch": 1.6091298145506419,
      "grad_norm": 0.48109135031700134,
      "learning_rate": 9.465862178469356e-05,
      "loss": 0.3471,
      "step": 846
    },
    {
      "epoch": 1.6110318592486923,
      "grad_norm": 0.41299277544021606,
      "learning_rate": 9.46522705620832e-05,
      "loss": 0.2896,
      "step": 847
    },
    {
      "epoch": 1.6129339039467427,
      "grad_norm": 0.4177182614803314,
      "learning_rate": 9.464591933947286e-05,
      "loss": 0.2519,
      "step": 848
    },
    {
      "epoch": 1.6148359486447932,
      "grad_norm": 0.36468592286109924,
      "learning_rate": 9.46395681168625e-05,
      "loss": 0.275,
      "step": 849
    },
    {
      "epoch": 1.6167379933428436,
      "grad_norm": 0.33025646209716797,
      "learning_rate": 9.463321689425215e-05,
      "loss": 0.234,
      "step": 850
    },
    {
      "epoch": 1.618640038040894,
      "grad_norm": 0.4377218186855316,
      "learning_rate": 9.462686567164179e-05,
      "loss": 0.2939,
      "step": 851
    },
    {
      "epoch": 1.6205420827389445,
      "grad_norm": 0.34059834480285645,
      "learning_rate": 9.462051444903144e-05,
      "loss": 0.2559,
      "step": 852
    },
    {
      "epoch": 1.6224441274369947,
      "grad_norm": 0.36525094509124756,
      "learning_rate": 9.46141632264211e-05,
      "loss": 0.2638,
      "step": 853
    },
    {
      "epoch": 1.6243461721350452,
      "grad_norm": 0.344927042722702,
      "learning_rate": 9.460781200381073e-05,
      "loss": 0.1906,
      "step": 854
    },
    {
      "epoch": 1.6262482168330956,
      "grad_norm": 0.4097568988800049,
      "learning_rate": 9.460146078120038e-05,
      "loss": 0.3143,
      "step": 855
    },
    {
      "epoch": 1.6281502615311458,
      "grad_norm": 0.32290300726890564,
      "learning_rate": 9.459510955859004e-05,
      "loss": 0.2734,
      "step": 856
    },
    {
      "epoch": 1.6300523062291963,
      "grad_norm": 0.3865107595920563,
      "learning_rate": 9.458875833597967e-05,
      "loss": 0.3012,
      "step": 857
    },
    {
      "epoch": 1.6319543509272467,
      "grad_norm": 0.3034641444683075,
      "learning_rate": 9.458240711336933e-05,
      "loss": 0.2164,
      "step": 858
    },
    {
      "epoch": 1.6338563956252972,
      "grad_norm": 0.3896719217300415,
      "learning_rate": 9.457605589075898e-05,
      "loss": 0.2577,
      "step": 859
    },
    {
      "epoch": 1.6357584403233476,
      "grad_norm": 0.35619622468948364,
      "learning_rate": 9.456970466814862e-05,
      "loss": 0.3076,
      "step": 860
    },
    {
      "epoch": 1.637660485021398,
      "grad_norm": 0.39600345492362976,
      "learning_rate": 9.456335344553827e-05,
      "loss": 0.4003,
      "step": 861
    },
    {
      "epoch": 1.6395625297194485,
      "grad_norm": 0.3511577248573303,
      "learning_rate": 9.455700222292792e-05,
      "loss": 0.2603,
      "step": 862
    },
    {
      "epoch": 1.641464574417499,
      "grad_norm": 0.44329899549484253,
      "learning_rate": 9.455065100031757e-05,
      "loss": 0.2921,
      "step": 863
    },
    {
      "epoch": 1.6433666191155494,
      "grad_norm": 0.3798992931842804,
      "learning_rate": 9.454429977770721e-05,
      "loss": 0.2897,
      "step": 864
    },
    {
      "epoch": 1.6452686638135996,
      "grad_norm": 0.38711193203926086,
      "learning_rate": 9.453794855509686e-05,
      "loss": 0.2791,
      "step": 865
    },
    {
      "epoch": 1.64717070851165,
      "grad_norm": 0.3537624478340149,
      "learning_rate": 9.453159733248651e-05,
      "loss": 0.2207,
      "step": 866
    },
    {
      "epoch": 1.6490727532097005,
      "grad_norm": 0.350455641746521,
      "learning_rate": 9.452524610987615e-05,
      "loss": 0.2595,
      "step": 867
    },
    {
      "epoch": 1.6509747979077507,
      "grad_norm": 0.35781386494636536,
      "learning_rate": 9.45188948872658e-05,
      "loss": 0.2618,
      "step": 868
    },
    {
      "epoch": 1.6528768426058011,
      "grad_norm": 0.4823295772075653,
      "learning_rate": 9.451254366465546e-05,
      "loss": 0.3174,
      "step": 869
    },
    {
      "epoch": 1.6547788873038516,
      "grad_norm": 0.31698495149612427,
      "learning_rate": 9.45061924420451e-05,
      "loss": 0.2165,
      "step": 870
    },
    {
      "epoch": 1.656680932001902,
      "grad_norm": 0.4576948583126068,
      "learning_rate": 9.449984121943475e-05,
      "loss": 0.2937,
      "step": 871
    },
    {
      "epoch": 1.6585829766999525,
      "grad_norm": 0.4196888506412506,
      "learning_rate": 9.44934899968244e-05,
      "loss": 0.2876,
      "step": 872
    },
    {
      "epoch": 1.660485021398003,
      "grad_norm": 0.48588597774505615,
      "learning_rate": 9.448713877421405e-05,
      "loss": 0.3433,
      "step": 873
    },
    {
      "epoch": 1.6623870660960534,
      "grad_norm": 0.427946537733078,
      "learning_rate": 9.448078755160369e-05,
      "loss": 0.3184,
      "step": 874
    },
    {
      "epoch": 1.6642891107941038,
      "grad_norm": 0.4138951897621155,
      "learning_rate": 9.447443632899333e-05,
      "loss": 0.2738,
      "step": 875
    },
    {
      "epoch": 1.666191155492154,
      "grad_norm": 0.36560842394828796,
      "learning_rate": 9.446808510638299e-05,
      "loss": 0.3029,
      "step": 876
    },
    {
      "epoch": 1.6680932001902045,
      "grad_norm": 0.42942315340042114,
      "learning_rate": 9.446173388377263e-05,
      "loss": 0.2888,
      "step": 877
    },
    {
      "epoch": 1.669995244888255,
      "grad_norm": 0.21167854964733124,
      "learning_rate": 9.445538266116227e-05,
      "loss": 0.1919,
      "step": 878
    },
    {
      "epoch": 1.6718972895863051,
      "grad_norm": 0.41339564323425293,
      "learning_rate": 9.444903143855193e-05,
      "loss": 0.2482,
      "step": 879
    },
    {
      "epoch": 1.6737993342843556,
      "grad_norm": 0.47189727425575256,
      "learning_rate": 9.444268021594157e-05,
      "loss": 0.328,
      "step": 880
    },
    {
      "epoch": 1.675701378982406,
      "grad_norm": 0.32868659496307373,
      "learning_rate": 9.443632899333122e-05,
      "loss": 0.1985,
      "step": 881
    },
    {
      "epoch": 1.6776034236804565,
      "grad_norm": 0.3501724898815155,
      "learning_rate": 9.442997777072086e-05,
      "loss": 0.2733,
      "step": 882
    },
    {
      "epoch": 1.679505468378507,
      "grad_norm": 0.37144583463668823,
      "learning_rate": 9.442362654811051e-05,
      "loss": 0.2293,
      "step": 883
    },
    {
      "epoch": 1.6814075130765573,
      "grad_norm": 0.36318424344062805,
      "learning_rate": 9.441727532550017e-05,
      "loss": 0.3521,
      "step": 884
    },
    {
      "epoch": 1.6833095577746078,
      "grad_norm": 0.4295286238193512,
      "learning_rate": 9.44109241028898e-05,
      "loss": 0.3113,
      "step": 885
    },
    {
      "epoch": 1.6852116024726582,
      "grad_norm": 0.3312181830406189,
      "learning_rate": 9.440457288027947e-05,
      "loss": 0.2818,
      "step": 886
    },
    {
      "epoch": 1.6871136471707087,
      "grad_norm": 0.3743634819984436,
      "learning_rate": 9.439822165766911e-05,
      "loss": 0.245,
      "step": 887
    },
    {
      "epoch": 1.6890156918687589,
      "grad_norm": 0.5934861898422241,
      "learning_rate": 9.439187043505875e-05,
      "loss": 0.3654,
      "step": 888
    },
    {
      "epoch": 1.6909177365668093,
      "grad_norm": 0.4149317741394043,
      "learning_rate": 9.43855192124484e-05,
      "loss": 0.2584,
      "step": 889
    },
    {
      "epoch": 1.6928197812648598,
      "grad_norm": 0.40615764260292053,
      "learning_rate": 9.437916798983805e-05,
      "loss": 0.2986,
      "step": 890
    },
    {
      "epoch": 1.69472182596291,
      "grad_norm": 0.37536385655403137,
      "learning_rate": 9.43728167672277e-05,
      "loss": 0.2813,
      "step": 891
    },
    {
      "epoch": 1.6966238706609604,
      "grad_norm": 0.41415923833847046,
      "learning_rate": 9.436646554461734e-05,
      "loss": 0.3333,
      "step": 892
    },
    {
      "epoch": 1.6985259153590109,
      "grad_norm": 0.30747082829475403,
      "learning_rate": 9.436011432200699e-05,
      "loss": 0.2143,
      "step": 893
    },
    {
      "epoch": 1.7004279600570613,
      "grad_norm": 0.44593873620033264,
      "learning_rate": 9.435376309939664e-05,
      "loss": 0.2834,
      "step": 894
    },
    {
      "epoch": 1.7023300047551118,
      "grad_norm": 0.3417704403400421,
      "learning_rate": 9.434741187678628e-05,
      "loss": 0.2265,
      "step": 895
    },
    {
      "epoch": 1.7042320494531622,
      "grad_norm": 0.3436511754989624,
      "learning_rate": 9.434106065417593e-05,
      "loss": 0.249,
      "step": 896
    },
    {
      "epoch": 1.7061340941512126,
      "grad_norm": 0.4569544494152069,
      "learning_rate": 9.433470943156559e-05,
      "loss": 0.3271,
      "step": 897
    },
    {
      "epoch": 1.708036138849263,
      "grad_norm": 0.3883751630783081,
      "learning_rate": 9.432835820895522e-05,
      "loss": 0.2673,
      "step": 898
    },
    {
      "epoch": 1.7099381835473135,
      "grad_norm": 0.3915776014328003,
      "learning_rate": 9.432200698634488e-05,
      "loss": 0.2313,
      "step": 899
    },
    {
      "epoch": 1.7118402282453637,
      "grad_norm": 0.3450072407722473,
      "learning_rate": 9.431565576373453e-05,
      "loss": 0.2726,
      "step": 900
    },
    {
      "epoch": 1.7137422729434142,
      "grad_norm": 0.3894912004470825,
      "learning_rate": 9.430930454112417e-05,
      "loss": 0.2607,
      "step": 901
    },
    {
      "epoch": 1.7156443176414644,
      "grad_norm": 0.3509180545806885,
      "learning_rate": 9.430295331851382e-05,
      "loss": 0.2781,
      "step": 902
    },
    {
      "epoch": 1.7175463623395149,
      "grad_norm": 0.5164948105812073,
      "learning_rate": 9.429660209590347e-05,
      "loss": 0.3619,
      "step": 903
    },
    {
      "epoch": 1.7194484070375653,
      "grad_norm": 0.4074023962020874,
      "learning_rate": 9.429025087329312e-05,
      "loss": 0.3116,
      "step": 904
    },
    {
      "epoch": 1.7213504517356157,
      "grad_norm": 0.4034394323825836,
      "learning_rate": 9.428389965068276e-05,
      "loss": 0.3155,
      "step": 905
    },
    {
      "epoch": 1.7232524964336662,
      "grad_norm": 0.32292982935905457,
      "learning_rate": 9.427754842807241e-05,
      "loss": 0.2171,
      "step": 906
    },
    {
      "epoch": 1.7251545411317166,
      "grad_norm": 0.368856817483902,
      "learning_rate": 9.427119720546206e-05,
      "loss": 0.3021,
      "step": 907
    },
    {
      "epoch": 1.727056585829767,
      "grad_norm": 0.34953123331069946,
      "learning_rate": 9.42648459828517e-05,
      "loss": 0.2701,
      "step": 908
    },
    {
      "epoch": 1.7289586305278175,
      "grad_norm": 0.37510743737220764,
      "learning_rate": 9.425849476024135e-05,
      "loss": 0.3216,
      "step": 909
    },
    {
      "epoch": 1.730860675225868,
      "grad_norm": 0.31331393122673035,
      "learning_rate": 9.4252143537631e-05,
      "loss": 0.2855,
      "step": 910
    },
    {
      "epoch": 1.7327627199239182,
      "grad_norm": 0.3806105852127075,
      "learning_rate": 9.424579231502064e-05,
      "loss": 0.3216,
      "step": 911
    },
    {
      "epoch": 1.7346647646219686,
      "grad_norm": 0.3693408668041229,
      "learning_rate": 9.42394410924103e-05,
      "loss": 0.2473,
      "step": 912
    },
    {
      "epoch": 1.736566809320019,
      "grad_norm": 0.2931939959526062,
      "learning_rate": 9.423308986979993e-05,
      "loss": 0.1873,
      "step": 913
    },
    {
      "epoch": 1.7384688540180693,
      "grad_norm": 0.4330272972583771,
      "learning_rate": 9.422673864718959e-05,
      "loss": 0.3078,
      "step": 914
    },
    {
      "epoch": 1.7403708987161197,
      "grad_norm": 0.4881534278392792,
      "learning_rate": 9.422038742457924e-05,
      "loss": 0.3771,
      "step": 915
    },
    {
      "epoch": 1.7422729434141702,
      "grad_norm": 0.3158344328403473,
      "learning_rate": 9.421403620196888e-05,
      "loss": 0.2813,
      "step": 916
    },
    {
      "epoch": 1.7441749881122206,
      "grad_norm": 0.4482041299343109,
      "learning_rate": 9.420768497935854e-05,
      "loss": 0.3872,
      "step": 917
    },
    {
      "epoch": 1.746077032810271,
      "grad_norm": 0.3493407070636749,
      "learning_rate": 9.420133375674818e-05,
      "loss": 0.2284,
      "step": 918
    },
    {
      "epoch": 1.7479790775083215,
      "grad_norm": 0.3753608763217926,
      "learning_rate": 9.419498253413782e-05,
      "loss": 0.254,
      "step": 919
    },
    {
      "epoch": 1.749881122206372,
      "grad_norm": 0.4550943374633789,
      "learning_rate": 9.418863131152747e-05,
      "loss": 0.3073,
      "step": 920
    },
    {
      "epoch": 1.7517831669044224,
      "grad_norm": 0.3239607810974121,
      "learning_rate": 9.418228008891712e-05,
      "loss": 0.2087,
      "step": 921
    },
    {
      "epoch": 1.7536852116024728,
      "grad_norm": 0.4610382616519928,
      "learning_rate": 9.417592886630677e-05,
      "loss": 0.3104,
      "step": 922
    },
    {
      "epoch": 1.755587256300523,
      "grad_norm": 0.4382965862751007,
      "learning_rate": 9.416957764369641e-05,
      "loss": 0.2583,
      "step": 923
    },
    {
      "epoch": 1.7574893009985735,
      "grad_norm": 0.31299924850463867,
      "learning_rate": 9.416322642108606e-05,
      "loss": 0.2033,
      "step": 924
    },
    {
      "epoch": 1.759391345696624,
      "grad_norm": 0.33872106671333313,
      "learning_rate": 9.415687519847571e-05,
      "loss": 0.2366,
      "step": 925
    },
    {
      "epoch": 1.7612933903946741,
      "grad_norm": 0.33771976828575134,
      "learning_rate": 9.415052397586535e-05,
      "loss": 0.3062,
      "step": 926
    },
    {
      "epoch": 1.7631954350927246,
      "grad_norm": 0.32810178399086,
      "learning_rate": 9.4144172753255e-05,
      "loss": 0.2264,
      "step": 927
    },
    {
      "epoch": 1.765097479790775,
      "grad_norm": 0.41518697142601013,
      "learning_rate": 9.413782153064466e-05,
      "loss": 0.2747,
      "step": 928
    },
    {
      "epoch": 1.7669995244888255,
      "grad_norm": 0.43647775053977966,
      "learning_rate": 9.41314703080343e-05,
      "loss": 0.3439,
      "step": 929
    },
    {
      "epoch": 1.768901569186876,
      "grad_norm": 0.2905902564525604,
      "learning_rate": 9.412511908542395e-05,
      "loss": 0.2327,
      "step": 930
    },
    {
      "epoch": 1.7708036138849264,
      "grad_norm": 0.38527336716651917,
      "learning_rate": 9.41187678628136e-05,
      "loss": 0.264,
      "step": 931
    },
    {
      "epoch": 1.7727056585829768,
      "grad_norm": 0.4135185182094574,
      "learning_rate": 9.411241664020324e-05,
      "loss": 0.3075,
      "step": 932
    },
    {
      "epoch": 1.7746077032810272,
      "grad_norm": 0.30278775095939636,
      "learning_rate": 9.410606541759289e-05,
      "loss": 0.1831,
      "step": 933
    },
    {
      "epoch": 1.7765097479790775,
      "grad_norm": 0.3687085509300232,
      "learning_rate": 9.409971419498254e-05,
      "loss": 0.2862,
      "step": 934
    },
    {
      "epoch": 1.778411792677128,
      "grad_norm": 0.3217594623565674,
      "learning_rate": 9.409336297237219e-05,
      "loss": 0.1975,
      "step": 935
    },
    {
      "epoch": 1.7803138373751783,
      "grad_norm": 0.3583223223686218,
      "learning_rate": 9.408701174976183e-05,
      "loss": 0.2345,
      "step": 936
    },
    {
      "epoch": 1.7822158820732286,
      "grad_norm": 0.4119435250759125,
      "learning_rate": 9.408066052715148e-05,
      "loss": 0.2916,
      "step": 937
    },
    {
      "epoch": 1.784117926771279,
      "grad_norm": 0.400728315114975,
      "learning_rate": 9.407430930454113e-05,
      "loss": 0.4505,
      "step": 938
    },
    {
      "epoch": 1.7860199714693294,
      "grad_norm": 0.3988611698150635,
      "learning_rate": 9.406795808193077e-05,
      "loss": 0.286,
      "step": 939
    },
    {
      "epoch": 1.78792201616738,
      "grad_norm": 0.4544796347618103,
      "learning_rate": 9.406160685932042e-05,
      "loss": 0.3268,
      "step": 940
    },
    {
      "epoch": 1.7898240608654303,
      "grad_norm": 0.3785744905471802,
      "learning_rate": 9.405525563671008e-05,
      "loss": 0.2532,
      "step": 941
    },
    {
      "epoch": 1.7917261055634808,
      "grad_norm": 0.4459128975868225,
      "learning_rate": 9.404890441409971e-05,
      "loss": 0.3348,
      "step": 942
    },
    {
      "epoch": 1.7936281502615312,
      "grad_norm": 0.3253449499607086,
      "learning_rate": 9.404255319148937e-05,
      "loss": 0.1945,
      "step": 943
    },
    {
      "epoch": 1.7955301949595817,
      "grad_norm": 0.4977390468120575,
      "learning_rate": 9.403620196887902e-05,
      "loss": 0.3,
      "step": 944
    },
    {
      "epoch": 1.797432239657632,
      "grad_norm": 0.46191859245300293,
      "learning_rate": 9.402985074626867e-05,
      "loss": 0.3638,
      "step": 945
    },
    {
      "epoch": 1.7993342843556823,
      "grad_norm": 0.38492342829704285,
      "learning_rate": 9.402349952365831e-05,
      "loss": 0.2566,
      "step": 946
    },
    {
      "epoch": 1.8012363290537328,
      "grad_norm": 0.34863540530204773,
      "learning_rate": 9.401714830104795e-05,
      "loss": 0.2321,
      "step": 947
    },
    {
      "epoch": 1.8031383737517832,
      "grad_norm": 0.3839346766471863,
      "learning_rate": 9.401079707843761e-05,
      "loss": 0.2751,
      "step": 948
    },
    {
      "epoch": 1.8050404184498334,
      "grad_norm": 0.36121171712875366,
      "learning_rate": 9.400444585582725e-05,
      "loss": 0.2492,
      "step": 949
    },
    {
      "epoch": 1.8069424631478839,
      "grad_norm": 0.3479311466217041,
      "learning_rate": 9.399809463321689e-05,
      "loss": 0.2436,
      "step": 950
    },
    {
      "epoch": 1.8088445078459343,
      "grad_norm": 0.35279884934425354,
      "learning_rate": 9.399174341060655e-05,
      "loss": 0.2718,
      "step": 951
    },
    {
      "epoch": 1.8107465525439848,
      "grad_norm": 0.43152448534965515,
      "learning_rate": 9.398539218799619e-05,
      "loss": 0.2739,
      "step": 952
    },
    {
      "epoch": 1.8126485972420352,
      "grad_norm": 0.3631283938884735,
      "learning_rate": 9.397904096538584e-05,
      "loss": 0.2239,
      "step": 953
    },
    {
      "epoch": 1.8145506419400856,
      "grad_norm": 0.4698762595653534,
      "learning_rate": 9.397268974277548e-05,
      "loss": 0.3247,
      "step": 954
    },
    {
      "epoch": 1.816452686638136,
      "grad_norm": 0.36629432439804077,
      "learning_rate": 9.396633852016513e-05,
      "loss": 0.2778,
      "step": 955
    },
    {
      "epoch": 1.8183547313361865,
      "grad_norm": 0.34220409393310547,
      "learning_rate": 9.395998729755479e-05,
      "loss": 0.2466,
      "step": 956
    },
    {
      "epoch": 1.820256776034237,
      "grad_norm": 0.3768969178199768,
      "learning_rate": 9.395363607494442e-05,
      "loss": 0.334,
      "step": 957
    },
    {
      "epoch": 1.8221588207322872,
      "grad_norm": 0.2891027629375458,
      "learning_rate": 9.394728485233409e-05,
      "loss": 0.206,
      "step": 958
    },
    {
      "epoch": 1.8240608654303376,
      "grad_norm": 0.2802363634109497,
      "learning_rate": 9.394093362972373e-05,
      "loss": 0.2566,
      "step": 959
    },
    {
      "epoch": 1.825962910128388,
      "grad_norm": 0.38722601532936096,
      "learning_rate": 9.393458240711337e-05,
      "loss": 0.2615,
      "step": 960
    },
    {
      "epoch": 1.8278649548264383,
      "grad_norm": 0.45663881301879883,
      "learning_rate": 9.392823118450302e-05,
      "loss": 0.3521,
      "step": 961
    },
    {
      "epoch": 1.8297669995244887,
      "grad_norm": 0.36096152663230896,
      "learning_rate": 9.392187996189267e-05,
      "loss": 0.2429,
      "step": 962
    },
    {
      "epoch": 1.8316690442225392,
      "grad_norm": 0.3237638473510742,
      "learning_rate": 9.391552873928232e-05,
      "loss": 0.2874,
      "step": 963
    },
    {
      "epoch": 1.8335710889205896,
      "grad_norm": 0.379863440990448,
      "learning_rate": 9.390917751667196e-05,
      "loss": 0.2504,
      "step": 964
    },
    {
      "epoch": 1.83547313361864,
      "grad_norm": 0.40816691517829895,
      "learning_rate": 9.390282629406161e-05,
      "loss": 0.2614,
      "step": 965
    },
    {
      "epoch": 1.8373751783166905,
      "grad_norm": 0.38382720947265625,
      "learning_rate": 9.389647507145126e-05,
      "loss": 0.2282,
      "step": 966
    },
    {
      "epoch": 1.839277223014741,
      "grad_norm": 0.328861266374588,
      "learning_rate": 9.38901238488409e-05,
      "loss": 0.1763,
      "step": 967
    },
    {
      "epoch": 1.8411792677127914,
      "grad_norm": 0.3471934497356415,
      "learning_rate": 9.388377262623055e-05,
      "loss": 0.2348,
      "step": 968
    },
    {
      "epoch": 1.8430813124108416,
      "grad_norm": 0.44112637639045715,
      "learning_rate": 9.38774214036202e-05,
      "loss": 0.3496,
      "step": 969
    },
    {
      "epoch": 1.844983357108892,
      "grad_norm": 0.4357364773750305,
      "learning_rate": 9.387107018100984e-05,
      "loss": 0.2832,
      "step": 970
    },
    {
      "epoch": 1.8468854018069425,
      "grad_norm": 0.4502738118171692,
      "learning_rate": 9.38647189583995e-05,
      "loss": 0.2862,
      "step": 971
    },
    {
      "epoch": 1.8487874465049927,
      "grad_norm": 0.3577602505683899,
      "learning_rate": 9.385836773578915e-05,
      "loss": 0.2019,
      "step": 972
    },
    {
      "epoch": 1.8506894912030432,
      "grad_norm": 0.36250707507133484,
      "learning_rate": 9.385201651317879e-05,
      "loss": 0.2936,
      "step": 973
    },
    {
      "epoch": 1.8525915359010936,
      "grad_norm": 0.44027233123779297,
      "learning_rate": 9.384566529056844e-05,
      "loss": 0.3004,
      "step": 974
    },
    {
      "epoch": 1.854493580599144,
      "grad_norm": 0.4500497877597809,
      "learning_rate": 9.383931406795809e-05,
      "loss": 0.3,
      "step": 975
    },
    {
      "epoch": 1.8563956252971945,
      "grad_norm": 0.3777524530887604,
      "learning_rate": 9.383296284534774e-05,
      "loss": 0.2535,
      "step": 976
    },
    {
      "epoch": 1.858297669995245,
      "grad_norm": 0.3377416431903839,
      "learning_rate": 9.382661162273738e-05,
      "loss": 0.2767,
      "step": 977
    },
    {
      "epoch": 1.8601997146932954,
      "grad_norm": 0.34563374519348145,
      "learning_rate": 9.382026040012702e-05,
      "loss": 0.1923,
      "step": 978
    },
    {
      "epoch": 1.8621017593913458,
      "grad_norm": 0.3025479018688202,
      "learning_rate": 9.381390917751668e-05,
      "loss": 0.2214,
      "step": 979
    },
    {
      "epoch": 1.8640038040893963,
      "grad_norm": 0.3614577054977417,
      "learning_rate": 9.380755795490632e-05,
      "loss": 0.299,
      "step": 980
    },
    {
      "epoch": 1.8659058487874465,
      "grad_norm": 0.34508028626441956,
      "learning_rate": 9.380120673229597e-05,
      "loss": 0.2201,
      "step": 981
    },
    {
      "epoch": 1.867807893485497,
      "grad_norm": 0.33169567584991455,
      "learning_rate": 9.379485550968563e-05,
      "loss": 0.2298,
      "step": 982
    },
    {
      "epoch": 1.8697099381835474,
      "grad_norm": 0.4361656904220581,
      "learning_rate": 9.378850428707526e-05,
      "loss": 0.3109,
      "step": 983
    },
    {
      "epoch": 1.8716119828815976,
      "grad_norm": 0.3832654654979706,
      "learning_rate": 9.378215306446492e-05,
      "loss": 0.2877,
      "step": 984
    },
    {
      "epoch": 1.873514027579648,
      "grad_norm": 0.3991541862487793,
      "learning_rate": 9.377580184185455e-05,
      "loss": 0.2755,
      "step": 985
    },
    {
      "epoch": 1.8754160722776985,
      "grad_norm": 0.6057716012001038,
      "learning_rate": 9.37694506192442e-05,
      "loss": 0.3665,
      "step": 986
    },
    {
      "epoch": 1.877318116975749,
      "grad_norm": 0.2887308895587921,
      "learning_rate": 9.376309939663386e-05,
      "loss": 0.2414,
      "step": 987
    },
    {
      "epoch": 1.8792201616737993,
      "grad_norm": 0.28379005193710327,
      "learning_rate": 9.37567481740235e-05,
      "loss": 0.1895,
      "step": 988
    },
    {
      "epoch": 1.8811222063718498,
      "grad_norm": 0.36071258783340454,
      "learning_rate": 9.375039695141316e-05,
      "loss": 0.2855,
      "step": 989
    },
    {
      "epoch": 1.8830242510699002,
      "grad_norm": 0.3872823119163513,
      "learning_rate": 9.37440457288028e-05,
      "loss": 0.3112,
      "step": 990
    },
    {
      "epoch": 1.8849262957679507,
      "grad_norm": 0.3761101961135864,
      "learning_rate": 9.373769450619244e-05,
      "loss": 0.2291,
      "step": 991
    },
    {
      "epoch": 1.886828340466001,
      "grad_norm": 0.404000461101532,
      "learning_rate": 9.373134328358209e-05,
      "loss": 0.2349,
      "step": 992
    },
    {
      "epoch": 1.8887303851640513,
      "grad_norm": 0.4787864089012146,
      "learning_rate": 9.372499206097174e-05,
      "loss": 0.3447,
      "step": 993
    },
    {
      "epoch": 1.8906324298621018,
      "grad_norm": 0.4898964762687683,
      "learning_rate": 9.37186408383614e-05,
      "loss": 0.3306,
      "step": 994
    },
    {
      "epoch": 1.892534474560152,
      "grad_norm": 0.3915330767631531,
      "learning_rate": 9.371228961575103e-05,
      "loss": 0.2896,
      "step": 995
    },
    {
      "epoch": 1.8944365192582024,
      "grad_norm": 0.4643494486808777,
      "learning_rate": 9.370593839314068e-05,
      "loss": 0.3131,
      "step": 996
    },
    {
      "epoch": 1.8963385639562529,
      "grad_norm": 0.39880135655403137,
      "learning_rate": 9.369958717053034e-05,
      "loss": 0.2598,
      "step": 997
    },
    {
      "epoch": 1.8982406086543033,
      "grad_norm": 0.3153114318847656,
      "learning_rate": 9.369323594791997e-05,
      "loss": 0.2429,
      "step": 998
    },
    {
      "epoch": 1.9001426533523538,
      "grad_norm": 0.4997500479221344,
      "learning_rate": 9.368688472530963e-05,
      "loss": 0.4179,
      "step": 999
    },
    {
      "epoch": 1.9020446980504042,
      "grad_norm": 0.3919009566307068,
      "learning_rate": 9.368053350269928e-05,
      "loss": 0.2468,
      "step": 1000
    },
    {
      "epoch": 1.9039467427484547,
      "grad_norm": 0.48444265127182007,
      "learning_rate": 9.367418228008892e-05,
      "loss": 0.3191,
      "step": 1001
    },
    {
      "epoch": 1.905848787446505,
      "grad_norm": 0.38168856501579285,
      "learning_rate": 9.366783105747857e-05,
      "loss": 0.2658,
      "step": 1002
    },
    {
      "epoch": 1.9077508321445555,
      "grad_norm": 0.47058162093162537,
      "learning_rate": 9.366147983486822e-05,
      "loss": 0.3392,
      "step": 1003
    },
    {
      "epoch": 1.9096528768426058,
      "grad_norm": 0.40145471692085266,
      "learning_rate": 9.365512861225786e-05,
      "loss": 0.2619,
      "step": 1004
    },
    {
      "epoch": 1.9115549215406562,
      "grad_norm": 0.6980530619621277,
      "learning_rate": 9.364877738964751e-05,
      "loss": 0.3111,
      "step": 1005
    },
    {
      "epoch": 1.9134569662387066,
      "grad_norm": 0.35878410935401917,
      "learning_rate": 9.364242616703716e-05,
      "loss": 0.3026,
      "step": 1006
    },
    {
      "epoch": 1.9153590109367569,
      "grad_norm": 0.3291071653366089,
      "learning_rate": 9.363607494442681e-05,
      "loss": 0.2813,
      "step": 1007
    },
    {
      "epoch": 1.9172610556348073,
      "grad_norm": 0.4286592900753021,
      "learning_rate": 9.362972372181645e-05,
      "loss": 0.2921,
      "step": 1008
    },
    {
      "epoch": 1.9191631003328578,
      "grad_norm": 0.2965177893638611,
      "learning_rate": 9.36233724992061e-05,
      "loss": 0.2373,
      "step": 1009
    },
    {
      "epoch": 1.9210651450309082,
      "grad_norm": 0.3153838515281677,
      "learning_rate": 9.361702127659576e-05,
      "loss": 0.2195,
      "step": 1010
    },
    {
      "epoch": 1.9229671897289586,
      "grad_norm": 0.4827108085155487,
      "learning_rate": 9.36106700539854e-05,
      "loss": 0.3127,
      "step": 1011
    },
    {
      "epoch": 1.924869234427009,
      "grad_norm": 0.43089860677719116,
      "learning_rate": 9.360431883137505e-05,
      "loss": 0.2687,
      "step": 1012
    },
    {
      "epoch": 1.9267712791250595,
      "grad_norm": 0.43147915601730347,
      "learning_rate": 9.35979676087647e-05,
      "loss": 0.3953,
      "step": 1013
    },
    {
      "epoch": 1.92867332382311,
      "grad_norm": 0.37924453616142273,
      "learning_rate": 9.359161638615434e-05,
      "loss": 0.2522,
      "step": 1014
    },
    {
      "epoch": 1.9305753685211604,
      "grad_norm": 0.34664931893348694,
      "learning_rate": 9.358526516354399e-05,
      "loss": 0.2048,
      "step": 1015
    },
    {
      "epoch": 1.9324774132192106,
      "grad_norm": 0.2877664566040039,
      "learning_rate": 9.357891394093364e-05,
      "loss": 0.1794,
      "step": 1016
    },
    {
      "epoch": 1.934379457917261,
      "grad_norm": 0.4924784302711487,
      "learning_rate": 9.357256271832329e-05,
      "loss": 0.2737,
      "step": 1017
    },
    {
      "epoch": 1.9362815026153115,
      "grad_norm": 0.36828553676605225,
      "learning_rate": 9.356621149571293e-05,
      "loss": 0.2761,
      "step": 1018
    },
    {
      "epoch": 1.9381835473133617,
      "grad_norm": 0.355372816324234,
      "learning_rate": 9.355986027310257e-05,
      "loss": 0.2647,
      "step": 1019
    },
    {
      "epoch": 1.9400855920114122,
      "grad_norm": 0.37469297647476196,
      "learning_rate": 9.355350905049223e-05,
      "loss": 0.2347,
      "step": 1020
    },
    {
      "epoch": 1.9419876367094626,
      "grad_norm": 0.44890064001083374,
      "learning_rate": 9.354715782788187e-05,
      "loss": 0.2581,
      "step": 1021
    },
    {
      "epoch": 1.943889681407513,
      "grad_norm": 0.355234295129776,
      "learning_rate": 9.354080660527151e-05,
      "loss": 0.2467,
      "step": 1022
    },
    {
      "epoch": 1.9457917261055635,
      "grad_norm": 0.463871568441391,
      "learning_rate": 9.353445538266116e-05,
      "loss": 0.2338,
      "step": 1023
    },
    {
      "epoch": 1.947693770803614,
      "grad_norm": 0.38206830620765686,
      "learning_rate": 9.352810416005081e-05,
      "loss": 0.2353,
      "step": 1024
    },
    {
      "epoch": 1.9495958155016644,
      "grad_norm": 0.37627413868904114,
      "learning_rate": 9.352175293744047e-05,
      "loss": 0.2375,
      "step": 1025
    },
    {
      "epoch": 1.9514978601997148,
      "grad_norm": 0.4191925823688507,
      "learning_rate": 9.35154017148301e-05,
      "loss": 0.2444,
      "step": 1026
    },
    {
      "epoch": 1.953399904897765,
      "grad_norm": 0.41149812936782837,
      "learning_rate": 9.350905049221976e-05,
      "loss": 0.2905,
      "step": 1027
    },
    {
      "epoch": 1.9553019495958155,
      "grad_norm": 0.329313725233078,
      "learning_rate": 9.350269926960941e-05,
      "loss": 0.2293,
      "step": 1028
    },
    {
      "epoch": 1.957203994293866,
      "grad_norm": 0.4160427749156952,
      "learning_rate": 9.349634804699905e-05,
      "loss": 0.2512,
      "step": 1029
    },
    {
      "epoch": 1.9591060389919162,
      "grad_norm": 0.4005848467350006,
      "learning_rate": 9.34899968243887e-05,
      "loss": 0.2446,
      "step": 1030
    },
    {
      "epoch": 1.9610080836899666,
      "grad_norm": 0.4497627019882202,
      "learning_rate": 9.348364560177835e-05,
      "loss": 0.3265,
      "step": 1031
    },
    {
      "epoch": 1.962910128388017,
      "grad_norm": 0.4275449216365814,
      "learning_rate": 9.347729437916799e-05,
      "loss": 0.302,
      "step": 1032
    },
    {
      "epoch": 1.9648121730860675,
      "grad_norm": 0.33947649598121643,
      "learning_rate": 9.347094315655764e-05,
      "loss": 0.1903,
      "step": 1033
    },
    {
      "epoch": 1.966714217784118,
      "grad_norm": 0.38422051072120667,
      "learning_rate": 9.346459193394729e-05,
      "loss": 0.2595,
      "step": 1034
    },
    {
      "epoch": 1.9686162624821684,
      "grad_norm": 0.35371389985084534,
      "learning_rate": 9.345824071133694e-05,
      "loss": 0.2284,
      "step": 1035
    },
    {
      "epoch": 1.9705183071802188,
      "grad_norm": 0.38803884387016296,
      "learning_rate": 9.345188948872658e-05,
      "loss": 0.3021,
      "step": 1036
    },
    {
      "epoch": 1.9724203518782693,
      "grad_norm": 0.38203269243240356,
      "learning_rate": 9.344553826611623e-05,
      "loss": 0.2863,
      "step": 1037
    },
    {
      "epoch": 1.9743223965763197,
      "grad_norm": 0.3267860412597656,
      "learning_rate": 9.343918704350588e-05,
      "loss": 0.226,
      "step": 1038
    },
    {
      "epoch": 1.97622444127437,
      "grad_norm": 0.39556884765625,
      "learning_rate": 9.343283582089552e-05,
      "loss": 0.2727,
      "step": 1039
    },
    {
      "epoch": 1.9781264859724204,
      "grad_norm": 0.4278768301010132,
      "learning_rate": 9.342648459828517e-05,
      "loss": 0.2723,
      "step": 1040
    },
    {
      "epoch": 1.9800285306704708,
      "grad_norm": 0.37279701232910156,
      "learning_rate": 9.342013337567483e-05,
      "loss": 0.2685,
      "step": 1041
    },
    {
      "epoch": 1.981930575368521,
      "grad_norm": 0.4421425759792328,
      "learning_rate": 9.341378215306447e-05,
      "loss": 0.2793,
      "step": 1042
    },
    {
      "epoch": 1.9838326200665715,
      "grad_norm": 0.4341887831687927,
      "learning_rate": 9.340743093045412e-05,
      "loss": 0.2752,
      "step": 1043
    },
    {
      "epoch": 1.985734664764622,
      "grad_norm": 0.42935600876808167,
      "learning_rate": 9.340107970784377e-05,
      "loss": 0.3127,
      "step": 1044
    },
    {
      "epoch": 1.9876367094626723,
      "grad_norm": 0.29476839303970337,
      "learning_rate": 9.339472848523341e-05,
      "loss": 0.1855,
      "step": 1045
    },
    {
      "epoch": 1.9895387541607228,
      "grad_norm": 0.43286338448524475,
      "learning_rate": 9.338837726262306e-05,
      "loss": 0.3109,
      "step": 1046
    },
    {
      "epoch": 1.9914407988587732,
      "grad_norm": 0.35097062587738037,
      "learning_rate": 9.338202604001271e-05,
      "loss": 0.2178,
      "step": 1047
    },
    {
      "epoch": 1.9933428435568237,
      "grad_norm": 0.3497145175933838,
      "learning_rate": 9.337567481740236e-05,
      "loss": 0.2372,
      "step": 1048
    },
    {
      "epoch": 1.9952448882548741,
      "grad_norm": 0.4399060904979706,
      "learning_rate": 9.3369323594792e-05,
      "loss": 0.3065,
      "step": 1049
    },
    {
      "epoch": 1.9971469329529246,
      "grad_norm": 0.43642693758010864,
      "learning_rate": 9.336297237218164e-05,
      "loss": 0.3099,
      "step": 1050
    },
    {
      "epoch": 1.9990489776509748,
      "grad_norm": 0.42969372868537903,
      "learning_rate": 9.33566211495713e-05,
      "loss": 0.2899,
      "step": 1051
    },
    {
      "epoch": 2.000951022349025,
      "grad_norm": 0.324709951877594,
      "learning_rate": 9.335026992696094e-05,
      "loss": 0.1977,
      "step": 1052
    },
    {
      "epoch": 2.0028530670470754,
      "grad_norm": 0.2254759967327118,
      "learning_rate": 9.33439187043506e-05,
      "loss": 0.1513,
      "step": 1053
    },
    {
      "epoch": 2.004755111745126,
      "grad_norm": 0.29324305057525635,
      "learning_rate": 9.333756748174025e-05,
      "loss": 0.1739,
      "step": 1054
    },
    {
      "epoch": 2.0066571564431763,
      "grad_norm": 0.2934301495552063,
      "learning_rate": 9.333121625912988e-05,
      "loss": 0.1788,
      "step": 1055
    },
    {
      "epoch": 2.0085592011412268,
      "grad_norm": 0.3355758786201477,
      "learning_rate": 9.332486503651954e-05,
      "loss": 0.1829,
      "step": 1056
    },
    {
      "epoch": 2.010461245839277,
      "grad_norm": 0.4047424793243408,
      "learning_rate": 9.331851381390917e-05,
      "loss": 0.2256,
      "step": 1057
    },
    {
      "epoch": 2.0123632905373277,
      "grad_norm": 0.38155117630958557,
      "learning_rate": 9.331216259129883e-05,
      "loss": 0.1992,
      "step": 1058
    },
    {
      "epoch": 2.014265335235378,
      "grad_norm": 0.4122423827648163,
      "learning_rate": 9.330581136868848e-05,
      "loss": 0.2222,
      "step": 1059
    },
    {
      "epoch": 2.0161673799334285,
      "grad_norm": 0.4098420739173889,
      "learning_rate": 9.329946014607812e-05,
      "loss": 0.1495,
      "step": 1060
    },
    {
      "epoch": 2.018069424631479,
      "grad_norm": 0.37494683265686035,
      "learning_rate": 9.329310892346778e-05,
      "loss": 0.1955,
      "step": 1061
    },
    {
      "epoch": 2.0199714693295294,
      "grad_norm": 0.4210919439792633,
      "learning_rate": 9.328675770085742e-05,
      "loss": 0.1851,
      "step": 1062
    },
    {
      "epoch": 2.02187351402758,
      "grad_norm": 0.415770560503006,
      "learning_rate": 9.328040647824706e-05,
      "loss": 0.209,
      "step": 1063
    },
    {
      "epoch": 2.02377555872563,
      "grad_norm": 0.38957807421684265,
      "learning_rate": 9.327405525563671e-05,
      "loss": 0.1597,
      "step": 1064
    },
    {
      "epoch": 2.0256776034236803,
      "grad_norm": 0.3568849563598633,
      "learning_rate": 9.326770403302636e-05,
      "loss": 0.1564,
      "step": 1065
    },
    {
      "epoch": 2.0275796481217307,
      "grad_norm": 0.4151419699192047,
      "learning_rate": 9.326135281041601e-05,
      "loss": 0.2213,
      "step": 1066
    },
    {
      "epoch": 2.029481692819781,
      "grad_norm": 0.437418669462204,
      "learning_rate": 9.325500158780565e-05,
      "loss": 0.2091,
      "step": 1067
    },
    {
      "epoch": 2.0313837375178316,
      "grad_norm": 0.45977523922920227,
      "learning_rate": 9.32486503651953e-05,
      "loss": 0.2044,
      "step": 1068
    },
    {
      "epoch": 2.033285782215882,
      "grad_norm": 0.3634967803955078,
      "learning_rate": 9.324229914258496e-05,
      "loss": 0.1575,
      "step": 1069
    },
    {
      "epoch": 2.0351878269139325,
      "grad_norm": 0.4348776638507843,
      "learning_rate": 9.32359479199746e-05,
      "loss": 0.1892,
      "step": 1070
    },
    {
      "epoch": 2.037089871611983,
      "grad_norm": 0.39220520853996277,
      "learning_rate": 9.322959669736425e-05,
      "loss": 0.1962,
      "step": 1071
    },
    {
      "epoch": 2.0389919163100334,
      "grad_norm": 0.4379669725894928,
      "learning_rate": 9.32232454747539e-05,
      "loss": 0.2201,
      "step": 1072
    },
    {
      "epoch": 2.040893961008084,
      "grad_norm": 0.31880828738212585,
      "learning_rate": 9.321689425214354e-05,
      "loss": 0.1471,
      "step": 1073
    },
    {
      "epoch": 2.0427960057061343,
      "grad_norm": 0.31966346502304077,
      "learning_rate": 9.321054302953319e-05,
      "loss": 0.1688,
      "step": 1074
    },
    {
      "epoch": 2.0446980504041843,
      "grad_norm": 0.38291382789611816,
      "learning_rate": 9.320419180692284e-05,
      "loss": 0.1797,
      "step": 1075
    },
    {
      "epoch": 2.0466000951022347,
      "grad_norm": 0.3871828615665436,
      "learning_rate": 9.319784058431248e-05,
      "loss": 0.2201,
      "step": 1076
    },
    {
      "epoch": 2.048502139800285,
      "grad_norm": 0.35201162099838257,
      "learning_rate": 9.319148936170213e-05,
      "loss": 0.1759,
      "step": 1077
    },
    {
      "epoch": 2.0504041844983356,
      "grad_norm": 0.32999902963638306,
      "learning_rate": 9.318513813909178e-05,
      "loss": 0.1676,
      "step": 1078
    },
    {
      "epoch": 2.052306229196386,
      "grad_norm": 0.38137802481651306,
      "learning_rate": 9.317878691648143e-05,
      "loss": 0.181,
      "step": 1079
    },
    {
      "epoch": 2.0542082738944365,
      "grad_norm": 0.28507858514785767,
      "learning_rate": 9.317243569387107e-05,
      "loss": 0.1333,
      "step": 1080
    },
    {
      "epoch": 2.056110318592487,
      "grad_norm": 0.511489987373352,
      "learning_rate": 9.316608447126071e-05,
      "loss": 0.271,
      "step": 1081
    },
    {
      "epoch": 2.0580123632905374,
      "grad_norm": 0.37042170763015747,
      "learning_rate": 9.315973324865038e-05,
      "loss": 0.2733,
      "step": 1082
    },
    {
      "epoch": 2.059914407988588,
      "grad_norm": 0.3986508548259735,
      "learning_rate": 9.315338202604001e-05,
      "loss": 0.1964,
      "step": 1083
    },
    {
      "epoch": 2.0618164526866383,
      "grad_norm": 0.37804266810417175,
      "learning_rate": 9.314703080342967e-05,
      "loss": 0.1601,
      "step": 1084
    },
    {
      "epoch": 2.0637184973846887,
      "grad_norm": 0.32077136635780334,
      "learning_rate": 9.314067958081932e-05,
      "loss": 0.1462,
      "step": 1085
    },
    {
      "epoch": 2.065620542082739,
      "grad_norm": 0.2813294231891632,
      "learning_rate": 9.313432835820896e-05,
      "loss": 0.1321,
      "step": 1086
    },
    {
      "epoch": 2.067522586780789,
      "grad_norm": 0.40840163826942444,
      "learning_rate": 9.312797713559861e-05,
      "loss": 0.1892,
      "step": 1087
    },
    {
      "epoch": 2.0694246314788396,
      "grad_norm": 0.3264133334159851,
      "learning_rate": 9.312162591298825e-05,
      "loss": 0.1415,
      "step": 1088
    },
    {
      "epoch": 2.07132667617689,
      "grad_norm": 0.4274674952030182,
      "learning_rate": 9.311527469037791e-05,
      "loss": 0.1813,
      "step": 1089
    },
    {
      "epoch": 2.0732287208749405,
      "grad_norm": 0.37283292412757874,
      "learning_rate": 9.310892346776755e-05,
      "loss": 0.1753,
      "step": 1090
    },
    {
      "epoch": 2.075130765572991,
      "grad_norm": 0.32638901472091675,
      "learning_rate": 9.310257224515719e-05,
      "loss": 0.1731,
      "step": 1091
    },
    {
      "epoch": 2.0770328102710414,
      "grad_norm": 0.3295043408870697,
      "learning_rate": 9.309622102254685e-05,
      "loss": 0.1934,
      "step": 1092
    },
    {
      "epoch": 2.078934854969092,
      "grad_norm": 0.34605681896209717,
      "learning_rate": 9.308986979993649e-05,
      "loss": 0.2556,
      "step": 1093
    },
    {
      "epoch": 2.0808368996671422,
      "grad_norm": 0.35646018385887146,
      "learning_rate": 9.308351857732613e-05,
      "loss": 0.1508,
      "step": 1094
    },
    {
      "epoch": 2.0827389443651927,
      "grad_norm": 0.3224691152572632,
      "learning_rate": 9.307716735471578e-05,
      "loss": 0.1592,
      "step": 1095
    },
    {
      "epoch": 2.084640989063243,
      "grad_norm": 0.3692566156387329,
      "learning_rate": 9.307081613210543e-05,
      "loss": 0.1555,
      "step": 1096
    },
    {
      "epoch": 2.0865430337612936,
      "grad_norm": 0.46436119079589844,
      "learning_rate": 9.306446490949509e-05,
      "loss": 0.2176,
      "step": 1097
    },
    {
      "epoch": 2.088445078459344,
      "grad_norm": 0.3176686465740204,
      "learning_rate": 9.305811368688472e-05,
      "loss": 0.1763,
      "step": 1098
    },
    {
      "epoch": 2.090347123157394,
      "grad_norm": 0.29192522168159485,
      "learning_rate": 9.305176246427438e-05,
      "loss": 0.1485,
      "step": 1099
    },
    {
      "epoch": 2.0922491678554445,
      "grad_norm": 0.34905532002449036,
      "learning_rate": 9.304541124166403e-05,
      "loss": 0.1657,
      "step": 1100
    },
    {
      "epoch": 2.094151212553495,
      "grad_norm": 0.4198562800884247,
      "learning_rate": 9.303906001905367e-05,
      "loss": 0.2077,
      "step": 1101
    },
    {
      "epoch": 2.0960532572515453,
      "grad_norm": 0.35974305868148804,
      "learning_rate": 9.303270879644332e-05,
      "loss": 0.1776,
      "step": 1102
    },
    {
      "epoch": 2.097955301949596,
      "grad_norm": 0.35371047258377075,
      "learning_rate": 9.302635757383297e-05,
      "loss": 0.1887,
      "step": 1103
    },
    {
      "epoch": 2.0998573466476462,
      "grad_norm": 0.30068957805633545,
      "learning_rate": 9.302000635122261e-05,
      "loss": 0.14,
      "step": 1104
    },
    {
      "epoch": 2.1017593913456967,
      "grad_norm": 0.31092819571495056,
      "learning_rate": 9.301365512861226e-05,
      "loss": 0.1603,
      "step": 1105
    },
    {
      "epoch": 2.103661436043747,
      "grad_norm": 0.3615265190601349,
      "learning_rate": 9.300730390600191e-05,
      "loss": 0.1791,
      "step": 1106
    },
    {
      "epoch": 2.1055634807417976,
      "grad_norm": 0.2767830491065979,
      "learning_rate": 9.300095268339156e-05,
      "loss": 0.1243,
      "step": 1107
    },
    {
      "epoch": 2.107465525439848,
      "grad_norm": 0.36988285183906555,
      "learning_rate": 9.29946014607812e-05,
      "loss": 0.1619,
      "step": 1108
    },
    {
      "epoch": 2.1093675701378984,
      "grad_norm": 0.6014404892921448,
      "learning_rate": 9.298825023817085e-05,
      "loss": 0.2635,
      "step": 1109
    },
    {
      "epoch": 2.1112696148359484,
      "grad_norm": 0.3621249794960022,
      "learning_rate": 9.29818990155605e-05,
      "loss": 0.1749,
      "step": 1110
    },
    {
      "epoch": 2.113171659533999,
      "grad_norm": 0.2977392077445984,
      "learning_rate": 9.297554779295014e-05,
      "loss": 0.1582,
      "step": 1111
    },
    {
      "epoch": 2.1150737042320493,
      "grad_norm": 0.3253994286060333,
      "learning_rate": 9.29691965703398e-05,
      "loss": 0.1787,
      "step": 1112
    },
    {
      "epoch": 2.1169757489300998,
      "grad_norm": 0.34662213921546936,
      "learning_rate": 9.296284534772945e-05,
      "loss": 0.1923,
      "step": 1113
    },
    {
      "epoch": 2.11887779362815,
      "grad_norm": 0.416458398103714,
      "learning_rate": 9.295649412511909e-05,
      "loss": 0.1941,
      "step": 1114
    },
    {
      "epoch": 2.1207798383262007,
      "grad_norm": 0.36649563908576965,
      "learning_rate": 9.295014290250874e-05,
      "loss": 0.2233,
      "step": 1115
    },
    {
      "epoch": 2.122681883024251,
      "grad_norm": 0.3445313274860382,
      "learning_rate": 9.294379167989839e-05,
      "loss": 0.1701,
      "step": 1116
    },
    {
      "epoch": 2.1245839277223015,
      "grad_norm": 0.38747549057006836,
      "learning_rate": 9.293744045728803e-05,
      "loss": 0.1707,
      "step": 1117
    },
    {
      "epoch": 2.126485972420352,
      "grad_norm": 0.4027896225452423,
      "learning_rate": 9.293108923467768e-05,
      "loss": 0.2086,
      "step": 1118
    },
    {
      "epoch": 2.1283880171184024,
      "grad_norm": 0.3629845976829529,
      "learning_rate": 9.292473801206733e-05,
      "loss": 0.1743,
      "step": 1119
    },
    {
      "epoch": 2.130290061816453,
      "grad_norm": 0.39419326186180115,
      "learning_rate": 9.291838678945698e-05,
      "loss": 0.1907,
      "step": 1120
    },
    {
      "epoch": 2.132192106514503,
      "grad_norm": 0.36944523453712463,
      "learning_rate": 9.291203556684662e-05,
      "loss": 0.1631,
      "step": 1121
    },
    {
      "epoch": 2.1340941512125533,
      "grad_norm": 0.4214774966239929,
      "learning_rate": 9.290568434423626e-05,
      "loss": 0.2397,
      "step": 1122
    },
    {
      "epoch": 2.1359961959106037,
      "grad_norm": 0.3092084228992462,
      "learning_rate": 9.289933312162593e-05,
      "loss": 0.1396,
      "step": 1123
    },
    {
      "epoch": 2.137898240608654,
      "grad_norm": 0.3649998605251312,
      "learning_rate": 9.289298189901556e-05,
      "loss": 0.1677,
      "step": 1124
    },
    {
      "epoch": 2.1398002853067046,
      "grad_norm": 0.4131282567977905,
      "learning_rate": 9.288663067640522e-05,
      "loss": 0.2049,
      "step": 1125
    },
    {
      "epoch": 2.141702330004755,
      "grad_norm": 0.4324544668197632,
      "learning_rate": 9.288027945379485e-05,
      "loss": 0.1757,
      "step": 1126
    },
    {
      "epoch": 2.1436043747028055,
      "grad_norm": 0.4258798658847809,
      "learning_rate": 9.28739282311845e-05,
      "loss": 0.199,
      "step": 1127
    },
    {
      "epoch": 2.145506419400856,
      "grad_norm": 0.4244062602519989,
      "learning_rate": 9.286757700857416e-05,
      "loss": 0.2006,
      "step": 1128
    },
    {
      "epoch": 2.1474084640989064,
      "grad_norm": 0.4003104865550995,
      "learning_rate": 9.28612257859638e-05,
      "loss": 0.2098,
      "step": 1129
    },
    {
      "epoch": 2.149310508796957,
      "grad_norm": 0.36191633343696594,
      "learning_rate": 9.285487456335345e-05,
      "loss": 0.1821,
      "step": 1130
    },
    {
      "epoch": 2.1512125534950073,
      "grad_norm": 0.47675448656082153,
      "learning_rate": 9.28485233407431e-05,
      "loss": 0.2083,
      "step": 1131
    },
    {
      "epoch": 2.1531145981930577,
      "grad_norm": 0.4418546259403229,
      "learning_rate": 9.284217211813274e-05,
      "loss": 0.2228,
      "step": 1132
    },
    {
      "epoch": 2.155016642891108,
      "grad_norm": 0.31201982498168945,
      "learning_rate": 9.283582089552239e-05,
      "loss": 0.1326,
      "step": 1133
    },
    {
      "epoch": 2.156918687589158,
      "grad_norm": 0.30012449622154236,
      "learning_rate": 9.282946967291204e-05,
      "loss": 0.1376,
      "step": 1134
    },
    {
      "epoch": 2.1588207322872086,
      "grad_norm": 0.3705848455429077,
      "learning_rate": 9.282311845030168e-05,
      "loss": 0.1719,
      "step": 1135
    },
    {
      "epoch": 2.160722776985259,
      "grad_norm": 0.4028238356113434,
      "learning_rate": 9.281676722769133e-05,
      "loss": 0.178,
      "step": 1136
    },
    {
      "epoch": 2.1626248216833095,
      "grad_norm": 0.38973838090896606,
      "learning_rate": 9.281041600508098e-05,
      "loss": 0.1875,
      "step": 1137
    },
    {
      "epoch": 2.16452686638136,
      "grad_norm": 0.3756285309791565,
      "learning_rate": 9.280406478247064e-05,
      "loss": 0.1883,
      "step": 1138
    },
    {
      "epoch": 2.1664289110794104,
      "grad_norm": 0.2721819579601288,
      "learning_rate": 9.279771355986027e-05,
      "loss": 0.1468,
      "step": 1139
    },
    {
      "epoch": 2.168330955777461,
      "grad_norm": 0.34547916054725647,
      "learning_rate": 9.279136233724993e-05,
      "loss": 0.2043,
      "step": 1140
    },
    {
      "epoch": 2.1702330004755113,
      "grad_norm": 0.44819575548171997,
      "learning_rate": 9.278501111463958e-05,
      "loss": 0.2029,
      "step": 1141
    },
    {
      "epoch": 2.1721350451735617,
      "grad_norm": 0.36632853746414185,
      "learning_rate": 9.277865989202922e-05,
      "loss": 0.1884,
      "step": 1142
    },
    {
      "epoch": 2.174037089871612,
      "grad_norm": 0.37020185589790344,
      "learning_rate": 9.277230866941887e-05,
      "loss": 0.1819,
      "step": 1143
    },
    {
      "epoch": 2.1759391345696626,
      "grad_norm": 0.4174460470676422,
      "learning_rate": 9.276595744680852e-05,
      "loss": 0.1918,
      "step": 1144
    },
    {
      "epoch": 2.1778411792677126,
      "grad_norm": 0.4120714068412781,
      "learning_rate": 9.275960622419816e-05,
      "loss": 0.2496,
      "step": 1145
    },
    {
      "epoch": 2.179743223965763,
      "grad_norm": 0.4350152909755707,
      "learning_rate": 9.275325500158781e-05,
      "loss": 0.1981,
      "step": 1146
    },
    {
      "epoch": 2.1816452686638135,
      "grad_norm": 0.35637348890304565,
      "learning_rate": 9.274690377897746e-05,
      "loss": 0.1639,
      "step": 1147
    },
    {
      "epoch": 2.183547313361864,
      "grad_norm": 0.34323298931121826,
      "learning_rate": 9.27405525563671e-05,
      "loss": 0.1761,
      "step": 1148
    },
    {
      "epoch": 2.1854493580599144,
      "grad_norm": 0.30730780959129333,
      "learning_rate": 9.273420133375675e-05,
      "loss": 0.1623,
      "step": 1149
    },
    {
      "epoch": 2.187351402757965,
      "grad_norm": 0.32239773869514465,
      "learning_rate": 9.27278501111464e-05,
      "loss": 0.1238,
      "step": 1150
    },
    {
      "epoch": 2.1892534474560152,
      "grad_norm": 0.35441848635673523,
      "learning_rate": 9.272149888853606e-05,
      "loss": 0.1578,
      "step": 1151
    },
    {
      "epoch": 2.1911554921540657,
      "grad_norm": 0.33287835121154785,
      "learning_rate": 9.27151476659257e-05,
      "loss": 0.1726,
      "step": 1152
    },
    {
      "epoch": 2.193057536852116,
      "grad_norm": 0.3281983435153961,
      "learning_rate": 9.270879644331533e-05,
      "loss": 0.1435,
      "step": 1153
    },
    {
      "epoch": 2.1949595815501666,
      "grad_norm": 0.31831398606300354,
      "learning_rate": 9.2702445220705e-05,
      "loss": 0.1585,
      "step": 1154
    },
    {
      "epoch": 2.196861626248217,
      "grad_norm": 0.43460169434547424,
      "learning_rate": 9.269609399809464e-05,
      "loss": 0.2121,
      "step": 1155
    },
    {
      "epoch": 2.198763670946267,
      "grad_norm": 0.3470516502857208,
      "learning_rate": 9.268974277548429e-05,
      "loss": 0.157,
      "step": 1156
    },
    {
      "epoch": 2.2006657156443175,
      "grad_norm": 0.3971126079559326,
      "learning_rate": 9.268339155287394e-05,
      "loss": 0.1738,
      "step": 1157
    },
    {
      "epoch": 2.202567760342368,
      "grad_norm": 0.39526277780532837,
      "learning_rate": 9.267704033026358e-05,
      "loss": 0.2117,
      "step": 1158
    },
    {
      "epoch": 2.2044698050404183,
      "grad_norm": 0.31649425625801086,
      "learning_rate": 9.267068910765323e-05,
      "loss": 0.1966,
      "step": 1159
    },
    {
      "epoch": 2.206371849738469,
      "grad_norm": 0.4104944169521332,
      "learning_rate": 9.266433788504287e-05,
      "loss": 0.2178,
      "step": 1160
    },
    {
      "epoch": 2.2082738944365192,
      "grad_norm": 0.3751467168331146,
      "learning_rate": 9.265798666243253e-05,
      "loss": 0.1921,
      "step": 1161
    },
    {
      "epoch": 2.2101759391345697,
      "grad_norm": 0.3348170816898346,
      "learning_rate": 9.265163543982217e-05,
      "loss": 0.1533,
      "step": 1162
    },
    {
      "epoch": 2.21207798383262,
      "grad_norm": 0.39907872676849365,
      "learning_rate": 9.264528421721181e-05,
      "loss": 0.1733,
      "step": 1163
    },
    {
      "epoch": 2.2139800285306706,
      "grad_norm": 0.45442381501197815,
      "learning_rate": 9.263893299460147e-05,
      "loss": 0.2065,
      "step": 1164
    },
    {
      "epoch": 2.215882073228721,
      "grad_norm": 0.37475696206092834,
      "learning_rate": 9.263258177199111e-05,
      "loss": 0.1914,
      "step": 1165
    },
    {
      "epoch": 2.2177841179267714,
      "grad_norm": 0.3757840394973755,
      "learning_rate": 9.262623054938075e-05,
      "loss": 0.1781,
      "step": 1166
    },
    {
      "epoch": 2.219686162624822,
      "grad_norm": 0.3655502200126648,
      "learning_rate": 9.26198793267704e-05,
      "loss": 0.1814,
      "step": 1167
    },
    {
      "epoch": 2.2215882073228723,
      "grad_norm": 0.4219561219215393,
      "learning_rate": 9.261352810416006e-05,
      "loss": 0.213,
      "step": 1168
    },
    {
      "epoch": 2.2234902520209223,
      "grad_norm": 0.3741750419139862,
      "learning_rate": 9.260717688154971e-05,
      "loss": 0.1782,
      "step": 1169
    },
    {
      "epoch": 2.2253922967189728,
      "grad_norm": 0.37189987301826477,
      "learning_rate": 9.260082565893935e-05,
      "loss": 0.1783,
      "step": 1170
    },
    {
      "epoch": 2.227294341417023,
      "grad_norm": 0.2988317608833313,
      "learning_rate": 9.2594474436329e-05,
      "loss": 0.1481,
      "step": 1171
    },
    {
      "epoch": 2.2291963861150736,
      "grad_norm": 0.38000479340553284,
      "learning_rate": 9.258812321371865e-05,
      "loss": 0.1843,
      "step": 1172
    },
    {
      "epoch": 2.231098430813124,
      "grad_norm": 0.30989545583724976,
      "learning_rate": 9.258177199110829e-05,
      "loss": 0.1487,
      "step": 1173
    },
    {
      "epoch": 2.2330004755111745,
      "grad_norm": 0.27984580397605896,
      "learning_rate": 9.257542076849794e-05,
      "loss": 0.1445,
      "step": 1174
    },
    {
      "epoch": 2.234902520209225,
      "grad_norm": 0.3828918933868408,
      "learning_rate": 9.256906954588759e-05,
      "loss": 0.1709,
      "step": 1175
    },
    {
      "epoch": 2.2368045649072754,
      "grad_norm": 0.33677807450294495,
      "learning_rate": 9.256271832327723e-05,
      "loss": 0.1656,
      "step": 1176
    },
    {
      "epoch": 2.238706609605326,
      "grad_norm": 0.37769967317581177,
      "learning_rate": 9.255636710066688e-05,
      "loss": 0.2101,
      "step": 1177
    },
    {
      "epoch": 2.2406086543033763,
      "grad_norm": 0.3978733420372009,
      "learning_rate": 9.255001587805653e-05,
      "loss": 0.215,
      "step": 1178
    },
    {
      "epoch": 2.2425106990014267,
      "grad_norm": 0.3774537146091461,
      "learning_rate": 9.254366465544618e-05,
      "loss": 0.1778,
      "step": 1179
    },
    {
      "epoch": 2.2444127436994767,
      "grad_norm": 0.4117525815963745,
      "learning_rate": 9.253731343283582e-05,
      "loss": 0.1801,
      "step": 1180
    },
    {
      "epoch": 2.246314788397527,
      "grad_norm": 0.41460955142974854,
      "learning_rate": 9.253096221022547e-05,
      "loss": 0.1939,
      "step": 1181
    },
    {
      "epoch": 2.2482168330955776,
      "grad_norm": 0.41124284267425537,
      "learning_rate": 9.252461098761513e-05,
      "loss": 0.1944,
      "step": 1182
    },
    {
      "epoch": 2.250118877793628,
      "grad_norm": 0.39252787828445435,
      "learning_rate": 9.251825976500476e-05,
      "loss": 0.2037,
      "step": 1183
    },
    {
      "epoch": 2.2520209224916785,
      "grad_norm": 0.4118300676345825,
      "learning_rate": 9.25119085423944e-05,
      "loss": 0.2067,
      "step": 1184
    },
    {
      "epoch": 2.253922967189729,
      "grad_norm": 0.43823009729385376,
      "learning_rate": 9.250555731978407e-05,
      "loss": 0.2093,
      "step": 1185
    },
    {
      "epoch": 2.2558250118877794,
      "grad_norm": 0.41397175192832947,
      "learning_rate": 9.249920609717371e-05,
      "loss": 0.195,
      "step": 1186
    },
    {
      "epoch": 2.25772705658583,
      "grad_norm": 0.4286901652812958,
      "learning_rate": 9.249285487456336e-05,
      "loss": 0.1777,
      "step": 1187
    },
    {
      "epoch": 2.2596291012838803,
      "grad_norm": 0.373329758644104,
      "learning_rate": 9.248650365195301e-05,
      "loss": 0.1759,
      "step": 1188
    },
    {
      "epoch": 2.2615311459819307,
      "grad_norm": 0.4786781072616577,
      "learning_rate": 9.248015242934265e-05,
      "loss": 0.2509,
      "step": 1189
    },
    {
      "epoch": 2.263433190679981,
      "grad_norm": 0.41533464193344116,
      "learning_rate": 9.24738012067323e-05,
      "loss": 0.1595,
      "step": 1190
    },
    {
      "epoch": 2.265335235378031,
      "grad_norm": 0.37687090039253235,
      "learning_rate": 9.246744998412194e-05,
      "loss": 0.19,
      "step": 1191
    },
    {
      "epoch": 2.2672372800760816,
      "grad_norm": 0.3623497188091278,
      "learning_rate": 9.24610987615116e-05,
      "loss": 0.1723,
      "step": 1192
    },
    {
      "epoch": 2.269139324774132,
      "grad_norm": 0.378251850605011,
      "learning_rate": 9.245474753890124e-05,
      "loss": 0.1773,
      "step": 1193
    },
    {
      "epoch": 2.2710413694721825,
      "grad_norm": 0.3755147457122803,
      "learning_rate": 9.244839631629088e-05,
      "loss": 0.1685,
      "step": 1194
    },
    {
      "epoch": 2.272943414170233,
      "grad_norm": 0.5196719765663147,
      "learning_rate": 9.244204509368055e-05,
      "loss": 0.2665,
      "step": 1195
    },
    {
      "epoch": 2.2748454588682834,
      "grad_norm": 0.4404764473438263,
      "learning_rate": 9.243569387107018e-05,
      "loss": 0.1956,
      "step": 1196
    },
    {
      "epoch": 2.276747503566334,
      "grad_norm": 0.47750818729400635,
      "learning_rate": 9.242934264845984e-05,
      "loss": 0.2164,
      "step": 1197
    },
    {
      "epoch": 2.2786495482643843,
      "grad_norm": 0.3968189060688019,
      "learning_rate": 9.242299142584947e-05,
      "loss": 0.2299,
      "step": 1198
    },
    {
      "epoch": 2.2805515929624347,
      "grad_norm": 0.4168682396411896,
      "learning_rate": 9.241664020323913e-05,
      "loss": 0.1924,
      "step": 1199
    },
    {
      "epoch": 2.282453637660485,
      "grad_norm": 0.3767165541648865,
      "learning_rate": 9.241028898062878e-05,
      "loss": 0.1868,
      "step": 1200
    },
    {
      "epoch": 2.2843556823585356,
      "grad_norm": 0.37699073553085327,
      "learning_rate": 9.240393775801842e-05,
      "loss": 0.1968,
      "step": 1201
    },
    {
      "epoch": 2.2862577270565856,
      "grad_norm": 0.4355759620666504,
      "learning_rate": 9.239758653540807e-05,
      "loss": 0.1988,
      "step": 1202
    },
    {
      "epoch": 2.2881597717546365,
      "grad_norm": 0.42668578028678894,
      "learning_rate": 9.239123531279772e-05,
      "loss": 0.1988,
      "step": 1203
    },
    {
      "epoch": 2.2900618164526865,
      "grad_norm": 0.44233736395835876,
      "learning_rate": 9.238488409018736e-05,
      "loss": 0.2128,
      "step": 1204
    },
    {
      "epoch": 2.291963861150737,
      "grad_norm": 0.31429731845855713,
      "learning_rate": 9.237853286757701e-05,
      "loss": 0.1527,
      "step": 1205
    },
    {
      "epoch": 2.2938659058487874,
      "grad_norm": 0.38366618752479553,
      "learning_rate": 9.237218164496666e-05,
      "loss": 0.1747,
      "step": 1206
    },
    {
      "epoch": 2.295767950546838,
      "grad_norm": 0.3685773015022278,
      "learning_rate": 9.23658304223563e-05,
      "loss": 0.183,
      "step": 1207
    },
    {
      "epoch": 2.2976699952448882,
      "grad_norm": 0.349924772977829,
      "learning_rate": 9.235947919974595e-05,
      "loss": 0.1641,
      "step": 1208
    },
    {
      "epoch": 2.2995720399429387,
      "grad_norm": 0.3128054738044739,
      "learning_rate": 9.23531279771356e-05,
      "loss": 0.1682,
      "step": 1209
    },
    {
      "epoch": 2.301474084640989,
      "grad_norm": 0.4457269608974457,
      "learning_rate": 9.234677675452526e-05,
      "loss": 0.1888,
      "step": 1210
    },
    {
      "epoch": 2.3033761293390396,
      "grad_norm": 0.37438902258872986,
      "learning_rate": 9.23404255319149e-05,
      "loss": 0.1612,
      "step": 1211
    },
    {
      "epoch": 2.30527817403709,
      "grad_norm": 0.3830793499946594,
      "learning_rate": 9.233407430930455e-05,
      "loss": 0.1825,
      "step": 1212
    },
    {
      "epoch": 2.3071802187351405,
      "grad_norm": 0.4047216773033142,
      "learning_rate": 9.23277230866942e-05,
      "loss": 0.1874,
      "step": 1213
    },
    {
      "epoch": 2.309082263433191,
      "grad_norm": 0.400716096162796,
      "learning_rate": 9.232137186408384e-05,
      "loss": 0.165,
      "step": 1214
    },
    {
      "epoch": 2.310984308131241,
      "grad_norm": 0.35491228103637695,
      "learning_rate": 9.231502064147349e-05,
      "loss": 0.1428,
      "step": 1215
    },
    {
      "epoch": 2.3128863528292913,
      "grad_norm": 0.3040875494480133,
      "learning_rate": 9.230866941886314e-05,
      "loss": 0.1315,
      "step": 1216
    },
    {
      "epoch": 2.314788397527342,
      "grad_norm": 0.40058350563049316,
      "learning_rate": 9.230231819625278e-05,
      "loss": 0.2016,
      "step": 1217
    },
    {
      "epoch": 2.316690442225392,
      "grad_norm": 0.33165568113327026,
      "learning_rate": 9.229596697364243e-05,
      "loss": 0.1668,
      "step": 1218
    },
    {
      "epoch": 2.3185924869234427,
      "grad_norm": 0.29281625151634216,
      "learning_rate": 9.228961575103208e-05,
      "loss": 0.1577,
      "step": 1219
    },
    {
      "epoch": 2.320494531621493,
      "grad_norm": 0.4083446264266968,
      "learning_rate": 9.228326452842172e-05,
      "loss": 0.174,
      "step": 1220
    },
    {
      "epoch": 2.3223965763195435,
      "grad_norm": 0.3308553695678711,
      "learning_rate": 9.227691330581137e-05,
      "loss": 0.21,
      "step": 1221
    },
    {
      "epoch": 2.324298621017594,
      "grad_norm": 0.4102175831794739,
      "learning_rate": 9.227056208320102e-05,
      "loss": 0.205,
      "step": 1222
    },
    {
      "epoch": 2.3262006657156444,
      "grad_norm": 0.48705750703811646,
      "learning_rate": 9.226421086059068e-05,
      "loss": 0.2544,
      "step": 1223
    },
    {
      "epoch": 2.328102710413695,
      "grad_norm": 0.3305780291557312,
      "learning_rate": 9.225785963798031e-05,
      "loss": 0.1786,
      "step": 1224
    },
    {
      "epoch": 2.3300047551117453,
      "grad_norm": 0.3046979308128357,
      "learning_rate": 9.225150841536995e-05,
      "loss": 0.1325,
      "step": 1225
    },
    {
      "epoch": 2.3319067998097953,
      "grad_norm": 0.4403087794780731,
      "learning_rate": 9.224515719275962e-05,
      "loss": 0.2288,
      "step": 1226
    },
    {
      "epoch": 2.3338088445078458,
      "grad_norm": 0.3797864317893982,
      "learning_rate": 9.223880597014926e-05,
      "loss": 0.2068,
      "step": 1227
    },
    {
      "epoch": 2.335710889205896,
      "grad_norm": 0.34793582558631897,
      "learning_rate": 9.223245474753891e-05,
      "loss": 0.182,
      "step": 1228
    },
    {
      "epoch": 2.3376129339039466,
      "grad_norm": 0.30754920840263367,
      "learning_rate": 9.222610352492856e-05,
      "loss": 0.144,
      "step": 1229
    },
    {
      "epoch": 2.339514978601997,
      "grad_norm": 0.4364961087703705,
      "learning_rate": 9.22197523023182e-05,
      "loss": 0.1824,
      "step": 1230
    },
    {
      "epoch": 2.3414170233000475,
      "grad_norm": 0.3395443260669708,
      "learning_rate": 9.221340107970785e-05,
      "loss": 0.1691,
      "step": 1231
    },
    {
      "epoch": 2.343319067998098,
      "grad_norm": 0.34626251459121704,
      "learning_rate": 9.220704985709749e-05,
      "loss": 0.2285,
      "step": 1232
    },
    {
      "epoch": 2.3452211126961484,
      "grad_norm": 0.316518098115921,
      "learning_rate": 9.220069863448715e-05,
      "loss": 0.1469,
      "step": 1233
    },
    {
      "epoch": 2.347123157394199,
      "grad_norm": 0.38813212513923645,
      "learning_rate": 9.219434741187679e-05,
      "loss": 0.1907,
      "step": 1234
    },
    {
      "epoch": 2.3490252020922493,
      "grad_norm": 0.3442121744155884,
      "learning_rate": 9.218799618926643e-05,
      "loss": 0.1398,
      "step": 1235
    },
    {
      "epoch": 2.3509272467902997,
      "grad_norm": 0.3373865783214569,
      "learning_rate": 9.218164496665608e-05,
      "loss": 0.1477,
      "step": 1236
    },
    {
      "epoch": 2.3528292914883497,
      "grad_norm": 0.39781641960144043,
      "learning_rate": 9.217529374404573e-05,
      "loss": 0.1766,
      "step": 1237
    },
    {
      "epoch": 2.3547313361864006,
      "grad_norm": 0.25478801131248474,
      "learning_rate": 9.216894252143537e-05,
      "loss": 0.1301,
      "step": 1238
    },
    {
      "epoch": 2.3566333808844506,
      "grad_norm": 0.350087970495224,
      "learning_rate": 9.216259129882502e-05,
      "loss": 0.161,
      "step": 1239
    },
    {
      "epoch": 2.358535425582501,
      "grad_norm": 0.4105963408946991,
      "learning_rate": 9.215624007621468e-05,
      "loss": 0.1887,
      "step": 1240
    },
    {
      "epoch": 2.3604374702805515,
      "grad_norm": 0.4141649007797241,
      "learning_rate": 9.214988885360433e-05,
      "loss": 0.333,
      "step": 1241
    },
    {
      "epoch": 2.362339514978602,
      "grad_norm": 0.4416482448577881,
      "learning_rate": 9.214353763099397e-05,
      "loss": 0.2329,
      "step": 1242
    },
    {
      "epoch": 2.3642415596766524,
      "grad_norm": 0.4285755753517151,
      "learning_rate": 9.213718640838362e-05,
      "loss": 0.2194,
      "step": 1243
    },
    {
      "epoch": 2.366143604374703,
      "grad_norm": 0.33636924624443054,
      "learning_rate": 9.213083518577327e-05,
      "loss": 0.1853,
      "step": 1244
    },
    {
      "epoch": 2.3680456490727533,
      "grad_norm": 0.40267783403396606,
      "learning_rate": 9.212448396316291e-05,
      "loss": 0.1837,
      "step": 1245
    },
    {
      "epoch": 2.3699476937708037,
      "grad_norm": 0.3251781463623047,
      "learning_rate": 9.211813274055256e-05,
      "loss": 0.1853,
      "step": 1246
    },
    {
      "epoch": 2.371849738468854,
      "grad_norm": 0.3559510111808777,
      "learning_rate": 9.211178151794221e-05,
      "loss": 0.1735,
      "step": 1247
    },
    {
      "epoch": 2.3737517831669046,
      "grad_norm": 0.3483911454677582,
      "learning_rate": 9.210543029533185e-05,
      "loss": 0.156,
      "step": 1248
    },
    {
      "epoch": 2.375653827864955,
      "grad_norm": 0.4093637764453888,
      "learning_rate": 9.20990790727215e-05,
      "loss": 0.2013,
      "step": 1249
    },
    {
      "epoch": 2.377555872563005,
      "grad_norm": 0.38886240124702454,
      "learning_rate": 9.209272785011115e-05,
      "loss": 0.1723,
      "step": 1250
    },
    {
      "epoch": 2.3794579172610555,
      "grad_norm": 0.3627004325389862,
      "learning_rate": 9.20863766275008e-05,
      "loss": 0.1639,
      "step": 1251
    },
    {
      "epoch": 2.381359961959106,
      "grad_norm": 0.33721840381622314,
      "learning_rate": 9.208002540489044e-05,
      "loss": 0.1613,
      "step": 1252
    },
    {
      "epoch": 2.3832620066571564,
      "grad_norm": 0.4337291121482849,
      "learning_rate": 9.20736741822801e-05,
      "loss": 0.2036,
      "step": 1253
    },
    {
      "epoch": 2.385164051355207,
      "grad_norm": 0.43212467432022095,
      "learning_rate": 9.206732295966975e-05,
      "loss": 0.1925,
      "step": 1254
    },
    {
      "epoch": 2.3870660960532573,
      "grad_norm": 0.3450334966182709,
      "learning_rate": 9.206097173705939e-05,
      "loss": 0.1489,
      "step": 1255
    },
    {
      "epoch": 2.3889681407513077,
      "grad_norm": 0.36295151710510254,
      "learning_rate": 9.205462051444902e-05,
      "loss": 0.1801,
      "step": 1256
    },
    {
      "epoch": 2.390870185449358,
      "grad_norm": 0.469532310962677,
      "learning_rate": 9.204826929183869e-05,
      "loss": 0.2163,
      "step": 1257
    },
    {
      "epoch": 2.3927722301474086,
      "grad_norm": 0.4618028402328491,
      "learning_rate": 9.204191806922833e-05,
      "loss": 0.2175,
      "step": 1258
    },
    {
      "epoch": 2.394674274845459,
      "grad_norm": 0.3891139030456543,
      "learning_rate": 9.203556684661798e-05,
      "loss": 0.1585,
      "step": 1259
    },
    {
      "epoch": 2.3965763195435095,
      "grad_norm": 0.4574741721153259,
      "learning_rate": 9.202921562400763e-05,
      "loss": 0.2545,
      "step": 1260
    },
    {
      "epoch": 2.3984783642415595,
      "grad_norm": 0.49759337306022644,
      "learning_rate": 9.202286440139727e-05,
      "loss": 0.2208,
      "step": 1261
    },
    {
      "epoch": 2.40038040893961,
      "grad_norm": 0.3180585503578186,
      "learning_rate": 9.201651317878692e-05,
      "loss": 0.157,
      "step": 1262
    },
    {
      "epoch": 2.4022824536376604,
      "grad_norm": 0.3678848147392273,
      "learning_rate": 9.201016195617656e-05,
      "loss": 0.1891,
      "step": 1263
    },
    {
      "epoch": 2.404184498335711,
      "grad_norm": 0.3016449809074402,
      "learning_rate": 9.200381073356623e-05,
      "loss": 0.1295,
      "step": 1264
    },
    {
      "epoch": 2.4060865430337612,
      "grad_norm": 0.522779643535614,
      "learning_rate": 9.199745951095586e-05,
      "loss": 0.2814,
      "step": 1265
    },
    {
      "epoch": 2.4079885877318117,
      "grad_norm": 0.45210519433021545,
      "learning_rate": 9.19911082883455e-05,
      "loss": 0.234,
      "step": 1266
    },
    {
      "epoch": 2.409890632429862,
      "grad_norm": 0.3812367022037506,
      "learning_rate": 9.198475706573517e-05,
      "loss": 0.2104,
      "step": 1267
    },
    {
      "epoch": 2.4117926771279126,
      "grad_norm": 0.3120013177394867,
      "learning_rate": 9.19784058431248e-05,
      "loss": 0.1511,
      "step": 1268
    },
    {
      "epoch": 2.413694721825963,
      "grad_norm": 0.34164851903915405,
      "learning_rate": 9.197205462051446e-05,
      "loss": 0.1607,
      "step": 1269
    },
    {
      "epoch": 2.4155967665240135,
      "grad_norm": 0.3127415180206299,
      "learning_rate": 9.19657033979041e-05,
      "loss": 0.143,
      "step": 1270
    },
    {
      "epoch": 2.417498811222064,
      "grad_norm": 0.4628545641899109,
      "learning_rate": 9.195935217529375e-05,
      "loss": 0.2187,
      "step": 1271
    },
    {
      "epoch": 2.419400855920114,
      "grad_norm": 0.3645714223384857,
      "learning_rate": 9.19530009526834e-05,
      "loss": 0.1648,
      "step": 1272
    },
    {
      "epoch": 2.4213029006181643,
      "grad_norm": 0.41127142310142517,
      "learning_rate": 9.194664973007304e-05,
      "loss": 0.1712,
      "step": 1273
    },
    {
      "epoch": 2.4232049453162148,
      "grad_norm": 0.48663556575775146,
      "learning_rate": 9.194029850746269e-05,
      "loss": 0.2713,
      "step": 1274
    },
    {
      "epoch": 2.425106990014265,
      "grad_norm": 0.3965604305267334,
      "learning_rate": 9.193394728485234e-05,
      "loss": 0.1766,
      "step": 1275
    },
    {
      "epoch": 2.4270090347123157,
      "grad_norm": 0.4565601646900177,
      "learning_rate": 9.192759606224198e-05,
      "loss": 0.1827,
      "step": 1276
    },
    {
      "epoch": 2.428911079410366,
      "grad_norm": 0.4272227883338928,
      "learning_rate": 9.192124483963163e-05,
      "loss": 0.1874,
      "step": 1277
    },
    {
      "epoch": 2.4308131241084165,
      "grad_norm": 0.42560452222824097,
      "learning_rate": 9.191489361702128e-05,
      "loss": 0.1829,
      "step": 1278
    },
    {
      "epoch": 2.432715168806467,
      "grad_norm": 0.30827009677886963,
      "learning_rate": 9.190854239441092e-05,
      "loss": 0.1747,
      "step": 1279
    },
    {
      "epoch": 2.4346172135045174,
      "grad_norm": 0.3780437707901001,
      "learning_rate": 9.190219117180057e-05,
      "loss": 0.1955,
      "step": 1280
    },
    {
      "epoch": 2.436519258202568,
      "grad_norm": 0.32639580965042114,
      "learning_rate": 9.189583994919023e-05,
      "loss": 0.1568,
      "step": 1281
    },
    {
      "epoch": 2.4384213029006183,
      "grad_norm": 0.37228289246559143,
      "learning_rate": 9.188948872657988e-05,
      "loss": 0.1871,
      "step": 1282
    },
    {
      "epoch": 2.4403233475986688,
      "grad_norm": 0.4045466482639313,
      "learning_rate": 9.188313750396952e-05,
      "loss": 0.2237,
      "step": 1283
    },
    {
      "epoch": 2.442225392296719,
      "grad_norm": 0.40609246492385864,
      "learning_rate": 9.187678628135917e-05,
      "loss": 0.2313,
      "step": 1284
    },
    {
      "epoch": 2.444127436994769,
      "grad_norm": 0.36473485827445984,
      "learning_rate": 9.187043505874882e-05,
      "loss": 0.2528,
      "step": 1285
    },
    {
      "epoch": 2.4460294816928196,
      "grad_norm": 0.4154009222984314,
      "learning_rate": 9.186408383613846e-05,
      "loss": 0.215,
      "step": 1286
    },
    {
      "epoch": 2.44793152639087,
      "grad_norm": 0.33488062024116516,
      "learning_rate": 9.185773261352811e-05,
      "loss": 0.1666,
      "step": 1287
    },
    {
      "epoch": 2.4498335710889205,
      "grad_norm": 0.392004132270813,
      "learning_rate": 9.185138139091776e-05,
      "loss": 0.2127,
      "step": 1288
    },
    {
      "epoch": 2.451735615786971,
      "grad_norm": 0.32925739884376526,
      "learning_rate": 9.18450301683074e-05,
      "loss": 0.1459,
      "step": 1289
    },
    {
      "epoch": 2.4536376604850214,
      "grad_norm": 0.3380909264087677,
      "learning_rate": 9.183867894569705e-05,
      "loss": 0.1482,
      "step": 1290
    },
    {
      "epoch": 2.455539705183072,
      "grad_norm": 0.47436705231666565,
      "learning_rate": 9.18323277230867e-05,
      "loss": 0.2652,
      "step": 1291
    },
    {
      "epoch": 2.4574417498811223,
      "grad_norm": 0.39543116092681885,
      "learning_rate": 9.182597650047634e-05,
      "loss": 0.1762,
      "step": 1292
    },
    {
      "epoch": 2.4593437945791727,
      "grad_norm": 0.4776802659034729,
      "learning_rate": 9.181962527786599e-05,
      "loss": 0.1967,
      "step": 1293
    },
    {
      "epoch": 2.461245839277223,
      "grad_norm": 0.37519994378089905,
      "learning_rate": 9.181327405525563e-05,
      "loss": 0.1909,
      "step": 1294
    },
    {
      "epoch": 2.4631478839752736,
      "grad_norm": 0.37666913866996765,
      "learning_rate": 9.18069228326453e-05,
      "loss": 0.1477,
      "step": 1295
    },
    {
      "epoch": 2.4650499286733236,
      "grad_norm": 0.3830261528491974,
      "learning_rate": 9.180057161003494e-05,
      "loss": 0.1825,
      "step": 1296
    },
    {
      "epoch": 2.466951973371374,
      "grad_norm": 0.4064732789993286,
      "learning_rate": 9.179422038742457e-05,
      "loss": 0.2,
      "step": 1297
    },
    {
      "epoch": 2.4688540180694245,
      "grad_norm": 0.318314790725708,
      "learning_rate": 9.178786916481424e-05,
      "loss": 0.1543,
      "step": 1298
    },
    {
      "epoch": 2.470756062767475,
      "grad_norm": 0.3804973065853119,
      "learning_rate": 9.178151794220388e-05,
      "loss": 0.2248,
      "step": 1299
    },
    {
      "epoch": 2.4726581074655254,
      "grad_norm": 0.4222256541252136,
      "learning_rate": 9.177516671959353e-05,
      "loss": 0.2037,
      "step": 1300
    },
    {
      "epoch": 2.474560152163576,
      "grad_norm": 0.4317629337310791,
      "learning_rate": 9.176881549698317e-05,
      "loss": 0.1914,
      "step": 1301
    },
    {
      "epoch": 2.4764621968616263,
      "grad_norm": 0.4674796760082245,
      "learning_rate": 9.176246427437282e-05,
      "loss": 0.212,
      "step": 1302
    },
    {
      "epoch": 2.4783642415596767,
      "grad_norm": 0.40157684683799744,
      "learning_rate": 9.175611305176247e-05,
      "loss": 0.1948,
      "step": 1303
    },
    {
      "epoch": 2.480266286257727,
      "grad_norm": 0.37824416160583496,
      "learning_rate": 9.174976182915211e-05,
      "loss": 0.1849,
      "step": 1304
    },
    {
      "epoch": 2.4821683309557776,
      "grad_norm": 0.5870863199234009,
      "learning_rate": 9.174341060654177e-05,
      "loss": 0.1586,
      "step": 1305
    },
    {
      "epoch": 2.484070375653828,
      "grad_norm": 0.3794877529144287,
      "learning_rate": 9.173705938393141e-05,
      "loss": 0.2162,
      "step": 1306
    },
    {
      "epoch": 2.485972420351878,
      "grad_norm": 0.40509578585624695,
      "learning_rate": 9.173070816132105e-05,
      "loss": 0.1895,
      "step": 1307
    },
    {
      "epoch": 2.4878744650499285,
      "grad_norm": 0.37314295768737793,
      "learning_rate": 9.17243569387107e-05,
      "loss": 0.1926,
      "step": 1308
    },
    {
      "epoch": 2.489776509747979,
      "grad_norm": 0.32264095544815063,
      "learning_rate": 9.171800571610035e-05,
      "loss": 0.1385,
      "step": 1309
    },
    {
      "epoch": 2.4916785544460294,
      "grad_norm": 0.43269702792167664,
      "learning_rate": 9.171165449348999e-05,
      "loss": 0.2189,
      "step": 1310
    },
    {
      "epoch": 2.49358059914408,
      "grad_norm": 0.330098956823349,
      "learning_rate": 9.170530327087964e-05,
      "loss": 0.168,
      "step": 1311
    },
    {
      "epoch": 2.4954826438421303,
      "grad_norm": 0.2726501524448395,
      "learning_rate": 9.16989520482693e-05,
      "loss": 0.1306,
      "step": 1312
    },
    {
      "epoch": 2.4973846885401807,
      "grad_norm": 0.27615344524383545,
      "learning_rate": 9.169260082565895e-05,
      "loss": 0.1361,
      "step": 1313
    },
    {
      "epoch": 2.499286733238231,
      "grad_norm": 0.3685866594314575,
      "learning_rate": 9.168624960304859e-05,
      "loss": 0.1901,
      "step": 1314
    },
    {
      "epoch": 2.5011887779362816,
      "grad_norm": 0.323897123336792,
      "learning_rate": 9.167989838043824e-05,
      "loss": 0.2608,
      "step": 1315
    },
    {
      "epoch": 2.503090822634332,
      "grad_norm": 0.6715079545974731,
      "learning_rate": 9.167354715782789e-05,
      "loss": 0.199,
      "step": 1316
    },
    {
      "epoch": 2.5049928673323825,
      "grad_norm": 0.32039186358451843,
      "learning_rate": 9.166719593521753e-05,
      "loss": 0.1723,
      "step": 1317
    },
    {
      "epoch": 2.5068949120304325,
      "grad_norm": 0.3974270224571228,
      "learning_rate": 9.166084471260718e-05,
      "loss": 0.1659,
      "step": 1318
    },
    {
      "epoch": 2.5087969567284834,
      "grad_norm": 0.3953278362751007,
      "learning_rate": 9.165449348999683e-05,
      "loss": 0.1879,
      "step": 1319
    },
    {
      "epoch": 2.5106990014265333,
      "grad_norm": 0.4061002731323242,
      "learning_rate": 9.164814226738647e-05,
      "loss": 0.1858,
      "step": 1320
    },
    {
      "epoch": 2.512601046124584,
      "grad_norm": 0.3816406726837158,
      "learning_rate": 9.164179104477612e-05,
      "loss": 0.1899,
      "step": 1321
    },
    {
      "epoch": 2.5145030908226342,
      "grad_norm": 0.3856441378593445,
      "learning_rate": 9.163543982216577e-05,
      "loss": 0.1727,
      "step": 1322
    },
    {
      "epoch": 2.5164051355206847,
      "grad_norm": 0.47267359495162964,
      "learning_rate": 9.162908859955543e-05,
      "loss": 0.2137,
      "step": 1323
    },
    {
      "epoch": 2.518307180218735,
      "grad_norm": 0.41764524579048157,
      "learning_rate": 9.162273737694506e-05,
      "loss": 0.2138,
      "step": 1324
    },
    {
      "epoch": 2.5202092249167856,
      "grad_norm": 0.42864158749580383,
      "learning_rate": 9.161638615433472e-05,
      "loss": 0.1919,
      "step": 1325
    },
    {
      "epoch": 2.522111269614836,
      "grad_norm": 0.5067504048347473,
      "learning_rate": 9.161003493172437e-05,
      "loss": 0.2068,
      "step": 1326
    },
    {
      "epoch": 2.5240133143128864,
      "grad_norm": 0.430951863527298,
      "learning_rate": 9.1603683709114e-05,
      "loss": 0.2195,
      "step": 1327
    },
    {
      "epoch": 2.525915359010937,
      "grad_norm": 0.37973999977111816,
      "learning_rate": 9.159733248650364e-05,
      "loss": 0.1799,
      "step": 1328
    },
    {
      "epoch": 2.527817403708987,
      "grad_norm": 0.362768292427063,
      "learning_rate": 9.159098126389331e-05,
      "loss": 0.1555,
      "step": 1329
    },
    {
      "epoch": 2.5297194484070378,
      "grad_norm": 0.41433513164520264,
      "learning_rate": 9.158463004128295e-05,
      "loss": 0.1958,
      "step": 1330
    },
    {
      "epoch": 2.5316214931050878,
      "grad_norm": 0.3091717064380646,
      "learning_rate": 9.15782788186726e-05,
      "loss": 0.1622,
      "step": 1331
    },
    {
      "epoch": 2.533523537803138,
      "grad_norm": 0.35242778062820435,
      "learning_rate": 9.157192759606225e-05,
      "loss": 0.1627,
      "step": 1332
    },
    {
      "epoch": 2.5354255825011887,
      "grad_norm": 0.38102760910987854,
      "learning_rate": 9.156557637345189e-05,
      "loss": 0.1663,
      "step": 1333
    },
    {
      "epoch": 2.537327627199239,
      "grad_norm": 0.4313855469226837,
      "learning_rate": 9.155922515084154e-05,
      "loss": 0.208,
      "step": 1334
    },
    {
      "epoch": 2.5392296718972895,
      "grad_norm": 0.33921730518341064,
      "learning_rate": 9.155287392823118e-05,
      "loss": 0.1572,
      "step": 1335
    },
    {
      "epoch": 2.54113171659534,
      "grad_norm": 0.3824930489063263,
      "learning_rate": 9.154652270562085e-05,
      "loss": 0.1986,
      "step": 1336
    },
    {
      "epoch": 2.5430337612933904,
      "grad_norm": 0.33059945702552795,
      "learning_rate": 9.154017148301048e-05,
      "loss": 0.156,
      "step": 1337
    },
    {
      "epoch": 2.544935805991441,
      "grad_norm": 0.4880346357822418,
      "learning_rate": 9.153382026040012e-05,
      "loss": 0.2319,
      "step": 1338
    },
    {
      "epoch": 2.5468378506894913,
      "grad_norm": 0.27151229977607727,
      "learning_rate": 9.152746903778979e-05,
      "loss": 0.128,
      "step": 1339
    },
    {
      "epoch": 2.5487398953875418,
      "grad_norm": 0.35515275597572327,
      "learning_rate": 9.152111781517943e-05,
      "loss": 0.1685,
      "step": 1340
    },
    {
      "epoch": 2.550641940085592,
      "grad_norm": 0.41455206274986267,
      "learning_rate": 9.151476659256908e-05,
      "loss": 0.2354,
      "step": 1341
    },
    {
      "epoch": 2.552543984783642,
      "grad_norm": 0.3215075731277466,
      "learning_rate": 9.150841536995872e-05,
      "loss": 0.1653,
      "step": 1342
    },
    {
      "epoch": 2.554446029481693,
      "grad_norm": 0.34158623218536377,
      "learning_rate": 9.150206414734837e-05,
      "loss": 0.1598,
      "step": 1343
    },
    {
      "epoch": 2.556348074179743,
      "grad_norm": 0.4195705056190491,
      "learning_rate": 9.149571292473802e-05,
      "loss": 0.228,
      "step": 1344
    },
    {
      "epoch": 2.5582501188777935,
      "grad_norm": 0.34753212332725525,
      "learning_rate": 9.148936170212766e-05,
      "loss": 0.1948,
      "step": 1345
    },
    {
      "epoch": 2.560152163575844,
      "grad_norm": 0.43792131543159485,
      "learning_rate": 9.148301047951731e-05,
      "loss": 0.2191,
      "step": 1346
    },
    {
      "epoch": 2.5620542082738944,
      "grad_norm": 0.35464513301849365,
      "learning_rate": 9.147665925690696e-05,
      "loss": 0.1555,
      "step": 1347
    },
    {
      "epoch": 2.563956252971945,
      "grad_norm": 0.50618976354599,
      "learning_rate": 9.14703080342966e-05,
      "loss": 0.2262,
      "step": 1348
    },
    {
      "epoch": 2.5658582976699953,
      "grad_norm": 0.3603616952896118,
      "learning_rate": 9.146395681168625e-05,
      "loss": 0.1647,
      "step": 1349
    },
    {
      "epoch": 2.5677603423680457,
      "grad_norm": 0.486316978931427,
      "learning_rate": 9.14576055890759e-05,
      "loss": 0.2052,
      "step": 1350
    },
    {
      "epoch": 2.569662387066096,
      "grad_norm": 0.45915400981903076,
      "learning_rate": 9.145125436646554e-05,
      "loss": 0.218,
      "step": 1351
    },
    {
      "epoch": 2.5715644317641466,
      "grad_norm": 0.3178432583808899,
      "learning_rate": 9.14449031438552e-05,
      "loss": 0.1453,
      "step": 1352
    },
    {
      "epoch": 2.5734664764621966,
      "grad_norm": 0.3939111828804016,
      "learning_rate": 9.143855192124485e-05,
      "loss": 0.1784,
      "step": 1353
    },
    {
      "epoch": 2.5753685211602475,
      "grad_norm": 0.3399297595024109,
      "learning_rate": 9.14322006986345e-05,
      "loss": 0.1644,
      "step": 1354
    },
    {
      "epoch": 2.5772705658582975,
      "grad_norm": 0.39880868792533875,
      "learning_rate": 9.142584947602414e-05,
      "loss": 0.2139,
      "step": 1355
    },
    {
      "epoch": 2.579172610556348,
      "grad_norm": 0.40534335374832153,
      "learning_rate": 9.141949825341379e-05,
      "loss": 0.1872,
      "step": 1356
    },
    {
      "epoch": 2.5810746552543984,
      "grad_norm": 0.3201380968093872,
      "learning_rate": 9.141314703080344e-05,
      "loss": 0.1557,
      "step": 1357
    },
    {
      "epoch": 2.582976699952449,
      "grad_norm": 0.31011682748794556,
      "learning_rate": 9.140679580819308e-05,
      "loss": 0.1301,
      "step": 1358
    },
    {
      "epoch": 2.5848787446504993,
      "grad_norm": 0.3697820007801056,
      "learning_rate": 9.140044458558273e-05,
      "loss": 0.1856,
      "step": 1359
    },
    {
      "epoch": 2.5867807893485497,
      "grad_norm": 0.291369765996933,
      "learning_rate": 9.139409336297238e-05,
      "loss": 0.1323,
      "step": 1360
    },
    {
      "epoch": 2.5886828340466,
      "grad_norm": 0.4111400842666626,
      "learning_rate": 9.138774214036202e-05,
      "loss": 0.2271,
      "step": 1361
    },
    {
      "epoch": 2.5905848787446506,
      "grad_norm": 0.4169454872608185,
      "learning_rate": 9.138139091775167e-05,
      "loss": 0.199,
      "step": 1362
    },
    {
      "epoch": 2.592486923442701,
      "grad_norm": 0.4209660589694977,
      "learning_rate": 9.137503969514132e-05,
      "loss": 0.2296,
      "step": 1363
    },
    {
      "epoch": 2.594388968140751,
      "grad_norm": 0.3968026041984558,
      "learning_rate": 9.136868847253096e-05,
      "loss": 0.2174,
      "step": 1364
    },
    {
      "epoch": 2.596291012838802,
      "grad_norm": 0.3477707803249359,
      "learning_rate": 9.136233724992061e-05,
      "loss": 0.1818,
      "step": 1365
    },
    {
      "epoch": 2.598193057536852,
      "grad_norm": 0.3979746699333191,
      "learning_rate": 9.135598602731025e-05,
      "loss": 0.2373,
      "step": 1366
    },
    {
      "epoch": 2.6000951022349024,
      "grad_norm": 0.32050615549087524,
      "learning_rate": 9.134963480469992e-05,
      "loss": 0.1562,
      "step": 1367
    },
    {
      "epoch": 2.601997146932953,
      "grad_norm": 0.4675930142402649,
      "learning_rate": 9.134328358208956e-05,
      "loss": 0.2942,
      "step": 1368
    },
    {
      "epoch": 2.6038991916310033,
      "grad_norm": 0.32259052991867065,
      "learning_rate": 9.13369323594792e-05,
      "loss": 0.1411,
      "step": 1369
    },
    {
      "epoch": 2.6058012363290537,
      "grad_norm": 0.3838285803794861,
      "learning_rate": 9.133058113686886e-05,
      "loss": 0.2098,
      "step": 1370
    },
    {
      "epoch": 2.607703281027104,
      "grad_norm": 0.4749825596809387,
      "learning_rate": 9.13242299142585e-05,
      "loss": 0.2621,
      "step": 1371
    },
    {
      "epoch": 2.6096053257251546,
      "grad_norm": 0.3093271255493164,
      "learning_rate": 9.131787869164815e-05,
      "loss": 0.1389,
      "step": 1372
    },
    {
      "epoch": 2.611507370423205,
      "grad_norm": 0.4896688461303711,
      "learning_rate": 9.131152746903779e-05,
      "loss": 0.2347,
      "step": 1373
    },
    {
      "epoch": 2.6134094151212555,
      "grad_norm": 0.39409998059272766,
      "learning_rate": 9.130517624642744e-05,
      "loss": 0.2224,
      "step": 1374
    },
    {
      "epoch": 2.615311459819306,
      "grad_norm": 0.39578184485435486,
      "learning_rate": 9.129882502381709e-05,
      "loss": 0.1963,
      "step": 1375
    },
    {
      "epoch": 2.6172135045173563,
      "grad_norm": 0.34999507665634155,
      "learning_rate": 9.129247380120673e-05,
      "loss": 0.1612,
      "step": 1376
    },
    {
      "epoch": 2.6191155492154063,
      "grad_norm": 0.33919695019721985,
      "learning_rate": 9.12861225785964e-05,
      "loss": 0.1813,
      "step": 1377
    },
    {
      "epoch": 2.6210175939134572,
      "grad_norm": 0.3273175060749054,
      "learning_rate": 9.127977135598603e-05,
      "loss": 0.1436,
      "step": 1378
    },
    {
      "epoch": 2.6229196386115072,
      "grad_norm": 0.4175270199775696,
      "learning_rate": 9.127342013337567e-05,
      "loss": 0.1832,
      "step": 1379
    },
    {
      "epoch": 2.6248216833095577,
      "grad_norm": 0.3580436408519745,
      "learning_rate": 9.126706891076532e-05,
      "loss": 0.1569,
      "step": 1380
    },
    {
      "epoch": 2.626723728007608,
      "grad_norm": 0.3683449625968933,
      "learning_rate": 9.126071768815498e-05,
      "loss": 0.1955,
      "step": 1381
    },
    {
      "epoch": 2.6286257727056586,
      "grad_norm": 0.3830251395702362,
      "learning_rate": 9.125436646554461e-05,
      "loss": 0.1626,
      "step": 1382
    },
    {
      "epoch": 2.630527817403709,
      "grad_norm": 0.3428569734096527,
      "learning_rate": 9.124801524293427e-05,
      "loss": 0.1477,
      "step": 1383
    },
    {
      "epoch": 2.6324298621017594,
      "grad_norm": 0.4621574878692627,
      "learning_rate": 9.124166402032392e-05,
      "loss": 0.1675,
      "step": 1384
    },
    {
      "epoch": 2.63433190679981,
      "grad_norm": 0.40000998973846436,
      "learning_rate": 9.123531279771357e-05,
      "loss": 0.1751,
      "step": 1385
    },
    {
      "epoch": 2.6362339514978603,
      "grad_norm": 0.4612349271774292,
      "learning_rate": 9.122896157510321e-05,
      "loss": 0.2165,
      "step": 1386
    },
    {
      "epoch": 2.6381359961959108,
      "grad_norm": 0.47919005155563354,
      "learning_rate": 9.122261035249286e-05,
      "loss": 0.2,
      "step": 1387
    },
    {
      "epoch": 2.6400380408939608,
      "grad_norm": 0.5020009875297546,
      "learning_rate": 9.121625912988251e-05,
      "loss": 0.1997,
      "step": 1388
    },
    {
      "epoch": 2.6419400855920117,
      "grad_norm": 0.4959258437156677,
      "learning_rate": 9.120990790727215e-05,
      "loss": 0.1903,
      "step": 1389
    },
    {
      "epoch": 2.6438421302900617,
      "grad_norm": 0.4882603585720062,
      "learning_rate": 9.12035566846618e-05,
      "loss": 0.2082,
      "step": 1390
    },
    {
      "epoch": 2.645744174988112,
      "grad_norm": 0.37479934096336365,
      "learning_rate": 9.119720546205145e-05,
      "loss": 0.179,
      "step": 1391
    },
    {
      "epoch": 2.6476462196861625,
      "grad_norm": 0.5104106068611145,
      "learning_rate": 9.119085423944109e-05,
      "loss": 0.2281,
      "step": 1392
    },
    {
      "epoch": 2.649548264384213,
      "grad_norm": 0.3893817663192749,
      "learning_rate": 9.118450301683074e-05,
      "loss": 0.2324,
      "step": 1393
    },
    {
      "epoch": 2.6514503090822634,
      "grad_norm": 0.35762450098991394,
      "learning_rate": 9.11781517942204e-05,
      "loss": 0.1933,
      "step": 1394
    },
    {
      "epoch": 2.653352353780314,
      "grad_norm": 0.37635737657546997,
      "learning_rate": 9.117180057161005e-05,
      "loss": 0.1869,
      "step": 1395
    },
    {
      "epoch": 2.6552543984783643,
      "grad_norm": 0.3230188488960266,
      "learning_rate": 9.116544934899969e-05,
      "loss": 0.1576,
      "step": 1396
    },
    {
      "epoch": 2.6571564431764148,
      "grad_norm": 0.3708724081516266,
      "learning_rate": 9.115909812638934e-05,
      "loss": 0.168,
      "step": 1397
    },
    {
      "epoch": 2.659058487874465,
      "grad_norm": 0.34403741359710693,
      "learning_rate": 9.115274690377899e-05,
      "loss": 0.2721,
      "step": 1398
    },
    {
      "epoch": 2.660960532572515,
      "grad_norm": 0.2812383770942688,
      "learning_rate": 9.114639568116863e-05,
      "loss": 0.1605,
      "step": 1399
    },
    {
      "epoch": 2.662862577270566,
      "grad_norm": 0.39116060733795166,
      "learning_rate": 9.114004445855827e-05,
      "loss": 0.1843,
      "step": 1400
    },
    {
      "epoch": 2.664764621968616,
      "grad_norm": 0.3641309440135956,
      "learning_rate": 9.113369323594793e-05,
      "loss": 0.1818,
      "step": 1401
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 0.4198780953884125,
      "learning_rate": 9.112734201333757e-05,
      "loss": 0.2044,
      "step": 1402
    },
    {
      "epoch": 2.668568711364717,
      "grad_norm": 0.3912922739982605,
      "learning_rate": 9.112099079072722e-05,
      "loss": 0.1881,
      "step": 1403
    },
    {
      "epoch": 2.6704707560627674,
      "grad_norm": 0.4235263764858246,
      "learning_rate": 9.111463956811686e-05,
      "loss": 0.2034,
      "step": 1404
    },
    {
      "epoch": 2.672372800760818,
      "grad_norm": 0.3731124699115753,
      "learning_rate": 9.110828834550651e-05,
      "loss": 0.1803,
      "step": 1405
    },
    {
      "epoch": 2.6742748454588683,
      "grad_norm": 0.3907954692840576,
      "learning_rate": 9.110193712289616e-05,
      "loss": 0.2074,
      "step": 1406
    },
    {
      "epoch": 2.6761768901569187,
      "grad_norm": 0.3954913914203644,
      "learning_rate": 9.10955859002858e-05,
      "loss": 0.1797,
      "step": 1407
    },
    {
      "epoch": 2.678078934854969,
      "grad_norm": 0.5066515207290649,
      "learning_rate": 9.108923467767547e-05,
      "loss": 0.2096,
      "step": 1408
    },
    {
      "epoch": 2.6799809795530196,
      "grad_norm": 0.4380313456058502,
      "learning_rate": 9.10828834550651e-05,
      "loss": 0.2064,
      "step": 1409
    },
    {
      "epoch": 2.68188302425107,
      "grad_norm": 0.3758976459503174,
      "learning_rate": 9.107653223245474e-05,
      "loss": 0.2076,
      "step": 1410
    },
    {
      "epoch": 2.6837850689491205,
      "grad_norm": 0.38098809123039246,
      "learning_rate": 9.10701810098444e-05,
      "loss": 0.1727,
      "step": 1411
    },
    {
      "epoch": 2.6856871136471705,
      "grad_norm": 0.4027041792869568,
      "learning_rate": 9.106382978723405e-05,
      "loss": 0.154,
      "step": 1412
    },
    {
      "epoch": 2.6875891583452214,
      "grad_norm": 0.307954877614975,
      "learning_rate": 9.10574785646237e-05,
      "loss": 0.1766,
      "step": 1413
    },
    {
      "epoch": 2.6894912030432714,
      "grad_norm": 0.4232465326786041,
      "learning_rate": 9.105112734201334e-05,
      "loss": 0.1866,
      "step": 1414
    },
    {
      "epoch": 2.691393247741322,
      "grad_norm": 0.4296838641166687,
      "learning_rate": 9.104477611940299e-05,
      "loss": 0.1813,
      "step": 1415
    },
    {
      "epoch": 2.6932952924393723,
      "grad_norm": 0.3334490954875946,
      "learning_rate": 9.103842489679264e-05,
      "loss": 0.1576,
      "step": 1416
    },
    {
      "epoch": 2.6951973371374227,
      "grad_norm": 0.42984020709991455,
      "learning_rate": 9.103207367418228e-05,
      "loss": 0.1945,
      "step": 1417
    },
    {
      "epoch": 2.697099381835473,
      "grad_norm": 0.4306494891643524,
      "learning_rate": 9.102572245157193e-05,
      "loss": 0.179,
      "step": 1418
    },
    {
      "epoch": 2.6990014265335236,
      "grad_norm": 0.38049131631851196,
      "learning_rate": 9.101937122896158e-05,
      "loss": 0.1951,
      "step": 1419
    },
    {
      "epoch": 2.700903471231574,
      "grad_norm": 0.3691817820072174,
      "learning_rate": 9.101302000635122e-05,
      "loss": 0.1725,
      "step": 1420
    },
    {
      "epoch": 2.7028055159296245,
      "grad_norm": 0.32240816950798035,
      "learning_rate": 9.100666878374087e-05,
      "loss": 0.1852,
      "step": 1421
    },
    {
      "epoch": 2.704707560627675,
      "grad_norm": 0.3735920488834381,
      "learning_rate": 9.100031756113053e-05,
      "loss": 0.1857,
      "step": 1422
    },
    {
      "epoch": 2.706609605325725,
      "grad_norm": 0.3693629801273346,
      "learning_rate": 9.099396633852016e-05,
      "loss": 0.1698,
      "step": 1423
    },
    {
      "epoch": 2.708511650023776,
      "grad_norm": 0.40846189856529236,
      "learning_rate": 9.098761511590982e-05,
      "loss": 0.2531,
      "step": 1424
    },
    {
      "epoch": 2.710413694721826,
      "grad_norm": 0.3387136161327362,
      "learning_rate": 9.098126389329947e-05,
      "loss": 0.152,
      "step": 1425
    },
    {
      "epoch": 2.7123157394198762,
      "grad_norm": 0.43113890290260315,
      "learning_rate": 9.097491267068912e-05,
      "loss": 0.1939,
      "step": 1426
    },
    {
      "epoch": 2.7142177841179267,
      "grad_norm": 0.45811060070991516,
      "learning_rate": 9.096856144807876e-05,
      "loss": 0.2217,
      "step": 1427
    },
    {
      "epoch": 2.716119828815977,
      "grad_norm": 0.3742765486240387,
      "learning_rate": 9.096221022546841e-05,
      "loss": 0.183,
      "step": 1428
    },
    {
      "epoch": 2.7180218735140276,
      "grad_norm": 0.39835286140441895,
      "learning_rate": 9.095585900285806e-05,
      "loss": 0.248,
      "step": 1429
    },
    {
      "epoch": 2.719923918212078,
      "grad_norm": 0.38528379797935486,
      "learning_rate": 9.09495077802477e-05,
      "loss": 0.1642,
      "step": 1430
    },
    {
      "epoch": 2.7218259629101285,
      "grad_norm": 0.4142857789993286,
      "learning_rate": 9.094315655763735e-05,
      "loss": 0.1817,
      "step": 1431
    },
    {
      "epoch": 2.723728007608179,
      "grad_norm": 0.4072723388671875,
      "learning_rate": 9.0936805335027e-05,
      "loss": 0.2017,
      "step": 1432
    },
    {
      "epoch": 2.7256300523062293,
      "grad_norm": 0.37081822752952576,
      "learning_rate": 9.093045411241664e-05,
      "loss": 0.2262,
      "step": 1433
    },
    {
      "epoch": 2.7275320970042793,
      "grad_norm": 0.3628768026828766,
      "learning_rate": 9.092410288980629e-05,
      "loss": 0.1714,
      "step": 1434
    },
    {
      "epoch": 2.7294341417023302,
      "grad_norm": 0.46637046337127686,
      "learning_rate": 9.091775166719594e-05,
      "loss": 0.3189,
      "step": 1435
    },
    {
      "epoch": 2.7313361864003802,
      "grad_norm": 0.2643025517463684,
      "learning_rate": 9.091140044458558e-05,
      "loss": 0.234,
      "step": 1436
    },
    {
      "epoch": 2.7332382310984307,
      "grad_norm": 0.36125344038009644,
      "learning_rate": 9.090504922197523e-05,
      "loss": 0.1981,
      "step": 1437
    },
    {
      "epoch": 2.735140275796481,
      "grad_norm": 0.3064311742782593,
      "learning_rate": 9.089869799936487e-05,
      "loss": 0.1644,
      "step": 1438
    },
    {
      "epoch": 2.7370423204945316,
      "grad_norm": 0.372164249420166,
      "learning_rate": 9.089234677675454e-05,
      "loss": 0.2023,
      "step": 1439
    },
    {
      "epoch": 2.738944365192582,
      "grad_norm": 0.346431165933609,
      "learning_rate": 9.088599555414418e-05,
      "loss": 0.1913,
      "step": 1440
    },
    {
      "epoch": 2.7408464098906324,
      "grad_norm": 0.3421470522880554,
      "learning_rate": 9.087964433153382e-05,
      "loss": 0.1599,
      "step": 1441
    },
    {
      "epoch": 2.742748454588683,
      "grad_norm": 0.33351296186447144,
      "learning_rate": 9.087329310892348e-05,
      "loss": 0.1775,
      "step": 1442
    },
    {
      "epoch": 2.7446504992867333,
      "grad_norm": 0.3450356721878052,
      "learning_rate": 9.086694188631312e-05,
      "loss": 0.199,
      "step": 1443
    },
    {
      "epoch": 2.7465525439847838,
      "grad_norm": 0.34339770674705505,
      "learning_rate": 9.086059066370277e-05,
      "loss": 0.1608,
      "step": 1444
    },
    {
      "epoch": 2.748454588682834,
      "grad_norm": 0.35941675305366516,
      "learning_rate": 9.085423944109241e-05,
      "loss": 0.1566,
      "step": 1445
    },
    {
      "epoch": 2.7503566333808847,
      "grad_norm": 0.396847665309906,
      "learning_rate": 9.084788821848206e-05,
      "loss": 0.1829,
      "step": 1446
    },
    {
      "epoch": 2.7522586780789347,
      "grad_norm": 0.3818894028663635,
      "learning_rate": 9.084153699587171e-05,
      "loss": 0.2017,
      "step": 1447
    },
    {
      "epoch": 2.754160722776985,
      "grad_norm": 0.46124422550201416,
      "learning_rate": 9.083518577326135e-05,
      "loss": 0.185,
      "step": 1448
    },
    {
      "epoch": 2.7560627674750355,
      "grad_norm": 0.4047834575176239,
      "learning_rate": 9.082883455065102e-05,
      "loss": 0.1848,
      "step": 1449
    },
    {
      "epoch": 2.757964812173086,
      "grad_norm": 0.5650888085365295,
      "learning_rate": 9.082248332804065e-05,
      "loss": 0.2274,
      "step": 1450
    },
    {
      "epoch": 2.7598668568711364,
      "grad_norm": 0.35878250002861023,
      "learning_rate": 9.081613210543029e-05,
      "loss": 0.1414,
      "step": 1451
    },
    {
      "epoch": 2.761768901569187,
      "grad_norm": 0.37223199009895325,
      "learning_rate": 9.080978088281994e-05,
      "loss": 0.1718,
      "step": 1452
    },
    {
      "epoch": 2.7636709462672373,
      "grad_norm": 0.34717050194740295,
      "learning_rate": 9.08034296602096e-05,
      "loss": 0.1719,
      "step": 1453
    },
    {
      "epoch": 2.7655729909652877,
      "grad_norm": 0.4706629812717438,
      "learning_rate": 9.079707843759923e-05,
      "loss": 0.1953,
      "step": 1454
    },
    {
      "epoch": 2.767475035663338,
      "grad_norm": 0.40658390522003174,
      "learning_rate": 9.079072721498889e-05,
      "loss": 0.1723,
      "step": 1455
    },
    {
      "epoch": 2.7693770803613886,
      "grad_norm": 0.5025349855422974,
      "learning_rate": 9.078437599237854e-05,
      "loss": 0.2122,
      "step": 1456
    },
    {
      "epoch": 2.771279125059439,
      "grad_norm": 0.4134734272956848,
      "learning_rate": 9.077802476976819e-05,
      "loss": 0.1872,
      "step": 1457
    },
    {
      "epoch": 2.773181169757489,
      "grad_norm": 0.4301147162914276,
      "learning_rate": 9.077167354715783e-05,
      "loss": 0.2102,
      "step": 1458
    },
    {
      "epoch": 2.77508321445554,
      "grad_norm": 0.4295254051685333,
      "learning_rate": 9.076532232454748e-05,
      "loss": 0.2132,
      "step": 1459
    },
    {
      "epoch": 2.77698525915359,
      "grad_norm": 0.40130358934402466,
      "learning_rate": 9.075897110193713e-05,
      "loss": 0.1891,
      "step": 1460
    },
    {
      "epoch": 2.7788873038516404,
      "grad_norm": 0.4124513268470764,
      "learning_rate": 9.075261987932677e-05,
      "loss": 0.204,
      "step": 1461
    },
    {
      "epoch": 2.780789348549691,
      "grad_norm": 0.3976169526576996,
      "learning_rate": 9.074626865671642e-05,
      "loss": 0.2016,
      "step": 1462
    },
    {
      "epoch": 2.7826913932477413,
      "grad_norm": 0.3949052095413208,
      "learning_rate": 9.073991743410607e-05,
      "loss": 0.1924,
      "step": 1463
    },
    {
      "epoch": 2.7845934379457917,
      "grad_norm": 0.4033919870853424,
      "learning_rate": 9.073356621149571e-05,
      "loss": 0.1967,
      "step": 1464
    },
    {
      "epoch": 2.786495482643842,
      "grad_norm": 0.32922443747520447,
      "learning_rate": 9.072721498888536e-05,
      "loss": 0.1639,
      "step": 1465
    },
    {
      "epoch": 2.7883975273418926,
      "grad_norm": 0.372179239988327,
      "learning_rate": 9.072086376627502e-05,
      "loss": 0.1783,
      "step": 1466
    },
    {
      "epoch": 2.790299572039943,
      "grad_norm": 0.45123547315597534,
      "learning_rate": 9.071451254366467e-05,
      "loss": 0.2573,
      "step": 1467
    },
    {
      "epoch": 2.7922016167379935,
      "grad_norm": 0.33130937814712524,
      "learning_rate": 9.07081613210543e-05,
      "loss": 0.1427,
      "step": 1468
    },
    {
      "epoch": 2.7941036614360435,
      "grad_norm": 0.4377565085887909,
      "learning_rate": 9.070181009844394e-05,
      "loss": 0.1915,
      "step": 1469
    },
    {
      "epoch": 2.7960057061340944,
      "grad_norm": 0.555698037147522,
      "learning_rate": 9.069545887583361e-05,
      "loss": 0.2495,
      "step": 1470
    },
    {
      "epoch": 2.7979077508321444,
      "grad_norm": 0.4749322831630707,
      "learning_rate": 9.068910765322325e-05,
      "loss": 0.1944,
      "step": 1471
    },
    {
      "epoch": 2.799809795530195,
      "grad_norm": 0.3543435335159302,
      "learning_rate": 9.068275643061289e-05,
      "loss": 0.1669,
      "step": 1472
    },
    {
      "epoch": 2.8017118402282453,
      "grad_norm": 0.33086055517196655,
      "learning_rate": 9.067640520800255e-05,
      "loss": 0.1792,
      "step": 1473
    },
    {
      "epoch": 2.8036138849262957,
      "grad_norm": 0.3898443281650543,
      "learning_rate": 9.067005398539219e-05,
      "loss": 0.171,
      "step": 1474
    },
    {
      "epoch": 2.805515929624346,
      "grad_norm": 0.4169894754886627,
      "learning_rate": 9.066370276278184e-05,
      "loss": 0.2057,
      "step": 1475
    },
    {
      "epoch": 2.8074179743223966,
      "grad_norm": 0.37259283661842346,
      "learning_rate": 9.065735154017148e-05,
      "loss": 0.1799,
      "step": 1476
    },
    {
      "epoch": 2.809320019020447,
      "grad_norm": 0.3892917037010193,
      "learning_rate": 9.065100031756113e-05,
      "loss": 0.1847,
      "step": 1477
    },
    {
      "epoch": 2.8112220637184975,
      "grad_norm": 0.5309971570968628,
      "learning_rate": 9.064464909495078e-05,
      "loss": 0.2462,
      "step": 1478
    },
    {
      "epoch": 2.813124108416548,
      "grad_norm": 0.3646765351295471,
      "learning_rate": 9.063829787234042e-05,
      "loss": 0.168,
      "step": 1479
    },
    {
      "epoch": 2.815026153114598,
      "grad_norm": 0.3424735963344574,
      "learning_rate": 9.063194664973009e-05,
      "loss": 0.1547,
      "step": 1480
    },
    {
      "epoch": 2.816928197812649,
      "grad_norm": 0.38415202498435974,
      "learning_rate": 9.062559542711973e-05,
      "loss": 0.2186,
      "step": 1481
    },
    {
      "epoch": 2.818830242510699,
      "grad_norm": 0.4032725393772125,
      "learning_rate": 9.061924420450936e-05,
      "loss": 0.1802,
      "step": 1482
    },
    {
      "epoch": 2.8207322872087492,
      "grad_norm": 0.35286685824394226,
      "learning_rate": 9.061289298189902e-05,
      "loss": 0.139,
      "step": 1483
    },
    {
      "epoch": 2.8226343319067997,
      "grad_norm": 0.35866954922676086,
      "learning_rate": 9.060654175928867e-05,
      "loss": 0.2022,
      "step": 1484
    },
    {
      "epoch": 2.82453637660485,
      "grad_norm": 0.36488500237464905,
      "learning_rate": 9.060019053667832e-05,
      "loss": 0.1816,
      "step": 1485
    },
    {
      "epoch": 2.8264384213029006,
      "grad_norm": 0.4557202160358429,
      "learning_rate": 9.059383931406796e-05,
      "loss": 0.1975,
      "step": 1486
    },
    {
      "epoch": 2.828340466000951,
      "grad_norm": 0.32717350125312805,
      "learning_rate": 9.058748809145761e-05,
      "loss": 0.1639,
      "step": 1487
    },
    {
      "epoch": 2.8302425106990015,
      "grad_norm": 0.41179734468460083,
      "learning_rate": 9.058113686884726e-05,
      "loss": 0.1841,
      "step": 1488
    },
    {
      "epoch": 2.832144555397052,
      "grad_norm": 0.3747973144054413,
      "learning_rate": 9.05747856462369e-05,
      "loss": 0.1678,
      "step": 1489
    },
    {
      "epoch": 2.8340466000951023,
      "grad_norm": 0.41899365186691284,
      "learning_rate": 9.056843442362655e-05,
      "loss": 0.2753,
      "step": 1490
    },
    {
      "epoch": 2.835948644793153,
      "grad_norm": 0.397416889667511,
      "learning_rate": 9.05620832010162e-05,
      "loss": 0.1658,
      "step": 1491
    },
    {
      "epoch": 2.8378506894912032,
      "grad_norm": 0.3874271810054779,
      "learning_rate": 9.055573197840584e-05,
      "loss": 0.1808,
      "step": 1492
    },
    {
      "epoch": 2.8397527341892532,
      "grad_norm": 0.3698302209377289,
      "learning_rate": 9.05493807557955e-05,
      "loss": 0.1869,
      "step": 1493
    },
    {
      "epoch": 2.841654778887304,
      "grad_norm": 0.3908369541168213,
      "learning_rate": 9.054302953318515e-05,
      "loss": 0.1866,
      "step": 1494
    },
    {
      "epoch": 2.843556823585354,
      "grad_norm": 0.5696883201599121,
      "learning_rate": 9.053667831057478e-05,
      "loss": 0.2083,
      "step": 1495
    },
    {
      "epoch": 2.8454588682834046,
      "grad_norm": 0.3560580611228943,
      "learning_rate": 9.053032708796444e-05,
      "loss": 0.1829,
      "step": 1496
    },
    {
      "epoch": 2.847360912981455,
      "grad_norm": 0.4369358718395233,
      "learning_rate": 9.052397586535409e-05,
      "loss": 0.2302,
      "step": 1497
    },
    {
      "epoch": 2.8492629576795054,
      "grad_norm": 0.4240768551826477,
      "learning_rate": 9.051762464274374e-05,
      "loss": 0.2204,
      "step": 1498
    },
    {
      "epoch": 2.851165002377556,
      "grad_norm": 0.4078483581542969,
      "learning_rate": 9.051127342013338e-05,
      "loss": 0.181,
      "step": 1499
    },
    {
      "epoch": 2.8530670470756063,
      "grad_norm": 0.4196905195713043,
      "learning_rate": 9.050492219752303e-05,
      "loss": 0.2147,
      "step": 1500
    },
    {
      "epoch": 2.8549690917736568,
      "grad_norm": 0.3858025372028351,
      "learning_rate": 9.049857097491268e-05,
      "loss": 0.1719,
      "step": 1501
    },
    {
      "epoch": 2.856871136471707,
      "grad_norm": 0.3923434019088745,
      "learning_rate": 9.049221975230232e-05,
      "loss": 0.1966,
      "step": 1502
    },
    {
      "epoch": 2.8587731811697576,
      "grad_norm": 0.38231122493743896,
      "learning_rate": 9.048586852969197e-05,
      "loss": 0.186,
      "step": 1503
    },
    {
      "epoch": 2.8606752258678076,
      "grad_norm": 0.3579331040382385,
      "learning_rate": 9.047951730708162e-05,
      "loss": 0.1777,
      "step": 1504
    },
    {
      "epoch": 2.8625772705658585,
      "grad_norm": 0.2968972623348236,
      "learning_rate": 9.047316608447126e-05,
      "loss": 0.1456,
      "step": 1505
    },
    {
      "epoch": 2.8644793152639085,
      "grad_norm": 0.3534374535083771,
      "learning_rate": 9.046681486186091e-05,
      "loss": 0.178,
      "step": 1506
    },
    {
      "epoch": 2.866381359961959,
      "grad_norm": 0.4368778169155121,
      "learning_rate": 9.046046363925057e-05,
      "loss": 0.2349,
      "step": 1507
    },
    {
      "epoch": 2.8682834046600094,
      "grad_norm": 0.43825942277908325,
      "learning_rate": 9.04541124166402e-05,
      "loss": 0.1857,
      "step": 1508
    },
    {
      "epoch": 2.87018544935806,
      "grad_norm": 0.35765841603279114,
      "learning_rate": 9.044776119402986e-05,
      "loss": 0.1787,
      "step": 1509
    },
    {
      "epoch": 2.8720874940561103,
      "grad_norm": 0.35496601462364197,
      "learning_rate": 9.04414099714195e-05,
      "loss": 0.1776,
      "step": 1510
    },
    {
      "epoch": 2.8739895387541607,
      "grad_norm": 0.39673030376434326,
      "learning_rate": 9.043505874880916e-05,
      "loss": 0.1916,
      "step": 1511
    },
    {
      "epoch": 2.875891583452211,
      "grad_norm": 0.3670983612537384,
      "learning_rate": 9.04287075261988e-05,
      "loss": 0.1726,
      "step": 1512
    },
    {
      "epoch": 2.8777936281502616,
      "grad_norm": 0.4254002273082733,
      "learning_rate": 9.042235630358844e-05,
      "loss": 0.224,
      "step": 1513
    },
    {
      "epoch": 2.879695672848312,
      "grad_norm": 0.37891489267349243,
      "learning_rate": 9.041600508097809e-05,
      "loss": 0.1805,
      "step": 1514
    },
    {
      "epoch": 2.881597717546362,
      "grad_norm": 0.33309099078178406,
      "learning_rate": 9.040965385836774e-05,
      "loss": 0.1442,
      "step": 1515
    },
    {
      "epoch": 2.883499762244413,
      "grad_norm": 0.4709990918636322,
      "learning_rate": 9.040330263575739e-05,
      "loss": 0.2342,
      "step": 1516
    },
    {
      "epoch": 2.885401806942463,
      "grad_norm": 0.41639766097068787,
      "learning_rate": 9.039695141314703e-05,
      "loss": 0.1772,
      "step": 1517
    },
    {
      "epoch": 2.8873038516405134,
      "grad_norm": 0.37914562225341797,
      "learning_rate": 9.039060019053668e-05,
      "loss": 0.1632,
      "step": 1518
    },
    {
      "epoch": 2.889205896338564,
      "grad_norm": 0.4000544250011444,
      "learning_rate": 9.038424896792633e-05,
      "loss": 0.1927,
      "step": 1519
    },
    {
      "epoch": 2.8911079410366143,
      "grad_norm": 0.42467859387397766,
      "learning_rate": 9.037789774531597e-05,
      "loss": 0.1789,
      "step": 1520
    },
    {
      "epoch": 2.8930099857346647,
      "grad_norm": 0.46945691108703613,
      "learning_rate": 9.037154652270562e-05,
      "loss": 0.197,
      "step": 1521
    },
    {
      "epoch": 2.894912030432715,
      "grad_norm": 0.43455827236175537,
      "learning_rate": 9.036519530009528e-05,
      "loss": 0.1877,
      "step": 1522
    },
    {
      "epoch": 2.8968140751307656,
      "grad_norm": 0.5169146656990051,
      "learning_rate": 9.035884407748491e-05,
      "loss": 0.2039,
      "step": 1523
    },
    {
      "epoch": 2.898716119828816,
      "grad_norm": 0.42767763137817383,
      "learning_rate": 9.035249285487457e-05,
      "loss": 0.2123,
      "step": 1524
    },
    {
      "epoch": 2.9006181645268665,
      "grad_norm": 0.40808382630348206,
      "learning_rate": 9.034614163226422e-05,
      "loss": 0.2617,
      "step": 1525
    },
    {
      "epoch": 2.902520209224917,
      "grad_norm": 0.3179365396499634,
      "learning_rate": 9.033979040965386e-05,
      "loss": 0.1548,
      "step": 1526
    },
    {
      "epoch": 2.9044222539229674,
      "grad_norm": 0.259781152009964,
      "learning_rate": 9.033343918704351e-05,
      "loss": 0.1299,
      "step": 1527
    },
    {
      "epoch": 2.9063242986210174,
      "grad_norm": 0.40235599875450134,
      "learning_rate": 9.032708796443316e-05,
      "loss": 0.1957,
      "step": 1528
    },
    {
      "epoch": 2.9082263433190683,
      "grad_norm": 0.3170933127403259,
      "learning_rate": 9.032073674182281e-05,
      "loss": 0.1594,
      "step": 1529
    },
    {
      "epoch": 2.9101283880171183,
      "grad_norm": 0.31572115421295166,
      "learning_rate": 9.031438551921245e-05,
      "loss": 0.1922,
      "step": 1530
    },
    {
      "epoch": 2.9120304327151687,
      "grad_norm": 0.4456964433193207,
      "learning_rate": 9.03080342966021e-05,
      "loss": 0.2459,
      "step": 1531
    },
    {
      "epoch": 2.913932477413219,
      "grad_norm": 0.3345606327056885,
      "learning_rate": 9.030168307399175e-05,
      "loss": 0.1708,
      "step": 1532
    },
    {
      "epoch": 2.9158345221112696,
      "grad_norm": 0.4247712790966034,
      "learning_rate": 9.029533185138139e-05,
      "loss": 0.227,
      "step": 1533
    },
    {
      "epoch": 2.91773656680932,
      "grad_norm": 0.3642347455024719,
      "learning_rate": 9.028898062877104e-05,
      "loss": 0.1971,
      "step": 1534
    },
    {
      "epoch": 2.9196386115073705,
      "grad_norm": 0.40530455112457275,
      "learning_rate": 9.02826294061607e-05,
      "loss": 0.1574,
      "step": 1535
    },
    {
      "epoch": 2.921540656205421,
      "grad_norm": 0.5143640637397766,
      "learning_rate": 9.027627818355033e-05,
      "loss": 0.207,
      "step": 1536
    },
    {
      "epoch": 2.9234427009034714,
      "grad_norm": 0.4270274043083191,
      "learning_rate": 9.026992696093999e-05,
      "loss": 0.1971,
      "step": 1537
    },
    {
      "epoch": 2.925344745601522,
      "grad_norm": 0.5170589685440063,
      "learning_rate": 9.026357573832964e-05,
      "loss": 0.2768,
      "step": 1538
    },
    {
      "epoch": 2.927246790299572,
      "grad_norm": 0.41313278675079346,
      "learning_rate": 9.025722451571929e-05,
      "loss": 0.1765,
      "step": 1539
    },
    {
      "epoch": 2.9291488349976227,
      "grad_norm": 0.4040130078792572,
      "learning_rate": 9.025087329310893e-05,
      "loss": 0.2002,
      "step": 1540
    },
    {
      "epoch": 2.9310508796956727,
      "grad_norm": 0.37281498312950134,
      "learning_rate": 9.024452207049857e-05,
      "loss": 0.1542,
      "step": 1541
    },
    {
      "epoch": 2.932952924393723,
      "grad_norm": 0.5352873802185059,
      "learning_rate": 9.023817084788823e-05,
      "loss": 0.2437,
      "step": 1542
    },
    {
      "epoch": 2.9348549690917736,
      "grad_norm": 0.4044128358364105,
      "learning_rate": 9.023181962527787e-05,
      "loss": 0.1721,
      "step": 1543
    },
    {
      "epoch": 2.936757013789824,
      "grad_norm": 0.35553574562072754,
      "learning_rate": 9.022546840266751e-05,
      "loss": 0.1838,
      "step": 1544
    },
    {
      "epoch": 2.9386590584878745,
      "grad_norm": 0.42568060755729675,
      "learning_rate": 9.021911718005717e-05,
      "loss": 0.2022,
      "step": 1545
    },
    {
      "epoch": 2.940561103185925,
      "grad_norm": 0.453700453042984,
      "learning_rate": 9.021276595744681e-05,
      "loss": 0.1866,
      "step": 1546
    },
    {
      "epoch": 2.9424631478839753,
      "grad_norm": 0.3909238576889038,
      "learning_rate": 9.020641473483646e-05,
      "loss": 0.1628,
      "step": 1547
    },
    {
      "epoch": 2.944365192582026,
      "grad_norm": 0.39725926518440247,
      "learning_rate": 9.02000635122261e-05,
      "loss": 0.217,
      "step": 1548
    },
    {
      "epoch": 2.9462672372800762,
      "grad_norm": 0.34860628843307495,
      "learning_rate": 9.019371228961575e-05,
      "loss": 0.1724,
      "step": 1549
    },
    {
      "epoch": 2.948169281978126,
      "grad_norm": 0.38813674449920654,
      "learning_rate": 9.01873610670054e-05,
      "loss": 0.2047,
      "step": 1550
    },
    {
      "epoch": 2.950071326676177,
      "grad_norm": 0.37160560488700867,
      "learning_rate": 9.018100984439504e-05,
      "loss": 0.2119,
      "step": 1551
    },
    {
      "epoch": 2.951973371374227,
      "grad_norm": 0.4166210889816284,
      "learning_rate": 9.017465862178471e-05,
      "loss": 0.2215,
      "step": 1552
    },
    {
      "epoch": 2.9538754160722775,
      "grad_norm": 0.3657042980194092,
      "learning_rate": 9.016830739917435e-05,
      "loss": 0.1924,
      "step": 1553
    },
    {
      "epoch": 2.955777460770328,
      "grad_norm": 0.37292999029159546,
      "learning_rate": 9.016195617656399e-05,
      "loss": 0.2329,
      "step": 1554
    },
    {
      "epoch": 2.9576795054683784,
      "grad_norm": 0.3373647928237915,
      "learning_rate": 9.015560495395364e-05,
      "loss": 0.2034,
      "step": 1555
    },
    {
      "epoch": 2.959581550166429,
      "grad_norm": 0.31643402576446533,
      "learning_rate": 9.014925373134329e-05,
      "loss": 0.1713,
      "step": 1556
    },
    {
      "epoch": 2.9614835948644793,
      "grad_norm": 0.3107222318649292,
      "learning_rate": 9.014290250873294e-05,
      "loss": 0.1511,
      "step": 1557
    },
    {
      "epoch": 2.9633856395625298,
      "grad_norm": 0.32063353061676025,
      "learning_rate": 9.013655128612258e-05,
      "loss": 0.1581,
      "step": 1558
    },
    {
      "epoch": 2.96528768426058,
      "grad_norm": 0.4035079777240753,
      "learning_rate": 9.013020006351223e-05,
      "loss": 0.2036,
      "step": 1559
    },
    {
      "epoch": 2.9671897289586306,
      "grad_norm": 0.28573077917099,
      "learning_rate": 9.012384884090188e-05,
      "loss": 0.1388,
      "step": 1560
    },
    {
      "epoch": 2.969091773656681,
      "grad_norm": 0.38853904604911804,
      "learning_rate": 9.011749761829152e-05,
      "loss": 0.1981,
      "step": 1561
    },
    {
      "epoch": 2.9709938183547315,
      "grad_norm": 0.39904823899269104,
      "learning_rate": 9.011114639568117e-05,
      "loss": 0.2249,
      "step": 1562
    },
    {
      "epoch": 2.9728958630527815,
      "grad_norm": 0.3704228103160858,
      "learning_rate": 9.010479517307082e-05,
      "loss": 0.2176,
      "step": 1563
    },
    {
      "epoch": 2.9747979077508324,
      "grad_norm": 0.3712176978588104,
      "learning_rate": 9.009844395046046e-05,
      "loss": 0.1685,
      "step": 1564
    },
    {
      "epoch": 2.9766999524488824,
      "grad_norm": 0.47927892208099365,
      "learning_rate": 9.009209272785011e-05,
      "loss": 0.2027,
      "step": 1565
    },
    {
      "epoch": 2.978601997146933,
      "grad_norm": 0.4230005443096161,
      "learning_rate": 9.008574150523977e-05,
      "loss": 0.212,
      "step": 1566
    },
    {
      "epoch": 2.9805040418449833,
      "grad_norm": 0.32152169942855835,
      "learning_rate": 9.00793902826294e-05,
      "loss": 0.1639,
      "step": 1567
    },
    {
      "epoch": 2.9824060865430337,
      "grad_norm": 0.42794153094291687,
      "learning_rate": 9.007303906001906e-05,
      "loss": 0.2143,
      "step": 1568
    },
    {
      "epoch": 2.984308131241084,
      "grad_norm": 0.37590306997299194,
      "learning_rate": 9.006668783740871e-05,
      "loss": 0.189,
      "step": 1569
    },
    {
      "epoch": 2.9862101759391346,
      "grad_norm": 0.3247901201248169,
      "learning_rate": 9.006033661479836e-05,
      "loss": 0.1616,
      "step": 1570
    },
    {
      "epoch": 2.988112220637185,
      "grad_norm": 0.36269792914390564,
      "learning_rate": 9.0053985392188e-05,
      "loss": 0.2037,
      "step": 1571
    },
    {
      "epoch": 2.9900142653352355,
      "grad_norm": 0.4436742067337036,
      "learning_rate": 9.004763416957764e-05,
      "loss": 0.202,
      "step": 1572
    },
    {
      "epoch": 2.991916310033286,
      "grad_norm": 0.45660001039505005,
      "learning_rate": 9.00412829469673e-05,
      "loss": 0.2298,
      "step": 1573
    },
    {
      "epoch": 2.993818354731336,
      "grad_norm": 0.3276821970939636,
      "learning_rate": 9.003493172435694e-05,
      "loss": 0.158,
      "step": 1574
    },
    {
      "epoch": 2.995720399429387,
      "grad_norm": 0.3427131175994873,
      "learning_rate": 9.002858050174659e-05,
      "loss": 0.1781,
      "step": 1575
    },
    {
      "epoch": 2.997622444127437,
      "grad_norm": 0.38842669129371643,
      "learning_rate": 9.002222927913624e-05,
      "loss": 0.1905,
      "step": 1576
    },
    {
      "epoch": 2.9995244888254873,
      "grad_norm": 0.4034234285354614,
      "learning_rate": 9.001587805652588e-05,
      "loss": 0.1989,
      "step": 1577
    },
    {
      "epoch": 3.0014265335235377,
      "grad_norm": 0.23682546615600586,
      "learning_rate": 9.000952683391553e-05,
      "loss": 0.0968,
      "step": 1578
    },
    {
      "epoch": 3.003328578221588,
      "grad_norm": 0.23321636021137238,
      "learning_rate": 9.000317561130517e-05,
      "loss": 0.1278,
      "step": 1579
    },
    {
      "epoch": 3.0052306229196386,
      "grad_norm": 0.2891576290130615,
      "learning_rate": 8.999682438869482e-05,
      "loss": 0.1297,
      "step": 1580
    },
    {
      "epoch": 3.007132667617689,
      "grad_norm": 0.30067315697669983,
      "learning_rate": 8.999047316608448e-05,
      "loss": 0.1216,
      "step": 1581
    },
    {
      "epoch": 3.0090347123157395,
      "grad_norm": 0.25676554441452026,
      "learning_rate": 8.998412194347411e-05,
      "loss": 0.1167,
      "step": 1582
    },
    {
      "epoch": 3.01093675701379,
      "grad_norm": 0.30124133825302124,
      "learning_rate": 8.997777072086378e-05,
      "loss": 0.1243,
      "step": 1583
    },
    {
      "epoch": 3.0128388017118404,
      "grad_norm": 0.30313733220100403,
      "learning_rate": 8.997141949825342e-05,
      "loss": 0.127,
      "step": 1584
    },
    {
      "epoch": 3.014740846409891,
      "grad_norm": 0.36067837476730347,
      "learning_rate": 8.996506827564306e-05,
      "loss": 0.1331,
      "step": 1585
    },
    {
      "epoch": 3.0166428911079413,
      "grad_norm": 0.3327738642692566,
      "learning_rate": 8.995871705303271e-05,
      "loss": 0.1304,
      "step": 1586
    },
    {
      "epoch": 3.0185449358059913,
      "grad_norm": 0.2918979525566101,
      "learning_rate": 8.995236583042236e-05,
      "loss": 0.1127,
      "step": 1587
    },
    {
      "epoch": 3.0204469805040417,
      "grad_norm": 0.40982192754745483,
      "learning_rate": 8.994601460781201e-05,
      "loss": 0.1283,
      "step": 1588
    },
    {
      "epoch": 3.022349025202092,
      "grad_norm": 0.37201565504074097,
      "learning_rate": 8.993966338520165e-05,
      "loss": 0.1198,
      "step": 1589
    },
    {
      "epoch": 3.0242510699001426,
      "grad_norm": 0.4271756708621979,
      "learning_rate": 8.99333121625913e-05,
      "loss": 0.1218,
      "step": 1590
    },
    {
      "epoch": 3.026153114598193,
      "grad_norm": 0.3430047035217285,
      "learning_rate": 8.992696093998095e-05,
      "loss": 0.1213,
      "step": 1591
    },
    {
      "epoch": 3.0280551592962435,
      "grad_norm": 0.3253467381000519,
      "learning_rate": 8.992060971737059e-05,
      "loss": 0.1124,
      "step": 1592
    },
    {
      "epoch": 3.029957203994294,
      "grad_norm": 0.38685157895088196,
      "learning_rate": 8.991425849476024e-05,
      "loss": 0.112,
      "step": 1593
    },
    {
      "epoch": 3.0318592486923444,
      "grad_norm": 0.36162498593330383,
      "learning_rate": 8.99079072721499e-05,
      "loss": 0.1061,
      "step": 1594
    },
    {
      "epoch": 3.033761293390395,
      "grad_norm": 0.32084980607032776,
      "learning_rate": 8.990155604953953e-05,
      "loss": 0.0965,
      "step": 1595
    },
    {
      "epoch": 3.0356633380884452,
      "grad_norm": 0.4037097096443176,
      "learning_rate": 8.989520482692919e-05,
      "loss": 0.1237,
      "step": 1596
    },
    {
      "epoch": 3.0375653827864957,
      "grad_norm": 0.23668204247951508,
      "learning_rate": 8.988885360431884e-05,
      "loss": 0.1778,
      "step": 1597
    },
    {
      "epoch": 3.0394674274845457,
      "grad_norm": 0.3448043167591095,
      "learning_rate": 8.988250238170848e-05,
      "loss": 0.1349,
      "step": 1598
    },
    {
      "epoch": 3.041369472182596,
      "grad_norm": 0.39455583691596985,
      "learning_rate": 8.987615115909813e-05,
      "loss": 0.1175,
      "step": 1599
    },
    {
      "epoch": 3.0432715168806466,
      "grad_norm": 0.39552587270736694,
      "learning_rate": 8.986979993648778e-05,
      "loss": 0.1296,
      "step": 1600
    },
    {
      "epoch": 3.045173561578697,
      "grad_norm": 0.36603817343711853,
      "learning_rate": 8.986344871387743e-05,
      "loss": 0.1392,
      "step": 1601
    },
    {
      "epoch": 3.0470756062767475,
      "grad_norm": 0.34084847569465637,
      "learning_rate": 8.985709749126707e-05,
      "loss": 0.1155,
      "step": 1602
    },
    {
      "epoch": 3.048977650974798,
      "grad_norm": 0.36548131704330444,
      "learning_rate": 8.985074626865672e-05,
      "loss": 0.1381,
      "step": 1603
    },
    {
      "epoch": 3.0508796956728483,
      "grad_norm": 0.30957910418510437,
      "learning_rate": 8.984439504604637e-05,
      "loss": 0.1123,
      "step": 1604
    },
    {
      "epoch": 3.0527817403708988,
      "grad_norm": 0.38922393321990967,
      "learning_rate": 8.983804382343601e-05,
      "loss": 0.1588,
      "step": 1605
    },
    {
      "epoch": 3.054683785068949,
      "grad_norm": 0.3416849672794342,
      "learning_rate": 8.983169260082566e-05,
      "loss": 0.1236,
      "step": 1606
    },
    {
      "epoch": 3.0565858297669997,
      "grad_norm": 0.31353771686553955,
      "learning_rate": 8.982534137821532e-05,
      "loss": 0.1025,
      "step": 1607
    },
    {
      "epoch": 3.05848787446505,
      "grad_norm": 0.36878702044487,
      "learning_rate": 8.981899015560495e-05,
      "loss": 0.1421,
      "step": 1608
    },
    {
      "epoch": 3.0603899191631005,
      "grad_norm": 0.38487425446510315,
      "learning_rate": 8.98126389329946e-05,
      "loss": 0.1223,
      "step": 1609
    },
    {
      "epoch": 3.0622919638611505,
      "grad_norm": 0.3435547649860382,
      "learning_rate": 8.980628771038426e-05,
      "loss": 0.1105,
      "step": 1610
    },
    {
      "epoch": 3.064194008559201,
      "grad_norm": 0.422198086977005,
      "learning_rate": 8.979993648777391e-05,
      "loss": 0.1368,
      "step": 1611
    },
    {
      "epoch": 3.0660960532572514,
      "grad_norm": 0.43352290987968445,
      "learning_rate": 8.979358526516355e-05,
      "loss": 0.1743,
      "step": 1612
    },
    {
      "epoch": 3.067998097955302,
      "grad_norm": 0.3885476887226105,
      "learning_rate": 8.978723404255319e-05,
      "loss": 0.1979,
      "step": 1613
    },
    {
      "epoch": 3.0699001426533523,
      "grad_norm": 0.3135451376438141,
      "learning_rate": 8.978088281994285e-05,
      "loss": 0.1105,
      "step": 1614
    },
    {
      "epoch": 3.0718021873514028,
      "grad_norm": 0.4184531271457672,
      "learning_rate": 8.977453159733249e-05,
      "loss": 0.1335,
      "step": 1615
    },
    {
      "epoch": 3.073704232049453,
      "grad_norm": 0.35463500022888184,
      "learning_rate": 8.976818037472213e-05,
      "loss": 0.1384,
      "step": 1616
    },
    {
      "epoch": 3.0756062767475036,
      "grad_norm": 0.33959662914276123,
      "learning_rate": 8.97618291521118e-05,
      "loss": 0.118,
      "step": 1617
    },
    {
      "epoch": 3.077508321445554,
      "grad_norm": 0.3295678198337555,
      "learning_rate": 8.975547792950143e-05,
      "loss": 0.1073,
      "step": 1618
    },
    {
      "epoch": 3.0794103661436045,
      "grad_norm": 0.32906121015548706,
      "learning_rate": 8.974912670689108e-05,
      "loss": 0.0992,
      "step": 1619
    },
    {
      "epoch": 3.081312410841655,
      "grad_norm": 0.2967415750026703,
      "learning_rate": 8.974277548428072e-05,
      "loss": 0.0901,
      "step": 1620
    },
    {
      "epoch": 3.0832144555397054,
      "grad_norm": 0.3415001928806305,
      "learning_rate": 8.973642426167037e-05,
      "loss": 0.1248,
      "step": 1621
    },
    {
      "epoch": 3.0851165002377554,
      "grad_norm": 0.2587614357471466,
      "learning_rate": 8.973007303906003e-05,
      "loss": 0.0872,
      "step": 1622
    },
    {
      "epoch": 3.087018544935806,
      "grad_norm": 0.3469274640083313,
      "learning_rate": 8.972372181644966e-05,
      "loss": 0.1147,
      "step": 1623
    },
    {
      "epoch": 3.0889205896338563,
      "grad_norm": 0.28534063696861267,
      "learning_rate": 8.971737059383932e-05,
      "loss": 0.1377,
      "step": 1624
    },
    {
      "epoch": 3.0908226343319067,
      "grad_norm": 0.3836195170879364,
      "learning_rate": 8.971101937122897e-05,
      "loss": 0.1242,
      "step": 1625
    },
    {
      "epoch": 3.092724679029957,
      "grad_norm": 0.40428081154823303,
      "learning_rate": 8.97046681486186e-05,
      "loss": 0.1017,
      "step": 1626
    },
    {
      "epoch": 3.0946267237280076,
      "grad_norm": 0.37237152457237244,
      "learning_rate": 8.969831692600826e-05,
      "loss": 0.1318,
      "step": 1627
    },
    {
      "epoch": 3.096528768426058,
      "grad_norm": 0.3669044077396393,
      "learning_rate": 8.969196570339791e-05,
      "loss": 0.1191,
      "step": 1628
    },
    {
      "epoch": 3.0984308131241085,
      "grad_norm": 0.36814671754837036,
      "learning_rate": 8.968561448078756e-05,
      "loss": 0.1227,
      "step": 1629
    },
    {
      "epoch": 3.100332857822159,
      "grad_norm": 0.3883667290210724,
      "learning_rate": 8.96792632581772e-05,
      "loss": 0.1556,
      "step": 1630
    },
    {
      "epoch": 3.1022349025202094,
      "grad_norm": 0.44517648220062256,
      "learning_rate": 8.967291203556685e-05,
      "loss": 0.1439,
      "step": 1631
    },
    {
      "epoch": 3.10413694721826,
      "grad_norm": 0.3230499029159546,
      "learning_rate": 8.96665608129565e-05,
      "loss": 0.117,
      "step": 1632
    },
    {
      "epoch": 3.10603899191631,
      "grad_norm": 0.2505279779434204,
      "learning_rate": 8.966020959034614e-05,
      "loss": 0.0945,
      "step": 1633
    },
    {
      "epoch": 3.1079410366143603,
      "grad_norm": 0.31753817200660706,
      "learning_rate": 8.96538583677358e-05,
      "loss": 0.1119,
      "step": 1634
    },
    {
      "epoch": 3.1098430813124107,
      "grad_norm": 0.34199607372283936,
      "learning_rate": 8.964750714512545e-05,
      "loss": 0.1508,
      "step": 1635
    },
    {
      "epoch": 3.111745126010461,
      "grad_norm": 0.39167290925979614,
      "learning_rate": 8.964115592251508e-05,
      "loss": 0.1422,
      "step": 1636
    },
    {
      "epoch": 3.1136471707085116,
      "grad_norm": 0.28108343482017517,
      "learning_rate": 8.963480469990474e-05,
      "loss": 0.0981,
      "step": 1637
    },
    {
      "epoch": 3.115549215406562,
      "grad_norm": 0.2806454598903656,
      "learning_rate": 8.962845347729439e-05,
      "loss": 0.1227,
      "step": 1638
    },
    {
      "epoch": 3.1174512601046125,
      "grad_norm": 0.3393970727920532,
      "learning_rate": 8.962210225468403e-05,
      "loss": 0.1419,
      "step": 1639
    },
    {
      "epoch": 3.119353304802663,
      "grad_norm": 0.3800428509712219,
      "learning_rate": 8.961575103207368e-05,
      "loss": 0.1323,
      "step": 1640
    },
    {
      "epoch": 3.1212553495007134,
      "grad_norm": 0.3849729299545288,
      "learning_rate": 8.960939980946333e-05,
      "loss": 0.1505,
      "step": 1641
    },
    {
      "epoch": 3.123157394198764,
      "grad_norm": 0.38189247250556946,
      "learning_rate": 8.960304858685298e-05,
      "loss": 0.1303,
      "step": 1642
    },
    {
      "epoch": 3.1250594388968143,
      "grad_norm": 0.3030915856361389,
      "learning_rate": 8.959669736424262e-05,
      "loss": 0.1141,
      "step": 1643
    },
    {
      "epoch": 3.1269614835948643,
      "grad_norm": 0.3842359185218811,
      "learning_rate": 8.959034614163226e-05,
      "loss": 0.1124,
      "step": 1644
    },
    {
      "epoch": 3.1288635282929147,
      "grad_norm": 0.3637976348400116,
      "learning_rate": 8.958399491902192e-05,
      "loss": 0.1275,
      "step": 1645
    },
    {
      "epoch": 3.130765572990965,
      "grad_norm": 0.2884964346885681,
      "learning_rate": 8.957764369641156e-05,
      "loss": 0.1065,
      "step": 1646
    },
    {
      "epoch": 3.1326676176890156,
      "grad_norm": 0.3866124749183655,
      "learning_rate": 8.957129247380121e-05,
      "loss": 0.1389,
      "step": 1647
    },
    {
      "epoch": 3.134569662387066,
      "grad_norm": 0.418950617313385,
      "learning_rate": 8.956494125119087e-05,
      "loss": 0.1406,
      "step": 1648
    },
    {
      "epoch": 3.1364717070851165,
      "grad_norm": 0.37514927983283997,
      "learning_rate": 8.95585900285805e-05,
      "loss": 0.1239,
      "step": 1649
    },
    {
      "epoch": 3.138373751783167,
      "grad_norm": 0.29558438062667847,
      "learning_rate": 8.955223880597016e-05,
      "loss": 0.1077,
      "step": 1650
    },
    {
      "epoch": 3.1402757964812174,
      "grad_norm": 0.3241124749183655,
      "learning_rate": 8.95458875833598e-05,
      "loss": 0.1254,
      "step": 1651
    },
    {
      "epoch": 3.142177841179268,
      "grad_norm": 0.40942251682281494,
      "learning_rate": 8.953953636074945e-05,
      "loss": 0.1388,
      "step": 1652
    },
    {
      "epoch": 3.1440798858773182,
      "grad_norm": 0.3899609446525574,
      "learning_rate": 8.95331851381391e-05,
      "loss": 0.1279,
      "step": 1653
    },
    {
      "epoch": 3.1459819305753687,
      "grad_norm": 0.37820303440093994,
      "learning_rate": 8.952683391552874e-05,
      "loss": 0.1146,
      "step": 1654
    },
    {
      "epoch": 3.147883975273419,
      "grad_norm": 0.3521963059902191,
      "learning_rate": 8.95204826929184e-05,
      "loss": 0.1337,
      "step": 1655
    },
    {
      "epoch": 3.1497860199714696,
      "grad_norm": 0.3292877674102783,
      "learning_rate": 8.951413147030804e-05,
      "loss": 0.1225,
      "step": 1656
    },
    {
      "epoch": 3.1516880646695196,
      "grad_norm": 0.28479406237602234,
      "learning_rate": 8.950778024769768e-05,
      "loss": 0.1006,
      "step": 1657
    },
    {
      "epoch": 3.15359010936757,
      "grad_norm": 0.2883979380130768,
      "learning_rate": 8.950142902508733e-05,
      "loss": 0.1114,
      "step": 1658
    },
    {
      "epoch": 3.1554921540656204,
      "grad_norm": 0.33744558691978455,
      "learning_rate": 8.949507780247698e-05,
      "loss": 0.1263,
      "step": 1659
    },
    {
      "epoch": 3.157394198763671,
      "grad_norm": 0.2845192551612854,
      "learning_rate": 8.948872657986663e-05,
      "loss": 0.1047,
      "step": 1660
    },
    {
      "epoch": 3.1592962434617213,
      "grad_norm": 0.3539939224720001,
      "learning_rate": 8.948237535725627e-05,
      "loss": 0.1183,
      "step": 1661
    },
    {
      "epoch": 3.1611982881597718,
      "grad_norm": 0.24927809834480286,
      "learning_rate": 8.947602413464592e-05,
      "loss": 0.0825,
      "step": 1662
    },
    {
      "epoch": 3.163100332857822,
      "grad_norm": 0.4059623181819916,
      "learning_rate": 8.946967291203558e-05,
      "loss": 0.1457,
      "step": 1663
    },
    {
      "epoch": 3.1650023775558727,
      "grad_norm": 0.3298782706260681,
      "learning_rate": 8.946332168942521e-05,
      "loss": 0.1226,
      "step": 1664
    },
    {
      "epoch": 3.166904422253923,
      "grad_norm": 0.3750251829624176,
      "learning_rate": 8.945697046681487e-05,
      "loss": 0.144,
      "step": 1665
    },
    {
      "epoch": 3.1688064669519735,
      "grad_norm": 0.40858665108680725,
      "learning_rate": 8.945061924420452e-05,
      "loss": 0.1426,
      "step": 1666
    },
    {
      "epoch": 3.170708511650024,
      "grad_norm": 0.38032254576683044,
      "learning_rate": 8.944426802159416e-05,
      "loss": 0.1479,
      "step": 1667
    },
    {
      "epoch": 3.172610556348074,
      "grad_norm": 0.3702940046787262,
      "learning_rate": 8.943791679898381e-05,
      "loss": 0.1262,
      "step": 1668
    },
    {
      "epoch": 3.1745126010461244,
      "grad_norm": 0.43061700463294983,
      "learning_rate": 8.943156557637346e-05,
      "loss": 0.1463,
      "step": 1669
    },
    {
      "epoch": 3.176414645744175,
      "grad_norm": 0.2968880832195282,
      "learning_rate": 8.94252143537631e-05,
      "loss": 0.1135,
      "step": 1670
    },
    {
      "epoch": 3.1783166904422253,
      "grad_norm": 0.28398388624191284,
      "learning_rate": 8.941886313115275e-05,
      "loss": 0.1137,
      "step": 1671
    },
    {
      "epoch": 3.1802187351402758,
      "grad_norm": 0.2764633595943451,
      "learning_rate": 8.94125119085424e-05,
      "loss": 0.0974,
      "step": 1672
    },
    {
      "epoch": 3.182120779838326,
      "grad_norm": 0.39509302377700806,
      "learning_rate": 8.940616068593205e-05,
      "loss": 0.1491,
      "step": 1673
    },
    {
      "epoch": 3.1840228245363766,
      "grad_norm": 0.2926827669143677,
      "learning_rate": 8.939980946332169e-05,
      "loss": 0.1207,
      "step": 1674
    },
    {
      "epoch": 3.185924869234427,
      "grad_norm": 0.35445713996887207,
      "learning_rate": 8.939345824071133e-05,
      "loss": 0.1252,
      "step": 1675
    },
    {
      "epoch": 3.1878269139324775,
      "grad_norm": 0.3183155059814453,
      "learning_rate": 8.9387107018101e-05,
      "loss": 0.1178,
      "step": 1676
    },
    {
      "epoch": 3.189728958630528,
      "grad_norm": 0.40158188343048096,
      "learning_rate": 8.938075579549063e-05,
      "loss": 0.1266,
      "step": 1677
    },
    {
      "epoch": 3.1916310033285784,
      "grad_norm": 0.33932897448539734,
      "learning_rate": 8.937440457288029e-05,
      "loss": 0.1321,
      "step": 1678
    },
    {
      "epoch": 3.1935330480266284,
      "grad_norm": 0.3436925411224365,
      "learning_rate": 8.936805335026994e-05,
      "loss": 0.1204,
      "step": 1679
    },
    {
      "epoch": 3.195435092724679,
      "grad_norm": 0.32970649003982544,
      "learning_rate": 8.936170212765958e-05,
      "loss": 0.1023,
      "step": 1680
    },
    {
      "epoch": 3.1973371374227293,
      "grad_norm": 0.3206690549850464,
      "learning_rate": 8.935535090504923e-05,
      "loss": 0.1011,
      "step": 1681
    },
    {
      "epoch": 3.1992391821207797,
      "grad_norm": 0.39323487877845764,
      "learning_rate": 8.934899968243887e-05,
      "loss": 0.1263,
      "step": 1682
    },
    {
      "epoch": 3.20114122681883,
      "grad_norm": 0.3755662143230438,
      "learning_rate": 8.934264845982853e-05,
      "loss": 0.1345,
      "step": 1683
    },
    {
      "epoch": 3.2030432715168806,
      "grad_norm": 0.3337384760379791,
      "learning_rate": 8.933629723721817e-05,
      "loss": 0.1094,
      "step": 1684
    },
    {
      "epoch": 3.204945316214931,
      "grad_norm": 0.35307517647743225,
      "learning_rate": 8.932994601460781e-05,
      "loss": 0.1244,
      "step": 1685
    },
    {
      "epoch": 3.2068473609129815,
      "grad_norm": 0.2809374928474426,
      "learning_rate": 8.932359479199747e-05,
      "loss": 0.0961,
      "step": 1686
    },
    {
      "epoch": 3.208749405611032,
      "grad_norm": 0.35939821600914,
      "learning_rate": 8.931724356938711e-05,
      "loss": 0.1294,
      "step": 1687
    },
    {
      "epoch": 3.2106514503090824,
      "grad_norm": 0.36626148223876953,
      "learning_rate": 8.931089234677675e-05,
      "loss": 0.141,
      "step": 1688
    },
    {
      "epoch": 3.212553495007133,
      "grad_norm": 0.31976842880249023,
      "learning_rate": 8.93045411241664e-05,
      "loss": 0.1058,
      "step": 1689
    },
    {
      "epoch": 3.2144555397051833,
      "grad_norm": 0.40340307354927063,
      "learning_rate": 8.929818990155605e-05,
      "loss": 0.142,
      "step": 1690
    },
    {
      "epoch": 3.2163575844032333,
      "grad_norm": 0.3481243848800659,
      "learning_rate": 8.92918386789457e-05,
      "loss": 0.1301,
      "step": 1691
    },
    {
      "epoch": 3.2182596291012837,
      "grad_norm": 0.41779786348342896,
      "learning_rate": 8.928548745633534e-05,
      "loss": 0.1531,
      "step": 1692
    },
    {
      "epoch": 3.220161673799334,
      "grad_norm": 0.33376792073249817,
      "learning_rate": 8.9279136233725e-05,
      "loss": 0.1397,
      "step": 1693
    },
    {
      "epoch": 3.2220637184973846,
      "grad_norm": 0.42083820700645447,
      "learning_rate": 8.927278501111465e-05,
      "loss": 0.1456,
      "step": 1694
    },
    {
      "epoch": 3.223965763195435,
      "grad_norm": 0.23268885910511017,
      "learning_rate": 8.926643378850429e-05,
      "loss": 0.1261,
      "step": 1695
    },
    {
      "epoch": 3.2258678078934855,
      "grad_norm": 0.3965808153152466,
      "learning_rate": 8.926008256589394e-05,
      "loss": 0.1454,
      "step": 1696
    },
    {
      "epoch": 3.227769852591536,
      "grad_norm": 0.40782594680786133,
      "learning_rate": 8.925373134328359e-05,
      "loss": 0.137,
      "step": 1697
    },
    {
      "epoch": 3.2296718972895864,
      "grad_norm": 0.37247705459594727,
      "learning_rate": 8.924738012067323e-05,
      "loss": 0.1227,
      "step": 1698
    },
    {
      "epoch": 3.231573941987637,
      "grad_norm": 0.5225626230239868,
      "learning_rate": 8.924102889806288e-05,
      "loss": 0.1596,
      "step": 1699
    },
    {
      "epoch": 3.2334759866856873,
      "grad_norm": 0.35236862301826477,
      "learning_rate": 8.923467767545253e-05,
      "loss": 0.1576,
      "step": 1700
    },
    {
      "epoch": 3.2353780313837377,
      "grad_norm": 0.3305290639400482,
      "learning_rate": 8.922832645284218e-05,
      "loss": 0.1114,
      "step": 1701
    },
    {
      "epoch": 3.237280076081788,
      "grad_norm": 0.37631455063819885,
      "learning_rate": 8.922197523023182e-05,
      "loss": 0.1278,
      "step": 1702
    },
    {
      "epoch": 3.239182120779838,
      "grad_norm": 0.3439154624938965,
      "learning_rate": 8.921562400762147e-05,
      "loss": 0.1658,
      "step": 1703
    },
    {
      "epoch": 3.2410841654778886,
      "grad_norm": 0.4184103310108185,
      "learning_rate": 8.920927278501112e-05,
      "loss": 0.1754,
      "step": 1704
    },
    {
      "epoch": 3.242986210175939,
      "grad_norm": 0.3708958029747009,
      "learning_rate": 8.920292156240076e-05,
      "loss": 0.148,
      "step": 1705
    },
    {
      "epoch": 3.2448882548739895,
      "grad_norm": 0.36626115441322327,
      "learning_rate": 8.919657033979041e-05,
      "loss": 0.152,
      "step": 1706
    },
    {
      "epoch": 3.24679029957204,
      "grad_norm": 0.3738412857055664,
      "learning_rate": 8.919021911718007e-05,
      "loss": 0.1432,
      "step": 1707
    },
    {
      "epoch": 3.2486923442700903,
      "grad_norm": 0.4470990002155304,
      "learning_rate": 8.91838678945697e-05,
      "loss": 0.1639,
      "step": 1708
    },
    {
      "epoch": 3.250594388968141,
      "grad_norm": 0.3332229554653168,
      "learning_rate": 8.917751667195936e-05,
      "loss": 0.1257,
      "step": 1709
    },
    {
      "epoch": 3.2524964336661912,
      "grad_norm": 0.3853921890258789,
      "learning_rate": 8.917116544934901e-05,
      "loss": 0.1262,
      "step": 1710
    },
    {
      "epoch": 3.2543984783642417,
      "grad_norm": 0.32993221282958984,
      "learning_rate": 8.916481422673865e-05,
      "loss": 0.1231,
      "step": 1711
    },
    {
      "epoch": 3.256300523062292,
      "grad_norm": 0.3631759285926819,
      "learning_rate": 8.91584630041283e-05,
      "loss": 0.148,
      "step": 1712
    },
    {
      "epoch": 3.2582025677603426,
      "grad_norm": 0.40394118428230286,
      "learning_rate": 8.915211178151795e-05,
      "loss": 0.1542,
      "step": 1713
    },
    {
      "epoch": 3.2601046124583926,
      "grad_norm": 0.3267883360385895,
      "learning_rate": 8.91457605589076e-05,
      "loss": 0.1411,
      "step": 1714
    },
    {
      "epoch": 3.262006657156443,
      "grad_norm": 0.3076201379299164,
      "learning_rate": 8.913940933629724e-05,
      "loss": 0.1189,
      "step": 1715
    },
    {
      "epoch": 3.2639087018544934,
      "grad_norm": 0.43854421377182007,
      "learning_rate": 8.913305811368688e-05,
      "loss": 0.1806,
      "step": 1716
    },
    {
      "epoch": 3.265810746552544,
      "grad_norm": 0.2679373621940613,
      "learning_rate": 8.912670689107654e-05,
      "loss": 0.1251,
      "step": 1717
    },
    {
      "epoch": 3.2677127912505943,
      "grad_norm": 0.35840150713920593,
      "learning_rate": 8.912035566846618e-05,
      "loss": 0.1276,
      "step": 1718
    },
    {
      "epoch": 3.2696148359486448,
      "grad_norm": 0.368457168340683,
      "learning_rate": 8.911400444585583e-05,
      "loss": 0.1312,
      "step": 1719
    },
    {
      "epoch": 3.271516880646695,
      "grad_norm": 0.3617841303348541,
      "learning_rate": 8.910765322324549e-05,
      "loss": 0.1165,
      "step": 1720
    },
    {
      "epoch": 3.2734189253447457,
      "grad_norm": 0.34482330083847046,
      "learning_rate": 8.910130200063512e-05,
      "loss": 0.1246,
      "step": 1721
    },
    {
      "epoch": 3.275320970042796,
      "grad_norm": 0.27358710765838623,
      "learning_rate": 8.909495077802478e-05,
      "loss": 0.1093,
      "step": 1722
    },
    {
      "epoch": 3.2772230147408465,
      "grad_norm": 0.40264174342155457,
      "learning_rate": 8.908859955541441e-05,
      "loss": 0.146,
      "step": 1723
    },
    {
      "epoch": 3.279125059438897,
      "grad_norm": 0.45845937728881836,
      "learning_rate": 8.908224833280407e-05,
      "loss": 0.1457,
      "step": 1724
    },
    {
      "epoch": 3.281027104136947,
      "grad_norm": 0.34490594267845154,
      "learning_rate": 8.907589711019372e-05,
      "loss": 0.1247,
      "step": 1725
    },
    {
      "epoch": 3.282929148834998,
      "grad_norm": 0.4256596267223358,
      "learning_rate": 8.906954588758336e-05,
      "loss": 0.1563,
      "step": 1726
    },
    {
      "epoch": 3.284831193533048,
      "grad_norm": 0.3607080280780792,
      "learning_rate": 8.906319466497302e-05,
      "loss": 0.1279,
      "step": 1727
    },
    {
      "epoch": 3.2867332382310983,
      "grad_norm": 0.30969080328941345,
      "learning_rate": 8.905684344236266e-05,
      "loss": 0.1238,
      "step": 1728
    },
    {
      "epoch": 3.2886352829291488,
      "grad_norm": 0.34044647216796875,
      "learning_rate": 8.90504922197523e-05,
      "loss": 0.1237,
      "step": 1729
    },
    {
      "epoch": 3.290537327627199,
      "grad_norm": 0.40037238597869873,
      "learning_rate": 8.904414099714195e-05,
      "loss": 0.1509,
      "step": 1730
    },
    {
      "epoch": 3.2924393723252496,
      "grad_norm": 0.3565572500228882,
      "learning_rate": 8.90377897745316e-05,
      "loss": 0.1251,
      "step": 1731
    },
    {
      "epoch": 3.2943414170233,
      "grad_norm": 0.33730757236480713,
      "learning_rate": 8.903143855192125e-05,
      "loss": 0.1527,
      "step": 1732
    },
    {
      "epoch": 3.2962434617213505,
      "grad_norm": 0.4168394207954407,
      "learning_rate": 8.902508732931089e-05,
      "loss": 0.1429,
      "step": 1733
    },
    {
      "epoch": 3.298145506419401,
      "grad_norm": 0.40814298391342163,
      "learning_rate": 8.901873610670054e-05,
      "loss": 0.1588,
      "step": 1734
    },
    {
      "epoch": 3.3000475511174514,
      "grad_norm": 0.42030104994773865,
      "learning_rate": 8.90123848840902e-05,
      "loss": 0.1495,
      "step": 1735
    },
    {
      "epoch": 3.301949595815502,
      "grad_norm": 0.3305467367172241,
      "learning_rate": 8.900603366147983e-05,
      "loss": 0.1239,
      "step": 1736
    },
    {
      "epoch": 3.3038516405135523,
      "grad_norm": 0.31360068917274475,
      "learning_rate": 8.899968243886949e-05,
      "loss": 0.108,
      "step": 1737
    },
    {
      "epoch": 3.3057536852116023,
      "grad_norm": 0.42463186383247375,
      "learning_rate": 8.899333121625914e-05,
      "loss": 0.1451,
      "step": 1738
    },
    {
      "epoch": 3.3076557299096527,
      "grad_norm": 0.3854060471057892,
      "learning_rate": 8.898697999364878e-05,
      "loss": 0.1638,
      "step": 1739
    },
    {
      "epoch": 3.309557774607703,
      "grad_norm": 0.46821728348731995,
      "learning_rate": 8.898062877103843e-05,
      "loss": 0.1718,
      "step": 1740
    },
    {
      "epoch": 3.3114598193057536,
      "grad_norm": 0.33078089356422424,
      "learning_rate": 8.897427754842808e-05,
      "loss": 0.1153,
      "step": 1741
    },
    {
      "epoch": 3.313361864003804,
      "grad_norm": 0.3746374249458313,
      "learning_rate": 8.896792632581772e-05,
      "loss": 0.1387,
      "step": 1742
    },
    {
      "epoch": 3.3152639087018545,
      "grad_norm": 0.33252257108688354,
      "learning_rate": 8.896157510320737e-05,
      "loss": 0.1218,
      "step": 1743
    },
    {
      "epoch": 3.317165953399905,
      "grad_norm": 0.3421841561794281,
      "learning_rate": 8.895522388059702e-05,
      "loss": 0.1376,
      "step": 1744
    },
    {
      "epoch": 3.3190679980979554,
      "grad_norm": 0.3410481810569763,
      "learning_rate": 8.894887265798667e-05,
      "loss": 0.1174,
      "step": 1745
    },
    {
      "epoch": 3.320970042796006,
      "grad_norm": 0.3556031882762909,
      "learning_rate": 8.894252143537631e-05,
      "loss": 0.1612,
      "step": 1746
    },
    {
      "epoch": 3.3228720874940563,
      "grad_norm": 0.35139304399490356,
      "learning_rate": 8.893617021276595e-05,
      "loss": 0.1371,
      "step": 1747
    },
    {
      "epoch": 3.3247741321921067,
      "grad_norm": 0.38646724820137024,
      "learning_rate": 8.892981899015562e-05,
      "loss": 0.1472,
      "step": 1748
    },
    {
      "epoch": 3.3266761768901567,
      "grad_norm": 0.40337100625038147,
      "learning_rate": 8.892346776754525e-05,
      "loss": 0.1938,
      "step": 1749
    },
    {
      "epoch": 3.328578221588207,
      "grad_norm": 0.2508182227611542,
      "learning_rate": 8.89171165449349e-05,
      "loss": 0.0987,
      "step": 1750
    },
    {
      "epoch": 3.3304802662862576,
      "grad_norm": 0.392284631729126,
      "learning_rate": 8.891076532232456e-05,
      "loss": 0.1448,
      "step": 1751
    },
    {
      "epoch": 3.332382310984308,
      "grad_norm": 0.25311291217803955,
      "learning_rate": 8.89044140997142e-05,
      "loss": 0.1227,
      "step": 1752
    },
    {
      "epoch": 3.3342843556823585,
      "grad_norm": 0.38591787219047546,
      "learning_rate": 8.889806287710385e-05,
      "loss": 0.1251,
      "step": 1753
    },
    {
      "epoch": 3.336186400380409,
      "grad_norm": 0.3149789869785309,
      "learning_rate": 8.889171165449349e-05,
      "loss": 0.1282,
      "step": 1754
    },
    {
      "epoch": 3.3380884450784594,
      "grad_norm": 0.4134093225002289,
      "learning_rate": 8.888536043188315e-05,
      "loss": 0.1509,
      "step": 1755
    },
    {
      "epoch": 3.33999048977651,
      "grad_norm": 0.3769814074039459,
      "learning_rate": 8.887900920927279e-05,
      "loss": 0.1283,
      "step": 1756
    },
    {
      "epoch": 3.3418925344745603,
      "grad_norm": 0.42259126901626587,
      "learning_rate": 8.887265798666243e-05,
      "loss": 0.1319,
      "step": 1757
    },
    {
      "epoch": 3.3437945791726107,
      "grad_norm": 0.4603644609451294,
      "learning_rate": 8.88663067640521e-05,
      "loss": 0.1427,
      "step": 1758
    },
    {
      "epoch": 3.345696623870661,
      "grad_norm": 0.3804812431335449,
      "learning_rate": 8.885995554144173e-05,
      "loss": 0.1479,
      "step": 1759
    },
    {
      "epoch": 3.347598668568711,
      "grad_norm": 0.42290598154067993,
      "learning_rate": 8.885360431883137e-05,
      "loss": 0.17,
      "step": 1760
    },
    {
      "epoch": 3.3495007132667616,
      "grad_norm": 0.3739291727542877,
      "learning_rate": 8.884725309622102e-05,
      "loss": 0.1297,
      "step": 1761
    },
    {
      "epoch": 3.351402757964812,
      "grad_norm": 0.36516469717025757,
      "learning_rate": 8.884090187361067e-05,
      "loss": 0.1294,
      "step": 1762
    },
    {
      "epoch": 3.3533048026628625,
      "grad_norm": 0.32364609837532043,
      "learning_rate": 8.883455065100033e-05,
      "loss": 0.1211,
      "step": 1763
    },
    {
      "epoch": 3.355206847360913,
      "grad_norm": 0.3903793394565582,
      "learning_rate": 8.882819942838996e-05,
      "loss": 0.1339,
      "step": 1764
    },
    {
      "epoch": 3.3571088920589633,
      "grad_norm": 0.3321349322795868,
      "learning_rate": 8.882184820577962e-05,
      "loss": 0.1229,
      "step": 1765
    },
    {
      "epoch": 3.359010936757014,
      "grad_norm": 0.3843282163143158,
      "learning_rate": 8.881549698316927e-05,
      "loss": 0.1425,
      "step": 1766
    },
    {
      "epoch": 3.3609129814550642,
      "grad_norm": 0.34259116649627686,
      "learning_rate": 8.88091457605589e-05,
      "loss": 0.1275,
      "step": 1767
    },
    {
      "epoch": 3.3628150261531147,
      "grad_norm": 0.335219144821167,
      "learning_rate": 8.880279453794856e-05,
      "loss": 0.1273,
      "step": 1768
    },
    {
      "epoch": 3.364717070851165,
      "grad_norm": 0.3495425879955292,
      "learning_rate": 8.879644331533821e-05,
      "loss": 0.1112,
      "step": 1769
    },
    {
      "epoch": 3.3666191155492156,
      "grad_norm": 0.430451899766922,
      "learning_rate": 8.879009209272785e-05,
      "loss": 0.1404,
      "step": 1770
    },
    {
      "epoch": 3.368521160247266,
      "grad_norm": 0.24980789422988892,
      "learning_rate": 8.87837408701175e-05,
      "loss": 0.1034,
      "step": 1771
    },
    {
      "epoch": 3.3704232049453164,
      "grad_norm": 0.4349839687347412,
      "learning_rate": 8.877738964750715e-05,
      "loss": 0.1371,
      "step": 1772
    },
    {
      "epoch": 3.3723252496433664,
      "grad_norm": 0.3427116572856903,
      "learning_rate": 8.87710384248968e-05,
      "loss": 0.1224,
      "step": 1773
    },
    {
      "epoch": 3.374227294341417,
      "grad_norm": 0.3835298418998718,
      "learning_rate": 8.876468720228644e-05,
      "loss": 0.1576,
      "step": 1774
    },
    {
      "epoch": 3.3761293390394673,
      "grad_norm": 0.3284079432487488,
      "learning_rate": 8.87583359796761e-05,
      "loss": 0.1039,
      "step": 1775
    },
    {
      "epoch": 3.3780313837375178,
      "grad_norm": 0.32109662890434265,
      "learning_rate": 8.875198475706575e-05,
      "loss": 0.1079,
      "step": 1776
    },
    {
      "epoch": 3.379933428435568,
      "grad_norm": 0.27259504795074463,
      "learning_rate": 8.874563353445538e-05,
      "loss": 0.0983,
      "step": 1777
    },
    {
      "epoch": 3.3818354731336187,
      "grad_norm": 0.3639247417449951,
      "learning_rate": 8.873928231184504e-05,
      "loss": 0.1297,
      "step": 1778
    },
    {
      "epoch": 3.383737517831669,
      "grad_norm": 0.3729754388332367,
      "learning_rate": 8.873293108923469e-05,
      "loss": 0.1419,
      "step": 1779
    },
    {
      "epoch": 3.3856395625297195,
      "grad_norm": 0.44657668471336365,
      "learning_rate": 8.872657986662433e-05,
      "loss": 0.1299,
      "step": 1780
    },
    {
      "epoch": 3.38754160722777,
      "grad_norm": 0.2924906611442566,
      "learning_rate": 8.872022864401398e-05,
      "loss": 0.109,
      "step": 1781
    },
    {
      "epoch": 3.3894436519258204,
      "grad_norm": 0.3643059730529785,
      "learning_rate": 8.871387742140363e-05,
      "loss": 0.1217,
      "step": 1782
    },
    {
      "epoch": 3.391345696623871,
      "grad_norm": 0.31588301062583923,
      "learning_rate": 8.870752619879327e-05,
      "loss": 0.1309,
      "step": 1783
    },
    {
      "epoch": 3.393247741321921,
      "grad_norm": 0.5099390149116516,
      "learning_rate": 8.870117497618292e-05,
      "loss": 0.3371,
      "step": 1784
    },
    {
      "epoch": 3.3951497860199713,
      "grad_norm": 0.3374120891094208,
      "learning_rate": 8.869482375357256e-05,
      "loss": 0.1341,
      "step": 1785
    },
    {
      "epoch": 3.3970518307180217,
      "grad_norm": 0.36739760637283325,
      "learning_rate": 8.868847253096222e-05,
      "loss": 0.135,
      "step": 1786
    },
    {
      "epoch": 3.398953875416072,
      "grad_norm": 0.36785241961479187,
      "learning_rate": 8.868212130835186e-05,
      "loss": 0.1402,
      "step": 1787
    },
    {
      "epoch": 3.4008559201141226,
      "grad_norm": 0.3834420442581177,
      "learning_rate": 8.86757700857415e-05,
      "loss": 0.132,
      "step": 1788
    },
    {
      "epoch": 3.402757964812173,
      "grad_norm": 0.40532076358795166,
      "learning_rate": 8.866941886313117e-05,
      "loss": 0.1491,
      "step": 1789
    },
    {
      "epoch": 3.4046600095102235,
      "grad_norm": 0.3840698003768921,
      "learning_rate": 8.86630676405208e-05,
      "loss": 0.1238,
      "step": 1790
    },
    {
      "epoch": 3.406562054208274,
      "grad_norm": 0.3948921859264374,
      "learning_rate": 8.865671641791046e-05,
      "loss": 0.1452,
      "step": 1791
    },
    {
      "epoch": 3.4084640989063244,
      "grad_norm": 0.30841973423957825,
      "learning_rate": 8.86503651953001e-05,
      "loss": 0.1152,
      "step": 1792
    },
    {
      "epoch": 3.410366143604375,
      "grad_norm": 0.3028883635997772,
      "learning_rate": 8.864401397268975e-05,
      "loss": 0.103,
      "step": 1793
    },
    {
      "epoch": 3.4122681883024253,
      "grad_norm": 0.3348149359226227,
      "learning_rate": 8.86376627500794e-05,
      "loss": 0.124,
      "step": 1794
    },
    {
      "epoch": 3.4141702330004753,
      "grad_norm": 0.397709459066391,
      "learning_rate": 8.863131152746904e-05,
      "loss": 0.1489,
      "step": 1795
    },
    {
      "epoch": 3.4160722776985257,
      "grad_norm": 0.33986514806747437,
      "learning_rate": 8.862496030485869e-05,
      "loss": 0.1243,
      "step": 1796
    },
    {
      "epoch": 3.417974322396576,
      "grad_norm": 0.3443019688129425,
      "learning_rate": 8.861860908224834e-05,
      "loss": 0.1206,
      "step": 1797
    },
    {
      "epoch": 3.4198763670946266,
      "grad_norm": 0.2696784734725952,
      "learning_rate": 8.861225785963798e-05,
      "loss": 0.0978,
      "step": 1798
    },
    {
      "epoch": 3.421778411792677,
      "grad_norm": 0.3711314797401428,
      "learning_rate": 8.860590663702763e-05,
      "loss": 0.1416,
      "step": 1799
    },
    {
      "epoch": 3.4236804564907275,
      "grad_norm": 0.4727902114391327,
      "learning_rate": 8.859955541441728e-05,
      "loss": 0.1749,
      "step": 1800
    },
    {
      "epoch": 3.425582501188778,
      "grad_norm": 0.39370161294937134,
      "learning_rate": 8.859320419180692e-05,
      "loss": 0.1516,
      "step": 1801
    },
    {
      "epoch": 3.4274845458868284,
      "grad_norm": 0.36975982785224915,
      "learning_rate": 8.858685296919657e-05,
      "loss": 0.1185,
      "step": 1802
    },
    {
      "epoch": 3.429386590584879,
      "grad_norm": 0.30827558040618896,
      "learning_rate": 8.858050174658622e-05,
      "loss": 0.1292,
      "step": 1803
    },
    {
      "epoch": 3.4312886352829293,
      "grad_norm": 0.3955543339252472,
      "learning_rate": 8.857415052397588e-05,
      "loss": 0.1484,
      "step": 1804
    },
    {
      "epoch": 3.4331906799809797,
      "grad_norm": 0.35280320048332214,
      "learning_rate": 8.856779930136551e-05,
      "loss": 0.1241,
      "step": 1805
    },
    {
      "epoch": 3.4350927246790297,
      "grad_norm": 0.4241807460784912,
      "learning_rate": 8.856144807875517e-05,
      "loss": 0.1663,
      "step": 1806
    },
    {
      "epoch": 3.4369947693770806,
      "grad_norm": 0.41491755843162537,
      "learning_rate": 8.855509685614482e-05,
      "loss": 0.1465,
      "step": 1807
    },
    {
      "epoch": 3.4388968140751306,
      "grad_norm": 0.3022492229938507,
      "learning_rate": 8.854874563353446e-05,
      "loss": 0.1132,
      "step": 1808
    },
    {
      "epoch": 3.440798858773181,
      "grad_norm": 0.3701956570148468,
      "learning_rate": 8.854239441092411e-05,
      "loss": 0.1525,
      "step": 1809
    },
    {
      "epoch": 3.4427009034712315,
      "grad_norm": 0.3692464232444763,
      "learning_rate": 8.853604318831376e-05,
      "loss": 0.1364,
      "step": 1810
    },
    {
      "epoch": 3.444602948169282,
      "grad_norm": 0.2783905267715454,
      "learning_rate": 8.85296919657034e-05,
      "loss": 0.1112,
      "step": 1811
    },
    {
      "epoch": 3.4465049928673324,
      "grad_norm": 0.26422539353370667,
      "learning_rate": 8.852334074309305e-05,
      "loss": 0.0871,
      "step": 1812
    },
    {
      "epoch": 3.448407037565383,
      "grad_norm": 0.3428441882133484,
      "learning_rate": 8.85169895204827e-05,
      "loss": 0.1397,
      "step": 1813
    },
    {
      "epoch": 3.4503090822634332,
      "grad_norm": 0.43042463064193726,
      "learning_rate": 8.851063829787234e-05,
      "loss": 0.1524,
      "step": 1814
    },
    {
      "epoch": 3.4522111269614837,
      "grad_norm": 0.4124317765235901,
      "learning_rate": 8.850428707526199e-05,
      "loss": 0.165,
      "step": 1815
    },
    {
      "epoch": 3.454113171659534,
      "grad_norm": 0.38967373967170715,
      "learning_rate": 8.849793585265164e-05,
      "loss": 0.129,
      "step": 1816
    },
    {
      "epoch": 3.4560152163575846,
      "grad_norm": 0.3426058292388916,
      "learning_rate": 8.84915846300413e-05,
      "loss": 0.1229,
      "step": 1817
    },
    {
      "epoch": 3.457917261055635,
      "grad_norm": 0.4571113884449005,
      "learning_rate": 8.848523340743093e-05,
      "loss": 0.1428,
      "step": 1818
    },
    {
      "epoch": 3.459819305753685,
      "grad_norm": 0.43344834446907043,
      "learning_rate": 8.847888218482057e-05,
      "loss": 0.1561,
      "step": 1819
    },
    {
      "epoch": 3.4617213504517355,
      "grad_norm": 0.36749354004859924,
      "learning_rate": 8.847253096221024e-05,
      "loss": 0.1313,
      "step": 1820
    },
    {
      "epoch": 3.463623395149786,
      "grad_norm": 0.36647292971611023,
      "learning_rate": 8.846617973959988e-05,
      "loss": 0.1278,
      "step": 1821
    },
    {
      "epoch": 3.4655254398478363,
      "grad_norm": 0.3204960525035858,
      "learning_rate": 8.845982851698953e-05,
      "loss": 0.11,
      "step": 1822
    },
    {
      "epoch": 3.467427484545887,
      "grad_norm": 0.366187185049057,
      "learning_rate": 8.845347729437918e-05,
      "loss": 0.1443,
      "step": 1823
    },
    {
      "epoch": 3.4693295292439372,
      "grad_norm": 0.4711836874485016,
      "learning_rate": 8.844712607176882e-05,
      "loss": 0.151,
      "step": 1824
    },
    {
      "epoch": 3.4712315739419877,
      "grad_norm": 0.35596373677253723,
      "learning_rate": 8.844077484915847e-05,
      "loss": 0.1246,
      "step": 1825
    },
    {
      "epoch": 3.473133618640038,
      "grad_norm": 0.41798681020736694,
      "learning_rate": 8.843442362654811e-05,
      "loss": 0.1575,
      "step": 1826
    },
    {
      "epoch": 3.4750356633380886,
      "grad_norm": 0.3631289303302765,
      "learning_rate": 8.842807240393777e-05,
      "loss": 0.1105,
      "step": 1827
    },
    {
      "epoch": 3.476937708036139,
      "grad_norm": 0.36891433596611023,
      "learning_rate": 8.842172118132741e-05,
      "loss": 0.146,
      "step": 1828
    },
    {
      "epoch": 3.4788397527341894,
      "grad_norm": 0.33271533250808716,
      "learning_rate": 8.841536995871705e-05,
      "loss": 0.1246,
      "step": 1829
    },
    {
      "epoch": 3.4807417974322394,
      "grad_norm": 0.2956920266151428,
      "learning_rate": 8.840901873610671e-05,
      "loss": 0.1181,
      "step": 1830
    },
    {
      "epoch": 3.48264384213029,
      "grad_norm": 0.3685608506202698,
      "learning_rate": 8.840266751349635e-05,
      "loss": 0.1338,
      "step": 1831
    },
    {
      "epoch": 3.4845458868283403,
      "grad_norm": 0.35031598806381226,
      "learning_rate": 8.839631629088599e-05,
      "loss": 0.1166,
      "step": 1832
    },
    {
      "epoch": 3.4864479315263908,
      "grad_norm": 0.5173628330230713,
      "learning_rate": 8.838996506827564e-05,
      "loss": 0.157,
      "step": 1833
    },
    {
      "epoch": 3.488349976224441,
      "grad_norm": 0.4643428921699524,
      "learning_rate": 8.83836138456653e-05,
      "loss": 0.1842,
      "step": 1834
    },
    {
      "epoch": 3.4902520209224916,
      "grad_norm": 0.3688521981239319,
      "learning_rate": 8.837726262305495e-05,
      "loss": 0.1375,
      "step": 1835
    },
    {
      "epoch": 3.492154065620542,
      "grad_norm": 0.3947365880012512,
      "learning_rate": 8.837091140044458e-05,
      "loss": 0.149,
      "step": 1836
    },
    {
      "epoch": 3.4940561103185925,
      "grad_norm": 0.35394486784935,
      "learning_rate": 8.836456017783424e-05,
      "loss": 0.1252,
      "step": 1837
    },
    {
      "epoch": 3.495958155016643,
      "grad_norm": 0.37168943881988525,
      "learning_rate": 8.835820895522389e-05,
      "loss": 0.1318,
      "step": 1838
    },
    {
      "epoch": 3.4978601997146934,
      "grad_norm": 0.37239521741867065,
      "learning_rate": 8.835185773261353e-05,
      "loss": 0.1214,
      "step": 1839
    },
    {
      "epoch": 3.499762244412744,
      "grad_norm": 0.36515411734580994,
      "learning_rate": 8.834550651000318e-05,
      "loss": 0.1412,
      "step": 1840
    },
    {
      "epoch": 3.501664289110794,
      "grad_norm": 0.38534054160118103,
      "learning_rate": 8.833915528739283e-05,
      "loss": 0.1334,
      "step": 1841
    },
    {
      "epoch": 3.5035663338088447,
      "grad_norm": 0.36949092149734497,
      "learning_rate": 8.833280406478247e-05,
      "loss": 0.1283,
      "step": 1842
    },
    {
      "epoch": 3.5054683785068947,
      "grad_norm": 0.39546898007392883,
      "learning_rate": 8.832645284217212e-05,
      "loss": 0.1471,
      "step": 1843
    },
    {
      "epoch": 3.507370423204945,
      "grad_norm": 0.34906435012817383,
      "learning_rate": 8.832010161956177e-05,
      "loss": 0.1386,
      "step": 1844
    },
    {
      "epoch": 3.5092724679029956,
      "grad_norm": 0.44590094685554504,
      "learning_rate": 8.831375039695142e-05,
      "loss": 0.157,
      "step": 1845
    },
    {
      "epoch": 3.511174512601046,
      "grad_norm": 0.3336107134819031,
      "learning_rate": 8.830739917434106e-05,
      "loss": 0.1435,
      "step": 1846
    },
    {
      "epoch": 3.5130765572990965,
      "grad_norm": 0.4013485610485077,
      "learning_rate": 8.830104795173071e-05,
      "loss": 0.1209,
      "step": 1847
    },
    {
      "epoch": 3.514978601997147,
      "grad_norm": 0.30285441875457764,
      "learning_rate": 8.829469672912037e-05,
      "loss": 0.108,
      "step": 1848
    },
    {
      "epoch": 3.5168806466951974,
      "grad_norm": 0.440489798784256,
      "learning_rate": 8.828834550651e-05,
      "loss": 0.1514,
      "step": 1849
    },
    {
      "epoch": 3.518782691393248,
      "grad_norm": 0.26309430599212646,
      "learning_rate": 8.828199428389964e-05,
      "loss": 0.0953,
      "step": 1850
    },
    {
      "epoch": 3.5206847360912983,
      "grad_norm": 0.548433244228363,
      "learning_rate": 8.827564306128931e-05,
      "loss": 0.1977,
      "step": 1851
    },
    {
      "epoch": 3.5225867807893483,
      "grad_norm": 0.4941021203994751,
      "learning_rate": 8.826929183867895e-05,
      "loss": 0.1268,
      "step": 1852
    },
    {
      "epoch": 3.524488825487399,
      "grad_norm": 0.3945002555847168,
      "learning_rate": 8.82629406160686e-05,
      "loss": 0.1304,
      "step": 1853
    },
    {
      "epoch": 3.526390870185449,
      "grad_norm": 0.3647942841053009,
      "learning_rate": 8.825658939345825e-05,
      "loss": 0.1454,
      "step": 1854
    },
    {
      "epoch": 3.5282929148834996,
      "grad_norm": 0.3890063762664795,
      "learning_rate": 8.825023817084789e-05,
      "loss": 0.1384,
      "step": 1855
    },
    {
      "epoch": 3.53019495958155,
      "grad_norm": 0.4001372456550598,
      "learning_rate": 8.824388694823754e-05,
      "loss": 0.1429,
      "step": 1856
    },
    {
      "epoch": 3.5320970042796005,
      "grad_norm": 0.407721608877182,
      "learning_rate": 8.823753572562718e-05,
      "loss": 0.1374,
      "step": 1857
    },
    {
      "epoch": 3.533999048977651,
      "grad_norm": 0.37832140922546387,
      "learning_rate": 8.823118450301684e-05,
      "loss": 0.1236,
      "step": 1858
    },
    {
      "epoch": 3.5359010936757014,
      "grad_norm": 0.35406047105789185,
      "learning_rate": 8.822483328040648e-05,
      "loss": 0.1306,
      "step": 1859
    },
    {
      "epoch": 3.537803138373752,
      "grad_norm": 0.2923578917980194,
      "learning_rate": 8.821848205779612e-05,
      "loss": 0.0986,
      "step": 1860
    },
    {
      "epoch": 3.5397051830718023,
      "grad_norm": 0.3824620544910431,
      "learning_rate": 8.821213083518579e-05,
      "loss": 0.1492,
      "step": 1861
    },
    {
      "epoch": 3.5416072277698527,
      "grad_norm": 0.38851413130760193,
      "learning_rate": 8.820577961257542e-05,
      "loss": 0.1612,
      "step": 1862
    },
    {
      "epoch": 3.543509272467903,
      "grad_norm": 0.3961692154407501,
      "learning_rate": 8.819942838996508e-05,
      "loss": 0.1525,
      "step": 1863
    },
    {
      "epoch": 3.5454113171659536,
      "grad_norm": 0.423235684633255,
      "learning_rate": 8.819307716735471e-05,
      "loss": 0.1514,
      "step": 1864
    },
    {
      "epoch": 3.5473133618640036,
      "grad_norm": 0.3355453610420227,
      "learning_rate": 8.818672594474437e-05,
      "loss": 0.1183,
      "step": 1865
    },
    {
      "epoch": 3.5492154065620545,
      "grad_norm": 0.44291865825653076,
      "learning_rate": 8.818037472213402e-05,
      "loss": 0.1457,
      "step": 1866
    },
    {
      "epoch": 3.5511174512601045,
      "grad_norm": 0.39356529712677,
      "learning_rate": 8.817402349952366e-05,
      "loss": 0.146,
      "step": 1867
    },
    {
      "epoch": 3.553019495958155,
      "grad_norm": 0.28863412141799927,
      "learning_rate": 8.816767227691331e-05,
      "loss": 0.1113,
      "step": 1868
    },
    {
      "epoch": 3.5549215406562054,
      "grad_norm": 0.3859669268131256,
      "learning_rate": 8.816132105430296e-05,
      "loss": 0.1234,
      "step": 1869
    },
    {
      "epoch": 3.556823585354256,
      "grad_norm": 0.3483799993991852,
      "learning_rate": 8.81549698316926e-05,
      "loss": 0.1324,
      "step": 1870
    },
    {
      "epoch": 3.5587256300523062,
      "grad_norm": 0.3053433299064636,
      "learning_rate": 8.814861860908225e-05,
      "loss": 0.1252,
      "step": 1871
    },
    {
      "epoch": 3.5606276747503567,
      "grad_norm": 0.44125038385391235,
      "learning_rate": 8.81422673864719e-05,
      "loss": 0.1627,
      "step": 1872
    },
    {
      "epoch": 3.562529719448407,
      "grad_norm": 0.35409316420555115,
      "learning_rate": 8.813591616386154e-05,
      "loss": 0.1312,
      "step": 1873
    },
    {
      "epoch": 3.5644317641464576,
      "grad_norm": 0.4219510853290558,
      "learning_rate": 8.812956494125119e-05,
      "loss": 0.1522,
      "step": 1874
    },
    {
      "epoch": 3.566333808844508,
      "grad_norm": 0.4153057932853699,
      "learning_rate": 8.812321371864084e-05,
      "loss": 0.1272,
      "step": 1875
    },
    {
      "epoch": 3.568235853542558,
      "grad_norm": 0.3225264549255371,
      "learning_rate": 8.81168624960305e-05,
      "loss": 0.1461,
      "step": 1876
    },
    {
      "epoch": 3.570137898240609,
      "grad_norm": 0.41065141558647156,
      "learning_rate": 8.811051127342013e-05,
      "loss": 0.1466,
      "step": 1877
    },
    {
      "epoch": 3.572039942938659,
      "grad_norm": 0.33854374289512634,
      "learning_rate": 8.810416005080979e-05,
      "loss": 0.2636,
      "step": 1878
    },
    {
      "epoch": 3.5739419876367093,
      "grad_norm": 0.4266054034233093,
      "learning_rate": 8.809780882819944e-05,
      "loss": 0.1546,
      "step": 1879
    },
    {
      "epoch": 3.57584403233476,
      "grad_norm": 0.32462188601493835,
      "learning_rate": 8.809145760558908e-05,
      "loss": 0.0992,
      "step": 1880
    },
    {
      "epoch": 3.5777460770328102,
      "grad_norm": 0.3243044912815094,
      "learning_rate": 8.808510638297873e-05,
      "loss": 0.127,
      "step": 1881
    },
    {
      "epoch": 3.5796481217308607,
      "grad_norm": 0.36742255091667175,
      "learning_rate": 8.807875516036838e-05,
      "loss": 0.1648,
      "step": 1882
    },
    {
      "epoch": 3.581550166428911,
      "grad_norm": 0.47478726506233215,
      "learning_rate": 8.807240393775802e-05,
      "loss": 0.1402,
      "step": 1883
    },
    {
      "epoch": 3.5834522111269616,
      "grad_norm": 0.29675087332725525,
      "learning_rate": 8.806605271514767e-05,
      "loss": 0.1102,
      "step": 1884
    },
    {
      "epoch": 3.585354255825012,
      "grad_norm": 0.26269370317459106,
      "learning_rate": 8.805970149253732e-05,
      "loss": 0.0926,
      "step": 1885
    },
    {
      "epoch": 3.5872563005230624,
      "grad_norm": 0.42690059542655945,
      "learning_rate": 8.805335026992696e-05,
      "loss": 0.1663,
      "step": 1886
    },
    {
      "epoch": 3.5891583452211124,
      "grad_norm": 0.4843170940876007,
      "learning_rate": 8.804699904731661e-05,
      "loss": 0.156,
      "step": 1887
    },
    {
      "epoch": 3.5910603899191633,
      "grad_norm": 0.4166446030139923,
      "learning_rate": 8.804064782470626e-05,
      "loss": 0.1556,
      "step": 1888
    },
    {
      "epoch": 3.5929624346172133,
      "grad_norm": 0.3265363872051239,
      "learning_rate": 8.803429660209592e-05,
      "loss": 0.122,
      "step": 1889
    },
    {
      "epoch": 3.5948644793152638,
      "grad_norm": 0.4674152433872223,
      "learning_rate": 8.802794537948555e-05,
      "loss": 0.1706,
      "step": 1890
    },
    {
      "epoch": 3.596766524013314,
      "grad_norm": 0.4072030782699585,
      "learning_rate": 8.802159415687519e-05,
      "loss": 0.1465,
      "step": 1891
    },
    {
      "epoch": 3.5986685687113646,
      "grad_norm": 0.4924727976322174,
      "learning_rate": 8.801524293426486e-05,
      "loss": 0.153,
      "step": 1892
    },
    {
      "epoch": 3.600570613409415,
      "grad_norm": 0.34262821078300476,
      "learning_rate": 8.80088917116545e-05,
      "loss": 0.1221,
      "step": 1893
    },
    {
      "epoch": 3.6024726581074655,
      "grad_norm": 0.3641190528869629,
      "learning_rate": 8.800254048904415e-05,
      "loss": 0.1146,
      "step": 1894
    },
    {
      "epoch": 3.604374702805516,
      "grad_norm": 0.3594358265399933,
      "learning_rate": 8.799618926643379e-05,
      "loss": 0.1198,
      "step": 1895
    },
    {
      "epoch": 3.6062767475035664,
      "grad_norm": 0.40045297145843506,
      "learning_rate": 8.798983804382344e-05,
      "loss": 0.2122,
      "step": 1896
    },
    {
      "epoch": 3.608178792201617,
      "grad_norm": 0.40417537093162537,
      "learning_rate": 8.798348682121309e-05,
      "loss": 0.1523,
      "step": 1897
    },
    {
      "epoch": 3.6100808368996673,
      "grad_norm": 0.3493559658527374,
      "learning_rate": 8.797713559860273e-05,
      "loss": 0.1105,
      "step": 1898
    },
    {
      "epoch": 3.6119828815977177,
      "grad_norm": 0.3540056645870209,
      "learning_rate": 8.79707843759924e-05,
      "loss": 0.1205,
      "step": 1899
    },
    {
      "epoch": 3.6138849262957677,
      "grad_norm": 0.4836410582065582,
      "learning_rate": 8.796443315338203e-05,
      "loss": 0.184,
      "step": 1900
    },
    {
      "epoch": 3.6157869709938186,
      "grad_norm": 0.34036317467689514,
      "learning_rate": 8.795808193077167e-05,
      "loss": 0.1313,
      "step": 1901
    },
    {
      "epoch": 3.6176890156918686,
      "grad_norm": 0.34924453496932983,
      "learning_rate": 8.795173070816132e-05,
      "loss": 0.1018,
      "step": 1902
    },
    {
      "epoch": 3.619591060389919,
      "grad_norm": 0.4308503270149231,
      "learning_rate": 8.794537948555097e-05,
      "loss": 0.1396,
      "step": 1903
    },
    {
      "epoch": 3.6214931050879695,
      "grad_norm": 0.44268596172332764,
      "learning_rate": 8.793902826294061e-05,
      "loss": 0.1377,
      "step": 1904
    },
    {
      "epoch": 3.62339514978602,
      "grad_norm": 0.36984702944755554,
      "learning_rate": 8.793267704033026e-05,
      "loss": 0.1343,
      "step": 1905
    },
    {
      "epoch": 3.6252971944840704,
      "grad_norm": 0.3913877606391907,
      "learning_rate": 8.792632581771992e-05,
      "loss": 0.1443,
      "step": 1906
    },
    {
      "epoch": 3.627199239182121,
      "grad_norm": 0.4213595986366272,
      "learning_rate": 8.791997459510957e-05,
      "loss": 0.1537,
      "step": 1907
    },
    {
      "epoch": 3.6291012838801713,
      "grad_norm": 0.4095703959465027,
      "learning_rate": 8.79136233724992e-05,
      "loss": 0.151,
      "step": 1908
    },
    {
      "epoch": 3.6310033285782217,
      "grad_norm": 0.366328626871109,
      "learning_rate": 8.790727214988886e-05,
      "loss": 0.1198,
      "step": 1909
    },
    {
      "epoch": 3.632905373276272,
      "grad_norm": 0.4124557375907898,
      "learning_rate": 8.790092092727851e-05,
      "loss": 0.1408,
      "step": 1910
    },
    {
      "epoch": 3.634807417974322,
      "grad_norm": 0.36249884963035583,
      "learning_rate": 8.789456970466815e-05,
      "loss": 0.2058,
      "step": 1911
    },
    {
      "epoch": 3.636709462672373,
      "grad_norm": 0.40580618381500244,
      "learning_rate": 8.78882184820578e-05,
      "loss": 0.1247,
      "step": 1912
    },
    {
      "epoch": 3.638611507370423,
      "grad_norm": 0.30640462040901184,
      "learning_rate": 8.788186725944745e-05,
      "loss": 0.1078,
      "step": 1913
    },
    {
      "epoch": 3.6405135520684735,
      "grad_norm": 0.4200808107852936,
      "learning_rate": 8.787551603683709e-05,
      "loss": 0.1572,
      "step": 1914
    },
    {
      "epoch": 3.642415596766524,
      "grad_norm": 0.43338900804519653,
      "learning_rate": 8.786916481422674e-05,
      "loss": 0.1606,
      "step": 1915
    },
    {
      "epoch": 3.6443176414645744,
      "grad_norm": 0.4340536296367645,
      "learning_rate": 8.78628135916164e-05,
      "loss": 0.1711,
      "step": 1916
    },
    {
      "epoch": 3.646219686162625,
      "grad_norm": 0.3239591419696808,
      "learning_rate": 8.785646236900605e-05,
      "loss": 0.1166,
      "step": 1917
    },
    {
      "epoch": 3.6481217308606753,
      "grad_norm": 0.3957262933254242,
      "learning_rate": 8.785011114639568e-05,
      "loss": 0.1605,
      "step": 1918
    },
    {
      "epoch": 3.6500237755587257,
      "grad_norm": 0.4386723041534424,
      "learning_rate": 8.784375992378534e-05,
      "loss": 0.1595,
      "step": 1919
    },
    {
      "epoch": 3.651925820256776,
      "grad_norm": 0.376113623380661,
      "learning_rate": 8.783740870117499e-05,
      "loss": 0.1708,
      "step": 1920
    },
    {
      "epoch": 3.6538278649548266,
      "grad_norm": 0.2861535847187042,
      "learning_rate": 8.783105747856463e-05,
      "loss": 0.1134,
      "step": 1921
    },
    {
      "epoch": 3.6557299096528766,
      "grad_norm": 0.3381497263908386,
      "learning_rate": 8.782470625595426e-05,
      "loss": 0.1522,
      "step": 1922
    },
    {
      "epoch": 3.6576319543509275,
      "grad_norm": 0.2682400047779083,
      "learning_rate": 8.781835503334393e-05,
      "loss": 0.1007,
      "step": 1923
    },
    {
      "epoch": 3.6595339990489775,
      "grad_norm": 0.4277699887752533,
      "learning_rate": 8.781200381073357e-05,
      "loss": 0.1757,
      "step": 1924
    },
    {
      "epoch": 3.661436043747028,
      "grad_norm": 0.3176470696926117,
      "learning_rate": 8.780565258812322e-05,
      "loss": 0.1186,
      "step": 1925
    },
    {
      "epoch": 3.6633380884450784,
      "grad_norm": 0.32315725088119507,
      "learning_rate": 8.779930136551287e-05,
      "loss": 0.1353,
      "step": 1926
    },
    {
      "epoch": 3.665240133143129,
      "grad_norm": 0.44492077827453613,
      "learning_rate": 8.779295014290251e-05,
      "loss": 0.1689,
      "step": 1927
    },
    {
      "epoch": 3.6671421778411792,
      "grad_norm": 0.33450883626937866,
      "learning_rate": 8.778659892029216e-05,
      "loss": 0.1171,
      "step": 1928
    },
    {
      "epoch": 3.6690442225392297,
      "grad_norm": 0.45678386092185974,
      "learning_rate": 8.77802476976818e-05,
      "loss": 0.1547,
      "step": 1929
    },
    {
      "epoch": 3.67094626723728,
      "grad_norm": 0.3756123185157776,
      "learning_rate": 8.777389647507147e-05,
      "loss": 0.1441,
      "step": 1930
    },
    {
      "epoch": 3.6728483119353306,
      "grad_norm": 0.30440792441368103,
      "learning_rate": 8.77675452524611e-05,
      "loss": 0.1034,
      "step": 1931
    },
    {
      "epoch": 3.674750356633381,
      "grad_norm": 0.38540956377983093,
      "learning_rate": 8.776119402985074e-05,
      "loss": 0.1456,
      "step": 1932
    },
    {
      "epoch": 3.6766524013314315,
      "grad_norm": 0.42409566044807434,
      "learning_rate": 8.775484280724041e-05,
      "loss": 0.1445,
      "step": 1933
    },
    {
      "epoch": 3.678554446029482,
      "grad_norm": 0.3903610408306122,
      "learning_rate": 8.774849158463005e-05,
      "loss": 0.1428,
      "step": 1934
    },
    {
      "epoch": 3.680456490727532,
      "grad_norm": 0.4002249836921692,
      "learning_rate": 8.77421403620197e-05,
      "loss": 0.1328,
      "step": 1935
    },
    {
      "epoch": 3.6823585354255823,
      "grad_norm": 0.37625521421432495,
      "learning_rate": 8.773578913940934e-05,
      "loss": 0.1271,
      "step": 1936
    },
    {
      "epoch": 3.6842605801236328,
      "grad_norm": 0.333882600069046,
      "learning_rate": 8.772943791679899e-05,
      "loss": 0.1209,
      "step": 1937
    },
    {
      "epoch": 3.686162624821683,
      "grad_norm": 0.3934018313884735,
      "learning_rate": 8.772308669418864e-05,
      "loss": 0.1383,
      "step": 1938
    },
    {
      "epoch": 3.6880646695197337,
      "grad_norm": 0.3329316973686218,
      "learning_rate": 8.771673547157828e-05,
      "loss": 0.1334,
      "step": 1939
    },
    {
      "epoch": 3.689966714217784,
      "grad_norm": 0.3686552047729492,
      "learning_rate": 8.771038424896793e-05,
      "loss": 0.1163,
      "step": 1940
    },
    {
      "epoch": 3.6918687589158345,
      "grad_norm": 0.35531577467918396,
      "learning_rate": 8.770403302635758e-05,
      "loss": 0.114,
      "step": 1941
    },
    {
      "epoch": 3.693770803613885,
      "grad_norm": 0.4164102375507355,
      "learning_rate": 8.769768180374722e-05,
      "loss": 0.1271,
      "step": 1942
    },
    {
      "epoch": 3.6956728483119354,
      "grad_norm": 0.4182850420475006,
      "learning_rate": 8.769133058113687e-05,
      "loss": 0.1343,
      "step": 1943
    },
    {
      "epoch": 3.697574893009986,
      "grad_norm": 0.3373199701309204,
      "learning_rate": 8.768497935852652e-05,
      "loss": 0.1424,
      "step": 1944
    },
    {
      "epoch": 3.6994769377080363,
      "grad_norm": 0.44398215413093567,
      "learning_rate": 8.767862813591616e-05,
      "loss": 0.1626,
      "step": 1945
    },
    {
      "epoch": 3.7013789824060863,
      "grad_norm": 0.2877051830291748,
      "learning_rate": 8.767227691330581e-05,
      "loss": 0.0941,
      "step": 1946
    },
    {
      "epoch": 3.703281027104137,
      "grad_norm": 0.30384746193885803,
      "learning_rate": 8.766592569069547e-05,
      "loss": 0.1239,
      "step": 1947
    },
    {
      "epoch": 3.705183071802187,
      "grad_norm": 0.41360363364219666,
      "learning_rate": 8.765957446808512e-05,
      "loss": 0.1567,
      "step": 1948
    },
    {
      "epoch": 3.7070851165002376,
      "grad_norm": 0.28865674138069153,
      "learning_rate": 8.765322324547476e-05,
      "loss": 0.1165,
      "step": 1949
    },
    {
      "epoch": 3.708987161198288,
      "grad_norm": 0.341654509305954,
      "learning_rate": 8.764687202286441e-05,
      "loss": 0.1199,
      "step": 1950
    },
    {
      "epoch": 3.7108892058963385,
      "grad_norm": 0.33211663365364075,
      "learning_rate": 8.764052080025406e-05,
      "loss": 0.1386,
      "step": 1951
    },
    {
      "epoch": 3.712791250594389,
      "grad_norm": 0.37999534606933594,
      "learning_rate": 8.76341695776437e-05,
      "loss": 0.1411,
      "step": 1952
    },
    {
      "epoch": 3.7146932952924394,
      "grad_norm": 0.3158533573150635,
      "learning_rate": 8.762781835503335e-05,
      "loss": 0.1082,
      "step": 1953
    },
    {
      "epoch": 3.71659533999049,
      "grad_norm": 0.42071765661239624,
      "learning_rate": 8.7621467132423e-05,
      "loss": 0.2395,
      "step": 1954
    },
    {
      "epoch": 3.7184973846885403,
      "grad_norm": 0.3723015785217285,
      "learning_rate": 8.761511590981264e-05,
      "loss": 0.1427,
      "step": 1955
    },
    {
      "epoch": 3.7203994293865907,
      "grad_norm": 0.31827929615974426,
      "learning_rate": 8.760876468720229e-05,
      "loss": 0.0983,
      "step": 1956
    },
    {
      "epoch": 3.7223014740846407,
      "grad_norm": 0.45022010803222656,
      "learning_rate": 8.760241346459194e-05,
      "loss": 0.1658,
      "step": 1957
    },
    {
      "epoch": 3.7242035187826916,
      "grad_norm": 0.4069976508617401,
      "learning_rate": 8.759606224198158e-05,
      "loss": 0.1277,
      "step": 1958
    },
    {
      "epoch": 3.7261055634807416,
      "grad_norm": 0.3239624500274658,
      "learning_rate": 8.758971101937123e-05,
      "loss": 0.1204,
      "step": 1959
    },
    {
      "epoch": 3.728007608178792,
      "grad_norm": 0.38038089871406555,
      "learning_rate": 8.758335979676087e-05,
      "loss": 0.1305,
      "step": 1960
    },
    {
      "epoch": 3.7299096528768425,
      "grad_norm": 0.44531160593032837,
      "learning_rate": 8.757700857415054e-05,
      "loss": 0.1504,
      "step": 1961
    },
    {
      "epoch": 3.731811697574893,
      "grad_norm": 0.380256712436676,
      "learning_rate": 8.757065735154017e-05,
      "loss": 0.1213,
      "step": 1962
    },
    {
      "epoch": 3.7337137422729434,
      "grad_norm": 0.39982911944389343,
      "learning_rate": 8.756430612892981e-05,
      "loss": 0.1255,
      "step": 1963
    },
    {
      "epoch": 3.735615786970994,
      "grad_norm": 0.39186495542526245,
      "learning_rate": 8.755795490631948e-05,
      "loss": 0.1459,
      "step": 1964
    },
    {
      "epoch": 3.7375178316690443,
      "grad_norm": 0.4191820025444031,
      "learning_rate": 8.755160368370912e-05,
      "loss": 0.1269,
      "step": 1965
    },
    {
      "epoch": 3.7394198763670947,
      "grad_norm": 0.3438499867916107,
      "learning_rate": 8.754525246109877e-05,
      "loss": 0.124,
      "step": 1966
    },
    {
      "epoch": 3.741321921065145,
      "grad_norm": 0.3626823127269745,
      "learning_rate": 8.753890123848841e-05,
      "loss": 0.1326,
      "step": 1967
    },
    {
      "epoch": 3.743223965763195,
      "grad_norm": 0.3823707103729248,
      "learning_rate": 8.753255001587806e-05,
      "loss": 0.1351,
      "step": 1968
    },
    {
      "epoch": 3.745126010461246,
      "grad_norm": 0.3537774980068207,
      "learning_rate": 8.752619879326771e-05,
      "loss": 0.1079,
      "step": 1969
    },
    {
      "epoch": 3.747028055159296,
      "grad_norm": 0.4008922576904297,
      "learning_rate": 8.751984757065735e-05,
      "loss": 0.1752,
      "step": 1970
    },
    {
      "epoch": 3.7489300998573465,
      "grad_norm": 0.3501138687133789,
      "learning_rate": 8.751349634804701e-05,
      "loss": 0.1296,
      "step": 1971
    },
    {
      "epoch": 3.750832144555397,
      "grad_norm": 0.3441070318222046,
      "learning_rate": 8.750714512543665e-05,
      "loss": 0.1161,
      "step": 1972
    },
    {
      "epoch": 3.7527341892534474,
      "grad_norm": 0.42847099900245667,
      "learning_rate": 8.750079390282629e-05,
      "loss": 0.1483,
      "step": 1973
    },
    {
      "epoch": 3.754636233951498,
      "grad_norm": 0.4879817068576813,
      "learning_rate": 8.749444268021594e-05,
      "loss": 0.1725,
      "step": 1974
    },
    {
      "epoch": 3.7565382786495483,
      "grad_norm": 0.32576873898506165,
      "learning_rate": 8.74880914576056e-05,
      "loss": 0.1211,
      "step": 1975
    },
    {
      "epoch": 3.7584403233475987,
      "grad_norm": 0.4470548927783966,
      "learning_rate": 8.748174023499523e-05,
      "loss": 0.155,
      "step": 1976
    },
    {
      "epoch": 3.760342368045649,
      "grad_norm": 0.506020724773407,
      "learning_rate": 8.747538901238488e-05,
      "loss": 0.1924,
      "step": 1977
    },
    {
      "epoch": 3.7622444127436996,
      "grad_norm": 0.3949258625507355,
      "learning_rate": 8.746903778977454e-05,
      "loss": 0.1365,
      "step": 1978
    },
    {
      "epoch": 3.76414645744175,
      "grad_norm": 0.381511390209198,
      "learning_rate": 8.746268656716419e-05,
      "loss": 0.1706,
      "step": 1979
    },
    {
      "epoch": 3.7660485021398005,
      "grad_norm": 0.32848381996154785,
      "learning_rate": 8.745633534455383e-05,
      "loss": 0.1302,
      "step": 1980
    },
    {
      "epoch": 3.7679505468378505,
      "grad_norm": 0.39011678099632263,
      "learning_rate": 8.744998412194348e-05,
      "loss": 0.1501,
      "step": 1981
    },
    {
      "epoch": 3.7698525915359014,
      "grad_norm": 0.35527095198631287,
      "learning_rate": 8.744363289933313e-05,
      "loss": 0.1218,
      "step": 1982
    },
    {
      "epoch": 3.7717546362339514,
      "grad_norm": 0.4448065459728241,
      "learning_rate": 8.743728167672277e-05,
      "loss": 0.1527,
      "step": 1983
    },
    {
      "epoch": 3.773656680932002,
      "grad_norm": 0.45173025131225586,
      "learning_rate": 8.743093045411242e-05,
      "loss": 0.1546,
      "step": 1984
    },
    {
      "epoch": 3.7755587256300522,
      "grad_norm": 0.3051410913467407,
      "learning_rate": 8.742457923150207e-05,
      "loss": 0.1176,
      "step": 1985
    },
    {
      "epoch": 3.7774607703281027,
      "grad_norm": 0.4559077322483063,
      "learning_rate": 8.741822800889171e-05,
      "loss": 0.1466,
      "step": 1986
    },
    {
      "epoch": 3.779362815026153,
      "grad_norm": 0.33901482820510864,
      "learning_rate": 8.741187678628136e-05,
      "loss": 0.1263,
      "step": 1987
    },
    {
      "epoch": 3.7812648597242036,
      "grad_norm": 0.3377963900566101,
      "learning_rate": 8.740552556367101e-05,
      "loss": 0.1029,
      "step": 1988
    },
    {
      "epoch": 3.783166904422254,
      "grad_norm": 0.3285292088985443,
      "learning_rate": 8.739917434106067e-05,
      "loss": 0.1256,
      "step": 1989
    },
    {
      "epoch": 3.7850689491203044,
      "grad_norm": 0.4042280614376068,
      "learning_rate": 8.73928231184503e-05,
      "loss": 0.1554,
      "step": 1990
    },
    {
      "epoch": 3.786970993818355,
      "grad_norm": 0.374153733253479,
      "learning_rate": 8.738647189583996e-05,
      "loss": 0.1109,
      "step": 1991
    },
    {
      "epoch": 3.788873038516405,
      "grad_norm": 0.3667593002319336,
      "learning_rate": 8.738012067322961e-05,
      "loss": 0.1014,
      "step": 1992
    },
    {
      "epoch": 3.7907750832144558,
      "grad_norm": 0.40893805027008057,
      "learning_rate": 8.737376945061925e-05,
      "loss": 0.137,
      "step": 1993
    },
    {
      "epoch": 3.7926771279125058,
      "grad_norm": 0.4428877830505371,
      "learning_rate": 8.736741822800888e-05,
      "loss": 0.1516,
      "step": 1994
    },
    {
      "epoch": 3.794579172610556,
      "grad_norm": 0.4404061734676361,
      "learning_rate": 8.736106700539855e-05,
      "loss": 0.155,
      "step": 1995
    },
    {
      "epoch": 3.7964812173086067,
      "grad_norm": 0.3298742473125458,
      "learning_rate": 8.735471578278819e-05,
      "loss": 0.1244,
      "step": 1996
    },
    {
      "epoch": 3.798383262006657,
      "grad_norm": 0.36190545558929443,
      "learning_rate": 8.734836456017784e-05,
      "loss": 0.148,
      "step": 1997
    },
    {
      "epoch": 3.8002853067047075,
      "grad_norm": 0.34386786818504333,
      "learning_rate": 8.734201333756749e-05,
      "loss": 0.1479,
      "step": 1998
    },
    {
      "epoch": 3.802187351402758,
      "grad_norm": 0.434257835149765,
      "learning_rate": 8.733566211495713e-05,
      "loss": 0.1624,
      "step": 1999
    },
    {
      "epoch": 3.8040893961008084,
      "grad_norm": 0.369232177734375,
      "learning_rate": 8.732931089234678e-05,
      "loss": 0.1297,
      "step": 2000
    },
    {
      "epoch": 3.805991440798859,
      "grad_norm": 0.31438469886779785,
      "learning_rate": 8.732295966973642e-05,
      "loss": 0.1074,
      "step": 2001
    },
    {
      "epoch": 3.8078934854969093,
      "grad_norm": 0.4128814935684204,
      "learning_rate": 8.731660844712609e-05,
      "loss": 0.1489,
      "step": 2002
    },
    {
      "epoch": 3.8097955301949593,
      "grad_norm": 0.2960624694824219,
      "learning_rate": 8.731025722451572e-05,
      "loss": 0.1063,
      "step": 2003
    },
    {
      "epoch": 3.81169757489301,
      "grad_norm": 0.35740041732788086,
      "learning_rate": 8.730390600190536e-05,
      "loss": 0.1438,
      "step": 2004
    },
    {
      "epoch": 3.81359961959106,
      "grad_norm": 0.3402657210826874,
      "learning_rate": 8.729755477929501e-05,
      "loss": 0.151,
      "step": 2005
    },
    {
      "epoch": 3.8155016642891106,
      "grad_norm": 0.3280869722366333,
      "learning_rate": 8.729120355668467e-05,
      "loss": 0.112,
      "step": 2006
    },
    {
      "epoch": 3.817403708987161,
      "grad_norm": 0.3747129440307617,
      "learning_rate": 8.728485233407432e-05,
      "loss": 0.1191,
      "step": 2007
    },
    {
      "epoch": 3.8193057536852115,
      "grad_norm": 0.3609796464443207,
      "learning_rate": 8.727850111146396e-05,
      "loss": 0.1373,
      "step": 2008
    },
    {
      "epoch": 3.821207798383262,
      "grad_norm": 0.38992708921432495,
      "learning_rate": 8.727214988885361e-05,
      "loss": 0.1474,
      "step": 2009
    },
    {
      "epoch": 3.8231098430813124,
      "grad_norm": 0.3531118929386139,
      "learning_rate": 8.726579866624326e-05,
      "loss": 0.1188,
      "step": 2010
    },
    {
      "epoch": 3.825011887779363,
      "grad_norm": 0.30585137009620667,
      "learning_rate": 8.72594474436329e-05,
      "loss": 0.1072,
      "step": 2011
    },
    {
      "epoch": 3.8269139324774133,
      "grad_norm": 0.40438538789749146,
      "learning_rate": 8.725309622102255e-05,
      "loss": 0.1527,
      "step": 2012
    },
    {
      "epoch": 3.8288159771754637,
      "grad_norm": 0.31290772557258606,
      "learning_rate": 8.72467449984122e-05,
      "loss": 0.1251,
      "step": 2013
    },
    {
      "epoch": 3.830718021873514,
      "grad_norm": 0.389160692691803,
      "learning_rate": 8.724039377580184e-05,
      "loss": 0.1387,
      "step": 2014
    },
    {
      "epoch": 3.8326200665715646,
      "grad_norm": 0.34139397740364075,
      "learning_rate": 8.723404255319149e-05,
      "loss": 0.1205,
      "step": 2015
    },
    {
      "epoch": 3.8345221112696146,
      "grad_norm": 0.4144088923931122,
      "learning_rate": 8.722769133058114e-05,
      "loss": 0.1493,
      "step": 2016
    },
    {
      "epoch": 3.8364241559676655,
      "grad_norm": 0.3793914318084717,
      "learning_rate": 8.722134010797078e-05,
      "loss": 0.1379,
      "step": 2017
    },
    {
      "epoch": 3.8383262006657155,
      "grad_norm": 0.3809344470500946,
      "learning_rate": 8.721498888536043e-05,
      "loss": 0.196,
      "step": 2018
    },
    {
      "epoch": 3.840228245363766,
      "grad_norm": 0.3764810860157013,
      "learning_rate": 8.720863766275009e-05,
      "loss": 0.1096,
      "step": 2019
    },
    {
      "epoch": 3.8421302900618164,
      "grad_norm": 0.47973567247390747,
      "learning_rate": 8.720228644013974e-05,
      "loss": 0.1195,
      "step": 2020
    },
    {
      "epoch": 3.844032334759867,
      "grad_norm": 0.4527863562107086,
      "learning_rate": 8.719593521752938e-05,
      "loss": 0.2112,
      "step": 2021
    },
    {
      "epoch": 3.8459343794579173,
      "grad_norm": 0.39066699147224426,
      "learning_rate": 8.718958399491903e-05,
      "loss": 0.1281,
      "step": 2022
    },
    {
      "epoch": 3.8478364241559677,
      "grad_norm": 0.37056446075439453,
      "learning_rate": 8.718323277230868e-05,
      "loss": 0.1519,
      "step": 2023
    },
    {
      "epoch": 3.849738468854018,
      "grad_norm": 0.516057550907135,
      "learning_rate": 8.717688154969832e-05,
      "loss": 0.1657,
      "step": 2024
    },
    {
      "epoch": 3.8516405135520686,
      "grad_norm": 0.3468872010707855,
      "learning_rate": 8.717053032708797e-05,
      "loss": 0.1408,
      "step": 2025
    },
    {
      "epoch": 3.853542558250119,
      "grad_norm": 0.5452744364738464,
      "learning_rate": 8.716417910447762e-05,
      "loss": 0.3173,
      "step": 2026
    },
    {
      "epoch": 3.855444602948169,
      "grad_norm": 0.4378301501274109,
      "learning_rate": 8.715782788186726e-05,
      "loss": 0.136,
      "step": 2027
    },
    {
      "epoch": 3.85734664764622,
      "grad_norm": 0.49818679690361023,
      "learning_rate": 8.715147665925691e-05,
      "loss": 0.233,
      "step": 2028
    },
    {
      "epoch": 3.85924869234427,
      "grad_norm": 0.4228188693523407,
      "learning_rate": 8.714512543664656e-05,
      "loss": 0.1485,
      "step": 2029
    },
    {
      "epoch": 3.8611507370423204,
      "grad_norm": 0.34110891819000244,
      "learning_rate": 8.71387742140362e-05,
      "loss": 0.1455,
      "step": 2030
    },
    {
      "epoch": 3.863052781740371,
      "grad_norm": 0.38667479157447815,
      "learning_rate": 8.713242299142585e-05,
      "loss": 0.1302,
      "step": 2031
    },
    {
      "epoch": 3.8649548264384213,
      "grad_norm": 0.3971845805644989,
      "learning_rate": 8.712607176881549e-05,
      "loss": 0.1562,
      "step": 2032
    },
    {
      "epoch": 3.8668568711364717,
      "grad_norm": 0.32637760043144226,
      "learning_rate": 8.711972054620516e-05,
      "loss": 0.1213,
      "step": 2033
    },
    {
      "epoch": 3.868758915834522,
      "grad_norm": 0.3475836217403412,
      "learning_rate": 8.71133693235948e-05,
      "loss": 0.1514,
      "step": 2034
    },
    {
      "epoch": 3.8706609605325726,
      "grad_norm": 0.37775367498397827,
      "learning_rate": 8.710701810098443e-05,
      "loss": 0.1672,
      "step": 2035
    },
    {
      "epoch": 3.872563005230623,
      "grad_norm": 0.4611580967903137,
      "learning_rate": 8.71006668783741e-05,
      "loss": 0.1977,
      "step": 2036
    },
    {
      "epoch": 3.8744650499286735,
      "grad_norm": 0.34681427478790283,
      "learning_rate": 8.709431565576374e-05,
      "loss": 0.127,
      "step": 2037
    },
    {
      "epoch": 3.8763670946267235,
      "grad_norm": 0.3547581732273102,
      "learning_rate": 8.708796443315339e-05,
      "loss": 0.1432,
      "step": 2038
    },
    {
      "epoch": 3.8782691393247744,
      "grad_norm": 0.3560992479324341,
      "learning_rate": 8.708161321054303e-05,
      "loss": 0.1269,
      "step": 2039
    },
    {
      "epoch": 3.8801711840228243,
      "grad_norm": 0.48965948820114136,
      "learning_rate": 8.707526198793268e-05,
      "loss": 0.1694,
      "step": 2040
    },
    {
      "epoch": 3.882073228720875,
      "grad_norm": 0.4042951464653015,
      "learning_rate": 8.706891076532233e-05,
      "loss": 0.1432,
      "step": 2041
    },
    {
      "epoch": 3.8839752734189252,
      "grad_norm": 0.40321534872055054,
      "learning_rate": 8.706255954271197e-05,
      "loss": 0.1206,
      "step": 2042
    },
    {
      "epoch": 3.8858773181169757,
      "grad_norm": 0.5154759883880615,
      "learning_rate": 8.705620832010164e-05,
      "loss": 0.2034,
      "step": 2043
    },
    {
      "epoch": 3.887779362815026,
      "grad_norm": 0.3707939684391022,
      "learning_rate": 8.704985709749127e-05,
      "loss": 0.1408,
      "step": 2044
    },
    {
      "epoch": 3.8896814075130766,
      "grad_norm": 0.46117648482322693,
      "learning_rate": 8.704350587488091e-05,
      "loss": 0.1921,
      "step": 2045
    },
    {
      "epoch": 3.891583452211127,
      "grad_norm": 0.4917357265949249,
      "learning_rate": 8.703715465227056e-05,
      "loss": 0.1684,
      "step": 2046
    },
    {
      "epoch": 3.8934854969091774,
      "grad_norm": 0.36523228883743286,
      "learning_rate": 8.703080342966022e-05,
      "loss": 0.1977,
      "step": 2047
    },
    {
      "epoch": 3.895387541607228,
      "grad_norm": 0.3557770550251007,
      "learning_rate": 8.702445220704985e-05,
      "loss": 0.1326,
      "step": 2048
    },
    {
      "epoch": 3.8972895863052783,
      "grad_norm": 0.2716139853000641,
      "learning_rate": 8.70181009844395e-05,
      "loss": 0.1119,
      "step": 2049
    },
    {
      "epoch": 3.8991916310033288,
      "grad_norm": 0.3266098201274872,
      "learning_rate": 8.701174976182916e-05,
      "loss": 0.1355,
      "step": 2050
    },
    {
      "epoch": 3.9010936757013788,
      "grad_norm": 0.4549683928489685,
      "learning_rate": 8.700539853921881e-05,
      "loss": 0.174,
      "step": 2051
    },
    {
      "epoch": 3.9029957203994297,
      "grad_norm": 0.3865867555141449,
      "learning_rate": 8.699904731660845e-05,
      "loss": 0.131,
      "step": 2052
    },
    {
      "epoch": 3.9048977650974797,
      "grad_norm": 0.4354785084724426,
      "learning_rate": 8.69926960939981e-05,
      "loss": 0.1497,
      "step": 2053
    },
    {
      "epoch": 3.90679980979553,
      "grad_norm": 0.38822686672210693,
      "learning_rate": 8.698634487138775e-05,
      "loss": 0.1272,
      "step": 2054
    },
    {
      "epoch": 3.9087018544935805,
      "grad_norm": 0.4395056366920471,
      "learning_rate": 8.697999364877739e-05,
      "loss": 0.1801,
      "step": 2055
    },
    {
      "epoch": 3.910603899191631,
      "grad_norm": 0.4310166835784912,
      "learning_rate": 8.697364242616704e-05,
      "loss": 0.1457,
      "step": 2056
    },
    {
      "epoch": 3.9125059438896814,
      "grad_norm": 0.42527538537979126,
      "learning_rate": 8.69672912035567e-05,
      "loss": 0.1827,
      "step": 2057
    },
    {
      "epoch": 3.914407988587732,
      "grad_norm": 0.41284388303756714,
      "learning_rate": 8.696093998094633e-05,
      "loss": 0.1588,
      "step": 2058
    },
    {
      "epoch": 3.9163100332857823,
      "grad_norm": 0.3561374247074127,
      "learning_rate": 8.695458875833598e-05,
      "loss": 0.138,
      "step": 2059
    },
    {
      "epoch": 3.9182120779838328,
      "grad_norm": 0.4057970941066742,
      "learning_rate": 8.694823753572564e-05,
      "loss": 0.1504,
      "step": 2060
    },
    {
      "epoch": 3.920114122681883,
      "grad_norm": 0.47292712330818176,
      "learning_rate": 8.694188631311529e-05,
      "loss": 0.1417,
      "step": 2061
    },
    {
      "epoch": 3.922016167379933,
      "grad_norm": 0.4207940995693207,
      "learning_rate": 8.693553509050493e-05,
      "loss": 0.1372,
      "step": 2062
    },
    {
      "epoch": 3.923918212077984,
      "grad_norm": 0.5482998490333557,
      "learning_rate": 8.692918386789456e-05,
      "loss": 0.1917,
      "step": 2063
    },
    {
      "epoch": 3.925820256776034,
      "grad_norm": 0.41113635897636414,
      "learning_rate": 8.692283264528423e-05,
      "loss": 0.1479,
      "step": 2064
    },
    {
      "epoch": 3.9277223014740845,
      "grad_norm": 0.3470059037208557,
      "learning_rate": 8.691648142267387e-05,
      "loss": 0.1235,
      "step": 2065
    },
    {
      "epoch": 3.929624346172135,
      "grad_norm": 0.4131185710430145,
      "learning_rate": 8.69101302000635e-05,
      "loss": 0.1476,
      "step": 2066
    },
    {
      "epoch": 3.9315263908701854,
      "grad_norm": 0.3750738501548767,
      "learning_rate": 8.690377897745317e-05,
      "loss": 0.1517,
      "step": 2067
    },
    {
      "epoch": 3.933428435568236,
      "grad_norm": 0.37411704659461975,
      "learning_rate": 8.689742775484281e-05,
      "loss": 0.1493,
      "step": 2068
    },
    {
      "epoch": 3.9353304802662863,
      "grad_norm": 0.4208986759185791,
      "learning_rate": 8.689107653223246e-05,
      "loss": 0.1558,
      "step": 2069
    },
    {
      "epoch": 3.9372325249643367,
      "grad_norm": 0.36959660053253174,
      "learning_rate": 8.68847253096221e-05,
      "loss": 0.1247,
      "step": 2070
    },
    {
      "epoch": 3.939134569662387,
      "grad_norm": 0.3977148234844208,
      "learning_rate": 8.687837408701175e-05,
      "loss": 0.1428,
      "step": 2071
    },
    {
      "epoch": 3.9410366143604376,
      "grad_norm": 0.40076392889022827,
      "learning_rate": 8.68720228644014e-05,
      "loss": 0.1652,
      "step": 2072
    },
    {
      "epoch": 3.9429386590584876,
      "grad_norm": 0.3828325569629669,
      "learning_rate": 8.686567164179104e-05,
      "loss": 0.1518,
      "step": 2073
    },
    {
      "epoch": 3.9448407037565385,
      "grad_norm": 0.35112518072128296,
      "learning_rate": 8.685932041918071e-05,
      "loss": 0.1303,
      "step": 2074
    },
    {
      "epoch": 3.9467427484545885,
      "grad_norm": 0.31564921140670776,
      "learning_rate": 8.685296919657035e-05,
      "loss": 0.1325,
      "step": 2075
    },
    {
      "epoch": 3.948644793152639,
      "grad_norm": 0.3110829293727875,
      "learning_rate": 8.684661797395998e-05,
      "loss": 0.0958,
      "step": 2076
    },
    {
      "epoch": 3.9505468378506894,
      "grad_norm": 0.41574040055274963,
      "learning_rate": 8.684026675134964e-05,
      "loss": 0.142,
      "step": 2077
    },
    {
      "epoch": 3.95244888254874,
      "grad_norm": 0.4371127188205719,
      "learning_rate": 8.683391552873929e-05,
      "loss": 0.1699,
      "step": 2078
    },
    {
      "epoch": 3.9543509272467903,
      "grad_norm": 0.41888341307640076,
      "learning_rate": 8.682756430612894e-05,
      "loss": 0.1467,
      "step": 2079
    },
    {
      "epoch": 3.9562529719448407,
      "grad_norm": 0.4013144373893738,
      "learning_rate": 8.682121308351858e-05,
      "loss": 0.1541,
      "step": 2080
    },
    {
      "epoch": 3.958155016642891,
      "grad_norm": 0.3627847135066986,
      "learning_rate": 8.681486186090823e-05,
      "loss": 0.1412,
      "step": 2081
    },
    {
      "epoch": 3.9600570613409416,
      "grad_norm": 0.34517934918403625,
      "learning_rate": 8.680851063829788e-05,
      "loss": 0.1302,
      "step": 2082
    },
    {
      "epoch": 3.961959106038992,
      "grad_norm": 0.409612238407135,
      "learning_rate": 8.680215941568752e-05,
      "loss": 0.1806,
      "step": 2083
    },
    {
      "epoch": 3.9638611507370425,
      "grad_norm": 0.37562572956085205,
      "learning_rate": 8.679580819307717e-05,
      "loss": 0.1305,
      "step": 2084
    },
    {
      "epoch": 3.965763195435093,
      "grad_norm": 0.30839917063713074,
      "learning_rate": 8.678945697046682e-05,
      "loss": 0.1179,
      "step": 2085
    },
    {
      "epoch": 3.967665240133143,
      "grad_norm": 0.4009683430194855,
      "learning_rate": 8.678310574785646e-05,
      "loss": 0.1392,
      "step": 2086
    },
    {
      "epoch": 3.969567284831194,
      "grad_norm": 0.5373052358627319,
      "learning_rate": 8.677675452524611e-05,
      "loss": 0.2366,
      "step": 2087
    },
    {
      "epoch": 3.971469329529244,
      "grad_norm": 0.44061073660850525,
      "learning_rate": 8.677040330263576e-05,
      "loss": 0.1541,
      "step": 2088
    },
    {
      "epoch": 3.9733713742272943,
      "grad_norm": 0.6880194544792175,
      "learning_rate": 8.67640520800254e-05,
      "loss": 0.1822,
      "step": 2089
    },
    {
      "epoch": 3.9752734189253447,
      "grad_norm": 0.4342186450958252,
      "learning_rate": 8.675770085741505e-05,
      "loss": 0.1398,
      "step": 2090
    },
    {
      "epoch": 3.977175463623395,
      "grad_norm": 0.3437482714653015,
      "learning_rate": 8.675134963480471e-05,
      "loss": 0.1407,
      "step": 2091
    },
    {
      "epoch": 3.9790775083214456,
      "grad_norm": 0.43729832768440247,
      "learning_rate": 8.674499841219436e-05,
      "loss": 0.1604,
      "step": 2092
    },
    {
      "epoch": 3.980979553019496,
      "grad_norm": 0.36654895544052124,
      "learning_rate": 8.6738647189584e-05,
      "loss": 0.1261,
      "step": 2093
    },
    {
      "epoch": 3.9828815977175465,
      "grad_norm": 0.40422323346138,
      "learning_rate": 8.673229596697365e-05,
      "loss": 0.1463,
      "step": 2094
    },
    {
      "epoch": 3.984783642415597,
      "grad_norm": 0.37436428666114807,
      "learning_rate": 8.67259447443633e-05,
      "loss": 0.1283,
      "step": 2095
    },
    {
      "epoch": 3.9866856871136473,
      "grad_norm": 0.4568138122558594,
      "learning_rate": 8.671959352175294e-05,
      "loss": 0.1735,
      "step": 2096
    },
    {
      "epoch": 3.9885877318116973,
      "grad_norm": 0.3864310681819916,
      "learning_rate": 8.671324229914259e-05,
      "loss": 0.1458,
      "step": 2097
    },
    {
      "epoch": 3.9904897765097482,
      "grad_norm": 0.3622378408908844,
      "learning_rate": 8.670689107653224e-05,
      "loss": 0.1333,
      "step": 2098
    },
    {
      "epoch": 3.9923918212077982,
      "grad_norm": 0.5126944780349731,
      "learning_rate": 8.670053985392188e-05,
      "loss": 0.1897,
      "step": 2099
    },
    {
      "epoch": 3.9942938659058487,
      "grad_norm": 0.3905584216117859,
      "learning_rate": 8.669418863131153e-05,
      "loss": 0.1743,
      "step": 2100
    },
    {
      "epoch": 3.996195910603899,
      "grad_norm": 0.4149746298789978,
      "learning_rate": 8.668783740870118e-05,
      "loss": 0.1686,
      "step": 2101
    },
    {
      "epoch": 3.9980979553019496,
      "grad_norm": 0.30447009205818176,
      "learning_rate": 8.668148618609082e-05,
      "loss": 0.1079,
      "step": 2102
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.533173143863678,
      "learning_rate": 8.667513496348047e-05,
      "loss": 0.1652,
      "step": 2103
    },
    {
      "epoch": 4.00190204469805,
      "grad_norm": 0.26669684052467346,
      "learning_rate": 8.666878374087011e-05,
      "loss": 0.1105,
      "step": 2104
    },
    {
      "epoch": 4.003804089396101,
      "grad_norm": 0.2511195242404938,
      "learning_rate": 8.666243251825978e-05,
      "loss": 0.1018,
      "step": 2105
    },
    {
      "epoch": 4.005706134094151,
      "grad_norm": 0.2838079035282135,
      "learning_rate": 8.665608129564942e-05,
      "loss": 0.0979,
      "step": 2106
    },
    {
      "epoch": 4.007608178792202,
      "grad_norm": 0.3789231479167938,
      "learning_rate": 8.664973007303905e-05,
      "loss": 0.1216,
      "step": 2107
    },
    {
      "epoch": 4.009510223490252,
      "grad_norm": 0.36412686109542847,
      "learning_rate": 8.664337885042872e-05,
      "loss": 0.0924,
      "step": 2108
    },
    {
      "epoch": 4.011412268188303,
      "grad_norm": 0.3399736285209656,
      "learning_rate": 8.663702762781836e-05,
      "loss": 0.1007,
      "step": 2109
    },
    {
      "epoch": 4.013314312886353,
      "grad_norm": 0.3104216456413269,
      "learning_rate": 8.663067640520801e-05,
      "loss": 0.1146,
      "step": 2110
    },
    {
      "epoch": 4.0152163575844035,
      "grad_norm": 0.33002039790153503,
      "learning_rate": 8.662432518259765e-05,
      "loss": 0.1112,
      "step": 2111
    },
    {
      "epoch": 4.0171184022824535,
      "grad_norm": 0.3158220946788788,
      "learning_rate": 8.66179739599873e-05,
      "loss": 0.0983,
      "step": 2112
    },
    {
      "epoch": 4.019020446980504,
      "grad_norm": 0.3281852900981903,
      "learning_rate": 8.661162273737695e-05,
      "loss": 0.1002,
      "step": 2113
    },
    {
      "epoch": 4.020922491678554,
      "grad_norm": 0.42810752987861633,
      "learning_rate": 8.660527151476659e-05,
      "loss": 0.145,
      "step": 2114
    },
    {
      "epoch": 4.022824536376604,
      "grad_norm": 0.343757301568985,
      "learning_rate": 8.659892029215624e-05,
      "loss": 0.1046,
      "step": 2115
    },
    {
      "epoch": 4.024726581074655,
      "grad_norm": 0.3978208601474762,
      "learning_rate": 8.65925690695459e-05,
      "loss": 0.1232,
      "step": 2116
    },
    {
      "epoch": 4.026628625772705,
      "grad_norm": 0.3716939687728882,
      "learning_rate": 8.658621784693553e-05,
      "loss": 0.1073,
      "step": 2117
    },
    {
      "epoch": 4.028530670470756,
      "grad_norm": 0.3938986659049988,
      "learning_rate": 8.657986662432518e-05,
      "loss": 0.1162,
      "step": 2118
    },
    {
      "epoch": 4.030432715168806,
      "grad_norm": 0.26515620946884155,
      "learning_rate": 8.657351540171484e-05,
      "loss": 0.0927,
      "step": 2119
    },
    {
      "epoch": 4.032334759866857,
      "grad_norm": 0.4481755197048187,
      "learning_rate": 8.656716417910447e-05,
      "loss": 0.1192,
      "step": 2120
    },
    {
      "epoch": 4.034236804564907,
      "grad_norm": 0.2902253568172455,
      "learning_rate": 8.656081295649413e-05,
      "loss": 0.0972,
      "step": 2121
    },
    {
      "epoch": 4.036138849262958,
      "grad_norm": 0.3764674961566925,
      "learning_rate": 8.655446173388378e-05,
      "loss": 0.1242,
      "step": 2122
    },
    {
      "epoch": 4.038040893961008,
      "grad_norm": 0.4040977954864502,
      "learning_rate": 8.654811051127343e-05,
      "loss": 0.1053,
      "step": 2123
    },
    {
      "epoch": 4.039942938659059,
      "grad_norm": 0.3967365026473999,
      "learning_rate": 8.654175928866307e-05,
      "loss": 0.1132,
      "step": 2124
    },
    {
      "epoch": 4.041844983357109,
      "grad_norm": 0.4135635197162628,
      "learning_rate": 8.653540806605272e-05,
      "loss": 0.1171,
      "step": 2125
    },
    {
      "epoch": 4.04374702805516,
      "grad_norm": 0.43473535776138306,
      "learning_rate": 8.652905684344237e-05,
      "loss": 0.1227,
      "step": 2126
    },
    {
      "epoch": 4.04564907275321,
      "grad_norm": 0.30436238646507263,
      "learning_rate": 8.652270562083201e-05,
      "loss": 0.0853,
      "step": 2127
    },
    {
      "epoch": 4.04755111745126,
      "grad_norm": 0.3265203535556793,
      "learning_rate": 8.651635439822166e-05,
      "loss": 0.1007,
      "step": 2128
    },
    {
      "epoch": 4.049453162149311,
      "grad_norm": 0.3733639121055603,
      "learning_rate": 8.651000317561131e-05,
      "loss": 0.1164,
      "step": 2129
    },
    {
      "epoch": 4.051355206847361,
      "grad_norm": 0.3707481324672699,
      "learning_rate": 8.650365195300095e-05,
      "loss": 0.1225,
      "step": 2130
    },
    {
      "epoch": 4.0532572515454115,
      "grad_norm": 0.39869242906570435,
      "learning_rate": 8.64973007303906e-05,
      "loss": 0.1127,
      "step": 2131
    },
    {
      "epoch": 4.0551592962434615,
      "grad_norm": 0.31656894087791443,
      "learning_rate": 8.649094950778026e-05,
      "loss": 0.0936,
      "step": 2132
    },
    {
      "epoch": 4.057061340941512,
      "grad_norm": 0.32848450541496277,
      "learning_rate": 8.648459828516991e-05,
      "loss": 0.1192,
      "step": 2133
    },
    {
      "epoch": 4.058963385639562,
      "grad_norm": 0.41309690475463867,
      "learning_rate": 8.647824706255955e-05,
      "loss": 0.1224,
      "step": 2134
    },
    {
      "epoch": 4.060865430337613,
      "grad_norm": 0.30171439051628113,
      "learning_rate": 8.647189583994918e-05,
      "loss": 0.1108,
      "step": 2135
    },
    {
      "epoch": 4.062767475035663,
      "grad_norm": 0.31793013215065,
      "learning_rate": 8.646554461733885e-05,
      "loss": 0.0958,
      "step": 2136
    },
    {
      "epoch": 4.064669519733714,
      "grad_norm": 0.3515986502170563,
      "learning_rate": 8.645919339472849e-05,
      "loss": 0.098,
      "step": 2137
    },
    {
      "epoch": 4.066571564431764,
      "grad_norm": 0.2572970390319824,
      "learning_rate": 8.645284217211813e-05,
      "loss": 0.0782,
      "step": 2138
    },
    {
      "epoch": 4.068473609129814,
      "grad_norm": 0.40460988879203796,
      "learning_rate": 8.644649094950779e-05,
      "loss": 0.111,
      "step": 2139
    },
    {
      "epoch": 4.070375653827865,
      "grad_norm": 0.25654932856559753,
      "learning_rate": 8.644013972689743e-05,
      "loss": 0.078,
      "step": 2140
    },
    {
      "epoch": 4.072277698525915,
      "grad_norm": 0.3793332278728485,
      "learning_rate": 8.643378850428708e-05,
      "loss": 0.1113,
      "step": 2141
    },
    {
      "epoch": 4.074179743223966,
      "grad_norm": 0.3457014560699463,
      "learning_rate": 8.642743728167672e-05,
      "loss": 0.1016,
      "step": 2142
    },
    {
      "epoch": 4.076081787922016,
      "grad_norm": 0.41619420051574707,
      "learning_rate": 8.642108605906637e-05,
      "loss": 0.1379,
      "step": 2143
    },
    {
      "epoch": 4.077983832620067,
      "grad_norm": 0.3582102656364441,
      "learning_rate": 8.641473483645602e-05,
      "loss": 0.1068,
      "step": 2144
    },
    {
      "epoch": 4.079885877318117,
      "grad_norm": 0.4142124652862549,
      "learning_rate": 8.640838361384566e-05,
      "loss": 0.1155,
      "step": 2145
    },
    {
      "epoch": 4.081787922016168,
      "grad_norm": 0.3544979393482208,
      "learning_rate": 8.640203239123533e-05,
      "loss": 0.0969,
      "step": 2146
    },
    {
      "epoch": 4.083689966714218,
      "grad_norm": 0.37561002373695374,
      "learning_rate": 8.639568116862497e-05,
      "loss": 0.1218,
      "step": 2147
    },
    {
      "epoch": 4.085592011412269,
      "grad_norm": 0.3568158447742462,
      "learning_rate": 8.63893299460146e-05,
      "loss": 0.1225,
      "step": 2148
    },
    {
      "epoch": 4.087494056110319,
      "grad_norm": 0.3126932382583618,
      "learning_rate": 8.638297872340426e-05,
      "loss": 0.084,
      "step": 2149
    },
    {
      "epoch": 4.089396100808369,
      "grad_norm": 0.4232020378112793,
      "learning_rate": 8.637662750079391e-05,
      "loss": 0.1155,
      "step": 2150
    },
    {
      "epoch": 4.0912981455064195,
      "grad_norm": 0.4121897518634796,
      "learning_rate": 8.637027627818356e-05,
      "loss": 0.1352,
      "step": 2151
    },
    {
      "epoch": 4.0932001902044695,
      "grad_norm": 0.3292025923728943,
      "learning_rate": 8.63639250555732e-05,
      "loss": 0.115,
      "step": 2152
    },
    {
      "epoch": 4.09510223490252,
      "grad_norm": 0.3273860514163971,
      "learning_rate": 8.635757383296285e-05,
      "loss": 0.1087,
      "step": 2153
    },
    {
      "epoch": 4.09700427960057,
      "grad_norm": 0.36760157346725464,
      "learning_rate": 8.63512226103525e-05,
      "loss": 0.1206,
      "step": 2154
    },
    {
      "epoch": 4.098906324298621,
      "grad_norm": 0.3717329502105713,
      "learning_rate": 8.634487138774214e-05,
      "loss": 0.1244,
      "step": 2155
    },
    {
      "epoch": 4.100808368996671,
      "grad_norm": 0.379068523645401,
      "learning_rate": 8.633852016513179e-05,
      "loss": 0.1048,
      "step": 2156
    },
    {
      "epoch": 4.102710413694722,
      "grad_norm": 0.30912551283836365,
      "learning_rate": 8.633216894252144e-05,
      "loss": 0.0838,
      "step": 2157
    },
    {
      "epoch": 4.104612458392772,
      "grad_norm": 0.3093559741973877,
      "learning_rate": 8.632581771991108e-05,
      "loss": 0.0948,
      "step": 2158
    },
    {
      "epoch": 4.106514503090823,
      "grad_norm": 0.2924623489379883,
      "learning_rate": 8.631946649730073e-05,
      "loss": 0.085,
      "step": 2159
    },
    {
      "epoch": 4.108416547788873,
      "grad_norm": 0.335437536239624,
      "learning_rate": 8.631311527469039e-05,
      "loss": 0.102,
      "step": 2160
    },
    {
      "epoch": 4.110318592486923,
      "grad_norm": 0.37450480461120605,
      "learning_rate": 8.630676405208002e-05,
      "loss": 0.1102,
      "step": 2161
    },
    {
      "epoch": 4.112220637184974,
      "grad_norm": 0.40548086166381836,
      "learning_rate": 8.630041282946968e-05,
      "loss": 0.1122,
      "step": 2162
    },
    {
      "epoch": 4.114122681883024,
      "grad_norm": 0.2255704551935196,
      "learning_rate": 8.629406160685933e-05,
      "loss": 0.0875,
      "step": 2163
    },
    {
      "epoch": 4.116024726581075,
      "grad_norm": 0.3774515390396118,
      "learning_rate": 8.628771038424898e-05,
      "loss": 0.1007,
      "step": 2164
    },
    {
      "epoch": 4.117926771279125,
      "grad_norm": 0.4410356879234314,
      "learning_rate": 8.628135916163862e-05,
      "loss": 0.1238,
      "step": 2165
    },
    {
      "epoch": 4.119828815977176,
      "grad_norm": 0.3007069230079651,
      "learning_rate": 8.627500793902826e-05,
      "loss": 0.0849,
      "step": 2166
    },
    {
      "epoch": 4.121730860675226,
      "grad_norm": 0.3165019750595093,
      "learning_rate": 8.626865671641792e-05,
      "loss": 0.0959,
      "step": 2167
    },
    {
      "epoch": 4.1236329053732765,
      "grad_norm": 0.3213941752910614,
      "learning_rate": 8.626230549380756e-05,
      "loss": 0.1011,
      "step": 2168
    },
    {
      "epoch": 4.1255349500713265,
      "grad_norm": 0.2742742598056793,
      "learning_rate": 8.625595427119721e-05,
      "loss": 0.0855,
      "step": 2169
    },
    {
      "epoch": 4.127436994769377,
      "grad_norm": 0.35063308477401733,
      "learning_rate": 8.624960304858686e-05,
      "loss": 0.1115,
      "step": 2170
    },
    {
      "epoch": 4.129339039467427,
      "grad_norm": 0.4272489845752716,
      "learning_rate": 8.62432518259765e-05,
      "loss": 0.1162,
      "step": 2171
    },
    {
      "epoch": 4.131241084165478,
      "grad_norm": 0.27256911993026733,
      "learning_rate": 8.623690060336615e-05,
      "loss": 0.1066,
      "step": 2172
    },
    {
      "epoch": 4.133143128863528,
      "grad_norm": 0.275309294462204,
      "learning_rate": 8.623054938075579e-05,
      "loss": 0.1029,
      "step": 2173
    },
    {
      "epoch": 4.135045173561578,
      "grad_norm": 0.2678431570529938,
      "learning_rate": 8.622419815814544e-05,
      "loss": 0.0836,
      "step": 2174
    },
    {
      "epoch": 4.136947218259629,
      "grad_norm": 0.3313474953174591,
      "learning_rate": 8.62178469355351e-05,
      "loss": 0.0925,
      "step": 2175
    },
    {
      "epoch": 4.138849262957679,
      "grad_norm": 0.2514117658138275,
      "learning_rate": 8.621149571292473e-05,
      "loss": 0.0905,
      "step": 2176
    },
    {
      "epoch": 4.14075130765573,
      "grad_norm": 0.2868940532207489,
      "learning_rate": 8.62051444903144e-05,
      "loss": 0.1057,
      "step": 2177
    },
    {
      "epoch": 4.14265335235378,
      "grad_norm": 0.3867243826389313,
      "learning_rate": 8.619879326770404e-05,
      "loss": 0.1151,
      "step": 2178
    },
    {
      "epoch": 4.144555397051831,
      "grad_norm": 0.3011827766895294,
      "learning_rate": 8.619244204509368e-05,
      "loss": 0.1152,
      "step": 2179
    },
    {
      "epoch": 4.146457441749881,
      "grad_norm": 0.33059659600257874,
      "learning_rate": 8.618609082248333e-05,
      "loss": 0.1121,
      "step": 2180
    },
    {
      "epoch": 4.148359486447932,
      "grad_norm": 0.45777612924575806,
      "learning_rate": 8.617973959987298e-05,
      "loss": 0.133,
      "step": 2181
    },
    {
      "epoch": 4.150261531145982,
      "grad_norm": 0.39224299788475037,
      "learning_rate": 8.617338837726263e-05,
      "loss": 0.1381,
      "step": 2182
    },
    {
      "epoch": 4.152163575844033,
      "grad_norm": 0.2813168168067932,
      "learning_rate": 8.616703715465227e-05,
      "loss": 0.0939,
      "step": 2183
    },
    {
      "epoch": 4.154065620542083,
      "grad_norm": 0.30850479006767273,
      "learning_rate": 8.616068593204192e-05,
      "loss": 0.1016,
      "step": 2184
    },
    {
      "epoch": 4.155967665240133,
      "grad_norm": 0.2755066156387329,
      "learning_rate": 8.615433470943157e-05,
      "loss": 0.1253,
      "step": 2185
    },
    {
      "epoch": 4.157869709938184,
      "grad_norm": 0.25375935435295105,
      "learning_rate": 8.614798348682121e-05,
      "loss": 0.088,
      "step": 2186
    },
    {
      "epoch": 4.159771754636234,
      "grad_norm": 0.27644097805023193,
      "learning_rate": 8.614163226421086e-05,
      "loss": 0.1053,
      "step": 2187
    },
    {
      "epoch": 4.1616737993342845,
      "grad_norm": 0.30916059017181396,
      "learning_rate": 8.613528104160052e-05,
      "loss": 0.1075,
      "step": 2188
    },
    {
      "epoch": 4.1635758440323345,
      "grad_norm": 0.3316441476345062,
      "learning_rate": 8.612892981899015e-05,
      "loss": 0.1087,
      "step": 2189
    },
    {
      "epoch": 4.165477888730385,
      "grad_norm": 0.27464917302131653,
      "learning_rate": 8.61225785963798e-05,
      "loss": 0.079,
      "step": 2190
    },
    {
      "epoch": 4.167379933428435,
      "grad_norm": 0.3684466779232025,
      "learning_rate": 8.611622737376946e-05,
      "loss": 0.1312,
      "step": 2191
    },
    {
      "epoch": 4.169281978126486,
      "grad_norm": 0.33914482593536377,
      "learning_rate": 8.61098761511591e-05,
      "loss": 0.0991,
      "step": 2192
    },
    {
      "epoch": 4.171184022824536,
      "grad_norm": 0.3610948324203491,
      "learning_rate": 8.610352492854875e-05,
      "loss": 0.1068,
      "step": 2193
    },
    {
      "epoch": 4.173086067522587,
      "grad_norm": 0.2824098765850067,
      "learning_rate": 8.60971737059384e-05,
      "loss": 0.0913,
      "step": 2194
    },
    {
      "epoch": 4.174988112220637,
      "grad_norm": 0.28685760498046875,
      "learning_rate": 8.609082248332805e-05,
      "loss": 0.098,
      "step": 2195
    },
    {
      "epoch": 4.176890156918688,
      "grad_norm": 0.44503989815711975,
      "learning_rate": 8.608447126071769e-05,
      "loss": 0.1441,
      "step": 2196
    },
    {
      "epoch": 4.178792201616738,
      "grad_norm": 0.4228593409061432,
      "learning_rate": 8.607812003810734e-05,
      "loss": 0.1228,
      "step": 2197
    },
    {
      "epoch": 4.180694246314788,
      "grad_norm": 0.34366467595100403,
      "learning_rate": 8.607176881549699e-05,
      "loss": 0.0969,
      "step": 2198
    },
    {
      "epoch": 4.182596291012839,
      "grad_norm": 0.3302469849586487,
      "learning_rate": 8.606541759288663e-05,
      "loss": 0.1093,
      "step": 2199
    },
    {
      "epoch": 4.184498335710889,
      "grad_norm": 0.316914826631546,
      "learning_rate": 8.605906637027628e-05,
      "loss": 0.096,
      "step": 2200
    },
    {
      "epoch": 4.18640038040894,
      "grad_norm": 0.3100655972957611,
      "learning_rate": 8.605271514766594e-05,
      "loss": 0.0902,
      "step": 2201
    },
    {
      "epoch": 4.18830242510699,
      "grad_norm": 0.2934771776199341,
      "learning_rate": 8.604636392505557e-05,
      "loss": 0.1011,
      "step": 2202
    },
    {
      "epoch": 4.190204469805041,
      "grad_norm": 0.32837802171707153,
      "learning_rate": 8.604001270244523e-05,
      "loss": 0.1284,
      "step": 2203
    },
    {
      "epoch": 4.192106514503091,
      "grad_norm": 0.3842618465423584,
      "learning_rate": 8.603366147983488e-05,
      "loss": 0.1072,
      "step": 2204
    },
    {
      "epoch": 4.194008559201142,
      "grad_norm": 0.29006102681159973,
      "learning_rate": 8.602731025722453e-05,
      "loss": 0.0919,
      "step": 2205
    },
    {
      "epoch": 4.195910603899192,
      "grad_norm": 0.31507110595703125,
      "learning_rate": 8.602095903461417e-05,
      "loss": 0.1103,
      "step": 2206
    },
    {
      "epoch": 4.1978126485972425,
      "grad_norm": 0.35961470007896423,
      "learning_rate": 8.60146078120038e-05,
      "loss": 0.1738,
      "step": 2207
    },
    {
      "epoch": 4.1997146932952925,
      "grad_norm": 0.34587833285331726,
      "learning_rate": 8.600825658939347e-05,
      "loss": 0.1096,
      "step": 2208
    },
    {
      "epoch": 4.2016167379933425,
      "grad_norm": 0.37271326780319214,
      "learning_rate": 8.600190536678311e-05,
      "loss": 0.1186,
      "step": 2209
    },
    {
      "epoch": 4.203518782691393,
      "grad_norm": 0.31880611181259155,
      "learning_rate": 8.599555414417275e-05,
      "loss": 0.1046,
      "step": 2210
    },
    {
      "epoch": 4.205420827389443,
      "grad_norm": 0.28906506299972534,
      "learning_rate": 8.598920292156241e-05,
      "loss": 0.0988,
      "step": 2211
    },
    {
      "epoch": 4.207322872087494,
      "grad_norm": 0.33470967411994934,
      "learning_rate": 8.598285169895205e-05,
      "loss": 0.1056,
      "step": 2212
    },
    {
      "epoch": 4.209224916785544,
      "grad_norm": 0.3186233341693878,
      "learning_rate": 8.59765004763417e-05,
      "loss": 0.1203,
      "step": 2213
    },
    {
      "epoch": 4.211126961483595,
      "grad_norm": 0.3465280532836914,
      "learning_rate": 8.597014925373134e-05,
      "loss": 0.1073,
      "step": 2214
    },
    {
      "epoch": 4.213029006181645,
      "grad_norm": 0.27451473474502563,
      "learning_rate": 8.596379803112099e-05,
      "loss": 0.0965,
      "step": 2215
    },
    {
      "epoch": 4.214931050879696,
      "grad_norm": 0.35004234313964844,
      "learning_rate": 8.595744680851064e-05,
      "loss": 0.1003,
      "step": 2216
    },
    {
      "epoch": 4.216833095577746,
      "grad_norm": 0.36494818329811096,
      "learning_rate": 8.595109558590028e-05,
      "loss": 0.1143,
      "step": 2217
    },
    {
      "epoch": 4.218735140275797,
      "grad_norm": 0.4278135597705841,
      "learning_rate": 8.594474436328995e-05,
      "loss": 0.1234,
      "step": 2218
    },
    {
      "epoch": 4.220637184973847,
      "grad_norm": 0.5124382972717285,
      "learning_rate": 8.593839314067959e-05,
      "loss": 0.1158,
      "step": 2219
    },
    {
      "epoch": 4.222539229671897,
      "grad_norm": 0.39850741624832153,
      "learning_rate": 8.593204191806923e-05,
      "loss": 0.1295,
      "step": 2220
    },
    {
      "epoch": 4.224441274369948,
      "grad_norm": 0.4141925573348999,
      "learning_rate": 8.592569069545888e-05,
      "loss": 0.1103,
      "step": 2221
    },
    {
      "epoch": 4.226343319067998,
      "grad_norm": 0.274980366230011,
      "learning_rate": 8.591933947284853e-05,
      "loss": 0.0927,
      "step": 2222
    },
    {
      "epoch": 4.228245363766049,
      "grad_norm": 0.4274260103702545,
      "learning_rate": 8.591298825023818e-05,
      "loss": 0.1248,
      "step": 2223
    },
    {
      "epoch": 4.230147408464099,
      "grad_norm": 0.39051416516304016,
      "learning_rate": 8.590663702762782e-05,
      "loss": 0.1068,
      "step": 2224
    },
    {
      "epoch": 4.2320494531621495,
      "grad_norm": 0.3913654685020447,
      "learning_rate": 8.590028580501747e-05,
      "loss": 0.1212,
      "step": 2225
    },
    {
      "epoch": 4.2339514978601995,
      "grad_norm": 0.33034393191337585,
      "learning_rate": 8.589393458240712e-05,
      "loss": 0.0875,
      "step": 2226
    },
    {
      "epoch": 4.23585354255825,
      "grad_norm": 0.405618280172348,
      "learning_rate": 8.588758335979676e-05,
      "loss": 0.1228,
      "step": 2227
    },
    {
      "epoch": 4.2377555872563,
      "grad_norm": 0.3220268189907074,
      "learning_rate": 8.588123213718641e-05,
      "loss": 0.1046,
      "step": 2228
    },
    {
      "epoch": 4.239657631954351,
      "grad_norm": 0.32537737488746643,
      "learning_rate": 8.587488091457606e-05,
      "loss": 0.0901,
      "step": 2229
    },
    {
      "epoch": 4.241559676652401,
      "grad_norm": 0.3968732953071594,
      "learning_rate": 8.58685296919657e-05,
      "loss": 0.1753,
      "step": 2230
    },
    {
      "epoch": 4.243461721350451,
      "grad_norm": 0.3441084325313568,
      "learning_rate": 8.586217846935535e-05,
      "loss": 0.1181,
      "step": 2231
    },
    {
      "epoch": 4.245363766048502,
      "grad_norm": 0.4014514684677124,
      "learning_rate": 8.5855827246745e-05,
      "loss": 0.1067,
      "step": 2232
    },
    {
      "epoch": 4.247265810746552,
      "grad_norm": 0.40167930722236633,
      "learning_rate": 8.584947602413464e-05,
      "loss": 0.1142,
      "step": 2233
    },
    {
      "epoch": 4.249167855444603,
      "grad_norm": 0.3604772984981537,
      "learning_rate": 8.58431248015243e-05,
      "loss": 0.108,
      "step": 2234
    },
    {
      "epoch": 4.251069900142653,
      "grad_norm": 0.4210832118988037,
      "learning_rate": 8.583677357891395e-05,
      "loss": 0.1161,
      "step": 2235
    },
    {
      "epoch": 4.252971944840704,
      "grad_norm": 0.34467047452926636,
      "learning_rate": 8.58304223563036e-05,
      "loss": 0.1187,
      "step": 2236
    },
    {
      "epoch": 4.254873989538754,
      "grad_norm": 0.8141130805015564,
      "learning_rate": 8.582407113369324e-05,
      "loss": 0.1766,
      "step": 2237
    },
    {
      "epoch": 4.256776034236805,
      "grad_norm": 0.28791263699531555,
      "learning_rate": 8.581771991108288e-05,
      "loss": 0.0953,
      "step": 2238
    },
    {
      "epoch": 4.258678078934855,
      "grad_norm": 0.2527415454387665,
      "learning_rate": 8.581136868847254e-05,
      "loss": 0.0847,
      "step": 2239
    },
    {
      "epoch": 4.260580123632906,
      "grad_norm": 0.2793647050857544,
      "learning_rate": 8.580501746586218e-05,
      "loss": 0.116,
      "step": 2240
    },
    {
      "epoch": 4.262482168330956,
      "grad_norm": 0.5324682593345642,
      "learning_rate": 8.579866624325183e-05,
      "loss": 0.1357,
      "step": 2241
    },
    {
      "epoch": 4.264384213029006,
      "grad_norm": 0.31979575753211975,
      "learning_rate": 8.579231502064148e-05,
      "loss": 0.1004,
      "step": 2242
    },
    {
      "epoch": 4.266286257727057,
      "grad_norm": 0.453645795583725,
      "learning_rate": 8.578596379803112e-05,
      "loss": 0.121,
      "step": 2243
    },
    {
      "epoch": 4.268188302425107,
      "grad_norm": 0.2688881754875183,
      "learning_rate": 8.577961257542077e-05,
      "loss": 0.0935,
      "step": 2244
    },
    {
      "epoch": 4.2700903471231575,
      "grad_norm": 0.30262473225593567,
      "learning_rate": 8.577326135281041e-05,
      "loss": 0.086,
      "step": 2245
    },
    {
      "epoch": 4.2719923918212075,
      "grad_norm": 0.4076935648918152,
      "learning_rate": 8.576691013020006e-05,
      "loss": 0.1075,
      "step": 2246
    },
    {
      "epoch": 4.273894436519258,
      "grad_norm": 0.5229641199111938,
      "learning_rate": 8.576055890758972e-05,
      "loss": 0.1585,
      "step": 2247
    },
    {
      "epoch": 4.275796481217308,
      "grad_norm": 0.3732607960700989,
      "learning_rate": 8.575420768497935e-05,
      "loss": 0.1065,
      "step": 2248
    },
    {
      "epoch": 4.277698525915359,
      "grad_norm": 0.39624014496803284,
      "learning_rate": 8.574785646236902e-05,
      "loss": 0.1229,
      "step": 2249
    },
    {
      "epoch": 4.279600570613409,
      "grad_norm": 0.47354966402053833,
      "learning_rate": 8.574150523975866e-05,
      "loss": 0.1574,
      "step": 2250
    },
    {
      "epoch": 4.28150261531146,
      "grad_norm": 0.35089337825775146,
      "learning_rate": 8.57351540171483e-05,
      "loss": 0.1098,
      "step": 2251
    },
    {
      "epoch": 4.28340466000951,
      "grad_norm": 0.3599602282047272,
      "learning_rate": 8.572880279453795e-05,
      "loss": 0.1136,
      "step": 2252
    },
    {
      "epoch": 4.285306704707561,
      "grad_norm": 0.4661259949207306,
      "learning_rate": 8.57224515719276e-05,
      "loss": 0.1297,
      "step": 2253
    },
    {
      "epoch": 4.287208749405611,
      "grad_norm": 0.27821779251098633,
      "learning_rate": 8.571610034931725e-05,
      "loss": 0.0974,
      "step": 2254
    },
    {
      "epoch": 4.289110794103661,
      "grad_norm": 0.3892570436000824,
      "learning_rate": 8.570974912670689e-05,
      "loss": 0.1362,
      "step": 2255
    },
    {
      "epoch": 4.291012838801712,
      "grad_norm": 0.3612288534641266,
      "learning_rate": 8.570339790409654e-05,
      "loss": 0.121,
      "step": 2256
    },
    {
      "epoch": 4.292914883499762,
      "grad_norm": 0.3542415499687195,
      "learning_rate": 8.56970466814862e-05,
      "loss": 0.1004,
      "step": 2257
    },
    {
      "epoch": 4.294816928197813,
      "grad_norm": 0.3457956910133362,
      "learning_rate": 8.569069545887583e-05,
      "loss": 0.1035,
      "step": 2258
    },
    {
      "epoch": 4.296718972895863,
      "grad_norm": 0.42984023690223694,
      "learning_rate": 8.568434423626548e-05,
      "loss": 0.1236,
      "step": 2259
    },
    {
      "epoch": 4.298621017593914,
      "grad_norm": 0.3002376854419708,
      "learning_rate": 8.567799301365514e-05,
      "loss": 0.0867,
      "step": 2260
    },
    {
      "epoch": 4.300523062291964,
      "grad_norm": 0.3134646415710449,
      "learning_rate": 8.567164179104477e-05,
      "loss": 0.0928,
      "step": 2261
    },
    {
      "epoch": 4.302425106990015,
      "grad_norm": 0.35177892446517944,
      "learning_rate": 8.566529056843443e-05,
      "loss": 0.1072,
      "step": 2262
    },
    {
      "epoch": 4.304327151688065,
      "grad_norm": 0.40704670548439026,
      "learning_rate": 8.565893934582408e-05,
      "loss": 0.1216,
      "step": 2263
    },
    {
      "epoch": 4.3062291963861155,
      "grad_norm": 0.40002110600471497,
      "learning_rate": 8.565258812321372e-05,
      "loss": 0.1153,
      "step": 2264
    },
    {
      "epoch": 4.3081312410841655,
      "grad_norm": 0.28185611963272095,
      "learning_rate": 8.564623690060337e-05,
      "loss": 0.0815,
      "step": 2265
    },
    {
      "epoch": 4.310033285782216,
      "grad_norm": 0.45204728841781616,
      "learning_rate": 8.563988567799302e-05,
      "loss": 0.1285,
      "step": 2266
    },
    {
      "epoch": 4.311935330480266,
      "grad_norm": 0.39130833745002747,
      "learning_rate": 8.563353445538267e-05,
      "loss": 0.1235,
      "step": 2267
    },
    {
      "epoch": 4.313837375178316,
      "grad_norm": 0.29855722188949585,
      "learning_rate": 8.562718323277231e-05,
      "loss": 0.0943,
      "step": 2268
    },
    {
      "epoch": 4.315739419876367,
      "grad_norm": 0.2964162826538086,
      "learning_rate": 8.562083201016196e-05,
      "loss": 0.1056,
      "step": 2269
    },
    {
      "epoch": 4.317641464574417,
      "grad_norm": 0.3408963978290558,
      "learning_rate": 8.561448078755161e-05,
      "loss": 0.1096,
      "step": 2270
    },
    {
      "epoch": 4.319543509272468,
      "grad_norm": 0.26335135102272034,
      "learning_rate": 8.560812956494125e-05,
      "loss": 0.1258,
      "step": 2271
    },
    {
      "epoch": 4.321445553970518,
      "grad_norm": 0.45781078934669495,
      "learning_rate": 8.56017783423309e-05,
      "loss": 0.1441,
      "step": 2272
    },
    {
      "epoch": 4.323347598668569,
      "grad_norm": 0.30225613713264465,
      "learning_rate": 8.559542711972056e-05,
      "loss": 0.0886,
      "step": 2273
    },
    {
      "epoch": 4.325249643366619,
      "grad_norm": 0.39499637484550476,
      "learning_rate": 8.55890758971102e-05,
      "loss": 0.108,
      "step": 2274
    },
    {
      "epoch": 4.32715168806467,
      "grad_norm": 0.25995761156082153,
      "learning_rate": 8.558272467449985e-05,
      "loss": 0.0832,
      "step": 2275
    },
    {
      "epoch": 4.32905373276272,
      "grad_norm": 0.4667019248008728,
      "learning_rate": 8.557637345188948e-05,
      "loss": 0.1376,
      "step": 2276
    },
    {
      "epoch": 4.330955777460771,
      "grad_norm": 0.6616588830947876,
      "learning_rate": 8.557002222927915e-05,
      "loss": 0.1402,
      "step": 2277
    },
    {
      "epoch": 4.332857822158821,
      "grad_norm": 0.362642765045166,
      "learning_rate": 8.556367100666879e-05,
      "loss": 0.1036,
      "step": 2278
    },
    {
      "epoch": 4.334759866856871,
      "grad_norm": 0.34205347299575806,
      "learning_rate": 8.555731978405843e-05,
      "loss": 0.0901,
      "step": 2279
    },
    {
      "epoch": 4.336661911554922,
      "grad_norm": 0.428653746843338,
      "learning_rate": 8.555096856144809e-05,
      "loss": 0.1291,
      "step": 2280
    },
    {
      "epoch": 4.338563956252972,
      "grad_norm": 0.31291234493255615,
      "learning_rate": 8.554461733883773e-05,
      "loss": 0.091,
      "step": 2281
    },
    {
      "epoch": 4.3404660009510225,
      "grad_norm": 0.33913081884384155,
      "learning_rate": 8.553826611622737e-05,
      "loss": 0.0844,
      "step": 2282
    },
    {
      "epoch": 4.3423680456490725,
      "grad_norm": 0.3302326500415802,
      "learning_rate": 8.553191489361702e-05,
      "loss": 0.0894,
      "step": 2283
    },
    {
      "epoch": 4.344270090347123,
      "grad_norm": 0.39421653747558594,
      "learning_rate": 8.552556367100667e-05,
      "loss": 0.1173,
      "step": 2284
    },
    {
      "epoch": 4.346172135045173,
      "grad_norm": 0.35651376843452454,
      "learning_rate": 8.551921244839632e-05,
      "loss": 0.0945,
      "step": 2285
    },
    {
      "epoch": 4.348074179743224,
      "grad_norm": 0.37059125304222107,
      "learning_rate": 8.551286122578596e-05,
      "loss": 0.1223,
      "step": 2286
    },
    {
      "epoch": 4.349976224441274,
      "grad_norm": 0.31241846084594727,
      "learning_rate": 8.550651000317561e-05,
      "loss": 0.1057,
      "step": 2287
    },
    {
      "epoch": 4.351878269139325,
      "grad_norm": 0.29532214999198914,
      "learning_rate": 8.550015878056527e-05,
      "loss": 0.1008,
      "step": 2288
    },
    {
      "epoch": 4.353780313837375,
      "grad_norm": 0.435973584651947,
      "learning_rate": 8.54938075579549e-05,
      "loss": 0.1258,
      "step": 2289
    },
    {
      "epoch": 4.355682358535425,
      "grad_norm": 0.3240755498409271,
      "learning_rate": 8.548745633534456e-05,
      "loss": 0.1383,
      "step": 2290
    },
    {
      "epoch": 4.357584403233476,
      "grad_norm": 0.3592849373817444,
      "learning_rate": 8.548110511273421e-05,
      "loss": 0.118,
      "step": 2291
    },
    {
      "epoch": 4.359486447931526,
      "grad_norm": 0.3495205342769623,
      "learning_rate": 8.547475389012385e-05,
      "loss": 0.1182,
      "step": 2292
    },
    {
      "epoch": 4.361388492629577,
      "grad_norm": 0.35103073716163635,
      "learning_rate": 8.54684026675135e-05,
      "loss": 0.1075,
      "step": 2293
    },
    {
      "epoch": 4.363290537327627,
      "grad_norm": 0.4233345091342926,
      "learning_rate": 8.546205144490315e-05,
      "loss": 0.1111,
      "step": 2294
    },
    {
      "epoch": 4.365192582025678,
      "grad_norm": 0.3999617099761963,
      "learning_rate": 8.54557002222928e-05,
      "loss": 0.1172,
      "step": 2295
    },
    {
      "epoch": 4.367094626723728,
      "grad_norm": 0.3122519254684448,
      "learning_rate": 8.544934899968244e-05,
      "loss": 0.0973,
      "step": 2296
    },
    {
      "epoch": 4.368996671421779,
      "grad_norm": 0.2844139039516449,
      "learning_rate": 8.544299777707209e-05,
      "loss": 0.0972,
      "step": 2297
    },
    {
      "epoch": 4.370898716119829,
      "grad_norm": 0.3841843008995056,
      "learning_rate": 8.543664655446174e-05,
      "loss": 0.1145,
      "step": 2298
    },
    {
      "epoch": 4.37280076081788,
      "grad_norm": 0.35272732377052307,
      "learning_rate": 8.543029533185138e-05,
      "loss": 0.1,
      "step": 2299
    },
    {
      "epoch": 4.37470280551593,
      "grad_norm": 0.3861033618450165,
      "learning_rate": 8.542394410924103e-05,
      "loss": 0.12,
      "step": 2300
    },
    {
      "epoch": 4.37660485021398,
      "grad_norm": 0.2895589768886566,
      "learning_rate": 8.541759288663069e-05,
      "loss": 0.0857,
      "step": 2301
    },
    {
      "epoch": 4.3785068949120305,
      "grad_norm": 0.4067385792732239,
      "learning_rate": 8.541124166402032e-05,
      "loss": 0.114,
      "step": 2302
    },
    {
      "epoch": 4.3804089396100805,
      "grad_norm": 0.3439483642578125,
      "learning_rate": 8.540489044140998e-05,
      "loss": 0.1218,
      "step": 2303
    },
    {
      "epoch": 4.382310984308131,
      "grad_norm": 0.273703396320343,
      "learning_rate": 8.539853921879963e-05,
      "loss": 0.0919,
      "step": 2304
    },
    {
      "epoch": 4.384213029006181,
      "grad_norm": 0.2975528836250305,
      "learning_rate": 8.539218799618927e-05,
      "loss": 0.0786,
      "step": 2305
    },
    {
      "epoch": 4.386115073704232,
      "grad_norm": 0.3109762370586395,
      "learning_rate": 8.538583677357892e-05,
      "loss": 0.1043,
      "step": 2306
    },
    {
      "epoch": 4.388017118402282,
      "grad_norm": 0.30896326899528503,
      "learning_rate": 8.537948555096857e-05,
      "loss": 0.0986,
      "step": 2307
    },
    {
      "epoch": 4.389919163100333,
      "grad_norm": 0.24300821125507355,
      "learning_rate": 8.537313432835822e-05,
      "loss": 0.0821,
      "step": 2308
    },
    {
      "epoch": 4.391821207798383,
      "grad_norm": 0.2907545566558838,
      "learning_rate": 8.536678310574786e-05,
      "loss": 0.0943,
      "step": 2309
    },
    {
      "epoch": 4.393723252496434,
      "grad_norm": 0.4220617115497589,
      "learning_rate": 8.53604318831375e-05,
      "loss": 0.1359,
      "step": 2310
    },
    {
      "epoch": 4.395625297194484,
      "grad_norm": 0.3436138331890106,
      "learning_rate": 8.535408066052716e-05,
      "loss": 0.1106,
      "step": 2311
    },
    {
      "epoch": 4.397527341892534,
      "grad_norm": 0.36533981561660767,
      "learning_rate": 8.53477294379168e-05,
      "loss": 0.1194,
      "step": 2312
    },
    {
      "epoch": 4.399429386590585,
      "grad_norm": 0.3554334044456482,
      "learning_rate": 8.534137821530645e-05,
      "loss": 0.1571,
      "step": 2313
    },
    {
      "epoch": 4.401331431288635,
      "grad_norm": 0.3670365512371063,
      "learning_rate": 8.53350269926961e-05,
      "loss": 0.1299,
      "step": 2314
    },
    {
      "epoch": 4.403233475986686,
      "grad_norm": 0.4539790451526642,
      "learning_rate": 8.532867577008574e-05,
      "loss": 0.1348,
      "step": 2315
    },
    {
      "epoch": 4.405135520684736,
      "grad_norm": 0.29808804392814636,
      "learning_rate": 8.53223245474754e-05,
      "loss": 0.1046,
      "step": 2316
    },
    {
      "epoch": 4.407037565382787,
      "grad_norm": 0.3486464321613312,
      "learning_rate": 8.531597332486503e-05,
      "loss": 0.1047,
      "step": 2317
    },
    {
      "epoch": 4.408939610080837,
      "grad_norm": 0.2947161793708801,
      "learning_rate": 8.530962210225469e-05,
      "loss": 0.0814,
      "step": 2318
    },
    {
      "epoch": 4.410841654778888,
      "grad_norm": 0.3321152627468109,
      "learning_rate": 8.530327087964434e-05,
      "loss": 0.1068,
      "step": 2319
    },
    {
      "epoch": 4.412743699476938,
      "grad_norm": 0.2441323846578598,
      "learning_rate": 8.529691965703398e-05,
      "loss": 0.0813,
      "step": 2320
    },
    {
      "epoch": 4.4146457441749885,
      "grad_norm": 0.37151622772216797,
      "learning_rate": 8.529056843442364e-05,
      "loss": 0.0995,
      "step": 2321
    },
    {
      "epoch": 4.4165477888730384,
      "grad_norm": 0.330240398645401,
      "learning_rate": 8.528421721181328e-05,
      "loss": 0.0999,
      "step": 2322
    },
    {
      "epoch": 4.418449833571089,
      "grad_norm": 0.38048794865608215,
      "learning_rate": 8.527786598920292e-05,
      "loss": 0.1065,
      "step": 2323
    },
    {
      "epoch": 4.420351878269139,
      "grad_norm": 0.3825136423110962,
      "learning_rate": 8.527151476659257e-05,
      "loss": 0.1021,
      "step": 2324
    },
    {
      "epoch": 4.422253922967189,
      "grad_norm": 0.3410681486129761,
      "learning_rate": 8.526516354398222e-05,
      "loss": 0.0899,
      "step": 2325
    },
    {
      "epoch": 4.42415596766524,
      "grad_norm": 0.33466002345085144,
      "learning_rate": 8.525881232137187e-05,
      "loss": 0.1051,
      "step": 2326
    },
    {
      "epoch": 4.42605801236329,
      "grad_norm": 0.3932620584964752,
      "learning_rate": 8.525246109876151e-05,
      "loss": 0.1156,
      "step": 2327
    },
    {
      "epoch": 4.427960057061341,
      "grad_norm": 0.31098031997680664,
      "learning_rate": 8.524610987615116e-05,
      "loss": 0.1026,
      "step": 2328
    },
    {
      "epoch": 4.429862101759391,
      "grad_norm": 0.3773583471775055,
      "learning_rate": 8.523975865354082e-05,
      "loss": 0.1113,
      "step": 2329
    },
    {
      "epoch": 4.431764146457442,
      "grad_norm": 0.33763033151626587,
      "learning_rate": 8.523340743093045e-05,
      "loss": 0.0941,
      "step": 2330
    },
    {
      "epoch": 4.433666191155492,
      "grad_norm": 0.23584803938865662,
      "learning_rate": 8.52270562083201e-05,
      "loss": 0.0777,
      "step": 2331
    },
    {
      "epoch": 4.435568235853543,
      "grad_norm": 0.3598161041736603,
      "learning_rate": 8.522070498570976e-05,
      "loss": 0.1173,
      "step": 2332
    },
    {
      "epoch": 4.437470280551593,
      "grad_norm": 0.3960074484348297,
      "learning_rate": 8.52143537630994e-05,
      "loss": 0.119,
      "step": 2333
    },
    {
      "epoch": 4.439372325249644,
      "grad_norm": 0.3260672092437744,
      "learning_rate": 8.520800254048905e-05,
      "loss": 0.1107,
      "step": 2334
    },
    {
      "epoch": 4.441274369947694,
      "grad_norm": 0.3651185929775238,
      "learning_rate": 8.52016513178787e-05,
      "loss": 0.0993,
      "step": 2335
    },
    {
      "epoch": 4.443176414645745,
      "grad_norm": 0.39154887199401855,
      "learning_rate": 8.519530009526834e-05,
      "loss": 0.1168,
      "step": 2336
    },
    {
      "epoch": 4.445078459343795,
      "grad_norm": 0.3429001569747925,
      "learning_rate": 8.518894887265799e-05,
      "loss": 0.1111,
      "step": 2337
    },
    {
      "epoch": 4.446980504041845,
      "grad_norm": 0.3407055735588074,
      "learning_rate": 8.518259765004764e-05,
      "loss": 0.1032,
      "step": 2338
    },
    {
      "epoch": 4.4488825487398955,
      "grad_norm": 0.3813023567199707,
      "learning_rate": 8.517624642743729e-05,
      "loss": 0.1077,
      "step": 2339
    },
    {
      "epoch": 4.4507845934379455,
      "grad_norm": 0.2836807370185852,
      "learning_rate": 8.516989520482693e-05,
      "loss": 0.0833,
      "step": 2340
    },
    {
      "epoch": 4.452686638135996,
      "grad_norm": 0.4083840250968933,
      "learning_rate": 8.516354398221657e-05,
      "loss": 0.1254,
      "step": 2341
    },
    {
      "epoch": 4.454588682834046,
      "grad_norm": 0.29835161566734314,
      "learning_rate": 8.515719275960623e-05,
      "loss": 0.1207,
      "step": 2342
    },
    {
      "epoch": 4.456490727532097,
      "grad_norm": 0.30677247047424316,
      "learning_rate": 8.515084153699587e-05,
      "loss": 0.0807,
      "step": 2343
    },
    {
      "epoch": 4.458392772230147,
      "grad_norm": 0.312853068113327,
      "learning_rate": 8.514449031438552e-05,
      "loss": 0.1174,
      "step": 2344
    },
    {
      "epoch": 4.460294816928198,
      "grad_norm": 0.431356281042099,
      "learning_rate": 8.513813909177518e-05,
      "loss": 0.1324,
      "step": 2345
    },
    {
      "epoch": 4.462196861626248,
      "grad_norm": 0.2785525918006897,
      "learning_rate": 8.513178786916482e-05,
      "loss": 0.1025,
      "step": 2346
    },
    {
      "epoch": 4.464098906324299,
      "grad_norm": 0.2919105291366577,
      "learning_rate": 8.512543664655447e-05,
      "loss": 0.1154,
      "step": 2347
    },
    {
      "epoch": 4.466000951022349,
      "grad_norm": 0.4356403350830078,
      "learning_rate": 8.51190854239441e-05,
      "loss": 0.1161,
      "step": 2348
    },
    {
      "epoch": 4.467902995720399,
      "grad_norm": 0.3411230146884918,
      "learning_rate": 8.511273420133377e-05,
      "loss": 0.1032,
      "step": 2349
    },
    {
      "epoch": 4.46980504041845,
      "grad_norm": 0.3335597515106201,
      "learning_rate": 8.510638297872341e-05,
      "loss": 0.1427,
      "step": 2350
    },
    {
      "epoch": 4.4717070851165,
      "grad_norm": 0.3813069760799408,
      "learning_rate": 8.510003175611305e-05,
      "loss": 0.1214,
      "step": 2351
    },
    {
      "epoch": 4.473609129814551,
      "grad_norm": 0.2616579830646515,
      "learning_rate": 8.509368053350271e-05,
      "loss": 0.0914,
      "step": 2352
    },
    {
      "epoch": 4.475511174512601,
      "grad_norm": 0.24161195755004883,
      "learning_rate": 8.508732931089235e-05,
      "loss": 0.0806,
      "step": 2353
    },
    {
      "epoch": 4.477413219210652,
      "grad_norm": 0.41089168190956116,
      "learning_rate": 8.508097808828199e-05,
      "loss": 0.1095,
      "step": 2354
    },
    {
      "epoch": 4.479315263908702,
      "grad_norm": 0.2930002510547638,
      "learning_rate": 8.507462686567164e-05,
      "loss": 0.0851,
      "step": 2355
    },
    {
      "epoch": 4.481217308606753,
      "grad_norm": 0.38217440247535706,
      "learning_rate": 8.506827564306129e-05,
      "loss": 0.106,
      "step": 2356
    },
    {
      "epoch": 4.483119353304803,
      "grad_norm": 0.4617588520050049,
      "learning_rate": 8.506192442045094e-05,
      "loss": 0.1269,
      "step": 2357
    },
    {
      "epoch": 4.4850213980028535,
      "grad_norm": 0.33491015434265137,
      "learning_rate": 8.505557319784058e-05,
      "loss": 0.1086,
      "step": 2358
    },
    {
      "epoch": 4.4869234427009035,
      "grad_norm": 0.31024834513664246,
      "learning_rate": 8.504922197523023e-05,
      "loss": 0.1039,
      "step": 2359
    },
    {
      "epoch": 4.4888254873989535,
      "grad_norm": 0.36780717968940735,
      "learning_rate": 8.504287075261989e-05,
      "loss": 0.1102,
      "step": 2360
    },
    {
      "epoch": 4.490727532097004,
      "grad_norm": 0.40606439113616943,
      "learning_rate": 8.503651953000952e-05,
      "loss": 0.13,
      "step": 2361
    },
    {
      "epoch": 4.492629576795054,
      "grad_norm": 0.4511033296585083,
      "learning_rate": 8.503016830739918e-05,
      "loss": 0.1182,
      "step": 2362
    },
    {
      "epoch": 4.494531621493105,
      "grad_norm": 0.36328256130218506,
      "learning_rate": 8.502381708478883e-05,
      "loss": 0.1024,
      "step": 2363
    },
    {
      "epoch": 4.496433666191155,
      "grad_norm": 0.3860591650009155,
      "learning_rate": 8.501746586217847e-05,
      "loss": 0.1019,
      "step": 2364
    },
    {
      "epoch": 4.498335710889206,
      "grad_norm": 0.46222564578056335,
      "learning_rate": 8.501111463956812e-05,
      "loss": 0.1132,
      "step": 2365
    },
    {
      "epoch": 4.500237755587256,
      "grad_norm": 0.3612005412578583,
      "learning_rate": 8.500476341695777e-05,
      "loss": 0.0963,
      "step": 2366
    },
    {
      "epoch": 4.502139800285307,
      "grad_norm": 0.43513086438179016,
      "learning_rate": 8.499841219434742e-05,
      "loss": 0.1109,
      "step": 2367
    },
    {
      "epoch": 4.504041844983357,
      "grad_norm": 0.2950316071510315,
      "learning_rate": 8.499206097173706e-05,
      "loss": 0.1124,
      "step": 2368
    },
    {
      "epoch": 4.505943889681408,
      "grad_norm": 0.36488962173461914,
      "learning_rate": 8.498570974912671e-05,
      "loss": 0.1,
      "step": 2369
    },
    {
      "epoch": 4.507845934379458,
      "grad_norm": 0.3592323064804077,
      "learning_rate": 8.497935852651636e-05,
      "loss": 0.0995,
      "step": 2370
    },
    {
      "epoch": 4.509747979077508,
      "grad_norm": 0.34753555059432983,
      "learning_rate": 8.4973007303906e-05,
      "loss": 0.1026,
      "step": 2371
    },
    {
      "epoch": 4.511650023775559,
      "grad_norm": 0.39495691657066345,
      "learning_rate": 8.496665608129565e-05,
      "loss": 0.1272,
      "step": 2372
    },
    {
      "epoch": 4.513552068473609,
      "grad_norm": 0.3553752601146698,
      "learning_rate": 8.49603048586853e-05,
      "loss": 0.1136,
      "step": 2373
    },
    {
      "epoch": 4.51545411317166,
      "grad_norm": 0.37848785519599915,
      "learning_rate": 8.495395363607494e-05,
      "loss": 0.1069,
      "step": 2374
    },
    {
      "epoch": 4.51735615786971,
      "grad_norm": 0.33565762639045715,
      "learning_rate": 8.49476024134646e-05,
      "loss": 0.1075,
      "step": 2375
    },
    {
      "epoch": 4.519258202567761,
      "grad_norm": 0.3359149694442749,
      "learning_rate": 8.494125119085425e-05,
      "loss": 0.098,
      "step": 2376
    },
    {
      "epoch": 4.521160247265811,
      "grad_norm": 0.3218232989311218,
      "learning_rate": 8.493489996824389e-05,
      "loss": 0.096,
      "step": 2377
    },
    {
      "epoch": 4.5230622919638614,
      "grad_norm": 0.3153054714202881,
      "learning_rate": 8.492854874563354e-05,
      "loss": 0.1015,
      "step": 2378
    },
    {
      "epoch": 4.5249643366619114,
      "grad_norm": 0.37637823820114136,
      "learning_rate": 8.492219752302319e-05,
      "loss": 0.1164,
      "step": 2379
    },
    {
      "epoch": 4.526866381359962,
      "grad_norm": 0.3270327150821686,
      "learning_rate": 8.491584630041284e-05,
      "loss": 0.1084,
      "step": 2380
    },
    {
      "epoch": 4.528768426058012,
      "grad_norm": 0.23998558521270752,
      "learning_rate": 8.490949507780248e-05,
      "loss": 0.0777,
      "step": 2381
    },
    {
      "epoch": 4.530670470756062,
      "grad_norm": 0.31294015049934387,
      "learning_rate": 8.490314385519212e-05,
      "loss": 0.0807,
      "step": 2382
    },
    {
      "epoch": 4.532572515454113,
      "grad_norm": 0.3305555582046509,
      "learning_rate": 8.489679263258178e-05,
      "loss": 0.1011,
      "step": 2383
    },
    {
      "epoch": 4.534474560152163,
      "grad_norm": 0.35641244053840637,
      "learning_rate": 8.489044140997142e-05,
      "loss": 0.11,
      "step": 2384
    },
    {
      "epoch": 4.536376604850214,
      "grad_norm": 0.3511948883533478,
      "learning_rate": 8.488409018736107e-05,
      "loss": 0.1009,
      "step": 2385
    },
    {
      "epoch": 4.538278649548264,
      "grad_norm": 0.3899917006492615,
      "learning_rate": 8.487773896475071e-05,
      "loss": 0.1285,
      "step": 2386
    },
    {
      "epoch": 4.540180694246315,
      "grad_norm": 0.4415057897567749,
      "learning_rate": 8.487138774214036e-05,
      "loss": 0.1434,
      "step": 2387
    },
    {
      "epoch": 4.542082738944365,
      "grad_norm": 0.42669907212257385,
      "learning_rate": 8.486503651953002e-05,
      "loss": 0.1201,
      "step": 2388
    },
    {
      "epoch": 4.543984783642416,
      "grad_norm": 0.27351129055023193,
      "learning_rate": 8.485868529691965e-05,
      "loss": 0.0761,
      "step": 2389
    },
    {
      "epoch": 4.545886828340466,
      "grad_norm": 0.31243595480918884,
      "learning_rate": 8.48523340743093e-05,
      "loss": 0.0909,
      "step": 2390
    },
    {
      "epoch": 4.547788873038517,
      "grad_norm": 0.36273542046546936,
      "learning_rate": 8.484598285169896e-05,
      "loss": 0.1156,
      "step": 2391
    },
    {
      "epoch": 4.549690917736567,
      "grad_norm": 0.3167242109775543,
      "learning_rate": 8.48396316290886e-05,
      "loss": 0.2065,
      "step": 2392
    },
    {
      "epoch": 4.551592962434617,
      "grad_norm": 0.3072797358036041,
      "learning_rate": 8.483328040647825e-05,
      "loss": 0.0939,
      "step": 2393
    },
    {
      "epoch": 4.553495007132668,
      "grad_norm": 0.32601553201675415,
      "learning_rate": 8.48269291838679e-05,
      "loss": 0.1052,
      "step": 2394
    },
    {
      "epoch": 4.555397051830718,
      "grad_norm": 0.41232773661613464,
      "learning_rate": 8.482057796125754e-05,
      "loss": 0.1207,
      "step": 2395
    },
    {
      "epoch": 4.5572990965287685,
      "grad_norm": 0.46499213576316833,
      "learning_rate": 8.481422673864719e-05,
      "loss": 0.1251,
      "step": 2396
    },
    {
      "epoch": 4.5592011412268185,
      "grad_norm": 0.3984009325504303,
      "learning_rate": 8.480787551603684e-05,
      "loss": 0.1317,
      "step": 2397
    },
    {
      "epoch": 4.561103185924869,
      "grad_norm": 0.3825131356716156,
      "learning_rate": 8.48015242934265e-05,
      "loss": 0.1273,
      "step": 2398
    },
    {
      "epoch": 4.563005230622919,
      "grad_norm": 0.39657148718833923,
      "learning_rate": 8.479517307081613e-05,
      "loss": 0.145,
      "step": 2399
    },
    {
      "epoch": 4.56490727532097,
      "grad_norm": 0.3764631748199463,
      "learning_rate": 8.478882184820578e-05,
      "loss": 0.1133,
      "step": 2400
    },
    {
      "epoch": 4.56680932001902,
      "grad_norm": 0.2968275249004364,
      "learning_rate": 8.478247062559544e-05,
      "loss": 0.0885,
      "step": 2401
    },
    {
      "epoch": 4.568711364717071,
      "grad_norm": 0.326856791973114,
      "learning_rate": 8.477611940298507e-05,
      "loss": 0.0923,
      "step": 2402
    },
    {
      "epoch": 4.570613409415121,
      "grad_norm": 0.38287606835365295,
      "learning_rate": 8.476976818037473e-05,
      "loss": 0.141,
      "step": 2403
    },
    {
      "epoch": 4.572515454113171,
      "grad_norm": 0.47493815422058105,
      "learning_rate": 8.476341695776438e-05,
      "loss": 0.1146,
      "step": 2404
    },
    {
      "epoch": 4.574417498811222,
      "grad_norm": 0.35078614950180054,
      "learning_rate": 8.475706573515402e-05,
      "loss": 0.1153,
      "step": 2405
    },
    {
      "epoch": 4.576319543509273,
      "grad_norm": 0.3837313950061798,
      "learning_rate": 8.475071451254367e-05,
      "loss": 0.1408,
      "step": 2406
    },
    {
      "epoch": 4.578221588207323,
      "grad_norm": 0.3800102472305298,
      "learning_rate": 8.474436328993332e-05,
      "loss": 0.1224,
      "step": 2407
    },
    {
      "epoch": 4.580123632905373,
      "grad_norm": 0.40831804275512695,
      "learning_rate": 8.473801206732296e-05,
      "loss": 0.1283,
      "step": 2408
    },
    {
      "epoch": 4.582025677603424,
      "grad_norm": 0.34854429960250854,
      "learning_rate": 8.473166084471261e-05,
      "loss": 0.101,
      "step": 2409
    },
    {
      "epoch": 4.583927722301474,
      "grad_norm": 0.3317374885082245,
      "learning_rate": 8.472530962210226e-05,
      "loss": 0.0986,
      "step": 2410
    },
    {
      "epoch": 4.585829766999525,
      "grad_norm": 0.3316230773925781,
      "learning_rate": 8.471895839949191e-05,
      "loss": 0.0955,
      "step": 2411
    },
    {
      "epoch": 4.587731811697575,
      "grad_norm": 0.3458825945854187,
      "learning_rate": 8.471260717688155e-05,
      "loss": 0.1246,
      "step": 2412
    },
    {
      "epoch": 4.589633856395626,
      "grad_norm": 0.2985215187072754,
      "learning_rate": 8.470625595427119e-05,
      "loss": 0.0904,
      "step": 2413
    },
    {
      "epoch": 4.591535901093676,
      "grad_norm": 0.5128130912780762,
      "learning_rate": 8.469990473166086e-05,
      "loss": 0.1119,
      "step": 2414
    },
    {
      "epoch": 4.5934379457917265,
      "grad_norm": 0.3538981080055237,
      "learning_rate": 8.46935535090505e-05,
      "loss": 0.1276,
      "step": 2415
    },
    {
      "epoch": 4.5953399904897765,
      "grad_norm": 0.24112893640995026,
      "learning_rate": 8.468720228644015e-05,
      "loss": 0.0813,
      "step": 2416
    },
    {
      "epoch": 4.597242035187827,
      "grad_norm": 0.34151947498321533,
      "learning_rate": 8.46808510638298e-05,
      "loss": 0.1214,
      "step": 2417
    },
    {
      "epoch": 4.599144079885877,
      "grad_norm": 0.3011094629764557,
      "learning_rate": 8.467449984121944e-05,
      "loss": 0.0955,
      "step": 2418
    },
    {
      "epoch": 4.601046124583927,
      "grad_norm": 0.45026248693466187,
      "learning_rate": 8.466814861860909e-05,
      "loss": 0.1309,
      "step": 2419
    },
    {
      "epoch": 4.602948169281978,
      "grad_norm": 0.38199952244758606,
      "learning_rate": 8.466179739599873e-05,
      "loss": 0.1229,
      "step": 2420
    },
    {
      "epoch": 4.604850213980028,
      "grad_norm": 0.44846484065055847,
      "learning_rate": 8.465544617338839e-05,
      "loss": 0.1254,
      "step": 2421
    },
    {
      "epoch": 4.606752258678079,
      "grad_norm": 0.29512494802474976,
      "learning_rate": 8.464909495077803e-05,
      "loss": 0.0874,
      "step": 2422
    },
    {
      "epoch": 4.608654303376129,
      "grad_norm": 0.34601306915283203,
      "learning_rate": 8.464274372816767e-05,
      "loss": 0.0928,
      "step": 2423
    },
    {
      "epoch": 4.61055634807418,
      "grad_norm": 0.4081529378890991,
      "learning_rate": 8.463639250555733e-05,
      "loss": 0.1161,
      "step": 2424
    },
    {
      "epoch": 4.61245839277223,
      "grad_norm": 0.39208075404167175,
      "learning_rate": 8.463004128294697e-05,
      "loss": 0.1124,
      "step": 2425
    },
    {
      "epoch": 4.614360437470281,
      "grad_norm": 0.2740732431411743,
      "learning_rate": 8.462369006033661e-05,
      "loss": 0.0698,
      "step": 2426
    },
    {
      "epoch": 4.616262482168331,
      "grad_norm": 0.37493231892585754,
      "learning_rate": 8.461733883772626e-05,
      "loss": 0.089,
      "step": 2427
    },
    {
      "epoch": 4.618164526866382,
      "grad_norm": 0.4912300407886505,
      "learning_rate": 8.461098761511591e-05,
      "loss": 0.1374,
      "step": 2428
    },
    {
      "epoch": 4.620066571564432,
      "grad_norm": 0.44587963819503784,
      "learning_rate": 8.460463639250557e-05,
      "loss": 0.1207,
      "step": 2429
    },
    {
      "epoch": 4.621968616262482,
      "grad_norm": 0.4140859544277191,
      "learning_rate": 8.45982851698952e-05,
      "loss": 0.1333,
      "step": 2430
    },
    {
      "epoch": 4.623870660960533,
      "grad_norm": 0.3500138223171234,
      "learning_rate": 8.459193394728486e-05,
      "loss": 0.1032,
      "step": 2431
    },
    {
      "epoch": 4.625772705658583,
      "grad_norm": 0.3875083327293396,
      "learning_rate": 8.458558272467451e-05,
      "loss": 0.1018,
      "step": 2432
    },
    {
      "epoch": 4.627674750356634,
      "grad_norm": 0.5065046548843384,
      "learning_rate": 8.457923150206415e-05,
      "loss": 0.125,
      "step": 2433
    },
    {
      "epoch": 4.629576795054684,
      "grad_norm": 0.2707502841949463,
      "learning_rate": 8.45728802794538e-05,
      "loss": 0.1002,
      "step": 2434
    },
    {
      "epoch": 4.6314788397527344,
      "grad_norm": 0.38502418994903564,
      "learning_rate": 8.456652905684345e-05,
      "loss": 0.1264,
      "step": 2435
    },
    {
      "epoch": 4.633380884450784,
      "grad_norm": 0.34822702407836914,
      "learning_rate": 8.456017783423309e-05,
      "loss": 0.1184,
      "step": 2436
    },
    {
      "epoch": 4.635282929148835,
      "grad_norm": 0.33620592951774597,
      "learning_rate": 8.455382661162274e-05,
      "loss": 0.1264,
      "step": 2437
    },
    {
      "epoch": 4.637184973846885,
      "grad_norm": 0.3064115345478058,
      "learning_rate": 8.454747538901239e-05,
      "loss": 0.1122,
      "step": 2438
    },
    {
      "epoch": 4.639087018544936,
      "grad_norm": 0.34428808093070984,
      "learning_rate": 8.454112416640204e-05,
      "loss": 0.1083,
      "step": 2439
    },
    {
      "epoch": 4.640989063242986,
      "grad_norm": 0.3312735855579376,
      "learning_rate": 8.453477294379168e-05,
      "loss": 0.1046,
      "step": 2440
    },
    {
      "epoch": 4.642891107941036,
      "grad_norm": 0.42405757308006287,
      "learning_rate": 8.452842172118133e-05,
      "loss": 0.1364,
      "step": 2441
    },
    {
      "epoch": 4.644793152639087,
      "grad_norm": 0.39682331681251526,
      "learning_rate": 8.452207049857099e-05,
      "loss": 0.1262,
      "step": 2442
    },
    {
      "epoch": 4.646695197337137,
      "grad_norm": 0.3447044789791107,
      "learning_rate": 8.451571927596062e-05,
      "loss": 0.1158,
      "step": 2443
    },
    {
      "epoch": 4.648597242035188,
      "grad_norm": 0.40121355652809143,
      "learning_rate": 8.450936805335026e-05,
      "loss": 0.1246,
      "step": 2444
    },
    {
      "epoch": 4.650499286733238,
      "grad_norm": 0.3898472785949707,
      "learning_rate": 8.450301683073993e-05,
      "loss": 0.1244,
      "step": 2445
    },
    {
      "epoch": 4.652401331431289,
      "grad_norm": 0.2964152991771698,
      "learning_rate": 8.449666560812957e-05,
      "loss": 0.0925,
      "step": 2446
    },
    {
      "epoch": 4.654303376129339,
      "grad_norm": 0.2836705446243286,
      "learning_rate": 8.449031438551922e-05,
      "loss": 0.101,
      "step": 2447
    },
    {
      "epoch": 4.65620542082739,
      "grad_norm": 0.3003692030906677,
      "learning_rate": 8.448396316290887e-05,
      "loss": 0.0922,
      "step": 2448
    },
    {
      "epoch": 4.65810746552544,
      "grad_norm": 0.5348609089851379,
      "learning_rate": 8.447761194029851e-05,
      "loss": 0.1735,
      "step": 2449
    },
    {
      "epoch": 4.660009510223491,
      "grad_norm": 0.3387379050254822,
      "learning_rate": 8.447126071768816e-05,
      "loss": 0.1126,
      "step": 2450
    },
    {
      "epoch": 4.661911554921541,
      "grad_norm": 0.30646830797195435,
      "learning_rate": 8.44649094950778e-05,
      "loss": 0.085,
      "step": 2451
    },
    {
      "epoch": 4.663813599619591,
      "grad_norm": 0.34434470534324646,
      "learning_rate": 8.445855827246746e-05,
      "loss": 0.1113,
      "step": 2452
    },
    {
      "epoch": 4.6657156443176415,
      "grad_norm": 0.38273414969444275,
      "learning_rate": 8.44522070498571e-05,
      "loss": 0.1135,
      "step": 2453
    },
    {
      "epoch": 4.6676176890156915,
      "grad_norm": 0.44843336939811707,
      "learning_rate": 8.444585582724674e-05,
      "loss": 0.1497,
      "step": 2454
    },
    {
      "epoch": 4.669519733713742,
      "grad_norm": 0.4575416147708893,
      "learning_rate": 8.44395046046364e-05,
      "loss": 0.1082,
      "step": 2455
    },
    {
      "epoch": 4.671421778411792,
      "grad_norm": 0.38473185896873474,
      "learning_rate": 8.443315338202604e-05,
      "loss": 0.1255,
      "step": 2456
    },
    {
      "epoch": 4.673323823109843,
      "grad_norm": 0.3839578926563263,
      "learning_rate": 8.44268021594157e-05,
      "loss": 0.1106,
      "step": 2457
    },
    {
      "epoch": 4.675225867807893,
      "grad_norm": 0.35472893714904785,
      "learning_rate": 8.442045093680533e-05,
      "loss": 0.1122,
      "step": 2458
    },
    {
      "epoch": 4.677127912505944,
      "grad_norm": 0.34224382042884827,
      "learning_rate": 8.441409971419499e-05,
      "loss": 0.0963,
      "step": 2459
    },
    {
      "epoch": 4.679029957203994,
      "grad_norm": 0.3992440104484558,
      "learning_rate": 8.440774849158464e-05,
      "loss": 0.1234,
      "step": 2460
    },
    {
      "epoch": 4.680932001902045,
      "grad_norm": 0.39441943168640137,
      "learning_rate": 8.440139726897428e-05,
      "loss": 0.11,
      "step": 2461
    },
    {
      "epoch": 4.682834046600095,
      "grad_norm": 0.43852171301841736,
      "learning_rate": 8.439504604636393e-05,
      "loss": 0.1361,
      "step": 2462
    },
    {
      "epoch": 4.684736091298145,
      "grad_norm": 0.35047483444213867,
      "learning_rate": 8.438869482375358e-05,
      "loss": 0.0981,
      "step": 2463
    },
    {
      "epoch": 4.686638135996196,
      "grad_norm": 0.3970755934715271,
      "learning_rate": 8.438234360114322e-05,
      "loss": 0.1196,
      "step": 2464
    },
    {
      "epoch": 4.688540180694246,
      "grad_norm": 0.2760510742664337,
      "learning_rate": 8.437599237853287e-05,
      "loss": 0.1035,
      "step": 2465
    },
    {
      "epoch": 4.690442225392297,
      "grad_norm": 0.26530909538269043,
      "learning_rate": 8.436964115592252e-05,
      "loss": 0.1589,
      "step": 2466
    },
    {
      "epoch": 4.692344270090347,
      "grad_norm": 0.2989928126335144,
      "learning_rate": 8.436328993331216e-05,
      "loss": 0.0945,
      "step": 2467
    },
    {
      "epoch": 4.694246314788398,
      "grad_norm": 0.42447128891944885,
      "learning_rate": 8.435693871070181e-05,
      "loss": 0.1433,
      "step": 2468
    },
    {
      "epoch": 4.696148359486448,
      "grad_norm": 0.4014334976673126,
      "learning_rate": 8.435058748809146e-05,
      "loss": 0.1242,
      "step": 2469
    },
    {
      "epoch": 4.698050404184499,
      "grad_norm": 0.3872852921485901,
      "learning_rate": 8.434423626548111e-05,
      "loss": 0.1195,
      "step": 2470
    },
    {
      "epoch": 4.699952448882549,
      "grad_norm": 0.3857705891132355,
      "learning_rate": 8.433788504287075e-05,
      "loss": 0.108,
      "step": 2471
    },
    {
      "epoch": 4.7018544935805995,
      "grad_norm": 0.3534420430660248,
      "learning_rate": 8.43315338202604e-05,
      "loss": 0.1218,
      "step": 2472
    },
    {
      "epoch": 4.7037565382786495,
      "grad_norm": 0.32009604573249817,
      "learning_rate": 8.432518259765006e-05,
      "loss": 0.1053,
      "step": 2473
    },
    {
      "epoch": 4.7056585829766995,
      "grad_norm": 0.2501387894153595,
      "learning_rate": 8.43188313750397e-05,
      "loss": 0.0668,
      "step": 2474
    },
    {
      "epoch": 4.70756062767475,
      "grad_norm": 0.3360025882720947,
      "learning_rate": 8.431248015242935e-05,
      "loss": 0.1119,
      "step": 2475
    },
    {
      "epoch": 4.709462672372801,
      "grad_norm": 0.31509891152381897,
      "learning_rate": 8.4306128929819e-05,
      "loss": 0.0955,
      "step": 2476
    },
    {
      "epoch": 4.711364717070851,
      "grad_norm": 0.42007285356521606,
      "learning_rate": 8.429977770720864e-05,
      "loss": 0.1441,
      "step": 2477
    },
    {
      "epoch": 4.713266761768901,
      "grad_norm": 0.39764338731765747,
      "learning_rate": 8.429342648459829e-05,
      "loss": 0.1175,
      "step": 2478
    },
    {
      "epoch": 4.715168806466952,
      "grad_norm": 0.33381861448287964,
      "learning_rate": 8.428707526198794e-05,
      "loss": 0.1199,
      "step": 2479
    },
    {
      "epoch": 4.717070851165002,
      "grad_norm": 0.2918257415294647,
      "learning_rate": 8.428072403937758e-05,
      "loss": 0.0796,
      "step": 2480
    },
    {
      "epoch": 4.718972895863053,
      "grad_norm": 0.42560750246047974,
      "learning_rate": 8.427437281676723e-05,
      "loss": 0.114,
      "step": 2481
    },
    {
      "epoch": 4.720874940561103,
      "grad_norm": 0.3700113594532013,
      "learning_rate": 8.426802159415688e-05,
      "loss": 0.1145,
      "step": 2482
    },
    {
      "epoch": 4.722776985259154,
      "grad_norm": 0.39171457290649414,
      "learning_rate": 8.426167037154653e-05,
      "loss": 0.128,
      "step": 2483
    },
    {
      "epoch": 4.724679029957204,
      "grad_norm": 0.3000270426273346,
      "learning_rate": 8.425531914893617e-05,
      "loss": 0.0932,
      "step": 2484
    },
    {
      "epoch": 4.726581074655254,
      "grad_norm": 0.2848623991012573,
      "learning_rate": 8.424896792632581e-05,
      "loss": 0.086,
      "step": 2485
    },
    {
      "epoch": 4.728483119353305,
      "grad_norm": 0.3404539227485657,
      "learning_rate": 8.424261670371548e-05,
      "loss": 0.0934,
      "step": 2486
    },
    {
      "epoch": 4.730385164051356,
      "grad_norm": 0.31609418988227844,
      "learning_rate": 8.423626548110511e-05,
      "loss": 0.0985,
      "step": 2487
    },
    {
      "epoch": 4.732287208749406,
      "grad_norm": 0.34037312865257263,
      "learning_rate": 8.422991425849477e-05,
      "loss": 0.1193,
      "step": 2488
    },
    {
      "epoch": 4.734189253447456,
      "grad_norm": 0.31899651885032654,
      "learning_rate": 8.422356303588442e-05,
      "loss": 0.1137,
      "step": 2489
    },
    {
      "epoch": 4.736091298145507,
      "grad_norm": 0.39307737350463867,
      "learning_rate": 8.421721181327406e-05,
      "loss": 0.1452,
      "step": 2490
    },
    {
      "epoch": 4.7379933428435566,
      "grad_norm": 0.26885175704956055,
      "learning_rate": 8.421086059066371e-05,
      "loss": 0.1025,
      "step": 2491
    },
    {
      "epoch": 4.739895387541607,
      "grad_norm": 0.23492799699306488,
      "learning_rate": 8.420450936805335e-05,
      "loss": 0.0821,
      "step": 2492
    },
    {
      "epoch": 4.741797432239657,
      "grad_norm": 0.30144715309143066,
      "learning_rate": 8.419815814544301e-05,
      "loss": 0.0924,
      "step": 2493
    },
    {
      "epoch": 4.743699476937708,
      "grad_norm": 0.3370392322540283,
      "learning_rate": 8.419180692283265e-05,
      "loss": 0.1281,
      "step": 2494
    },
    {
      "epoch": 4.745601521635758,
      "grad_norm": 0.3939819633960724,
      "learning_rate": 8.418545570022229e-05,
      "loss": 0.1115,
      "step": 2495
    },
    {
      "epoch": 4.747503566333809,
      "grad_norm": 0.7242825627326965,
      "learning_rate": 8.417910447761194e-05,
      "loss": 0.1038,
      "step": 2496
    },
    {
      "epoch": 4.749405611031859,
      "grad_norm": 0.3430320620536804,
      "learning_rate": 8.417275325500159e-05,
      "loss": 0.107,
      "step": 2497
    },
    {
      "epoch": 4.75130765572991,
      "grad_norm": 0.37956321239471436,
      "learning_rate": 8.416640203239123e-05,
      "loss": 0.1203,
      "step": 2498
    },
    {
      "epoch": 4.75320970042796,
      "grad_norm": 0.3118121027946472,
      "learning_rate": 8.416005080978088e-05,
      "loss": 0.0961,
      "step": 2499
    },
    {
      "epoch": 4.75511174512601,
      "grad_norm": 0.3842122554779053,
      "learning_rate": 8.415369958717053e-05,
      "loss": 0.1095,
      "step": 2500
    },
    {
      "epoch": 4.757013789824061,
      "grad_norm": 0.36103618144989014,
      "learning_rate": 8.414734836456019e-05,
      "loss": 0.107,
      "step": 2501
    },
    {
      "epoch": 4.758915834522111,
      "grad_norm": 0.4404369592666626,
      "learning_rate": 8.414099714194982e-05,
      "loss": 0.0972,
      "step": 2502
    },
    {
      "epoch": 4.760817879220162,
      "grad_norm": 0.45303696393966675,
      "learning_rate": 8.413464591933948e-05,
      "loss": 0.1286,
      "step": 2503
    },
    {
      "epoch": 4.762719923918212,
      "grad_norm": 0.36196044087409973,
      "learning_rate": 8.412829469672913e-05,
      "loss": 0.1095,
      "step": 2504
    },
    {
      "epoch": 4.764621968616263,
      "grad_norm": 0.49001795053482056,
      "learning_rate": 8.412194347411877e-05,
      "loss": 0.1578,
      "step": 2505
    },
    {
      "epoch": 4.766524013314313,
      "grad_norm": 0.32446369528770447,
      "learning_rate": 8.411559225150842e-05,
      "loss": 0.0991,
      "step": 2506
    },
    {
      "epoch": 4.768426058012364,
      "grad_norm": 0.3021388053894043,
      "learning_rate": 8.410924102889807e-05,
      "loss": 0.0902,
      "step": 2507
    },
    {
      "epoch": 4.770328102710414,
      "grad_norm": 0.28912147879600525,
      "learning_rate": 8.410288980628771e-05,
      "loss": 0.106,
      "step": 2508
    },
    {
      "epoch": 4.7722301474084645,
      "grad_norm": 0.40766748785972595,
      "learning_rate": 8.409653858367736e-05,
      "loss": 0.1155,
      "step": 2509
    },
    {
      "epoch": 4.7741321921065145,
      "grad_norm": 0.5005617737770081,
      "learning_rate": 8.409018736106701e-05,
      "loss": 0.1674,
      "step": 2510
    },
    {
      "epoch": 4.7760342368045645,
      "grad_norm": 0.4575154781341553,
      "learning_rate": 8.408383613845666e-05,
      "loss": 0.1639,
      "step": 2511
    },
    {
      "epoch": 4.777936281502615,
      "grad_norm": 0.4962354302406311,
      "learning_rate": 8.40774849158463e-05,
      "loss": 0.1336,
      "step": 2512
    },
    {
      "epoch": 4.779838326200665,
      "grad_norm": 0.4569809138774872,
      "learning_rate": 8.407113369323595e-05,
      "loss": 0.1323,
      "step": 2513
    },
    {
      "epoch": 4.781740370898716,
      "grad_norm": 0.34369999170303345,
      "learning_rate": 8.40647824706256e-05,
      "loss": 0.1171,
      "step": 2514
    },
    {
      "epoch": 4.783642415596766,
      "grad_norm": 0.3565669655799866,
      "learning_rate": 8.405843124801524e-05,
      "loss": 0.1159,
      "step": 2515
    },
    {
      "epoch": 4.785544460294817,
      "grad_norm": 0.24039465188980103,
      "learning_rate": 8.405208002540488e-05,
      "loss": 0.0976,
      "step": 2516
    },
    {
      "epoch": 4.787446504992867,
      "grad_norm": 0.37532779574394226,
      "learning_rate": 8.404572880279455e-05,
      "loss": 0.1129,
      "step": 2517
    },
    {
      "epoch": 4.789348549690918,
      "grad_norm": 0.334505170583725,
      "learning_rate": 8.403937758018419e-05,
      "loss": 0.1016,
      "step": 2518
    },
    {
      "epoch": 4.791250594388968,
      "grad_norm": 0.43082761764526367,
      "learning_rate": 8.403302635757384e-05,
      "loss": 0.1307,
      "step": 2519
    },
    {
      "epoch": 4.793152639087019,
      "grad_norm": 0.4381292760372162,
      "learning_rate": 8.402667513496349e-05,
      "loss": 0.1137,
      "step": 2520
    },
    {
      "epoch": 4.795054683785069,
      "grad_norm": 0.4337981045246124,
      "learning_rate": 8.402032391235313e-05,
      "loss": 0.1281,
      "step": 2521
    },
    {
      "epoch": 4.796956728483119,
      "grad_norm": 0.4429587721824646,
      "learning_rate": 8.401397268974278e-05,
      "loss": 0.1191,
      "step": 2522
    },
    {
      "epoch": 4.79885877318117,
      "grad_norm": 0.4298746883869171,
      "learning_rate": 8.400762146713242e-05,
      "loss": 0.1367,
      "step": 2523
    },
    {
      "epoch": 4.80076081787922,
      "grad_norm": 0.42826715111732483,
      "learning_rate": 8.400127024452208e-05,
      "loss": 0.1222,
      "step": 2524
    },
    {
      "epoch": 4.802662862577271,
      "grad_norm": 0.37338751554489136,
      "learning_rate": 8.399491902191172e-05,
      "loss": 0.1048,
      "step": 2525
    },
    {
      "epoch": 4.804564907275321,
      "grad_norm": 0.38671061396598816,
      "learning_rate": 8.398856779930136e-05,
      "loss": 0.1154,
      "step": 2526
    },
    {
      "epoch": 4.806466951973372,
      "grad_norm": 0.3544102907180786,
      "learning_rate": 8.398221657669103e-05,
      "loss": 0.1055,
      "step": 2527
    },
    {
      "epoch": 4.808368996671422,
      "grad_norm": 0.38023364543914795,
      "learning_rate": 8.397586535408066e-05,
      "loss": 0.1117,
      "step": 2528
    },
    {
      "epoch": 4.8102710413694725,
      "grad_norm": 0.3622092008590698,
      "learning_rate": 8.396951413147032e-05,
      "loss": 0.1099,
      "step": 2529
    },
    {
      "epoch": 4.8121730860675225,
      "grad_norm": 0.692039966583252,
      "learning_rate": 8.396316290885995e-05,
      "loss": 0.1335,
      "step": 2530
    },
    {
      "epoch": 4.814075130765573,
      "grad_norm": 0.35321712493896484,
      "learning_rate": 8.39568116862496e-05,
      "loss": 0.1175,
      "step": 2531
    },
    {
      "epoch": 4.815977175463623,
      "grad_norm": 0.37036386132240295,
      "learning_rate": 8.395046046363926e-05,
      "loss": 0.1253,
      "step": 2532
    },
    {
      "epoch": 4.817879220161673,
      "grad_norm": 0.42249128222465515,
      "learning_rate": 8.39441092410289e-05,
      "loss": 0.1163,
      "step": 2533
    },
    {
      "epoch": 4.819781264859724,
      "grad_norm": 0.3563583195209503,
      "learning_rate": 8.393775801841855e-05,
      "loss": 0.1597,
      "step": 2534
    },
    {
      "epoch": 4.821683309557774,
      "grad_norm": 0.39946305751800537,
      "learning_rate": 8.39314067958082e-05,
      "loss": 0.1156,
      "step": 2535
    },
    {
      "epoch": 4.823585354255825,
      "grad_norm": 0.31761807203292847,
      "learning_rate": 8.392505557319784e-05,
      "loss": 0.0946,
      "step": 2536
    },
    {
      "epoch": 4.825487398953875,
      "grad_norm": 0.4180295765399933,
      "learning_rate": 8.391870435058749e-05,
      "loss": 0.1271,
      "step": 2537
    },
    {
      "epoch": 4.827389443651926,
      "grad_norm": 0.36158043146133423,
      "learning_rate": 8.391235312797714e-05,
      "loss": 0.106,
      "step": 2538
    },
    {
      "epoch": 4.829291488349976,
      "grad_norm": 0.4044169783592224,
      "learning_rate": 8.390600190536678e-05,
      "loss": 0.1094,
      "step": 2539
    },
    {
      "epoch": 4.831193533048027,
      "grad_norm": 0.3362937569618225,
      "learning_rate": 8.389965068275643e-05,
      "loss": 0.078,
      "step": 2540
    },
    {
      "epoch": 4.833095577746077,
      "grad_norm": 0.3558341860771179,
      "learning_rate": 8.389329946014608e-05,
      "loss": 0.1125,
      "step": 2541
    },
    {
      "epoch": 4.834997622444128,
      "grad_norm": 0.44893354177474976,
      "learning_rate": 8.388694823753574e-05,
      "loss": 0.1393,
      "step": 2542
    },
    {
      "epoch": 4.836899667142178,
      "grad_norm": 0.3790888488292694,
      "learning_rate": 8.388059701492537e-05,
      "loss": 0.1312,
      "step": 2543
    },
    {
      "epoch": 4.838801711840228,
      "grad_norm": 0.24070213735103607,
      "learning_rate": 8.387424579231503e-05,
      "loss": 0.0772,
      "step": 2544
    },
    {
      "epoch": 4.840703756538279,
      "grad_norm": 0.4367123246192932,
      "learning_rate": 8.386789456970468e-05,
      "loss": 0.1227,
      "step": 2545
    },
    {
      "epoch": 4.842605801236329,
      "grad_norm": 0.3168450891971588,
      "learning_rate": 8.386154334709432e-05,
      "loss": 0.0928,
      "step": 2546
    },
    {
      "epoch": 4.8445078459343796,
      "grad_norm": 0.36236846446990967,
      "learning_rate": 8.385519212448397e-05,
      "loss": 0.0997,
      "step": 2547
    },
    {
      "epoch": 4.8464098906324296,
      "grad_norm": 0.31763169169425964,
      "learning_rate": 8.384884090187362e-05,
      "loss": 0.1093,
      "step": 2548
    },
    {
      "epoch": 4.84831193533048,
      "grad_norm": 0.3502260148525238,
      "learning_rate": 8.384248967926326e-05,
      "loss": 0.1299,
      "step": 2549
    },
    {
      "epoch": 4.85021398002853,
      "grad_norm": 0.3593395948410034,
      "learning_rate": 8.383613845665291e-05,
      "loss": 0.1066,
      "step": 2550
    },
    {
      "epoch": 4.852116024726581,
      "grad_norm": 0.39665883779525757,
      "learning_rate": 8.382978723404256e-05,
      "loss": 0.1267,
      "step": 2551
    },
    {
      "epoch": 4.854018069424631,
      "grad_norm": 0.4395765960216522,
      "learning_rate": 8.38234360114322e-05,
      "loss": 0.174,
      "step": 2552
    },
    {
      "epoch": 4.855920114122682,
      "grad_norm": 0.3507075607776642,
      "learning_rate": 8.381708478882185e-05,
      "loss": 0.0953,
      "step": 2553
    },
    {
      "epoch": 4.857822158820732,
      "grad_norm": 0.3769589364528656,
      "learning_rate": 8.381073356621149e-05,
      "loss": 0.1395,
      "step": 2554
    },
    {
      "epoch": 4.859724203518782,
      "grad_norm": 0.30503159761428833,
      "learning_rate": 8.380438234360116e-05,
      "loss": 0.0937,
      "step": 2555
    },
    {
      "epoch": 4.861626248216833,
      "grad_norm": 0.39943060278892517,
      "learning_rate": 8.37980311209908e-05,
      "loss": 0.1103,
      "step": 2556
    },
    {
      "epoch": 4.863528292914884,
      "grad_norm": 0.36200422048568726,
      "learning_rate": 8.379167989838043e-05,
      "loss": 0.1135,
      "step": 2557
    },
    {
      "epoch": 4.865430337612934,
      "grad_norm": 0.3811735510826111,
      "learning_rate": 8.37853286757701e-05,
      "loss": 0.1265,
      "step": 2558
    },
    {
      "epoch": 4.867332382310984,
      "grad_norm": 0.42090871930122375,
      "learning_rate": 8.377897745315974e-05,
      "loss": 0.1339,
      "step": 2559
    },
    {
      "epoch": 4.869234427009035,
      "grad_norm": 0.41796380281448364,
      "learning_rate": 8.377262623054939e-05,
      "loss": 0.1136,
      "step": 2560
    },
    {
      "epoch": 4.871136471707085,
      "grad_norm": 0.33189094066619873,
      "learning_rate": 8.376627500793903e-05,
      "loss": 0.0923,
      "step": 2561
    },
    {
      "epoch": 4.873038516405136,
      "grad_norm": 0.46369072794914246,
      "learning_rate": 8.375992378532868e-05,
      "loss": 0.1236,
      "step": 2562
    },
    {
      "epoch": 4.874940561103186,
      "grad_norm": 0.27973759174346924,
      "learning_rate": 8.375357256271833e-05,
      "loss": 0.0933,
      "step": 2563
    },
    {
      "epoch": 4.876842605801237,
      "grad_norm": 0.39309409260749817,
      "learning_rate": 8.374722134010797e-05,
      "loss": 0.1135,
      "step": 2564
    },
    {
      "epoch": 4.878744650499287,
      "grad_norm": 0.43652641773223877,
      "learning_rate": 8.374087011749763e-05,
      "loss": 0.136,
      "step": 2565
    },
    {
      "epoch": 4.8806466951973375,
      "grad_norm": 0.30485180020332336,
      "learning_rate": 8.373451889488727e-05,
      "loss": 0.0894,
      "step": 2566
    },
    {
      "epoch": 4.8825487398953875,
      "grad_norm": 0.40164196491241455,
      "learning_rate": 8.372816767227691e-05,
      "loss": 0.1235,
      "step": 2567
    },
    {
      "epoch": 4.884450784593438,
      "grad_norm": 0.3442533314228058,
      "learning_rate": 8.372181644966656e-05,
      "loss": 0.1222,
      "step": 2568
    },
    {
      "epoch": 4.886352829291488,
      "grad_norm": 0.38092851638793945,
      "learning_rate": 8.371546522705621e-05,
      "loss": 0.1135,
      "step": 2569
    },
    {
      "epoch": 4.888254873989538,
      "grad_norm": 0.37114188075065613,
      "learning_rate": 8.370911400444585e-05,
      "loss": 0.1181,
      "step": 2570
    },
    {
      "epoch": 4.890156918687589,
      "grad_norm": 0.35971492528915405,
      "learning_rate": 8.37027627818355e-05,
      "loss": 0.1247,
      "step": 2571
    },
    {
      "epoch": 4.892058963385639,
      "grad_norm": 0.25756967067718506,
      "learning_rate": 8.369641155922516e-05,
      "loss": 0.0929,
      "step": 2572
    },
    {
      "epoch": 4.89396100808369,
      "grad_norm": 0.4541129171848297,
      "learning_rate": 8.369006033661481e-05,
      "loss": 0.142,
      "step": 2573
    },
    {
      "epoch": 4.89586305278174,
      "grad_norm": 0.48526903986930847,
      "learning_rate": 8.368370911400445e-05,
      "loss": 0.1612,
      "step": 2574
    },
    {
      "epoch": 4.897765097479791,
      "grad_norm": 0.31703343987464905,
      "learning_rate": 8.36773578913941e-05,
      "loss": 0.1135,
      "step": 2575
    },
    {
      "epoch": 4.899667142177841,
      "grad_norm": 0.2969724237918854,
      "learning_rate": 8.367100666878375e-05,
      "loss": 0.1148,
      "step": 2576
    },
    {
      "epoch": 4.901569186875892,
      "grad_norm": 0.37165188789367676,
      "learning_rate": 8.366465544617339e-05,
      "loss": 0.1066,
      "step": 2577
    },
    {
      "epoch": 4.903471231573942,
      "grad_norm": 0.2899304926395416,
      "learning_rate": 8.365830422356304e-05,
      "loss": 0.0896,
      "step": 2578
    },
    {
      "epoch": 4.905373276271993,
      "grad_norm": 0.3420521914958954,
      "learning_rate": 8.365195300095269e-05,
      "loss": 0.0929,
      "step": 2579
    },
    {
      "epoch": 4.907275320970043,
      "grad_norm": 0.48174387216567993,
      "learning_rate": 8.364560177834233e-05,
      "loss": 0.1422,
      "step": 2580
    },
    {
      "epoch": 4.909177365668093,
      "grad_norm": 0.3492242693901062,
      "learning_rate": 8.363925055573198e-05,
      "loss": 0.1116,
      "step": 2581
    },
    {
      "epoch": 4.911079410366144,
      "grad_norm": 0.367914080619812,
      "learning_rate": 8.363289933312163e-05,
      "loss": 0.1139,
      "step": 2582
    },
    {
      "epoch": 4.912981455064194,
      "grad_norm": 0.32939612865448,
      "learning_rate": 8.362654811051129e-05,
      "loss": 0.1175,
      "step": 2583
    },
    {
      "epoch": 4.914883499762245,
      "grad_norm": 0.3939587473869324,
      "learning_rate": 8.362019688790092e-05,
      "loss": 0.1263,
      "step": 2584
    },
    {
      "epoch": 4.916785544460295,
      "grad_norm": 0.36641520261764526,
      "learning_rate": 8.361384566529058e-05,
      "loss": 0.1219,
      "step": 2585
    },
    {
      "epoch": 4.9186875891583455,
      "grad_norm": 0.2804834544658661,
      "learning_rate": 8.360749444268023e-05,
      "loss": 0.0839,
      "step": 2586
    },
    {
      "epoch": 4.9205896338563955,
      "grad_norm": 0.310461163520813,
      "learning_rate": 8.360114322006987e-05,
      "loss": 0.0949,
      "step": 2587
    },
    {
      "epoch": 4.922491678554446,
      "grad_norm": 0.34361201524734497,
      "learning_rate": 8.35947919974595e-05,
      "loss": 0.1167,
      "step": 2588
    },
    {
      "epoch": 4.924393723252496,
      "grad_norm": 0.3348811864852905,
      "learning_rate": 8.358844077484917e-05,
      "loss": 0.1035,
      "step": 2589
    },
    {
      "epoch": 4.926295767950547,
      "grad_norm": 0.24014593660831451,
      "learning_rate": 8.358208955223881e-05,
      "loss": 0.1413,
      "step": 2590
    },
    {
      "epoch": 4.928197812648597,
      "grad_norm": 0.4338441491127014,
      "learning_rate": 8.357573832962846e-05,
      "loss": 0.1186,
      "step": 2591
    },
    {
      "epoch": 4.930099857346647,
      "grad_norm": 0.3601210415363312,
      "learning_rate": 8.356938710701811e-05,
      "loss": 0.1014,
      "step": 2592
    },
    {
      "epoch": 4.932001902044698,
      "grad_norm": 0.2996499538421631,
      "learning_rate": 8.356303588440775e-05,
      "loss": 0.0906,
      "step": 2593
    },
    {
      "epoch": 4.933903946742748,
      "grad_norm": 0.30851230025291443,
      "learning_rate": 8.35566846617974e-05,
      "loss": 0.0806,
      "step": 2594
    },
    {
      "epoch": 4.935805991440799,
      "grad_norm": 0.22290165722370148,
      "learning_rate": 8.355033343918704e-05,
      "loss": 0.0728,
      "step": 2595
    },
    {
      "epoch": 4.937708036138849,
      "grad_norm": 0.28518247604370117,
      "learning_rate": 8.35439822165767e-05,
      "loss": 0.0894,
      "step": 2596
    },
    {
      "epoch": 4.9396100808369,
      "grad_norm": 0.424231618642807,
      "learning_rate": 8.353763099396634e-05,
      "loss": 0.1157,
      "step": 2597
    },
    {
      "epoch": 4.94151212553495,
      "grad_norm": 0.5748564600944519,
      "learning_rate": 8.353127977135598e-05,
      "loss": 0.1777,
      "step": 2598
    },
    {
      "epoch": 4.943414170233001,
      "grad_norm": 0.39010798931121826,
      "learning_rate": 8.352492854874565e-05,
      "loss": 0.104,
      "step": 2599
    },
    {
      "epoch": 4.945316214931051,
      "grad_norm": 0.40491625666618347,
      "learning_rate": 8.351857732613529e-05,
      "loss": 0.115,
      "step": 2600
    },
    {
      "epoch": 4.947218259629102,
      "grad_norm": 0.3881874084472656,
      "learning_rate": 8.351222610352494e-05,
      "loss": 0.1125,
      "step": 2601
    },
    {
      "epoch": 4.949120304327152,
      "grad_norm": 0.4075947403907776,
      "learning_rate": 8.350587488091458e-05,
      "loss": 0.1376,
      "step": 2602
    },
    {
      "epoch": 4.951022349025202,
      "grad_norm": 0.4263762831687927,
      "learning_rate": 8.349952365830423e-05,
      "loss": 0.1214,
      "step": 2603
    },
    {
      "epoch": 4.9529243937232525,
      "grad_norm": 0.4403824806213379,
      "learning_rate": 8.349317243569388e-05,
      "loss": 0.1212,
      "step": 2604
    },
    {
      "epoch": 4.9548264384213025,
      "grad_norm": 0.41958004236221313,
      "learning_rate": 8.348682121308352e-05,
      "loss": 0.1197,
      "step": 2605
    },
    {
      "epoch": 4.956728483119353,
      "grad_norm": 0.3664645850658417,
      "learning_rate": 8.348046999047317e-05,
      "loss": 0.1208,
      "step": 2606
    },
    {
      "epoch": 4.958630527817403,
      "grad_norm": 0.3618158996105194,
      "learning_rate": 8.347411876786282e-05,
      "loss": 0.1241,
      "step": 2607
    },
    {
      "epoch": 4.960532572515454,
      "grad_norm": 0.3135223686695099,
      "learning_rate": 8.346776754525246e-05,
      "loss": 0.0807,
      "step": 2608
    },
    {
      "epoch": 4.962434617213504,
      "grad_norm": 0.3673211932182312,
      "learning_rate": 8.346141632264211e-05,
      "loss": 0.1188,
      "step": 2609
    },
    {
      "epoch": 4.964336661911555,
      "grad_norm": 0.34168919920921326,
      "learning_rate": 8.345506510003176e-05,
      "loss": 0.1113,
      "step": 2610
    },
    {
      "epoch": 4.966238706609605,
      "grad_norm": 0.3807981312274933,
      "learning_rate": 8.34487138774214e-05,
      "loss": 0.1243,
      "step": 2611
    },
    {
      "epoch": 4.968140751307656,
      "grad_norm": 0.35833629965782166,
      "learning_rate": 8.344236265481105e-05,
      "loss": 0.1175,
      "step": 2612
    },
    {
      "epoch": 4.970042796005706,
      "grad_norm": 0.4410795569419861,
      "learning_rate": 8.34360114322007e-05,
      "loss": 0.1174,
      "step": 2613
    },
    {
      "epoch": 4.971944840703756,
      "grad_norm": 0.27122291922569275,
      "learning_rate": 8.342966020959036e-05,
      "loss": 0.1062,
      "step": 2614
    },
    {
      "epoch": 4.973846885401807,
      "grad_norm": 0.3411978483200073,
      "learning_rate": 8.342330898698e-05,
      "loss": 0.1274,
      "step": 2615
    },
    {
      "epoch": 4.975748930099857,
      "grad_norm": 0.36536306142807007,
      "learning_rate": 8.341695776436965e-05,
      "loss": 0.1182,
      "step": 2616
    },
    {
      "epoch": 4.977650974797908,
      "grad_norm": 0.3873109221458435,
      "learning_rate": 8.34106065417593e-05,
      "loss": 0.1043,
      "step": 2617
    },
    {
      "epoch": 4.979553019495958,
      "grad_norm": 0.30192115902900696,
      "learning_rate": 8.340425531914894e-05,
      "loss": 0.0984,
      "step": 2618
    },
    {
      "epoch": 4.981455064194009,
      "grad_norm": 0.37886565923690796,
      "learning_rate": 8.339790409653859e-05,
      "loss": 0.1161,
      "step": 2619
    },
    {
      "epoch": 4.983357108892059,
      "grad_norm": 0.34957846999168396,
      "learning_rate": 8.339155287392824e-05,
      "loss": 0.1083,
      "step": 2620
    },
    {
      "epoch": 4.98525915359011,
      "grad_norm": 0.3169527053833008,
      "learning_rate": 8.338520165131788e-05,
      "loss": 0.088,
      "step": 2621
    },
    {
      "epoch": 4.98716119828816,
      "grad_norm": 0.41983914375305176,
      "learning_rate": 8.337885042870753e-05,
      "loss": 0.1158,
      "step": 2622
    },
    {
      "epoch": 4.9890632429862105,
      "grad_norm": 0.3467552661895752,
      "learning_rate": 8.337249920609718e-05,
      "loss": 0.0958,
      "step": 2623
    },
    {
      "epoch": 4.9909652876842605,
      "grad_norm": 0.3872130513191223,
      "learning_rate": 8.336614798348682e-05,
      "loss": 0.1012,
      "step": 2624
    },
    {
      "epoch": 4.9928673323823105,
      "grad_norm": 0.2966238856315613,
      "learning_rate": 8.335979676087647e-05,
      "loss": 0.0913,
      "step": 2625
    },
    {
      "epoch": 4.994769377080361,
      "grad_norm": 0.4195917248725891,
      "learning_rate": 8.335344553826611e-05,
      "loss": 0.1124,
      "step": 2626
    },
    {
      "epoch": 4.996671421778412,
      "grad_norm": 0.39411017298698425,
      "learning_rate": 8.334709431565578e-05,
      "loss": 0.1286,
      "step": 2627
    },
    {
      "epoch": 4.998573466476462,
      "grad_norm": 0.3783339262008667,
      "learning_rate": 8.334074309304541e-05,
      "loss": 0.1084,
      "step": 2628
    },
    {
      "epoch": 5.000475511174512,
      "grad_norm": 0.4462081789970398,
      "learning_rate": 8.333439187043505e-05,
      "loss": 0.118,
      "step": 2629
    },
    {
      "epoch": 5.002377555872563,
      "grad_norm": 0.26820147037506104,
      "learning_rate": 8.332804064782472e-05,
      "loss": 0.1318,
      "step": 2630
    },
    {
      "epoch": 5.004279600570613,
      "grad_norm": 0.27382394671440125,
      "learning_rate": 8.332168942521436e-05,
      "loss": 0.0847,
      "step": 2631
    },
    {
      "epoch": 5.006181645268664,
      "grad_norm": 0.2846938371658325,
      "learning_rate": 8.331533820260401e-05,
      "loss": 0.0897,
      "step": 2632
    },
    {
      "epoch": 5.008083689966714,
      "grad_norm": 0.27744022011756897,
      "learning_rate": 8.330898697999365e-05,
      "loss": 0.0809,
      "step": 2633
    },
    {
      "epoch": 5.009985734664765,
      "grad_norm": 0.23713688552379608,
      "learning_rate": 8.33026357573833e-05,
      "loss": 0.0768,
      "step": 2634
    },
    {
      "epoch": 5.011887779362815,
      "grad_norm": 0.4143311381340027,
      "learning_rate": 8.329628453477295e-05,
      "loss": 0.0924,
      "step": 2635
    },
    {
      "epoch": 5.013789824060866,
      "grad_norm": 0.315565288066864,
      "learning_rate": 8.328993331216259e-05,
      "loss": 0.082,
      "step": 2636
    },
    {
      "epoch": 5.015691868758916,
      "grad_norm": 0.3401367664337158,
      "learning_rate": 8.328358208955225e-05,
      "loss": 0.1119,
      "step": 2637
    },
    {
      "epoch": 5.017593913456966,
      "grad_norm": 0.30635321140289307,
      "learning_rate": 8.327723086694189e-05,
      "loss": 0.1126,
      "step": 2638
    },
    {
      "epoch": 5.019495958155017,
      "grad_norm": 0.28435221314430237,
      "learning_rate": 8.327087964433153e-05,
      "loss": 0.0782,
      "step": 2639
    },
    {
      "epoch": 5.021398002853067,
      "grad_norm": 0.36568841338157654,
      "learning_rate": 8.326452842172118e-05,
      "loss": 0.1068,
      "step": 2640
    },
    {
      "epoch": 5.023300047551118,
      "grad_norm": 0.31536993384361267,
      "learning_rate": 8.325817719911083e-05,
      "loss": 0.0999,
      "step": 2641
    },
    {
      "epoch": 5.025202092249168,
      "grad_norm": 0.30222904682159424,
      "learning_rate": 8.325182597650047e-05,
      "loss": 0.09,
      "step": 2642
    },
    {
      "epoch": 5.0271041369472185,
      "grad_norm": 0.27157366275787354,
      "learning_rate": 8.324547475389012e-05,
      "loss": 0.0877,
      "step": 2643
    },
    {
      "epoch": 5.0290061816452685,
      "grad_norm": 0.239767923951149,
      "learning_rate": 8.323912353127978e-05,
      "loss": 0.0774,
      "step": 2644
    },
    {
      "epoch": 5.030908226343319,
      "grad_norm": 0.3949924111366272,
      "learning_rate": 8.323277230866943e-05,
      "loss": 0.0971,
      "step": 2645
    },
    {
      "epoch": 5.032810271041369,
      "grad_norm": 0.384383887052536,
      "learning_rate": 8.322642108605907e-05,
      "loss": 0.093,
      "step": 2646
    },
    {
      "epoch": 5.03471231573942,
      "grad_norm": 0.38407453894615173,
      "learning_rate": 8.322006986344872e-05,
      "loss": 0.097,
      "step": 2647
    },
    {
      "epoch": 5.03661436043747,
      "grad_norm": 0.2176978439092636,
      "learning_rate": 8.321371864083837e-05,
      "loss": 0.061,
      "step": 2648
    },
    {
      "epoch": 5.038516405135521,
      "grad_norm": 0.4389532208442688,
      "learning_rate": 8.320736741822801e-05,
      "loss": 0.1065,
      "step": 2649
    },
    {
      "epoch": 5.040418449833571,
      "grad_norm": 0.2318810671567917,
      "learning_rate": 8.320101619561766e-05,
      "loss": 0.0698,
      "step": 2650
    },
    {
      "epoch": 5.042320494531621,
      "grad_norm": 0.3554401993751526,
      "learning_rate": 8.319466497300731e-05,
      "loss": 0.1052,
      "step": 2651
    },
    {
      "epoch": 5.044222539229672,
      "grad_norm": 0.4346822500228882,
      "learning_rate": 8.318831375039695e-05,
      "loss": 0.1108,
      "step": 2652
    },
    {
      "epoch": 5.046124583927722,
      "grad_norm": 0.34418416023254395,
      "learning_rate": 8.31819625277866e-05,
      "loss": 0.1003,
      "step": 2653
    },
    {
      "epoch": 5.048026628625773,
      "grad_norm": 0.35901421308517456,
      "learning_rate": 8.317561130517625e-05,
      "loss": 0.0786,
      "step": 2654
    },
    {
      "epoch": 5.049928673323823,
      "grad_norm": 0.2725570797920227,
      "learning_rate": 8.31692600825659e-05,
      "loss": 0.0901,
      "step": 2655
    },
    {
      "epoch": 5.051830718021874,
      "grad_norm": 0.3747507333755493,
      "learning_rate": 8.316290885995554e-05,
      "loss": 0.0922,
      "step": 2656
    },
    {
      "epoch": 5.053732762719924,
      "grad_norm": 0.3192732036113739,
      "learning_rate": 8.315655763734518e-05,
      "loss": 0.0834,
      "step": 2657
    },
    {
      "epoch": 5.055634807417975,
      "grad_norm": 0.32220616936683655,
      "learning_rate": 8.315020641473485e-05,
      "loss": 0.0976,
      "step": 2658
    },
    {
      "epoch": 5.057536852116025,
      "grad_norm": 0.29179829359054565,
      "learning_rate": 8.314385519212449e-05,
      "loss": 0.0865,
      "step": 2659
    },
    {
      "epoch": 5.0594388968140755,
      "grad_norm": 0.337753564119339,
      "learning_rate": 8.313750396951412e-05,
      "loss": 0.0934,
      "step": 2660
    },
    {
      "epoch": 5.0613409415121255,
      "grad_norm": 0.2385604828596115,
      "learning_rate": 8.313115274690379e-05,
      "loss": 0.0897,
      "step": 2661
    },
    {
      "epoch": 5.0632429862101755,
      "grad_norm": 0.3223874568939209,
      "learning_rate": 8.312480152429343e-05,
      "loss": 0.113,
      "step": 2662
    },
    {
      "epoch": 5.065145030908226,
      "grad_norm": 0.2641155421733856,
      "learning_rate": 8.311845030168308e-05,
      "loss": 0.0784,
      "step": 2663
    },
    {
      "epoch": 5.067047075606276,
      "grad_norm": 0.3063553273677826,
      "learning_rate": 8.311209907907272e-05,
      "loss": 0.0895,
      "step": 2664
    },
    {
      "epoch": 5.068949120304327,
      "grad_norm": 0.28479015827178955,
      "learning_rate": 8.310574785646237e-05,
      "loss": 0.0781,
      "step": 2665
    },
    {
      "epoch": 5.070851165002377,
      "grad_norm": 0.25443270802497864,
      "learning_rate": 8.309939663385202e-05,
      "loss": 0.0816,
      "step": 2666
    },
    {
      "epoch": 5.072753209700428,
      "grad_norm": 0.3532758951187134,
      "learning_rate": 8.309304541124166e-05,
      "loss": 0.1004,
      "step": 2667
    },
    {
      "epoch": 5.074655254398478,
      "grad_norm": 0.26470455527305603,
      "learning_rate": 8.308669418863133e-05,
      "loss": 0.0959,
      "step": 2668
    },
    {
      "epoch": 5.076557299096529,
      "grad_norm": 0.3608281910419464,
      "learning_rate": 8.308034296602096e-05,
      "loss": 0.0925,
      "step": 2669
    },
    {
      "epoch": 5.078459343794579,
      "grad_norm": 0.223192036151886,
      "learning_rate": 8.30739917434106e-05,
      "loss": 0.0652,
      "step": 2670
    },
    {
      "epoch": 5.08036138849263,
      "grad_norm": 0.36757364869117737,
      "learning_rate": 8.306764052080025e-05,
      "loss": 0.0847,
      "step": 2671
    },
    {
      "epoch": 5.08226343319068,
      "grad_norm": 0.3613258898258209,
      "learning_rate": 8.30612892981899e-05,
      "loss": 0.1133,
      "step": 2672
    },
    {
      "epoch": 5.08416547788873,
      "grad_norm": 0.3451091945171356,
      "learning_rate": 8.305493807557956e-05,
      "loss": 0.0851,
      "step": 2673
    },
    {
      "epoch": 5.086067522586781,
      "grad_norm": 0.39230048656463623,
      "learning_rate": 8.30485868529692e-05,
      "loss": 0.1264,
      "step": 2674
    },
    {
      "epoch": 5.087969567284831,
      "grad_norm": 0.30938684940338135,
      "learning_rate": 8.304223563035885e-05,
      "loss": 0.1009,
      "step": 2675
    },
    {
      "epoch": 5.089871611982882,
      "grad_norm": 0.29073426127433777,
      "learning_rate": 8.30358844077485e-05,
      "loss": 0.0884,
      "step": 2676
    },
    {
      "epoch": 5.091773656680932,
      "grad_norm": 0.32096439599990845,
      "learning_rate": 8.302953318513814e-05,
      "loss": 0.0944,
      "step": 2677
    },
    {
      "epoch": 5.093675701378983,
      "grad_norm": 0.3321041464805603,
      "learning_rate": 8.302318196252779e-05,
      "loss": 0.0945,
      "step": 2678
    },
    {
      "epoch": 5.095577746077033,
      "grad_norm": 0.4814242720603943,
      "learning_rate": 8.301683073991744e-05,
      "loss": 0.1222,
      "step": 2679
    },
    {
      "epoch": 5.0974797907750835,
      "grad_norm": 0.5268080830574036,
      "learning_rate": 8.301047951730708e-05,
      "loss": 0.1096,
      "step": 2680
    },
    {
      "epoch": 5.0993818354731335,
      "grad_norm": 0.29475218057632446,
      "learning_rate": 8.300412829469673e-05,
      "loss": 0.0794,
      "step": 2681
    },
    {
      "epoch": 5.101283880171184,
      "grad_norm": 0.2624375820159912,
      "learning_rate": 8.299777707208638e-05,
      "loss": 0.0745,
      "step": 2682
    },
    {
      "epoch": 5.103185924869234,
      "grad_norm": 0.3305892050266266,
      "learning_rate": 8.299142584947602e-05,
      "loss": 0.0931,
      "step": 2683
    },
    {
      "epoch": 5.105087969567284,
      "grad_norm": 0.3345329165458679,
      "learning_rate": 8.298507462686567e-05,
      "loss": 0.0979,
      "step": 2684
    },
    {
      "epoch": 5.106990014265335,
      "grad_norm": 0.37109294533729553,
      "learning_rate": 8.297872340425533e-05,
      "loss": 0.0908,
      "step": 2685
    },
    {
      "epoch": 5.108892058963385,
      "grad_norm": 0.28300270438194275,
      "learning_rate": 8.297237218164498e-05,
      "loss": 0.0817,
      "step": 2686
    },
    {
      "epoch": 5.110794103661436,
      "grad_norm": 0.29176023602485657,
      "learning_rate": 8.296602095903462e-05,
      "loss": 0.0818,
      "step": 2687
    },
    {
      "epoch": 5.112696148359486,
      "grad_norm": 0.38012629747390747,
      "learning_rate": 8.295966973642427e-05,
      "loss": 0.1113,
      "step": 2688
    },
    {
      "epoch": 5.114598193057537,
      "grad_norm": 0.3183780610561371,
      "learning_rate": 8.295331851381392e-05,
      "loss": 0.114,
      "step": 2689
    },
    {
      "epoch": 5.116500237755587,
      "grad_norm": 0.34422457218170166,
      "learning_rate": 8.294696729120356e-05,
      "loss": 0.1029,
      "step": 2690
    },
    {
      "epoch": 5.118402282453638,
      "grad_norm": 0.29582858085632324,
      "learning_rate": 8.294061606859321e-05,
      "loss": 0.0921,
      "step": 2691
    },
    {
      "epoch": 5.120304327151688,
      "grad_norm": 0.29475873708724976,
      "learning_rate": 8.293426484598286e-05,
      "loss": 0.0699,
      "step": 2692
    },
    {
      "epoch": 5.122206371849739,
      "grad_norm": 0.24619872868061066,
      "learning_rate": 8.29279136233725e-05,
      "loss": 0.0729,
      "step": 2693
    },
    {
      "epoch": 5.124108416547789,
      "grad_norm": 0.2696808874607086,
      "learning_rate": 8.292156240076215e-05,
      "loss": 0.0907,
      "step": 2694
    },
    {
      "epoch": 5.12601046124584,
      "grad_norm": 0.3869519829750061,
      "learning_rate": 8.29152111781518e-05,
      "loss": 0.1007,
      "step": 2695
    },
    {
      "epoch": 5.12791250594389,
      "grad_norm": 0.3821418881416321,
      "learning_rate": 8.290885995554144e-05,
      "loss": 0.1093,
      "step": 2696
    },
    {
      "epoch": 5.12981455064194,
      "grad_norm": 0.5121551752090454,
      "learning_rate": 8.29025087329311e-05,
      "loss": 0.1344,
      "step": 2697
    },
    {
      "epoch": 5.131716595339991,
      "grad_norm": 0.2786102890968323,
      "learning_rate": 8.289615751032073e-05,
      "loss": 0.0883,
      "step": 2698
    },
    {
      "epoch": 5.133618640038041,
      "grad_norm": 0.31683072447776794,
      "learning_rate": 8.28898062877104e-05,
      "loss": 0.0901,
      "step": 2699
    },
    {
      "epoch": 5.1355206847360915,
      "grad_norm": 0.318406879901886,
      "learning_rate": 8.288345506510004e-05,
      "loss": 0.0924,
      "step": 2700
    },
    {
      "epoch": 5.1374227294341415,
      "grad_norm": 0.3161453604698181,
      "learning_rate": 8.287710384248967e-05,
      "loss": 0.0863,
      "step": 2701
    },
    {
      "epoch": 5.139324774132192,
      "grad_norm": 0.4410136342048645,
      "learning_rate": 8.287075261987934e-05,
      "loss": 0.1014,
      "step": 2702
    },
    {
      "epoch": 5.141226818830242,
      "grad_norm": 0.3554822504520416,
      "learning_rate": 8.286440139726898e-05,
      "loss": 0.0905,
      "step": 2703
    },
    {
      "epoch": 5.143128863528293,
      "grad_norm": 0.3166615664958954,
      "learning_rate": 8.285805017465863e-05,
      "loss": 0.0814,
      "step": 2704
    },
    {
      "epoch": 5.145030908226343,
      "grad_norm": 0.314028263092041,
      "learning_rate": 8.285169895204827e-05,
      "loss": 0.0819,
      "step": 2705
    },
    {
      "epoch": 5.146932952924394,
      "grad_norm": 0.3088028132915497,
      "learning_rate": 8.284534772943792e-05,
      "loss": 0.1025,
      "step": 2706
    },
    {
      "epoch": 5.148834997622444,
      "grad_norm": 0.30459243059158325,
      "learning_rate": 8.283899650682757e-05,
      "loss": 0.0919,
      "step": 2707
    },
    {
      "epoch": 5.150737042320494,
      "grad_norm": 0.31617748737335205,
      "learning_rate": 8.283264528421721e-05,
      "loss": 0.0828,
      "step": 2708
    },
    {
      "epoch": 5.152639087018545,
      "grad_norm": 0.38467937707901,
      "learning_rate": 8.282629406160688e-05,
      "loss": 0.1875,
      "step": 2709
    },
    {
      "epoch": 5.154541131716595,
      "grad_norm": 0.31181344389915466,
      "learning_rate": 8.281994283899651e-05,
      "loss": 0.0836,
      "step": 2710
    },
    {
      "epoch": 5.156443176414646,
      "grad_norm": 0.2717238962650299,
      "learning_rate": 8.281359161638615e-05,
      "loss": 0.1023,
      "step": 2711
    },
    {
      "epoch": 5.158345221112696,
      "grad_norm": 0.24646538496017456,
      "learning_rate": 8.28072403937758e-05,
      "loss": 0.073,
      "step": 2712
    },
    {
      "epoch": 5.160247265810747,
      "grad_norm": 0.28425633907318115,
      "learning_rate": 8.280088917116546e-05,
      "loss": 0.082,
      "step": 2713
    },
    {
      "epoch": 5.162149310508797,
      "grad_norm": 0.24602612853050232,
      "learning_rate": 8.27945379485551e-05,
      "loss": 0.0748,
      "step": 2714
    },
    {
      "epoch": 5.164051355206848,
      "grad_norm": 0.3181002736091614,
      "learning_rate": 8.278818672594475e-05,
      "loss": 0.0915,
      "step": 2715
    },
    {
      "epoch": 5.165953399904898,
      "grad_norm": 0.3696095049381256,
      "learning_rate": 8.27818355033344e-05,
      "loss": 0.1079,
      "step": 2716
    },
    {
      "epoch": 5.1678554446029485,
      "grad_norm": 0.3242207467556,
      "learning_rate": 8.277548428072405e-05,
      "loss": 0.0963,
      "step": 2717
    },
    {
      "epoch": 5.1697574893009985,
      "grad_norm": 0.3968127965927124,
      "learning_rate": 8.276913305811369e-05,
      "loss": 0.0799,
      "step": 2718
    },
    {
      "epoch": 5.171659533999049,
      "grad_norm": 0.2539953887462616,
      "learning_rate": 8.276278183550334e-05,
      "loss": 0.0678,
      "step": 2719
    },
    {
      "epoch": 5.173561578697099,
      "grad_norm": 0.34455937147140503,
      "learning_rate": 8.275643061289299e-05,
      "loss": 0.1237,
      "step": 2720
    },
    {
      "epoch": 5.175463623395149,
      "grad_norm": 0.29677414894104004,
      "learning_rate": 8.275007939028263e-05,
      "loss": 0.1007,
      "step": 2721
    },
    {
      "epoch": 5.1773656680932,
      "grad_norm": 0.2782600224018097,
      "learning_rate": 8.274372816767228e-05,
      "loss": 0.0945,
      "step": 2722
    },
    {
      "epoch": 5.17926771279125,
      "grad_norm": 0.3580763339996338,
      "learning_rate": 8.273737694506193e-05,
      "loss": 0.0931,
      "step": 2723
    },
    {
      "epoch": 5.181169757489301,
      "grad_norm": 0.19728906452655792,
      "learning_rate": 8.273102572245157e-05,
      "loss": 0.0695,
      "step": 2724
    },
    {
      "epoch": 5.183071802187351,
      "grad_norm": 0.2223612666130066,
      "learning_rate": 8.272467449984122e-05,
      "loss": 0.0631,
      "step": 2725
    },
    {
      "epoch": 5.184973846885402,
      "grad_norm": 0.2530241906642914,
      "learning_rate": 8.271832327723088e-05,
      "loss": 0.068,
      "step": 2726
    },
    {
      "epoch": 5.186875891583452,
      "grad_norm": 0.605760931968689,
      "learning_rate": 8.271197205462053e-05,
      "loss": 0.1698,
      "step": 2727
    },
    {
      "epoch": 5.188777936281503,
      "grad_norm": 0.39937227964401245,
      "learning_rate": 8.270562083201017e-05,
      "loss": 0.1197,
      "step": 2728
    },
    {
      "epoch": 5.190679980979553,
      "grad_norm": 0.380604088306427,
      "learning_rate": 8.26992696093998e-05,
      "loss": 0.1024,
      "step": 2729
    },
    {
      "epoch": 5.192582025677604,
      "grad_norm": 0.3906427323818207,
      "learning_rate": 8.269291838678947e-05,
      "loss": 0.1103,
      "step": 2730
    },
    {
      "epoch": 5.194484070375654,
      "grad_norm": 0.25769877433776855,
      "learning_rate": 8.268656716417911e-05,
      "loss": 0.0716,
      "step": 2731
    },
    {
      "epoch": 5.196386115073704,
      "grad_norm": 0.27115434408187866,
      "learning_rate": 8.268021594156875e-05,
      "loss": 0.0748,
      "step": 2732
    },
    {
      "epoch": 5.198288159771755,
      "grad_norm": 0.31713467836380005,
      "learning_rate": 8.267386471895841e-05,
      "loss": 0.1119,
      "step": 2733
    },
    {
      "epoch": 5.200190204469805,
      "grad_norm": 0.27098706364631653,
      "learning_rate": 8.266751349634805e-05,
      "loss": 0.0916,
      "step": 2734
    },
    {
      "epoch": 5.202092249167856,
      "grad_norm": 0.26350903511047363,
      "learning_rate": 8.26611622737377e-05,
      "loss": 0.1107,
      "step": 2735
    },
    {
      "epoch": 5.203994293865906,
      "grad_norm": 0.3183554410934448,
      "learning_rate": 8.265481105112734e-05,
      "loss": 0.0864,
      "step": 2736
    },
    {
      "epoch": 5.2058963385639565,
      "grad_norm": 0.33910661935806274,
      "learning_rate": 8.264845982851699e-05,
      "loss": 0.0975,
      "step": 2737
    },
    {
      "epoch": 5.2077983832620065,
      "grad_norm": 0.23509973287582397,
      "learning_rate": 8.264210860590664e-05,
      "loss": 0.0637,
      "step": 2738
    },
    {
      "epoch": 5.209700427960057,
      "grad_norm": 0.31973764300346375,
      "learning_rate": 8.263575738329628e-05,
      "loss": 0.0974,
      "step": 2739
    },
    {
      "epoch": 5.211602472658107,
      "grad_norm": 0.42649564146995544,
      "learning_rate": 8.262940616068595e-05,
      "loss": 0.1126,
      "step": 2740
    },
    {
      "epoch": 5.213504517356158,
      "grad_norm": 0.2706199288368225,
      "learning_rate": 8.262305493807558e-05,
      "loss": 0.0855,
      "step": 2741
    },
    {
      "epoch": 5.215406562054208,
      "grad_norm": 0.4610327184200287,
      "learning_rate": 8.261670371546522e-05,
      "loss": 0.1214,
      "step": 2742
    },
    {
      "epoch": 5.217308606752258,
      "grad_norm": 0.2792705297470093,
      "learning_rate": 8.261035249285488e-05,
      "loss": 0.0798,
      "step": 2743
    },
    {
      "epoch": 5.219210651450309,
      "grad_norm": 0.2638397514820099,
      "learning_rate": 8.260400127024453e-05,
      "loss": 0.0837,
      "step": 2744
    },
    {
      "epoch": 5.221112696148359,
      "grad_norm": 0.2619345188140869,
      "learning_rate": 8.259765004763418e-05,
      "loss": 0.0781,
      "step": 2745
    },
    {
      "epoch": 5.22301474084641,
      "grad_norm": 0.32482966780662537,
      "learning_rate": 8.259129882502382e-05,
      "loss": 0.0867,
      "step": 2746
    },
    {
      "epoch": 5.22491678554446,
      "grad_norm": 0.3487076759338379,
      "learning_rate": 8.258494760241347e-05,
      "loss": 0.1038,
      "step": 2747
    },
    {
      "epoch": 5.226818830242511,
      "grad_norm": 0.37396374344825745,
      "learning_rate": 8.257859637980312e-05,
      "loss": 0.1041,
      "step": 2748
    },
    {
      "epoch": 5.228720874940561,
      "grad_norm": 0.310883492231369,
      "learning_rate": 8.257224515719276e-05,
      "loss": 0.0958,
      "step": 2749
    },
    {
      "epoch": 5.230622919638612,
      "grad_norm": 0.33855023980140686,
      "learning_rate": 8.256589393458241e-05,
      "loss": 0.1001,
      "step": 2750
    },
    {
      "epoch": 5.232524964336662,
      "grad_norm": 0.295162558555603,
      "learning_rate": 8.255954271197206e-05,
      "loss": 0.084,
      "step": 2751
    },
    {
      "epoch": 5.234427009034713,
      "grad_norm": 0.22703345119953156,
      "learning_rate": 8.25531914893617e-05,
      "loss": 0.0777,
      "step": 2752
    },
    {
      "epoch": 5.236329053732763,
      "grad_norm": 0.23589177429676056,
      "learning_rate": 8.254684026675135e-05,
      "loss": 0.0781,
      "step": 2753
    },
    {
      "epoch": 5.238231098430813,
      "grad_norm": 0.2537785768508911,
      "learning_rate": 8.2540489044141e-05,
      "loss": 0.1024,
      "step": 2754
    },
    {
      "epoch": 5.240133143128864,
      "grad_norm": 0.3413544297218323,
      "learning_rate": 8.253413782153064e-05,
      "loss": 0.0792,
      "step": 2755
    },
    {
      "epoch": 5.242035187826914,
      "grad_norm": 0.28834807872772217,
      "learning_rate": 8.25277865989203e-05,
      "loss": 0.0969,
      "step": 2756
    },
    {
      "epoch": 5.2439372325249645,
      "grad_norm": 0.2617645263671875,
      "learning_rate": 8.252143537630995e-05,
      "loss": 0.0906,
      "step": 2757
    },
    {
      "epoch": 5.2458392772230145,
      "grad_norm": 0.283772736787796,
      "learning_rate": 8.25150841536996e-05,
      "loss": 0.085,
      "step": 2758
    },
    {
      "epoch": 5.247741321921065,
      "grad_norm": 0.2892938256263733,
      "learning_rate": 8.250873293108924e-05,
      "loss": 0.0971,
      "step": 2759
    },
    {
      "epoch": 5.249643366619115,
      "grad_norm": 0.30989378690719604,
      "learning_rate": 8.250238170847889e-05,
      "loss": 0.0832,
      "step": 2760
    },
    {
      "epoch": 5.251545411317166,
      "grad_norm": 0.3129521906375885,
      "learning_rate": 8.249603048586854e-05,
      "loss": 0.0767,
      "step": 2761
    },
    {
      "epoch": 5.253447456015216,
      "grad_norm": 0.2838345170021057,
      "learning_rate": 8.248967926325818e-05,
      "loss": 0.0864,
      "step": 2762
    },
    {
      "epoch": 5.255349500713267,
      "grad_norm": 0.3472948968410492,
      "learning_rate": 8.248332804064783e-05,
      "loss": 0.0927,
      "step": 2763
    },
    {
      "epoch": 5.257251545411317,
      "grad_norm": 0.32811862230300903,
      "learning_rate": 8.247697681803748e-05,
      "loss": 0.1022,
      "step": 2764
    },
    {
      "epoch": 5.259153590109367,
      "grad_norm": 0.2726902961730957,
      "learning_rate": 8.247062559542712e-05,
      "loss": 0.0989,
      "step": 2765
    },
    {
      "epoch": 5.261055634807418,
      "grad_norm": 0.26436948776245117,
      "learning_rate": 8.246427437281677e-05,
      "loss": 0.087,
      "step": 2766
    },
    {
      "epoch": 5.262957679505468,
      "grad_norm": 0.25681155920028687,
      "learning_rate": 8.245792315020641e-05,
      "loss": 0.0844,
      "step": 2767
    },
    {
      "epoch": 5.264859724203519,
      "grad_norm": 0.3114239275455475,
      "learning_rate": 8.245157192759606e-05,
      "loss": 0.1081,
      "step": 2768
    },
    {
      "epoch": 5.266761768901569,
      "grad_norm": 0.31467998027801514,
      "learning_rate": 8.244522070498571e-05,
      "loss": 0.103,
      "step": 2769
    },
    {
      "epoch": 5.26866381359962,
      "grad_norm": 0.32647088170051575,
      "learning_rate": 8.243886948237535e-05,
      "loss": 0.0987,
      "step": 2770
    },
    {
      "epoch": 5.27056585829767,
      "grad_norm": 0.3176961839199066,
      "learning_rate": 8.243251825976502e-05,
      "loss": 0.0995,
      "step": 2771
    },
    {
      "epoch": 5.272467902995721,
      "grad_norm": 0.29948559403419495,
      "learning_rate": 8.242616703715466e-05,
      "loss": 0.0886,
      "step": 2772
    },
    {
      "epoch": 5.274369947693771,
      "grad_norm": 0.2150421142578125,
      "learning_rate": 8.24198158145443e-05,
      "loss": 0.0859,
      "step": 2773
    },
    {
      "epoch": 5.2762719923918215,
      "grad_norm": 0.276875764131546,
      "learning_rate": 8.241346459193395e-05,
      "loss": 0.0985,
      "step": 2774
    },
    {
      "epoch": 5.2781740370898715,
      "grad_norm": 0.2445184886455536,
      "learning_rate": 8.24071133693236e-05,
      "loss": 0.1443,
      "step": 2775
    },
    {
      "epoch": 5.280076081787922,
      "grad_norm": 0.2786940932273865,
      "learning_rate": 8.240076214671325e-05,
      "loss": 0.0922,
      "step": 2776
    },
    {
      "epoch": 5.281978126485972,
      "grad_norm": 0.330771267414093,
      "learning_rate": 8.239441092410289e-05,
      "loss": 0.082,
      "step": 2777
    },
    {
      "epoch": 5.283880171184022,
      "grad_norm": 0.3194374442100525,
      "learning_rate": 8.238805970149254e-05,
      "loss": 0.0972,
      "step": 2778
    },
    {
      "epoch": 5.285782215882073,
      "grad_norm": 0.27137935161590576,
      "learning_rate": 8.238170847888219e-05,
      "loss": 0.0731,
      "step": 2779
    },
    {
      "epoch": 5.287684260580123,
      "grad_norm": 0.3671146035194397,
      "learning_rate": 8.237535725627183e-05,
      "loss": 0.0755,
      "step": 2780
    },
    {
      "epoch": 5.289586305278174,
      "grad_norm": 0.4442805051803589,
      "learning_rate": 8.236900603366148e-05,
      "loss": 0.1041,
      "step": 2781
    },
    {
      "epoch": 5.291488349976224,
      "grad_norm": 0.3038801848888397,
      "learning_rate": 8.236265481105113e-05,
      "loss": 0.0921,
      "step": 2782
    },
    {
      "epoch": 5.293390394674275,
      "grad_norm": 0.31325793266296387,
      "learning_rate": 8.235630358844077e-05,
      "loss": 0.0972,
      "step": 2783
    },
    {
      "epoch": 5.295292439372325,
      "grad_norm": 0.28008419275283813,
      "learning_rate": 8.234995236583042e-05,
      "loss": 0.0797,
      "step": 2784
    },
    {
      "epoch": 5.297194484070376,
      "grad_norm": 0.3274953067302704,
      "learning_rate": 8.234360114322008e-05,
      "loss": 0.0863,
      "step": 2785
    },
    {
      "epoch": 5.299096528768426,
      "grad_norm": 0.40671971440315247,
      "learning_rate": 8.233724992060971e-05,
      "loss": 0.1018,
      "step": 2786
    },
    {
      "epoch": 5.300998573466477,
      "grad_norm": 0.43156853318214417,
      "learning_rate": 8.233089869799937e-05,
      "loss": 0.1142,
      "step": 2787
    },
    {
      "epoch": 5.302900618164527,
      "grad_norm": 0.28940603137016296,
      "learning_rate": 8.232454747538902e-05,
      "loss": 0.0647,
      "step": 2788
    },
    {
      "epoch": 5.304802662862578,
      "grad_norm": 0.4310309886932373,
      "learning_rate": 8.231819625277867e-05,
      "loss": 0.1124,
      "step": 2789
    },
    {
      "epoch": 5.306704707560628,
      "grad_norm": 0.3912397027015686,
      "learning_rate": 8.231184503016831e-05,
      "loss": 0.0995,
      "step": 2790
    },
    {
      "epoch": 5.308606752258678,
      "grad_norm": 0.33847618103027344,
      "learning_rate": 8.230549380755796e-05,
      "loss": 0.0926,
      "step": 2791
    },
    {
      "epoch": 5.310508796956729,
      "grad_norm": 0.30807194113731384,
      "learning_rate": 8.229914258494761e-05,
      "loss": 0.0961,
      "step": 2792
    },
    {
      "epoch": 5.312410841654779,
      "grad_norm": 0.29250118136405945,
      "learning_rate": 8.229279136233725e-05,
      "loss": 0.0774,
      "step": 2793
    },
    {
      "epoch": 5.3143128863528295,
      "grad_norm": 0.34272050857543945,
      "learning_rate": 8.22864401397269e-05,
      "loss": 0.0756,
      "step": 2794
    },
    {
      "epoch": 5.3162149310508795,
      "grad_norm": 0.3028480112552643,
      "learning_rate": 8.228008891711655e-05,
      "loss": 0.0936,
      "step": 2795
    },
    {
      "epoch": 5.31811697574893,
      "grad_norm": 0.35978344082832336,
      "learning_rate": 8.227373769450619e-05,
      "loss": 0.0856,
      "step": 2796
    },
    {
      "epoch": 5.32001902044698,
      "grad_norm": 0.24763593077659607,
      "learning_rate": 8.226738647189584e-05,
      "loss": 0.0683,
      "step": 2797
    },
    {
      "epoch": 5.321921065145031,
      "grad_norm": 0.34561780095100403,
      "learning_rate": 8.22610352492855e-05,
      "loss": 0.0887,
      "step": 2798
    },
    {
      "epoch": 5.323823109843081,
      "grad_norm": 0.34023576974868774,
      "learning_rate": 8.225468402667515e-05,
      "loss": 0.1141,
      "step": 2799
    },
    {
      "epoch": 5.325725154541132,
      "grad_norm": 0.32317525148391724,
      "learning_rate": 8.224833280406479e-05,
      "loss": 0.0902,
      "step": 2800
    },
    {
      "epoch": 5.327627199239182,
      "grad_norm": 0.3947696089744568,
      "learning_rate": 8.224198158145442e-05,
      "loss": 0.1152,
      "step": 2801
    },
    {
      "epoch": 5.329529243937232,
      "grad_norm": 0.2948441803455353,
      "learning_rate": 8.223563035884409e-05,
      "loss": 0.1001,
      "step": 2802
    },
    {
      "epoch": 5.331431288635283,
      "grad_norm": 0.42355984449386597,
      "learning_rate": 8.222927913623373e-05,
      "loss": 0.0999,
      "step": 2803
    },
    {
      "epoch": 5.333333333333333,
      "grad_norm": 0.22205117344856262,
      "learning_rate": 8.222292791362337e-05,
      "loss": 0.0788,
      "step": 2804
    },
    {
      "epoch": 5.335235378031384,
      "grad_norm": 0.28190529346466064,
      "learning_rate": 8.221657669101303e-05,
      "loss": 0.0857,
      "step": 2805
    },
    {
      "epoch": 5.337137422729434,
      "grad_norm": 0.272397518157959,
      "learning_rate": 8.221022546840267e-05,
      "loss": 0.0804,
      "step": 2806
    },
    {
      "epoch": 5.339039467427485,
      "grad_norm": 0.2642618417739868,
      "learning_rate": 8.220387424579232e-05,
      "loss": 0.0768,
      "step": 2807
    },
    {
      "epoch": 5.340941512125535,
      "grad_norm": 0.3142416477203369,
      "learning_rate": 8.219752302318196e-05,
      "loss": 0.0985,
      "step": 2808
    },
    {
      "epoch": 5.342843556823586,
      "grad_norm": 0.3162345588207245,
      "learning_rate": 8.219117180057161e-05,
      "loss": 0.0842,
      "step": 2809
    },
    {
      "epoch": 5.344745601521636,
      "grad_norm": 0.3445020318031311,
      "learning_rate": 8.218482057796126e-05,
      "loss": 0.0948,
      "step": 2810
    },
    {
      "epoch": 5.346647646219687,
      "grad_norm": 0.3281369209289551,
      "learning_rate": 8.21784693553509e-05,
      "loss": 0.1059,
      "step": 2811
    },
    {
      "epoch": 5.348549690917737,
      "grad_norm": 0.36346635222435,
      "learning_rate": 8.217211813274057e-05,
      "loss": 0.1098,
      "step": 2812
    },
    {
      "epoch": 5.350451735615787,
      "grad_norm": 0.29653385281562805,
      "learning_rate": 8.21657669101302e-05,
      "loss": 0.0994,
      "step": 2813
    },
    {
      "epoch": 5.3523537803138375,
      "grad_norm": 0.19076651334762573,
      "learning_rate": 8.215941568751984e-05,
      "loss": 0.0636,
      "step": 2814
    },
    {
      "epoch": 5.3542558250118875,
      "grad_norm": 0.23119299113750458,
      "learning_rate": 8.21530644649095e-05,
      "loss": 0.0755,
      "step": 2815
    },
    {
      "epoch": 5.356157869709938,
      "grad_norm": 0.21026362478733063,
      "learning_rate": 8.214671324229915e-05,
      "loss": 0.0714,
      "step": 2816
    },
    {
      "epoch": 5.358059914407988,
      "grad_norm": 0.19502770900726318,
      "learning_rate": 8.21403620196888e-05,
      "loss": 0.0817,
      "step": 2817
    },
    {
      "epoch": 5.359961959106039,
      "grad_norm": 0.2644445598125458,
      "learning_rate": 8.213401079707844e-05,
      "loss": 0.0723,
      "step": 2818
    },
    {
      "epoch": 5.361864003804089,
      "grad_norm": 0.34537041187286377,
      "learning_rate": 8.212765957446809e-05,
      "loss": 0.1039,
      "step": 2819
    },
    {
      "epoch": 5.36376604850214,
      "grad_norm": 0.28133493661880493,
      "learning_rate": 8.212130835185774e-05,
      "loss": 0.0906,
      "step": 2820
    },
    {
      "epoch": 5.36566809320019,
      "grad_norm": 0.2598632872104645,
      "learning_rate": 8.211495712924738e-05,
      "loss": 0.0848,
      "step": 2821
    },
    {
      "epoch": 5.367570137898241,
      "grad_norm": 0.2941516041755676,
      "learning_rate": 8.210860590663703e-05,
      "loss": 0.0951,
      "step": 2822
    },
    {
      "epoch": 5.369472182596291,
      "grad_norm": 0.617328941822052,
      "learning_rate": 8.210225468402668e-05,
      "loss": 0.1256,
      "step": 2823
    },
    {
      "epoch": 5.371374227294341,
      "grad_norm": 0.3117159903049469,
      "learning_rate": 8.209590346141632e-05,
      "loss": 0.0971,
      "step": 2824
    },
    {
      "epoch": 5.373276271992392,
      "grad_norm": 0.28506991267204285,
      "learning_rate": 8.208955223880597e-05,
      "loss": 0.0706,
      "step": 2825
    },
    {
      "epoch": 5.375178316690442,
      "grad_norm": 0.3072415888309479,
      "learning_rate": 8.208320101619563e-05,
      "loss": 0.0939,
      "step": 2826
    },
    {
      "epoch": 5.377080361388493,
      "grad_norm": 0.3019329309463501,
      "learning_rate": 8.207684979358526e-05,
      "loss": 0.117,
      "step": 2827
    },
    {
      "epoch": 5.378982406086543,
      "grad_norm": 0.3263597786426544,
      "learning_rate": 8.207049857097492e-05,
      "loss": 0.1013,
      "step": 2828
    },
    {
      "epoch": 5.380884450784594,
      "grad_norm": 0.2426205277442932,
      "learning_rate": 8.206414734836457e-05,
      "loss": 0.09,
      "step": 2829
    },
    {
      "epoch": 5.382786495482644,
      "grad_norm": 0.3131352663040161,
      "learning_rate": 8.205779612575422e-05,
      "loss": 0.0831,
      "step": 2830
    },
    {
      "epoch": 5.3846885401806945,
      "grad_norm": 0.2705948054790497,
      "learning_rate": 8.205144490314386e-05,
      "loss": 0.0825,
      "step": 2831
    },
    {
      "epoch": 5.3865905848787445,
      "grad_norm": 0.24609902501106262,
      "learning_rate": 8.20450936805335e-05,
      "loss": 0.0941,
      "step": 2832
    },
    {
      "epoch": 5.388492629576795,
      "grad_norm": 0.2822064757347107,
      "learning_rate": 8.203874245792316e-05,
      "loss": 0.0868,
      "step": 2833
    },
    {
      "epoch": 5.390394674274845,
      "grad_norm": 0.3406517207622528,
      "learning_rate": 8.20323912353128e-05,
      "loss": 0.083,
      "step": 2834
    },
    {
      "epoch": 5.392296718972895,
      "grad_norm": 0.38472169637680054,
      "learning_rate": 8.202604001270245e-05,
      "loss": 0.101,
      "step": 2835
    },
    {
      "epoch": 5.394198763670946,
      "grad_norm": 0.39235028624534607,
      "learning_rate": 8.20196887900921e-05,
      "loss": 0.1109,
      "step": 2836
    },
    {
      "epoch": 5.396100808368996,
      "grad_norm": 0.2799689769744873,
      "learning_rate": 8.201333756748174e-05,
      "loss": 0.087,
      "step": 2837
    },
    {
      "epoch": 5.398002853067047,
      "grad_norm": 0.32398489117622375,
      "learning_rate": 8.20069863448714e-05,
      "loss": 0.0994,
      "step": 2838
    },
    {
      "epoch": 5.399904897765097,
      "grad_norm": 0.29752808809280396,
      "learning_rate": 8.200063512226103e-05,
      "loss": 0.093,
      "step": 2839
    },
    {
      "epoch": 5.401806942463148,
      "grad_norm": 0.2725158631801605,
      "learning_rate": 8.199428389965068e-05,
      "loss": 0.1066,
      "step": 2840
    },
    {
      "epoch": 5.403708987161198,
      "grad_norm": 0.2493102252483368,
      "learning_rate": 8.198793267704034e-05,
      "loss": 0.0659,
      "step": 2841
    },
    {
      "epoch": 5.405611031859249,
      "grad_norm": 0.3313886225223541,
      "learning_rate": 8.198158145442997e-05,
      "loss": 0.093,
      "step": 2842
    },
    {
      "epoch": 5.407513076557299,
      "grad_norm": 0.3171900808811188,
      "learning_rate": 8.197523023181964e-05,
      "loss": 0.1016,
      "step": 2843
    },
    {
      "epoch": 5.40941512125535,
      "grad_norm": 0.2522033452987671,
      "learning_rate": 8.196887900920928e-05,
      "loss": 0.0713,
      "step": 2844
    },
    {
      "epoch": 5.4113171659534,
      "grad_norm": 0.26137956976890564,
      "learning_rate": 8.196252778659892e-05,
      "loss": 0.0756,
      "step": 2845
    },
    {
      "epoch": 5.41321921065145,
      "grad_norm": 0.30233171582221985,
      "learning_rate": 8.195617656398857e-05,
      "loss": 0.0849,
      "step": 2846
    },
    {
      "epoch": 5.415121255349501,
      "grad_norm": 0.3242833614349365,
      "learning_rate": 8.194982534137822e-05,
      "loss": 0.1057,
      "step": 2847
    },
    {
      "epoch": 5.417023300047551,
      "grad_norm": 0.28022679686546326,
      "learning_rate": 8.194347411876787e-05,
      "loss": 0.0612,
      "step": 2848
    },
    {
      "epoch": 5.418925344745602,
      "grad_norm": 0.31969156861305237,
      "learning_rate": 8.193712289615751e-05,
      "loss": 0.0869,
      "step": 2849
    },
    {
      "epoch": 5.420827389443652,
      "grad_norm": 0.3627922832965851,
      "learning_rate": 8.193077167354716e-05,
      "loss": 0.0886,
      "step": 2850
    },
    {
      "epoch": 5.4227294341417025,
      "grad_norm": 0.4317481219768524,
      "learning_rate": 8.192442045093681e-05,
      "loss": 0.1014,
      "step": 2851
    },
    {
      "epoch": 5.4246314788397525,
      "grad_norm": 0.3192225992679596,
      "learning_rate": 8.191806922832645e-05,
      "loss": 0.084,
      "step": 2852
    },
    {
      "epoch": 5.426533523537803,
      "grad_norm": 0.28119418025016785,
      "learning_rate": 8.19117180057161e-05,
      "loss": 0.0849,
      "step": 2853
    },
    {
      "epoch": 5.428435568235853,
      "grad_norm": 0.25419628620147705,
      "learning_rate": 8.190536678310576e-05,
      "loss": 0.0849,
      "step": 2854
    },
    {
      "epoch": 5.430337612933904,
      "grad_norm": 0.3957049250602722,
      "learning_rate": 8.18990155604954e-05,
      "loss": 0.1057,
      "step": 2855
    },
    {
      "epoch": 5.432239657631954,
      "grad_norm": 0.31903624534606934,
      "learning_rate": 8.189266433788505e-05,
      "loss": 0.0958,
      "step": 2856
    },
    {
      "epoch": 5.434141702330005,
      "grad_norm": 0.2692996561527252,
      "learning_rate": 8.18863131152747e-05,
      "loss": 0.0982,
      "step": 2857
    },
    {
      "epoch": 5.436043747028055,
      "grad_norm": 0.3083006739616394,
      "learning_rate": 8.187996189266434e-05,
      "loss": 0.1022,
      "step": 2858
    },
    {
      "epoch": 5.437945791726105,
      "grad_norm": 0.4186379909515381,
      "learning_rate": 8.187361067005399e-05,
      "loss": 0.1278,
      "step": 2859
    },
    {
      "epoch": 5.439847836424156,
      "grad_norm": 0.3124418258666992,
      "learning_rate": 8.186725944744364e-05,
      "loss": 0.0892,
      "step": 2860
    },
    {
      "epoch": 5.441749881122206,
      "grad_norm": 0.2959184944629669,
      "learning_rate": 8.186090822483329e-05,
      "loss": 0.0897,
      "step": 2861
    },
    {
      "epoch": 5.443651925820257,
      "grad_norm": 0.3044806718826294,
      "learning_rate": 8.185455700222293e-05,
      "loss": 0.0902,
      "step": 2862
    },
    {
      "epoch": 5.445553970518307,
      "grad_norm": 0.3201037347316742,
      "learning_rate": 8.184820577961258e-05,
      "loss": 0.1028,
      "step": 2863
    },
    {
      "epoch": 5.447456015216358,
      "grad_norm": 0.226481631398201,
      "learning_rate": 8.184185455700223e-05,
      "loss": 0.0701,
      "step": 2864
    },
    {
      "epoch": 5.449358059914408,
      "grad_norm": 0.29510584473609924,
      "learning_rate": 8.183550333439187e-05,
      "loss": 0.0954,
      "step": 2865
    },
    {
      "epoch": 5.451260104612459,
      "grad_norm": 0.3167766332626343,
      "learning_rate": 8.182915211178152e-05,
      "loss": 0.0907,
      "step": 2866
    },
    {
      "epoch": 5.453162149310509,
      "grad_norm": 0.3047981858253479,
      "learning_rate": 8.182280088917117e-05,
      "loss": 0.0978,
      "step": 2867
    },
    {
      "epoch": 5.45506419400856,
      "grad_norm": 0.3672785460948944,
      "learning_rate": 8.181644966656081e-05,
      "loss": 0.0992,
      "step": 2868
    },
    {
      "epoch": 5.45696623870661,
      "grad_norm": 0.3375677466392517,
      "learning_rate": 8.181009844395046e-05,
      "loss": 0.0933,
      "step": 2869
    },
    {
      "epoch": 5.4588682834046605,
      "grad_norm": 0.3718927204608917,
      "learning_rate": 8.180374722134012e-05,
      "loss": 0.0983,
      "step": 2870
    },
    {
      "epoch": 5.4607703281027105,
      "grad_norm": 0.2880299687385559,
      "learning_rate": 8.179739599872977e-05,
      "loss": 0.0937,
      "step": 2871
    },
    {
      "epoch": 5.4626723728007605,
      "grad_norm": 0.5487256646156311,
      "learning_rate": 8.179104477611941e-05,
      "loss": 0.0926,
      "step": 2872
    },
    {
      "epoch": 5.464574417498811,
      "grad_norm": 0.29540562629699707,
      "learning_rate": 8.178469355350905e-05,
      "loss": 0.0827,
      "step": 2873
    },
    {
      "epoch": 5.466476462196861,
      "grad_norm": 0.3226901590824127,
      "learning_rate": 8.177834233089871e-05,
      "loss": 0.0718,
      "step": 2874
    },
    {
      "epoch": 5.468378506894912,
      "grad_norm": 0.311902791261673,
      "learning_rate": 8.177199110828835e-05,
      "loss": 0.0761,
      "step": 2875
    },
    {
      "epoch": 5.470280551592962,
      "grad_norm": 0.39249876141548157,
      "learning_rate": 8.176563988567799e-05,
      "loss": 0.1199,
      "step": 2876
    },
    {
      "epoch": 5.472182596291013,
      "grad_norm": 0.32737264037132263,
      "learning_rate": 8.175928866306764e-05,
      "loss": 0.0756,
      "step": 2877
    },
    {
      "epoch": 5.474084640989063,
      "grad_norm": 0.4456403851509094,
      "learning_rate": 8.175293744045729e-05,
      "loss": 0.0903,
      "step": 2878
    },
    {
      "epoch": 5.475986685687114,
      "grad_norm": 0.2985632121562958,
      "learning_rate": 8.174658621784694e-05,
      "loss": 0.0854,
      "step": 2879
    },
    {
      "epoch": 5.477888730385164,
      "grad_norm": 0.34640341997146606,
      "learning_rate": 8.174023499523658e-05,
      "loss": 0.1007,
      "step": 2880
    },
    {
      "epoch": 5.479790775083215,
      "grad_norm": 0.3404572010040283,
      "learning_rate": 8.173388377262623e-05,
      "loss": 0.126,
      "step": 2881
    },
    {
      "epoch": 5.481692819781265,
      "grad_norm": 0.3574541509151459,
      "learning_rate": 8.172753255001588e-05,
      "loss": 0.102,
      "step": 2882
    },
    {
      "epoch": 5.483594864479315,
      "grad_norm": 0.23681578040122986,
      "learning_rate": 8.172118132740552e-05,
      "loss": 0.0768,
      "step": 2883
    },
    {
      "epoch": 5.485496909177366,
      "grad_norm": 0.30854761600494385,
      "learning_rate": 8.171483010479517e-05,
      "loss": 0.0777,
      "step": 2884
    },
    {
      "epoch": 5.487398953875416,
      "grad_norm": 0.35504138469696045,
      "learning_rate": 8.170847888218483e-05,
      "loss": 0.1114,
      "step": 2885
    },
    {
      "epoch": 5.489300998573467,
      "grad_norm": 0.26217421889305115,
      "learning_rate": 8.170212765957446e-05,
      "loss": 0.0982,
      "step": 2886
    },
    {
      "epoch": 5.491203043271517,
      "grad_norm": 0.3383858799934387,
      "learning_rate": 8.169577643696412e-05,
      "loss": 0.1318,
      "step": 2887
    },
    {
      "epoch": 5.4931050879695675,
      "grad_norm": 0.33602502942085266,
      "learning_rate": 8.168942521435377e-05,
      "loss": 0.1095,
      "step": 2888
    },
    {
      "epoch": 5.4950071326676175,
      "grad_norm": 0.3534327745437622,
      "learning_rate": 8.168307399174342e-05,
      "loss": 0.0891,
      "step": 2889
    },
    {
      "epoch": 5.496909177365668,
      "grad_norm": 0.28556129336357117,
      "learning_rate": 8.167672276913306e-05,
      "loss": 0.0984,
      "step": 2890
    },
    {
      "epoch": 5.498811222063718,
      "grad_norm": 0.32809069752693176,
      "learning_rate": 8.167037154652271e-05,
      "loss": 0.1075,
      "step": 2891
    },
    {
      "epoch": 5.500713266761769,
      "grad_norm": 0.608711838722229,
      "learning_rate": 8.166402032391236e-05,
      "loss": 0.0796,
      "step": 2892
    },
    {
      "epoch": 5.502615311459819,
      "grad_norm": 0.20012949407100677,
      "learning_rate": 8.1657669101302e-05,
      "loss": 0.0573,
      "step": 2893
    },
    {
      "epoch": 5.504517356157869,
      "grad_norm": 0.2836569547653198,
      "learning_rate": 8.165131787869165e-05,
      "loss": 0.1029,
      "step": 2894
    },
    {
      "epoch": 5.50641940085592,
      "grad_norm": 0.28194811940193176,
      "learning_rate": 8.16449666560813e-05,
      "loss": 0.0995,
      "step": 2895
    },
    {
      "epoch": 5.50832144555397,
      "grad_norm": 0.28990745544433594,
      "learning_rate": 8.163861543347094e-05,
      "loss": 0.1038,
      "step": 2896
    },
    {
      "epoch": 5.510223490252021,
      "grad_norm": 0.3534923791885376,
      "learning_rate": 8.16322642108606e-05,
      "loss": 0.0933,
      "step": 2897
    },
    {
      "epoch": 5.512125534950071,
      "grad_norm": 0.4764708876609802,
      "learning_rate": 8.162591298825025e-05,
      "loss": 0.1351,
      "step": 2898
    },
    {
      "epoch": 5.514027579648122,
      "grad_norm": 0.2705288827419281,
      "learning_rate": 8.161956176563988e-05,
      "loss": 0.1027,
      "step": 2899
    },
    {
      "epoch": 5.515929624346172,
      "grad_norm": 0.36976736783981323,
      "learning_rate": 8.161321054302954e-05,
      "loss": 0.0959,
      "step": 2900
    },
    {
      "epoch": 5.517831669044223,
      "grad_norm": 0.3459687829017639,
      "learning_rate": 8.160685932041919e-05,
      "loss": 0.1124,
      "step": 2901
    },
    {
      "epoch": 5.519733713742273,
      "grad_norm": 0.4014488458633423,
      "learning_rate": 8.160050809780884e-05,
      "loss": 0.1079,
      "step": 2902
    },
    {
      "epoch": 5.521635758440324,
      "grad_norm": 0.39544427394866943,
      "learning_rate": 8.159415687519848e-05,
      "loss": 0.1004,
      "step": 2903
    },
    {
      "epoch": 5.523537803138374,
      "grad_norm": 0.30223849415779114,
      "learning_rate": 8.158780565258812e-05,
      "loss": 0.0859,
      "step": 2904
    },
    {
      "epoch": 5.525439847836424,
      "grad_norm": 0.2351481020450592,
      "learning_rate": 8.158145442997778e-05,
      "loss": 0.0716,
      "step": 2905
    },
    {
      "epoch": 5.527341892534475,
      "grad_norm": 0.40507906675338745,
      "learning_rate": 8.157510320736742e-05,
      "loss": 0.1118,
      "step": 2906
    },
    {
      "epoch": 5.529243937232525,
      "grad_norm": 0.2664635181427002,
      "learning_rate": 8.156875198475707e-05,
      "loss": 0.0956,
      "step": 2907
    },
    {
      "epoch": 5.5311459819305755,
      "grad_norm": 0.296220988035202,
      "learning_rate": 8.156240076214672e-05,
      "loss": 0.1011,
      "step": 2908
    },
    {
      "epoch": 5.5330480266286255,
      "grad_norm": 0.34762975573539734,
      "learning_rate": 8.155604953953636e-05,
      "loss": 0.0936,
      "step": 2909
    },
    {
      "epoch": 5.534950071326676,
      "grad_norm": 0.42042768001556396,
      "learning_rate": 8.154969831692601e-05,
      "loss": 0.1104,
      "step": 2910
    },
    {
      "epoch": 5.536852116024726,
      "grad_norm": 0.33530357480049133,
      "learning_rate": 8.154334709431565e-05,
      "loss": 0.1481,
      "step": 2911
    },
    {
      "epoch": 5.538754160722777,
      "grad_norm": 0.37931686639785767,
      "learning_rate": 8.15369958717053e-05,
      "loss": 0.114,
      "step": 2912
    },
    {
      "epoch": 5.540656205420827,
      "grad_norm": 0.3276258707046509,
      "learning_rate": 8.153064464909496e-05,
      "loss": 0.1009,
      "step": 2913
    },
    {
      "epoch": 5.542558250118878,
      "grad_norm": 0.29436194896698,
      "learning_rate": 8.15242934264846e-05,
      "loss": 0.0914,
      "step": 2914
    },
    {
      "epoch": 5.544460294816928,
      "grad_norm": 0.32761478424072266,
      "learning_rate": 8.151794220387426e-05,
      "loss": 0.1162,
      "step": 2915
    },
    {
      "epoch": 5.546362339514978,
      "grad_norm": 0.6013909578323364,
      "learning_rate": 8.15115909812639e-05,
      "loss": 0.1271,
      "step": 2916
    },
    {
      "epoch": 5.548264384213029,
      "grad_norm": 0.2207658290863037,
      "learning_rate": 8.150523975865354e-05,
      "loss": 0.0861,
      "step": 2917
    },
    {
      "epoch": 5.550166428911079,
      "grad_norm": 0.25360438227653503,
      "learning_rate": 8.149888853604319e-05,
      "loss": 0.055,
      "step": 2918
    },
    {
      "epoch": 5.55206847360913,
      "grad_norm": 0.2537856101989746,
      "learning_rate": 8.149253731343284e-05,
      "loss": 0.0749,
      "step": 2919
    },
    {
      "epoch": 5.55397051830718,
      "grad_norm": 0.29756224155426025,
      "learning_rate": 8.148618609082249e-05,
      "loss": 0.082,
      "step": 2920
    },
    {
      "epoch": 5.555872563005231,
      "grad_norm": 0.41644203662872314,
      "learning_rate": 8.147983486821213e-05,
      "loss": 0.1885,
      "step": 2921
    },
    {
      "epoch": 5.557774607703281,
      "grad_norm": 0.32786139845848083,
      "learning_rate": 8.147348364560178e-05,
      "loss": 0.0822,
      "step": 2922
    },
    {
      "epoch": 5.559676652401332,
      "grad_norm": 0.30312976241111755,
      "learning_rate": 8.146713242299143e-05,
      "loss": 0.0863,
      "step": 2923
    },
    {
      "epoch": 5.561578697099382,
      "grad_norm": 0.34702423214912415,
      "learning_rate": 8.146078120038107e-05,
      "loss": 0.1012,
      "step": 2924
    },
    {
      "epoch": 5.563480741797433,
      "grad_norm": 0.24299529194831848,
      "learning_rate": 8.145442997777072e-05,
      "loss": 0.0739,
      "step": 2925
    },
    {
      "epoch": 5.565382786495483,
      "grad_norm": 0.30519744753837585,
      "learning_rate": 8.144807875516038e-05,
      "loss": 0.0953,
      "step": 2926
    },
    {
      "epoch": 5.567284831193533,
      "grad_norm": 0.32798048853874207,
      "learning_rate": 8.144172753255001e-05,
      "loss": 0.1077,
      "step": 2927
    },
    {
      "epoch": 5.5691868758915835,
      "grad_norm": 0.38108792901039124,
      "learning_rate": 8.143537630993967e-05,
      "loss": 0.1246,
      "step": 2928
    },
    {
      "epoch": 5.571088920589634,
      "grad_norm": 0.480277955532074,
      "learning_rate": 8.142902508732932e-05,
      "loss": 0.0909,
      "step": 2929
    },
    {
      "epoch": 5.572990965287684,
      "grad_norm": 0.3235543668270111,
      "learning_rate": 8.142267386471896e-05,
      "loss": 0.0925,
      "step": 2930
    },
    {
      "epoch": 5.574893009985734,
      "grad_norm": 0.34970083832740784,
      "learning_rate": 8.141632264210861e-05,
      "loss": 0.0908,
      "step": 2931
    },
    {
      "epoch": 5.576795054683785,
      "grad_norm": 0.2239646017551422,
      "learning_rate": 8.140997141949826e-05,
      "loss": 0.0723,
      "step": 2932
    },
    {
      "epoch": 5.578697099381835,
      "grad_norm": 0.33661478757858276,
      "learning_rate": 8.140362019688791e-05,
      "loss": 0.0922,
      "step": 2933
    },
    {
      "epoch": 5.580599144079886,
      "grad_norm": 0.4126195013523102,
      "learning_rate": 8.139726897427755e-05,
      "loss": 0.0943,
      "step": 2934
    },
    {
      "epoch": 5.582501188777936,
      "grad_norm": 0.3538368046283722,
      "learning_rate": 8.139091775166719e-05,
      "loss": 0.0955,
      "step": 2935
    },
    {
      "epoch": 5.584403233475987,
      "grad_norm": 0.3369925320148468,
      "learning_rate": 8.138456652905685e-05,
      "loss": 0.1166,
      "step": 2936
    },
    {
      "epoch": 5.586305278174037,
      "grad_norm": 0.27559757232666016,
      "learning_rate": 8.137821530644649e-05,
      "loss": 0.0808,
      "step": 2937
    },
    {
      "epoch": 5.588207322872088,
      "grad_norm": 0.30875301361083984,
      "learning_rate": 8.137186408383614e-05,
      "loss": 0.1024,
      "step": 2938
    },
    {
      "epoch": 5.590109367570138,
      "grad_norm": 0.2719765305519104,
      "learning_rate": 8.13655128612258e-05,
      "loss": 0.0897,
      "step": 2939
    },
    {
      "epoch": 5.592011412268189,
      "grad_norm": 0.3717488646507263,
      "learning_rate": 8.135916163861543e-05,
      "loss": 0.0966,
      "step": 2940
    },
    {
      "epoch": 5.593913456966239,
      "grad_norm": 0.2868727743625641,
      "learning_rate": 8.135281041600509e-05,
      "loss": 0.0802,
      "step": 2941
    },
    {
      "epoch": 5.595815501664289,
      "grad_norm": 0.28469622135162354,
      "learning_rate": 8.134645919339472e-05,
      "loss": 0.0989,
      "step": 2942
    },
    {
      "epoch": 5.59771754636234,
      "grad_norm": 0.34950777888298035,
      "learning_rate": 8.134010797078439e-05,
      "loss": 0.0939,
      "step": 2943
    },
    {
      "epoch": 5.59961959106039,
      "grad_norm": 0.23884734511375427,
      "learning_rate": 8.133375674817403e-05,
      "loss": 0.0845,
      "step": 2944
    },
    {
      "epoch": 5.6015216357584405,
      "grad_norm": 0.34531524777412415,
      "learning_rate": 8.132740552556367e-05,
      "loss": 0.0978,
      "step": 2945
    },
    {
      "epoch": 5.6034236804564905,
      "grad_norm": 0.26655009388923645,
      "learning_rate": 8.132105430295333e-05,
      "loss": 0.0723,
      "step": 2946
    },
    {
      "epoch": 5.605325725154541,
      "grad_norm": 0.23067669570446014,
      "learning_rate": 8.131470308034297e-05,
      "loss": 0.078,
      "step": 2947
    },
    {
      "epoch": 5.607227769852591,
      "grad_norm": 0.4221946597099304,
      "learning_rate": 8.130835185773261e-05,
      "loss": 0.1085,
      "step": 2948
    },
    {
      "epoch": 5.609129814550642,
      "grad_norm": 0.3162672519683838,
      "learning_rate": 8.130200063512226e-05,
      "loss": 0.0893,
      "step": 2949
    },
    {
      "epoch": 5.611031859248692,
      "grad_norm": 0.32246506214141846,
      "learning_rate": 8.129564941251191e-05,
      "loss": 0.0836,
      "step": 2950
    },
    {
      "epoch": 5.612933903946743,
      "grad_norm": 0.31152433156967163,
      "learning_rate": 8.128929818990156e-05,
      "loss": 0.0806,
      "step": 2951
    },
    {
      "epoch": 5.614835948644793,
      "grad_norm": 0.3357599377632141,
      "learning_rate": 8.12829469672912e-05,
      "loss": 0.0958,
      "step": 2952
    },
    {
      "epoch": 5.616737993342843,
      "grad_norm": 0.3181150555610657,
      "learning_rate": 8.127659574468085e-05,
      "loss": 0.0855,
      "step": 2953
    },
    {
      "epoch": 5.618640038040894,
      "grad_norm": 0.3397297263145447,
      "learning_rate": 8.12702445220705e-05,
      "loss": 0.1022,
      "step": 2954
    },
    {
      "epoch": 5.620542082738944,
      "grad_norm": 0.31479981541633606,
      "learning_rate": 8.126389329946014e-05,
      "loss": 0.0774,
      "step": 2955
    },
    {
      "epoch": 5.622444127436995,
      "grad_norm": 0.26667311787605286,
      "learning_rate": 8.12575420768498e-05,
      "loss": 0.0756,
      "step": 2956
    },
    {
      "epoch": 5.624346172135045,
      "grad_norm": 0.2729688882827759,
      "learning_rate": 8.125119085423945e-05,
      "loss": 0.0721,
      "step": 2957
    },
    {
      "epoch": 5.626248216833096,
      "grad_norm": 0.24858340620994568,
      "learning_rate": 8.124483963162909e-05,
      "loss": 0.0711,
      "step": 2958
    },
    {
      "epoch": 5.628150261531146,
      "grad_norm": 0.3526616096496582,
      "learning_rate": 8.123848840901874e-05,
      "loss": 0.084,
      "step": 2959
    },
    {
      "epoch": 5.630052306229197,
      "grad_norm": 0.2841814458370209,
      "learning_rate": 8.123213718640839e-05,
      "loss": 0.0999,
      "step": 2960
    },
    {
      "epoch": 5.631954350927247,
      "grad_norm": 0.2419266402721405,
      "learning_rate": 8.122578596379804e-05,
      "loss": 0.0662,
      "step": 2961
    },
    {
      "epoch": 5.633856395625298,
      "grad_norm": 0.34861576557159424,
      "learning_rate": 8.121943474118768e-05,
      "loss": 0.0804,
      "step": 2962
    },
    {
      "epoch": 5.635758440323348,
      "grad_norm": 0.42378073930740356,
      "learning_rate": 8.121308351857733e-05,
      "loss": 0.1006,
      "step": 2963
    },
    {
      "epoch": 5.637660485021398,
      "grad_norm": 0.41002216935157776,
      "learning_rate": 8.120673229596698e-05,
      "loss": 0.101,
      "step": 2964
    },
    {
      "epoch": 5.6395625297194485,
      "grad_norm": 0.2810782194137573,
      "learning_rate": 8.120038107335662e-05,
      "loss": 0.0702,
      "step": 2965
    },
    {
      "epoch": 5.6414645744174985,
      "grad_norm": 0.5979880094528198,
      "learning_rate": 8.119402985074627e-05,
      "loss": 0.149,
      "step": 2966
    },
    {
      "epoch": 5.643366619115549,
      "grad_norm": 0.26545101404190063,
      "learning_rate": 8.118767862813593e-05,
      "loss": 0.0998,
      "step": 2967
    },
    {
      "epoch": 5.645268663813599,
      "grad_norm": 0.37219372391700745,
      "learning_rate": 8.118132740552556e-05,
      "loss": 0.0984,
      "step": 2968
    },
    {
      "epoch": 5.64717070851165,
      "grad_norm": 0.3255815804004669,
      "learning_rate": 8.117497618291522e-05,
      "loss": 0.0931,
      "step": 2969
    },
    {
      "epoch": 5.6490727532097,
      "grad_norm": 0.4053998589515686,
      "learning_rate": 8.116862496030487e-05,
      "loss": 0.1134,
      "step": 2970
    },
    {
      "epoch": 5.650974797907751,
      "grad_norm": 0.3078075647354126,
      "learning_rate": 8.11622737376945e-05,
      "loss": 0.0908,
      "step": 2971
    },
    {
      "epoch": 5.652876842605801,
      "grad_norm": 0.28989338874816895,
      "learning_rate": 8.115592251508416e-05,
      "loss": 0.1144,
      "step": 2972
    },
    {
      "epoch": 5.654778887303852,
      "grad_norm": 0.3299599289894104,
      "learning_rate": 8.114957129247381e-05,
      "loss": 0.1064,
      "step": 2973
    },
    {
      "epoch": 5.656680932001902,
      "grad_norm": 0.4091484844684601,
      "learning_rate": 8.114322006986346e-05,
      "loss": 0.1126,
      "step": 2974
    },
    {
      "epoch": 5.658582976699952,
      "grad_norm": 0.264202356338501,
      "learning_rate": 8.11368688472531e-05,
      "loss": 0.0796,
      "step": 2975
    },
    {
      "epoch": 5.660485021398003,
      "grad_norm": 0.32306942343711853,
      "learning_rate": 8.113051762464274e-05,
      "loss": 0.0955,
      "step": 2976
    },
    {
      "epoch": 5.662387066096053,
      "grad_norm": 0.3152141273021698,
      "learning_rate": 8.11241664020324e-05,
      "loss": 0.1098,
      "step": 2977
    },
    {
      "epoch": 5.664289110794104,
      "grad_norm": 0.35286226868629456,
      "learning_rate": 8.111781517942204e-05,
      "loss": 0.1081,
      "step": 2978
    },
    {
      "epoch": 5.666191155492154,
      "grad_norm": 0.35619062185287476,
      "learning_rate": 8.111146395681169e-05,
      "loss": 0.11,
      "step": 2979
    },
    {
      "epoch": 5.668093200190205,
      "grad_norm": 0.4397338330745697,
      "learning_rate": 8.110511273420135e-05,
      "loss": 0.0886,
      "step": 2980
    },
    {
      "epoch": 5.669995244888255,
      "grad_norm": 0.23792682588100433,
      "learning_rate": 8.109876151159098e-05,
      "loss": 0.0834,
      "step": 2981
    },
    {
      "epoch": 5.671897289586306,
      "grad_norm": 0.30805590748786926,
      "learning_rate": 8.109241028898064e-05,
      "loss": 0.0816,
      "step": 2982
    },
    {
      "epoch": 5.673799334284356,
      "grad_norm": 0.3652699589729309,
      "learning_rate": 8.108605906637027e-05,
      "loss": 0.0995,
      "step": 2983
    },
    {
      "epoch": 5.6757013789824065,
      "grad_norm": 0.2952606976032257,
      "learning_rate": 8.107970784375993e-05,
      "loss": 0.102,
      "step": 2984
    },
    {
      "epoch": 5.6776034236804565,
      "grad_norm": 0.3017944395542145,
      "learning_rate": 8.107335662114958e-05,
      "loss": 0.0939,
      "step": 2985
    },
    {
      "epoch": 5.6795054683785064,
      "grad_norm": 0.3887818157672882,
      "learning_rate": 8.106700539853922e-05,
      "loss": 0.1168,
      "step": 2986
    },
    {
      "epoch": 5.681407513076557,
      "grad_norm": 0.3510635793209076,
      "learning_rate": 8.106065417592887e-05,
      "loss": 0.0877,
      "step": 2987
    },
    {
      "epoch": 5.683309557774607,
      "grad_norm": 0.4994543790817261,
      "learning_rate": 8.105430295331852e-05,
      "loss": 0.1296,
      "step": 2988
    },
    {
      "epoch": 5.685211602472658,
      "grad_norm": 0.43380653858184814,
      "learning_rate": 8.104795173070816e-05,
      "loss": 0.1293,
      "step": 2989
    },
    {
      "epoch": 5.687113647170708,
      "grad_norm": 0.33307018876075745,
      "learning_rate": 8.104160050809781e-05,
      "loss": 0.1161,
      "step": 2990
    },
    {
      "epoch": 5.689015691868759,
      "grad_norm": 0.2537522315979004,
      "learning_rate": 8.103524928548746e-05,
      "loss": 0.0864,
      "step": 2991
    },
    {
      "epoch": 5.690917736566809,
      "grad_norm": 0.269766628742218,
      "learning_rate": 8.102889806287711e-05,
      "loss": 0.0895,
      "step": 2992
    },
    {
      "epoch": 5.69281978126486,
      "grad_norm": 0.4431968331336975,
      "learning_rate": 8.102254684026675e-05,
      "loss": 0.1144,
      "step": 2993
    },
    {
      "epoch": 5.69472182596291,
      "grad_norm": 0.3709297180175781,
      "learning_rate": 8.10161956176564e-05,
      "loss": 0.1371,
      "step": 2994
    },
    {
      "epoch": 5.696623870660961,
      "grad_norm": 0.23933501541614532,
      "learning_rate": 8.100984439504605e-05,
      "loss": 0.0765,
      "step": 2995
    },
    {
      "epoch": 5.698525915359011,
      "grad_norm": 0.2999648451805115,
      "learning_rate": 8.100349317243569e-05,
      "loss": 0.0925,
      "step": 2996
    },
    {
      "epoch": 5.700427960057061,
      "grad_norm": 0.3433458209037781,
      "learning_rate": 8.099714194982535e-05,
      "loss": 0.1018,
      "step": 2997
    },
    {
      "epoch": 5.702330004755112,
      "grad_norm": 0.30743271112442017,
      "learning_rate": 8.0990790727215e-05,
      "loss": 0.0818,
      "step": 2998
    },
    {
      "epoch": 5.704232049453163,
      "grad_norm": 0.28858688473701477,
      "learning_rate": 8.098443950460464e-05,
      "loss": 0.089,
      "step": 2999
    },
    {
      "epoch": 5.706134094151213,
      "grad_norm": 0.3105423152446747,
      "learning_rate": 8.097808828199429e-05,
      "loss": 0.0859,
      "step": 3000
    },
    {
      "epoch": 5.708036138849263,
      "grad_norm": 0.348749041557312,
      "learning_rate": 8.097173705938394e-05,
      "loss": 0.1047,
      "step": 3001
    },
    {
      "epoch": 5.7099381835473135,
      "grad_norm": 0.2492302805185318,
      "learning_rate": 8.096538583677358e-05,
      "loss": 0.0874,
      "step": 3002
    },
    {
      "epoch": 5.7118402282453635,
      "grad_norm": 0.351367324590683,
      "learning_rate": 8.095903461416323e-05,
      "loss": 0.104,
      "step": 3003
    },
    {
      "epoch": 5.713742272943414,
      "grad_norm": 0.31233423948287964,
      "learning_rate": 8.095268339155288e-05,
      "loss": 0.0812,
      "step": 3004
    },
    {
      "epoch": 5.715644317641464,
      "grad_norm": 0.3879316747188568,
      "learning_rate": 8.094633216894253e-05,
      "loss": 0.1078,
      "step": 3005
    },
    {
      "epoch": 5.717546362339515,
      "grad_norm": 0.3155204653739929,
      "learning_rate": 8.093998094633217e-05,
      "loss": 0.0831,
      "step": 3006
    },
    {
      "epoch": 5.719448407037565,
      "grad_norm": 0.37463921308517456,
      "learning_rate": 8.093362972372181e-05,
      "loss": 0.1045,
      "step": 3007
    },
    {
      "epoch": 5.721350451735615,
      "grad_norm": 0.37191277742385864,
      "learning_rate": 8.092727850111147e-05,
      "loss": 0.1002,
      "step": 3008
    },
    {
      "epoch": 5.723252496433666,
      "grad_norm": 0.3215421736240387,
      "learning_rate": 8.092092727850111e-05,
      "loss": 0.1033,
      "step": 3009
    },
    {
      "epoch": 5.725154541131717,
      "grad_norm": 0.290348619222641,
      "learning_rate": 8.091457605589076e-05,
      "loss": 0.0767,
      "step": 3010
    },
    {
      "epoch": 5.727056585829767,
      "grad_norm": 0.2508927285671234,
      "learning_rate": 8.090822483328042e-05,
      "loss": 0.0676,
      "step": 3011
    },
    {
      "epoch": 5.728958630527817,
      "grad_norm": 0.7272295355796814,
      "learning_rate": 8.090187361067005e-05,
      "loss": 0.1755,
      "step": 3012
    },
    {
      "epoch": 5.730860675225868,
      "grad_norm": 0.3547666072845459,
      "learning_rate": 8.08955223880597e-05,
      "loss": 0.1002,
      "step": 3013
    },
    {
      "epoch": 5.732762719923918,
      "grad_norm": 0.26749446988105774,
      "learning_rate": 8.088917116544935e-05,
      "loss": 0.0795,
      "step": 3014
    },
    {
      "epoch": 5.734664764621969,
      "grad_norm": 0.2167988121509552,
      "learning_rate": 8.088281994283901e-05,
      "loss": 0.0704,
      "step": 3015
    },
    {
      "epoch": 5.736566809320019,
      "grad_norm": 0.23982734978199005,
      "learning_rate": 8.087646872022865e-05,
      "loss": 0.0724,
      "step": 3016
    },
    {
      "epoch": 5.73846885401807,
      "grad_norm": 0.3214990794658661,
      "learning_rate": 8.087011749761829e-05,
      "loss": 0.0884,
      "step": 3017
    },
    {
      "epoch": 5.74037089871612,
      "grad_norm": 0.37602269649505615,
      "learning_rate": 8.086376627500795e-05,
      "loss": 0.0956,
      "step": 3018
    },
    {
      "epoch": 5.742272943414171,
      "grad_norm": 0.28181731700897217,
      "learning_rate": 8.085741505239759e-05,
      "loss": 0.0787,
      "step": 3019
    },
    {
      "epoch": 5.744174988112221,
      "grad_norm": 0.24445931613445282,
      "learning_rate": 8.085106382978723e-05,
      "loss": 0.0804,
      "step": 3020
    },
    {
      "epoch": 5.7460770328102715,
      "grad_norm": 0.3743366003036499,
      "learning_rate": 8.084471260717688e-05,
      "loss": 0.1172,
      "step": 3021
    },
    {
      "epoch": 5.7479790775083215,
      "grad_norm": 0.2813834547996521,
      "learning_rate": 8.083836138456653e-05,
      "loss": 0.0849,
      "step": 3022
    },
    {
      "epoch": 5.7498811222063715,
      "grad_norm": 0.35081392526626587,
      "learning_rate": 8.083201016195618e-05,
      "loss": 0.1163,
      "step": 3023
    },
    {
      "epoch": 5.751783166904422,
      "grad_norm": 0.4236094653606415,
      "learning_rate": 8.082565893934582e-05,
      "loss": 0.0978,
      "step": 3024
    },
    {
      "epoch": 5.753685211602472,
      "grad_norm": 0.38664132356643677,
      "learning_rate": 8.081930771673547e-05,
      "loss": 0.1077,
      "step": 3025
    },
    {
      "epoch": 5.755587256300523,
      "grad_norm": 0.3874824047088623,
      "learning_rate": 8.081295649412513e-05,
      "loss": 0.0991,
      "step": 3026
    },
    {
      "epoch": 5.757489300998573,
      "grad_norm": 0.32273393869400024,
      "learning_rate": 8.080660527151476e-05,
      "loss": 0.0859,
      "step": 3027
    },
    {
      "epoch": 5.759391345696624,
      "grad_norm": 0.39935457706451416,
      "learning_rate": 8.080025404890442e-05,
      "loss": 0.108,
      "step": 3028
    },
    {
      "epoch": 5.761293390394674,
      "grad_norm": 0.24157053232192993,
      "learning_rate": 8.079390282629407e-05,
      "loss": 0.0774,
      "step": 3029
    },
    {
      "epoch": 5.763195435092725,
      "grad_norm": 0.38274866342544556,
      "learning_rate": 8.07875516036837e-05,
      "loss": 0.1171,
      "step": 3030
    },
    {
      "epoch": 5.765097479790775,
      "grad_norm": 0.34954944252967834,
      "learning_rate": 8.078120038107336e-05,
      "loss": 0.0714,
      "step": 3031
    },
    {
      "epoch": 5.766999524488826,
      "grad_norm": 0.35465356707572937,
      "learning_rate": 8.077484915846301e-05,
      "loss": 0.0857,
      "step": 3032
    },
    {
      "epoch": 5.768901569186876,
      "grad_norm": 0.32265377044677734,
      "learning_rate": 8.076849793585266e-05,
      "loss": 0.0933,
      "step": 3033
    },
    {
      "epoch": 5.770803613884926,
      "grad_norm": 0.3461415469646454,
      "learning_rate": 8.07621467132423e-05,
      "loss": 0.0872,
      "step": 3034
    },
    {
      "epoch": 5.772705658582977,
      "grad_norm": 0.358465313911438,
      "learning_rate": 8.075579549063195e-05,
      "loss": 0.1205,
      "step": 3035
    },
    {
      "epoch": 5.774607703281027,
      "grad_norm": 0.3661046326160431,
      "learning_rate": 8.07494442680216e-05,
      "loss": 0.1167,
      "step": 3036
    },
    {
      "epoch": 5.776509747979078,
      "grad_norm": 0.31511178612709045,
      "learning_rate": 8.074309304541124e-05,
      "loss": 0.1017,
      "step": 3037
    },
    {
      "epoch": 5.778411792677128,
      "grad_norm": 0.27341797947883606,
      "learning_rate": 8.07367418228009e-05,
      "loss": 0.0708,
      "step": 3038
    },
    {
      "epoch": 5.780313837375179,
      "grad_norm": 0.3324090242385864,
      "learning_rate": 8.073039060019055e-05,
      "loss": 0.0835,
      "step": 3039
    },
    {
      "epoch": 5.782215882073229,
      "grad_norm": 0.2953234314918518,
      "learning_rate": 8.072403937758018e-05,
      "loss": 0.0991,
      "step": 3040
    },
    {
      "epoch": 5.7841179267712794,
      "grad_norm": 0.324093759059906,
      "learning_rate": 8.071768815496984e-05,
      "loss": 0.0999,
      "step": 3041
    },
    {
      "epoch": 5.7860199714693294,
      "grad_norm": 0.4137260317802429,
      "learning_rate": 8.071133693235949e-05,
      "loss": 0.1338,
      "step": 3042
    },
    {
      "epoch": 5.78792201616738,
      "grad_norm": 0.247357577085495,
      "learning_rate": 8.070498570974913e-05,
      "loss": 0.0792,
      "step": 3043
    },
    {
      "epoch": 5.78982406086543,
      "grad_norm": 0.40995171666145325,
      "learning_rate": 8.069863448713878e-05,
      "loss": 0.136,
      "step": 3044
    },
    {
      "epoch": 5.79172610556348,
      "grad_norm": 0.34627994894981384,
      "learning_rate": 8.069228326452842e-05,
      "loss": 0.0996,
      "step": 3045
    },
    {
      "epoch": 5.793628150261531,
      "grad_norm": 0.25772425532341003,
      "learning_rate": 8.068593204191808e-05,
      "loss": 0.0858,
      "step": 3046
    },
    {
      "epoch": 5.795530194959581,
      "grad_norm": 0.3984861671924591,
      "learning_rate": 8.067958081930772e-05,
      "loss": 0.1134,
      "step": 3047
    },
    {
      "epoch": 5.797432239657632,
      "grad_norm": 0.35580113530158997,
      "learning_rate": 8.067322959669736e-05,
      "loss": 0.0912,
      "step": 3048
    },
    {
      "epoch": 5.799334284355682,
      "grad_norm": 0.2826536297798157,
      "learning_rate": 8.066687837408702e-05,
      "loss": 0.1028,
      "step": 3049
    },
    {
      "epoch": 5.801236329053733,
      "grad_norm": 0.26871412992477417,
      "learning_rate": 8.066052715147666e-05,
      "loss": 0.0884,
      "step": 3050
    },
    {
      "epoch": 5.803138373751783,
      "grad_norm": 0.3068493902683258,
      "learning_rate": 8.065417592886631e-05,
      "loss": 0.0921,
      "step": 3051
    },
    {
      "epoch": 5.805040418449834,
      "grad_norm": 0.31987568736076355,
      "learning_rate": 8.064782470625595e-05,
      "loss": 0.0899,
      "step": 3052
    },
    {
      "epoch": 5.806942463147884,
      "grad_norm": 0.25358253717422485,
      "learning_rate": 8.06414734836456e-05,
      "loss": 0.0961,
      "step": 3053
    },
    {
      "epoch": 5.808844507845935,
      "grad_norm": 0.2783609926700592,
      "learning_rate": 8.063512226103526e-05,
      "loss": 0.0943,
      "step": 3054
    },
    {
      "epoch": 5.810746552543985,
      "grad_norm": 0.30739203095436096,
      "learning_rate": 8.06287710384249e-05,
      "loss": 0.0962,
      "step": 3055
    },
    {
      "epoch": 5.812648597242035,
      "grad_norm": 0.2969946265220642,
      "learning_rate": 8.062241981581455e-05,
      "loss": 0.0924,
      "step": 3056
    },
    {
      "epoch": 5.814550641940086,
      "grad_norm": 0.4024634063243866,
      "learning_rate": 8.06160685932042e-05,
      "loss": 0.0973,
      "step": 3057
    },
    {
      "epoch": 5.816452686638136,
      "grad_norm": 0.390010267496109,
      "learning_rate": 8.060971737059384e-05,
      "loss": 0.1082,
      "step": 3058
    },
    {
      "epoch": 5.8183547313361865,
      "grad_norm": 0.45062586665153503,
      "learning_rate": 8.060336614798349e-05,
      "loss": 0.1212,
      "step": 3059
    },
    {
      "epoch": 5.8202567760342365,
      "grad_norm": 0.26907825469970703,
      "learning_rate": 8.059701492537314e-05,
      "loss": 0.0929,
      "step": 3060
    },
    {
      "epoch": 5.822158820732287,
      "grad_norm": 0.2780208885669708,
      "learning_rate": 8.059066370276278e-05,
      "loss": 0.0931,
      "step": 3061
    },
    {
      "epoch": 5.824060865430337,
      "grad_norm": 0.3321477472782135,
      "learning_rate": 8.058431248015243e-05,
      "loss": 0.106,
      "step": 3062
    },
    {
      "epoch": 5.825962910128388,
      "grad_norm": 0.2702338397502899,
      "learning_rate": 8.057796125754208e-05,
      "loss": 0.0804,
      "step": 3063
    },
    {
      "epoch": 5.827864954826438,
      "grad_norm": 0.27758532762527466,
      "learning_rate": 8.057161003493173e-05,
      "loss": 0.1032,
      "step": 3064
    },
    {
      "epoch": 5.829766999524489,
      "grad_norm": 0.39256051182746887,
      "learning_rate": 8.056525881232137e-05,
      "loss": 0.1187,
      "step": 3065
    },
    {
      "epoch": 5.831669044222539,
      "grad_norm": 0.3004806637763977,
      "learning_rate": 8.055890758971102e-05,
      "loss": 0.0988,
      "step": 3066
    },
    {
      "epoch": 5.833571088920589,
      "grad_norm": 0.3810843229293823,
      "learning_rate": 8.055255636710068e-05,
      "loss": 0.1141,
      "step": 3067
    },
    {
      "epoch": 5.83547313361864,
      "grad_norm": 0.27720025181770325,
      "learning_rate": 8.054620514449031e-05,
      "loss": 0.0898,
      "step": 3068
    },
    {
      "epoch": 5.83737517831669,
      "grad_norm": 0.3405880630016327,
      "learning_rate": 8.053985392187997e-05,
      "loss": 0.1195,
      "step": 3069
    },
    {
      "epoch": 5.839277223014741,
      "grad_norm": 0.23189480602741241,
      "learning_rate": 8.053350269926962e-05,
      "loss": 0.0824,
      "step": 3070
    },
    {
      "epoch": 5.841179267712791,
      "grad_norm": 0.2764407694339752,
      "learning_rate": 8.052715147665926e-05,
      "loss": 0.1146,
      "step": 3071
    },
    {
      "epoch": 5.843081312410842,
      "grad_norm": 0.34092894196510315,
      "learning_rate": 8.052080025404891e-05,
      "loss": 0.0994,
      "step": 3072
    },
    {
      "epoch": 5.844983357108892,
      "grad_norm": 0.46230098605155945,
      "learning_rate": 8.051444903143856e-05,
      "loss": 0.136,
      "step": 3073
    },
    {
      "epoch": 5.846885401806943,
      "grad_norm": 0.2688174545764923,
      "learning_rate": 8.05080978088282e-05,
      "loss": 0.1094,
      "step": 3074
    },
    {
      "epoch": 5.848787446504993,
      "grad_norm": 0.3978007137775421,
      "learning_rate": 8.050174658621785e-05,
      "loss": 0.1037,
      "step": 3075
    },
    {
      "epoch": 5.850689491203044,
      "grad_norm": 0.29960164427757263,
      "learning_rate": 8.04953953636075e-05,
      "loss": 0.0939,
      "step": 3076
    },
    {
      "epoch": 5.852591535901094,
      "grad_norm": 0.3292900025844574,
      "learning_rate": 8.048904414099715e-05,
      "loss": 0.0874,
      "step": 3077
    },
    {
      "epoch": 5.854493580599144,
      "grad_norm": 0.33958300948143005,
      "learning_rate": 8.048269291838679e-05,
      "loss": 0.0976,
      "step": 3078
    },
    {
      "epoch": 5.8563956252971945,
      "grad_norm": 0.3058733642101288,
      "learning_rate": 8.047634169577643e-05,
      "loss": 0.0968,
      "step": 3079
    },
    {
      "epoch": 5.858297669995245,
      "grad_norm": 0.3476194143295288,
      "learning_rate": 8.04699904731661e-05,
      "loss": 0.0853,
      "step": 3080
    },
    {
      "epoch": 5.860199714693295,
      "grad_norm": 0.34195569157600403,
      "learning_rate": 8.046363925055573e-05,
      "loss": 0.0905,
      "step": 3081
    },
    {
      "epoch": 5.862101759391345,
      "grad_norm": 0.3744758367538452,
      "learning_rate": 8.045728802794539e-05,
      "loss": 0.103,
      "step": 3082
    },
    {
      "epoch": 5.864003804089396,
      "grad_norm": 0.3824380934238434,
      "learning_rate": 8.045093680533504e-05,
      "loss": 0.0941,
      "step": 3083
    },
    {
      "epoch": 5.865905848787446,
      "grad_norm": 0.33374378085136414,
      "learning_rate": 8.044458558272468e-05,
      "loss": 0.1087,
      "step": 3084
    },
    {
      "epoch": 5.867807893485497,
      "grad_norm": 0.33335942029953003,
      "learning_rate": 8.043823436011433e-05,
      "loss": 0.0997,
      "step": 3085
    },
    {
      "epoch": 5.869709938183547,
      "grad_norm": 0.3410753309726715,
      "learning_rate": 8.043188313750397e-05,
      "loss": 0.0801,
      "step": 3086
    },
    {
      "epoch": 5.871611982881598,
      "grad_norm": 0.31030407547950745,
      "learning_rate": 8.042553191489363e-05,
      "loss": 0.0963,
      "step": 3087
    },
    {
      "epoch": 5.873514027579648,
      "grad_norm": 0.29095837473869324,
      "learning_rate": 8.041918069228327e-05,
      "loss": 0.0929,
      "step": 3088
    },
    {
      "epoch": 5.875416072277699,
      "grad_norm": 0.31219327449798584,
      "learning_rate": 8.041282946967291e-05,
      "loss": 0.0989,
      "step": 3089
    },
    {
      "epoch": 5.877318116975749,
      "grad_norm": 0.3598634898662567,
      "learning_rate": 8.040647824706257e-05,
      "loss": 0.0974,
      "step": 3090
    },
    {
      "epoch": 5.8792201616738,
      "grad_norm": 0.4326852858066559,
      "learning_rate": 8.040012702445221e-05,
      "loss": 0.1409,
      "step": 3091
    },
    {
      "epoch": 5.88112220637185,
      "grad_norm": 0.5164662003517151,
      "learning_rate": 8.039377580184185e-05,
      "loss": 0.126,
      "step": 3092
    },
    {
      "epoch": 5.8830242510699,
      "grad_norm": 0.26032760739326477,
      "learning_rate": 8.03874245792315e-05,
      "loss": 0.0768,
      "step": 3093
    },
    {
      "epoch": 5.884926295767951,
      "grad_norm": 0.3444868326187134,
      "learning_rate": 8.038107335662115e-05,
      "loss": 0.1065,
      "step": 3094
    },
    {
      "epoch": 5.886828340466001,
      "grad_norm": 0.37405380606651306,
      "learning_rate": 8.03747221340108e-05,
      "loss": 0.0984,
      "step": 3095
    },
    {
      "epoch": 5.888730385164052,
      "grad_norm": 0.36833861470222473,
      "learning_rate": 8.036837091140044e-05,
      "loss": 0.0999,
      "step": 3096
    },
    {
      "epoch": 5.890632429862102,
      "grad_norm": 0.3146866261959076,
      "learning_rate": 8.03620196887901e-05,
      "loss": 0.0915,
      "step": 3097
    },
    {
      "epoch": 5.8925344745601524,
      "grad_norm": 0.3376007378101349,
      "learning_rate": 8.035566846617975e-05,
      "loss": 0.0781,
      "step": 3098
    },
    {
      "epoch": 5.8944365192582024,
      "grad_norm": 0.3204367458820343,
      "learning_rate": 8.034931724356939e-05,
      "loss": 0.1133,
      "step": 3099
    },
    {
      "epoch": 5.896338563956253,
      "grad_norm": 0.21730051934719086,
      "learning_rate": 8.034296602095904e-05,
      "loss": 0.061,
      "step": 3100
    },
    {
      "epoch": 5.898240608654303,
      "grad_norm": 0.26351940631866455,
      "learning_rate": 8.033661479834869e-05,
      "loss": 0.0789,
      "step": 3101
    },
    {
      "epoch": 5.900142653352354,
      "grad_norm": 0.3514474034309387,
      "learning_rate": 8.033026357573833e-05,
      "loss": 0.1141,
      "step": 3102
    },
    {
      "epoch": 5.902044698050404,
      "grad_norm": 0.351646363735199,
      "learning_rate": 8.032391235312798e-05,
      "loss": 0.089,
      "step": 3103
    },
    {
      "epoch": 5.903946742748454,
      "grad_norm": 0.3225652873516083,
      "learning_rate": 8.031756113051763e-05,
      "loss": 0.1012,
      "step": 3104
    },
    {
      "epoch": 5.905848787446505,
      "grad_norm": 0.3405236601829529,
      "learning_rate": 8.031120990790728e-05,
      "loss": 0.0851,
      "step": 3105
    },
    {
      "epoch": 5.907750832144555,
      "grad_norm": 0.39399632811546326,
      "learning_rate": 8.030485868529692e-05,
      "loss": 0.0915,
      "step": 3106
    },
    {
      "epoch": 5.909652876842606,
      "grad_norm": 0.4024227559566498,
      "learning_rate": 8.029850746268657e-05,
      "loss": 0.1077,
      "step": 3107
    },
    {
      "epoch": 5.911554921540656,
      "grad_norm": 0.3052838146686554,
      "learning_rate": 8.029215624007623e-05,
      "loss": 0.0787,
      "step": 3108
    },
    {
      "epoch": 5.913456966238707,
      "grad_norm": 0.4465295970439911,
      "learning_rate": 8.028580501746586e-05,
      "loss": 0.0963,
      "step": 3109
    },
    {
      "epoch": 5.915359010936757,
      "grad_norm": 0.28723543882369995,
      "learning_rate": 8.02794537948555e-05,
      "loss": 0.0752,
      "step": 3110
    },
    {
      "epoch": 5.917261055634808,
      "grad_norm": 0.4216603934764862,
      "learning_rate": 8.027310257224517e-05,
      "loss": 0.1018,
      "step": 3111
    },
    {
      "epoch": 5.919163100332858,
      "grad_norm": 0.2752178907394409,
      "learning_rate": 8.02667513496348e-05,
      "loss": 0.0817,
      "step": 3112
    },
    {
      "epoch": 5.921065145030909,
      "grad_norm": 0.27371007204055786,
      "learning_rate": 8.026040012702446e-05,
      "loss": 0.0783,
      "step": 3113
    },
    {
      "epoch": 5.922967189728959,
      "grad_norm": 0.3314353823661804,
      "learning_rate": 8.025404890441411e-05,
      "loss": 0.0891,
      "step": 3114
    },
    {
      "epoch": 5.924869234427009,
      "grad_norm": 0.35603460669517517,
      "learning_rate": 8.024769768180375e-05,
      "loss": 0.104,
      "step": 3115
    },
    {
      "epoch": 5.9267712791250595,
      "grad_norm": 0.3468729853630066,
      "learning_rate": 8.02413464591934e-05,
      "loss": 0.0974,
      "step": 3116
    },
    {
      "epoch": 5.9286733238231095,
      "grad_norm": 0.3321313261985779,
      "learning_rate": 8.023499523658304e-05,
      "loss": 0.093,
      "step": 3117
    },
    {
      "epoch": 5.93057536852116,
      "grad_norm": 0.36478671431541443,
      "learning_rate": 8.02286440139727e-05,
      "loss": 0.0979,
      "step": 3118
    },
    {
      "epoch": 5.93247741321921,
      "grad_norm": 0.32760557532310486,
      "learning_rate": 8.022229279136234e-05,
      "loss": 0.1148,
      "step": 3119
    },
    {
      "epoch": 5.934379457917261,
      "grad_norm": 0.3271568715572357,
      "learning_rate": 8.021594156875198e-05,
      "loss": 0.0959,
      "step": 3120
    },
    {
      "epoch": 5.936281502615311,
      "grad_norm": 0.3014872968196869,
      "learning_rate": 8.020959034614164e-05,
      "loss": 0.0942,
      "step": 3121
    },
    {
      "epoch": 5.938183547313362,
      "grad_norm": 0.30579501390457153,
      "learning_rate": 8.020323912353128e-05,
      "loss": 0.0826,
      "step": 3122
    },
    {
      "epoch": 5.940085592011412,
      "grad_norm": 0.42316779494285583,
      "learning_rate": 8.019688790092093e-05,
      "loss": 0.1025,
      "step": 3123
    },
    {
      "epoch": 5.941987636709463,
      "grad_norm": 0.30180487036705017,
      "learning_rate": 8.019053667831057e-05,
      "loss": 0.0905,
      "step": 3124
    },
    {
      "epoch": 5.943889681407513,
      "grad_norm": 0.3880978226661682,
      "learning_rate": 8.018418545570023e-05,
      "loss": 0.1066,
      "step": 3125
    },
    {
      "epoch": 5.945791726105563,
      "grad_norm": 0.3437277674674988,
      "learning_rate": 8.017783423308988e-05,
      "loss": 0.0973,
      "step": 3126
    },
    {
      "epoch": 5.947693770803614,
      "grad_norm": 0.3143709897994995,
      "learning_rate": 8.017148301047952e-05,
      "loss": 0.0947,
      "step": 3127
    },
    {
      "epoch": 5.949595815501664,
      "grad_norm": 0.2899588644504547,
      "learning_rate": 8.016513178786917e-05,
      "loss": 0.0979,
      "step": 3128
    },
    {
      "epoch": 5.951497860199715,
      "grad_norm": 0.3774805963039398,
      "learning_rate": 8.015878056525882e-05,
      "loss": 0.2451,
      "step": 3129
    },
    {
      "epoch": 5.953399904897765,
      "grad_norm": 0.38204115629196167,
      "learning_rate": 8.015242934264846e-05,
      "loss": 0.1064,
      "step": 3130
    },
    {
      "epoch": 5.955301949595816,
      "grad_norm": 0.25636038184165955,
      "learning_rate": 8.014607812003811e-05,
      "loss": 0.0767,
      "step": 3131
    },
    {
      "epoch": 5.957203994293866,
      "grad_norm": 0.4159661829471588,
      "learning_rate": 8.013972689742776e-05,
      "loss": 0.0946,
      "step": 3132
    },
    {
      "epoch": 5.959106038991917,
      "grad_norm": 0.38069942593574524,
      "learning_rate": 8.01333756748174e-05,
      "loss": 0.1257,
      "step": 3133
    },
    {
      "epoch": 5.961008083689967,
      "grad_norm": 0.3810632526874542,
      "learning_rate": 8.012702445220705e-05,
      "loss": 0.0761,
      "step": 3134
    },
    {
      "epoch": 5.9629101283880175,
      "grad_norm": 0.30504754185676575,
      "learning_rate": 8.01206732295967e-05,
      "loss": 0.0933,
      "step": 3135
    },
    {
      "epoch": 5.9648121730860675,
      "grad_norm": 0.24171409010887146,
      "learning_rate": 8.011432200698635e-05,
      "loss": 0.0851,
      "step": 3136
    },
    {
      "epoch": 5.9667142177841175,
      "grad_norm": 0.32692912220954895,
      "learning_rate": 8.010797078437599e-05,
      "loss": 0.0836,
      "step": 3137
    },
    {
      "epoch": 5.968616262482168,
      "grad_norm": 0.2763676643371582,
      "learning_rate": 8.010161956176564e-05,
      "loss": 0.0778,
      "step": 3138
    },
    {
      "epoch": 5.970518307180218,
      "grad_norm": 0.42881324887275696,
      "learning_rate": 8.00952683391553e-05,
      "loss": 0.1088,
      "step": 3139
    },
    {
      "epoch": 5.972420351878269,
      "grad_norm": 0.3442386984825134,
      "learning_rate": 8.008891711654493e-05,
      "loss": 0.0993,
      "step": 3140
    },
    {
      "epoch": 5.974322396576319,
      "grad_norm": 0.3938843607902527,
      "learning_rate": 8.008256589393459e-05,
      "loss": 0.1337,
      "step": 3141
    },
    {
      "epoch": 5.97622444127437,
      "grad_norm": 0.33586129546165466,
      "learning_rate": 8.007621467132424e-05,
      "loss": 0.1002,
      "step": 3142
    },
    {
      "epoch": 5.97812648597242,
      "grad_norm": 0.3166608214378357,
      "learning_rate": 8.006986344871388e-05,
      "loss": 0.099,
      "step": 3143
    },
    {
      "epoch": 5.980028530670471,
      "grad_norm": 0.31059524416923523,
      "learning_rate": 8.006351222610353e-05,
      "loss": 0.1011,
      "step": 3144
    },
    {
      "epoch": 5.981930575368521,
      "grad_norm": 0.33092477917671204,
      "learning_rate": 8.005716100349318e-05,
      "loss": 0.0785,
      "step": 3145
    },
    {
      "epoch": 5.983832620066572,
      "grad_norm": 0.4692544639110565,
      "learning_rate": 8.005080978088282e-05,
      "loss": 0.1377,
      "step": 3146
    },
    {
      "epoch": 5.985734664764622,
      "grad_norm": 0.42482489347457886,
      "learning_rate": 8.004445855827247e-05,
      "loss": 0.1286,
      "step": 3147
    },
    {
      "epoch": 5.987636709462672,
      "grad_norm": 0.3942713737487793,
      "learning_rate": 8.003810733566212e-05,
      "loss": 0.1085,
      "step": 3148
    },
    {
      "epoch": 5.989538754160723,
      "grad_norm": 0.4134069085121155,
      "learning_rate": 8.003175611305177e-05,
      "loss": 0.1149,
      "step": 3149
    },
    {
      "epoch": 5.991440798858774,
      "grad_norm": 0.30162832140922546,
      "learning_rate": 8.002540489044141e-05,
      "loss": 0.0945,
      "step": 3150
    },
    {
      "epoch": 5.993342843556824,
      "grad_norm": 0.38140755891799927,
      "learning_rate": 8.001905366783105e-05,
      "loss": 0.103,
      "step": 3151
    },
    {
      "epoch": 5.995244888254874,
      "grad_norm": 0.2811982035636902,
      "learning_rate": 8.001270244522072e-05,
      "loss": 0.0877,
      "step": 3152
    },
    {
      "epoch": 5.997146932952925,
      "grad_norm": 0.3001856207847595,
      "learning_rate": 8.000635122261035e-05,
      "loss": 0.0895,
      "step": 3153
    },
    {
      "epoch": 5.999048977650975,
      "grad_norm": 0.2948606610298157,
      "learning_rate": 8e-05,
      "loss": 0.1035,
      "step": 3154
    },
    {
      "epoch": 6.000951022349025,
      "grad_norm": 0.3746136426925659,
      "learning_rate": 7.999364877738964e-05,
      "loss": 0.1152,
      "step": 3155
    },
    {
      "epoch": 6.002853067047075,
      "grad_norm": 0.21500763297080994,
      "learning_rate": 7.99872975547793e-05,
      "loss": 0.0769,
      "step": 3156
    },
    {
      "epoch": 6.004755111745126,
      "grad_norm": 0.249894380569458,
      "learning_rate": 7.998094633216895e-05,
      "loss": 0.0665,
      "step": 3157
    },
    {
      "epoch": 6.006657156443176,
      "grad_norm": 0.3538912832736969,
      "learning_rate": 7.997459510955859e-05,
      "loss": 0.1028,
      "step": 3158
    },
    {
      "epoch": 6.008559201141227,
      "grad_norm": 0.20869600772857666,
      "learning_rate": 7.996824388694825e-05,
      "loss": 0.0594,
      "step": 3159
    },
    {
      "epoch": 6.010461245839277,
      "grad_norm": 0.38950517773628235,
      "learning_rate": 7.996189266433789e-05,
      "loss": 0.0932,
      "step": 3160
    },
    {
      "epoch": 6.012363290537327,
      "grad_norm": 0.33621397614479065,
      "learning_rate": 7.995554144172753e-05,
      "loss": 0.0939,
      "step": 3161
    },
    {
      "epoch": 6.014265335235378,
      "grad_norm": 0.3625282943248749,
      "learning_rate": 7.994919021911718e-05,
      "loss": 0.107,
      "step": 3162
    },
    {
      "epoch": 6.016167379933428,
      "grad_norm": 0.32474133372306824,
      "learning_rate": 7.994283899650683e-05,
      "loss": 0.1042,
      "step": 3163
    },
    {
      "epoch": 6.018069424631479,
      "grad_norm": 0.1935519427061081,
      "learning_rate": 7.993648777389647e-05,
      "loss": 0.0735,
      "step": 3164
    },
    {
      "epoch": 6.019971469329529,
      "grad_norm": 0.3521823287010193,
      "learning_rate": 7.993013655128612e-05,
      "loss": 0.0945,
      "step": 3165
    },
    {
      "epoch": 6.02187351402758,
      "grad_norm": 0.22658848762512207,
      "learning_rate": 7.992378532867577e-05,
      "loss": 0.059,
      "step": 3166
    },
    {
      "epoch": 6.02377555872563,
      "grad_norm": 0.2659030556678772,
      "learning_rate": 7.991743410606543e-05,
      "loss": 0.0803,
      "step": 3167
    },
    {
      "epoch": 6.025677603423681,
      "grad_norm": 0.3843698799610138,
      "learning_rate": 7.991108288345506e-05,
      "loss": 0.1053,
      "step": 3168
    },
    {
      "epoch": 6.027579648121731,
      "grad_norm": 0.23756776750087738,
      "learning_rate": 7.990473166084472e-05,
      "loss": 0.0749,
      "step": 3169
    },
    {
      "epoch": 6.029481692819782,
      "grad_norm": 0.25975948572158813,
      "learning_rate": 7.989838043823437e-05,
      "loss": 0.0879,
      "step": 3170
    },
    {
      "epoch": 6.031383737517832,
      "grad_norm": 0.32162702083587646,
      "learning_rate": 7.9892029215624e-05,
      "loss": 0.0886,
      "step": 3171
    },
    {
      "epoch": 6.0332857822158825,
      "grad_norm": 0.3066456615924835,
      "learning_rate": 7.988567799301366e-05,
      "loss": 0.0712,
      "step": 3172
    },
    {
      "epoch": 6.0351878269139325,
      "grad_norm": 0.2621491253376007,
      "learning_rate": 7.987932677040331e-05,
      "loss": 0.0594,
      "step": 3173
    },
    {
      "epoch": 6.0370898716119825,
      "grad_norm": 0.16599467396736145,
      "learning_rate": 7.987297554779295e-05,
      "loss": 0.0584,
      "step": 3174
    },
    {
      "epoch": 6.038991916310033,
      "grad_norm": 0.36765870451927185,
      "learning_rate": 7.98666243251826e-05,
      "loss": 0.0789,
      "step": 3175
    },
    {
      "epoch": 6.040893961008083,
      "grad_norm": 0.3242380619049072,
      "learning_rate": 7.986027310257225e-05,
      "loss": 0.0977,
      "step": 3176
    },
    {
      "epoch": 6.042796005706134,
      "grad_norm": 0.4075121283531189,
      "learning_rate": 7.98539218799619e-05,
      "loss": 0.09,
      "step": 3177
    },
    {
      "epoch": 6.044698050404184,
      "grad_norm": 0.28770434856414795,
      "learning_rate": 7.984757065735154e-05,
      "loss": 0.0748,
      "step": 3178
    },
    {
      "epoch": 6.046600095102235,
      "grad_norm": 0.32635626196861267,
      "learning_rate": 7.98412194347412e-05,
      "loss": 0.088,
      "step": 3179
    },
    {
      "epoch": 6.048502139800285,
      "grad_norm": 0.254139244556427,
      "learning_rate": 7.983486821213085e-05,
      "loss": 0.0718,
      "step": 3180
    },
    {
      "epoch": 6.050404184498336,
      "grad_norm": 0.349336713552475,
      "learning_rate": 7.982851698952048e-05,
      "loss": 0.118,
      "step": 3181
    },
    {
      "epoch": 6.052306229196386,
      "grad_norm": 0.33583179116249084,
      "learning_rate": 7.982216576691012e-05,
      "loss": 0.1126,
      "step": 3182
    },
    {
      "epoch": 6.054208273894437,
      "grad_norm": 0.25572314858436584,
      "learning_rate": 7.981581454429979e-05,
      "loss": 0.1009,
      "step": 3183
    },
    {
      "epoch": 6.056110318592487,
      "grad_norm": 0.2622678577899933,
      "learning_rate": 7.980946332168943e-05,
      "loss": 0.0628,
      "step": 3184
    },
    {
      "epoch": 6.058012363290537,
      "grad_norm": 0.24747587740421295,
      "learning_rate": 7.980311209907908e-05,
      "loss": 0.0657,
      "step": 3185
    },
    {
      "epoch": 6.059914407988588,
      "grad_norm": 0.27714699506759644,
      "learning_rate": 7.979676087646873e-05,
      "loss": 0.062,
      "step": 3186
    },
    {
      "epoch": 6.061816452686638,
      "grad_norm": 0.27411553263664246,
      "learning_rate": 7.979040965385837e-05,
      "loss": 0.082,
      "step": 3187
    },
    {
      "epoch": 6.063718497384689,
      "grad_norm": 0.3901764452457428,
      "learning_rate": 7.978405843124802e-05,
      "loss": 0.1028,
      "step": 3188
    },
    {
      "epoch": 6.065620542082739,
      "grad_norm": 0.20082591474056244,
      "learning_rate": 7.977770720863766e-05,
      "loss": 0.0619,
      "step": 3189
    },
    {
      "epoch": 6.06752258678079,
      "grad_norm": 0.30728980898857117,
      "learning_rate": 7.977135598602732e-05,
      "loss": 0.0786,
      "step": 3190
    },
    {
      "epoch": 6.06942463147884,
      "grad_norm": 0.3012641966342926,
      "learning_rate": 7.976500476341696e-05,
      "loss": 0.0725,
      "step": 3191
    },
    {
      "epoch": 6.0713266761768905,
      "grad_norm": 0.21619060635566711,
      "learning_rate": 7.97586535408066e-05,
      "loss": 0.0595,
      "step": 3192
    },
    {
      "epoch": 6.0732287208749405,
      "grad_norm": 0.2747598886489868,
      "learning_rate": 7.975230231819627e-05,
      "loss": 0.0675,
      "step": 3193
    },
    {
      "epoch": 6.075130765572991,
      "grad_norm": 0.3214286267757416,
      "learning_rate": 7.97459510955859e-05,
      "loss": 0.0902,
      "step": 3194
    },
    {
      "epoch": 6.077032810271041,
      "grad_norm": 0.37599408626556396,
      "learning_rate": 7.973959987297556e-05,
      "loss": 0.0794,
      "step": 3195
    },
    {
      "epoch": 6.078934854969091,
      "grad_norm": 0.2802245318889618,
      "learning_rate": 7.97332486503652e-05,
      "loss": 0.0691,
      "step": 3196
    },
    {
      "epoch": 6.080836899667142,
      "grad_norm": 0.22154775261878967,
      "learning_rate": 7.972689742775485e-05,
      "loss": 0.0848,
      "step": 3197
    },
    {
      "epoch": 6.082738944365192,
      "grad_norm": 0.4010308086872101,
      "learning_rate": 7.97205462051445e-05,
      "loss": 0.0838,
      "step": 3198
    },
    {
      "epoch": 6.084640989063243,
      "grad_norm": 0.32006803154945374,
      "learning_rate": 7.971419498253414e-05,
      "loss": 0.1033,
      "step": 3199
    },
    {
      "epoch": 6.086543033761293,
      "grad_norm": 0.2201933115720749,
      "learning_rate": 7.970784375992379e-05,
      "loss": 0.0648,
      "step": 3200
    },
    {
      "epoch": 6.088445078459344,
      "grad_norm": 0.2613593637943268,
      "learning_rate": 7.970149253731344e-05,
      "loss": 0.0743,
      "step": 3201
    },
    {
      "epoch": 6.090347123157394,
      "grad_norm": 0.26748836040496826,
      "learning_rate": 7.969514131470308e-05,
      "loss": 0.0774,
      "step": 3202
    },
    {
      "epoch": 6.092249167855445,
      "grad_norm": 0.26437023282051086,
      "learning_rate": 7.968879009209273e-05,
      "loss": 0.0915,
      "step": 3203
    },
    {
      "epoch": 6.094151212553495,
      "grad_norm": 0.30303144454956055,
      "learning_rate": 7.968243886948238e-05,
      "loss": 0.1109,
      "step": 3204
    },
    {
      "epoch": 6.096053257251546,
      "grad_norm": 0.22862769663333893,
      "learning_rate": 7.967608764687202e-05,
      "loss": 0.0659,
      "step": 3205
    },
    {
      "epoch": 6.097955301949596,
      "grad_norm": 0.22851070761680603,
      "learning_rate": 7.966973642426167e-05,
      "loss": 0.0664,
      "step": 3206
    },
    {
      "epoch": 6.099857346647646,
      "grad_norm": 0.3395095467567444,
      "learning_rate": 7.966338520165132e-05,
      "loss": 0.1202,
      "step": 3207
    },
    {
      "epoch": 6.101759391345697,
      "grad_norm": 0.33721911907196045,
      "learning_rate": 7.965703397904098e-05,
      "loss": 0.0971,
      "step": 3208
    },
    {
      "epoch": 6.103661436043747,
      "grad_norm": 0.28860366344451904,
      "learning_rate": 7.965068275643061e-05,
      "loss": 0.0876,
      "step": 3209
    },
    {
      "epoch": 6.1055634807417976,
      "grad_norm": 0.3022976815700531,
      "learning_rate": 7.964433153382027e-05,
      "loss": 0.072,
      "step": 3210
    },
    {
      "epoch": 6.1074655254398476,
      "grad_norm": 0.27039656043052673,
      "learning_rate": 7.963798031120992e-05,
      "loss": 0.0857,
      "step": 3211
    },
    {
      "epoch": 6.109367570137898,
      "grad_norm": 0.28560107946395874,
      "learning_rate": 7.963162908859956e-05,
      "loss": 0.0703,
      "step": 3212
    },
    {
      "epoch": 6.111269614835948,
      "grad_norm": 0.16948960721492767,
      "learning_rate": 7.962527786598921e-05,
      "loss": 0.0574,
      "step": 3213
    },
    {
      "epoch": 6.113171659533999,
      "grad_norm": 0.31261399388313293,
      "learning_rate": 7.961892664337886e-05,
      "loss": 0.098,
      "step": 3214
    },
    {
      "epoch": 6.115073704232049,
      "grad_norm": 0.2879314720630646,
      "learning_rate": 7.96125754207685e-05,
      "loss": 0.0768,
      "step": 3215
    },
    {
      "epoch": 6.1169757489301,
      "grad_norm": 0.3212367296218872,
      "learning_rate": 7.960622419815815e-05,
      "loss": 0.0816,
      "step": 3216
    },
    {
      "epoch": 6.11887779362815,
      "grad_norm": 0.306856244802475,
      "learning_rate": 7.95998729755478e-05,
      "loss": 0.0706,
      "step": 3217
    },
    {
      "epoch": 6.120779838326201,
      "grad_norm": 0.30228865146636963,
      "learning_rate": 7.959352175293744e-05,
      "loss": 0.0842,
      "step": 3218
    },
    {
      "epoch": 6.122681883024251,
      "grad_norm": 0.40820634365081787,
      "learning_rate": 7.958717053032709e-05,
      "loss": 0.0831,
      "step": 3219
    },
    {
      "epoch": 6.124583927722301,
      "grad_norm": 0.2787196934223175,
      "learning_rate": 7.958081930771673e-05,
      "loss": 0.0817,
      "step": 3220
    },
    {
      "epoch": 6.126485972420352,
      "grad_norm": 0.26536598801612854,
      "learning_rate": 7.95744680851064e-05,
      "loss": 0.0868,
      "step": 3221
    },
    {
      "epoch": 6.128388017118402,
      "grad_norm": 0.33307042717933655,
      "learning_rate": 7.956811686249603e-05,
      "loss": 0.1104,
      "step": 3222
    },
    {
      "epoch": 6.130290061816453,
      "grad_norm": 0.29908856749534607,
      "learning_rate": 7.956176563988567e-05,
      "loss": 0.0981,
      "step": 3223
    },
    {
      "epoch": 6.132192106514503,
      "grad_norm": 0.25544753670692444,
      "learning_rate": 7.955541441727534e-05,
      "loss": 0.0635,
      "step": 3224
    },
    {
      "epoch": 6.134094151212554,
      "grad_norm": 0.458103746175766,
      "learning_rate": 7.954906319466498e-05,
      "loss": 0.1149,
      "step": 3225
    },
    {
      "epoch": 6.135996195910604,
      "grad_norm": 0.23199333250522614,
      "learning_rate": 7.954271197205463e-05,
      "loss": 0.0806,
      "step": 3226
    },
    {
      "epoch": 6.137898240608655,
      "grad_norm": 0.25189074873924255,
      "learning_rate": 7.953636074944427e-05,
      "loss": 0.0782,
      "step": 3227
    },
    {
      "epoch": 6.139800285306705,
      "grad_norm": 0.3281899094581604,
      "learning_rate": 7.953000952683392e-05,
      "loss": 0.0706,
      "step": 3228
    },
    {
      "epoch": 6.1417023300047555,
      "grad_norm": 0.309741348028183,
      "learning_rate": 7.952365830422357e-05,
      "loss": 0.0676,
      "step": 3229
    },
    {
      "epoch": 6.1436043747028055,
      "grad_norm": 0.4716850519180298,
      "learning_rate": 7.951730708161321e-05,
      "loss": 0.1398,
      "step": 3230
    },
    {
      "epoch": 6.1455064194008555,
      "grad_norm": 0.28038209676742554,
      "learning_rate": 7.951095585900287e-05,
      "loss": 0.1256,
      "step": 3231
    },
    {
      "epoch": 6.147408464098906,
      "grad_norm": 0.2801377773284912,
      "learning_rate": 7.950460463639251e-05,
      "loss": 0.063,
      "step": 3232
    },
    {
      "epoch": 6.149310508796956,
      "grad_norm": 0.2686994671821594,
      "learning_rate": 7.949825341378215e-05,
      "loss": 0.0735,
      "step": 3233
    },
    {
      "epoch": 6.151212553495007,
      "grad_norm": 0.30351752042770386,
      "learning_rate": 7.94919021911718e-05,
      "loss": 0.0856,
      "step": 3234
    },
    {
      "epoch": 6.153114598193057,
      "grad_norm": 0.2922990024089813,
      "learning_rate": 7.948555096856145e-05,
      "loss": 0.0883,
      "step": 3235
    },
    {
      "epoch": 6.155016642891108,
      "grad_norm": 0.41719111800193787,
      "learning_rate": 7.947919974595109e-05,
      "loss": 0.0905,
      "step": 3236
    },
    {
      "epoch": 6.156918687589158,
      "grad_norm": 0.24950455129146576,
      "learning_rate": 7.947284852334074e-05,
      "loss": 0.0596,
      "step": 3237
    },
    {
      "epoch": 6.158820732287209,
      "grad_norm": 0.2344481348991394,
      "learning_rate": 7.94664973007304e-05,
      "loss": 0.0789,
      "step": 3238
    },
    {
      "epoch": 6.160722776985259,
      "grad_norm": 0.21180817484855652,
      "learning_rate": 7.946014607812005e-05,
      "loss": 0.0616,
      "step": 3239
    },
    {
      "epoch": 6.16262482168331,
      "grad_norm": 0.2386842519044876,
      "learning_rate": 7.945379485550969e-05,
      "loss": 0.086,
      "step": 3240
    },
    {
      "epoch": 6.16452686638136,
      "grad_norm": 0.21530573070049286,
      "learning_rate": 7.944744363289934e-05,
      "loss": 0.0704,
      "step": 3241
    },
    {
      "epoch": 6.166428911079411,
      "grad_norm": 0.24444977939128876,
      "learning_rate": 7.944109241028899e-05,
      "loss": 0.0711,
      "step": 3242
    },
    {
      "epoch": 6.168330955777461,
      "grad_norm": 0.24185262620449066,
      "learning_rate": 7.943474118767863e-05,
      "loss": 0.0705,
      "step": 3243
    },
    {
      "epoch": 6.170233000475511,
      "grad_norm": 0.28293880820274353,
      "learning_rate": 7.942838996506828e-05,
      "loss": 0.0709,
      "step": 3244
    },
    {
      "epoch": 6.172135045173562,
      "grad_norm": 0.2779384255409241,
      "learning_rate": 7.942203874245793e-05,
      "loss": 0.0821,
      "step": 3245
    },
    {
      "epoch": 6.174037089871612,
      "grad_norm": 0.509385883808136,
      "learning_rate": 7.941568751984757e-05,
      "loss": 0.1144,
      "step": 3246
    },
    {
      "epoch": 6.175939134569663,
      "grad_norm": 0.2250889390707016,
      "learning_rate": 7.940933629723722e-05,
      "loss": 0.0576,
      "step": 3247
    },
    {
      "epoch": 6.177841179267713,
      "grad_norm": 0.21779869496822357,
      "learning_rate": 7.940298507462687e-05,
      "loss": 0.0871,
      "step": 3248
    },
    {
      "epoch": 6.1797432239657635,
      "grad_norm": 0.3144400715827942,
      "learning_rate": 7.939663385201652e-05,
      "loss": 0.0832,
      "step": 3249
    },
    {
      "epoch": 6.1816452686638135,
      "grad_norm": 0.2595573663711548,
      "learning_rate": 7.939028262940616e-05,
      "loss": 0.078,
      "step": 3250
    },
    {
      "epoch": 6.183547313361864,
      "grad_norm": 0.4201616942882538,
      "learning_rate": 7.938393140679582e-05,
      "loss": 0.1123,
      "step": 3251
    },
    {
      "epoch": 6.185449358059914,
      "grad_norm": 0.2986547350883484,
      "learning_rate": 7.937758018418547e-05,
      "loss": 0.0911,
      "step": 3252
    },
    {
      "epoch": 6.187351402757965,
      "grad_norm": 0.3912739157676697,
      "learning_rate": 7.93712289615751e-05,
      "loss": 0.0904,
      "step": 3253
    },
    {
      "epoch": 6.189253447456015,
      "grad_norm": 0.2589069902896881,
      "learning_rate": 7.936487773896474e-05,
      "loss": 0.081,
      "step": 3254
    },
    {
      "epoch": 6.191155492154065,
      "grad_norm": 0.3324070870876312,
      "learning_rate": 7.935852651635441e-05,
      "loss": 0.0897,
      "step": 3255
    },
    {
      "epoch": 6.193057536852116,
      "grad_norm": 0.2300262153148651,
      "learning_rate": 7.935217529374405e-05,
      "loss": 0.0658,
      "step": 3256
    },
    {
      "epoch": 6.194959581550166,
      "grad_norm": 0.19878318905830383,
      "learning_rate": 7.93458240711337e-05,
      "loss": 0.0579,
      "step": 3257
    },
    {
      "epoch": 6.196861626248217,
      "grad_norm": 0.3048309087753296,
      "learning_rate": 7.933947284852334e-05,
      "loss": 0.0719,
      "step": 3258
    },
    {
      "epoch": 6.198763670946267,
      "grad_norm": 0.3125574290752411,
      "learning_rate": 7.933312162591299e-05,
      "loss": 0.0707,
      "step": 3259
    },
    {
      "epoch": 6.200665715644318,
      "grad_norm": 0.2814660668373108,
      "learning_rate": 7.932677040330264e-05,
      "loss": 0.0761,
      "step": 3260
    },
    {
      "epoch": 6.202567760342368,
      "grad_norm": 0.22039173543453217,
      "learning_rate": 7.932041918069228e-05,
      "loss": 0.0634,
      "step": 3261
    },
    {
      "epoch": 6.204469805040419,
      "grad_norm": 0.2500283420085907,
      "learning_rate": 7.931406795808194e-05,
      "loss": 0.0746,
      "step": 3262
    },
    {
      "epoch": 6.206371849738469,
      "grad_norm": 0.28570613265037537,
      "learning_rate": 7.930771673547158e-05,
      "loss": 0.0809,
      "step": 3263
    },
    {
      "epoch": 6.20827389443652,
      "grad_norm": 0.4383977949619293,
      "learning_rate": 7.930136551286122e-05,
      "loss": 0.1058,
      "step": 3264
    },
    {
      "epoch": 6.21017593913457,
      "grad_norm": 0.24185074865818024,
      "learning_rate": 7.929501429025087e-05,
      "loss": 0.0737,
      "step": 3265
    },
    {
      "epoch": 6.21207798383262,
      "grad_norm": 0.3835535943508148,
      "learning_rate": 7.928866306764052e-05,
      "loss": 0.1063,
      "step": 3266
    },
    {
      "epoch": 6.2139800285306706,
      "grad_norm": 0.23498846590518951,
      "learning_rate": 7.928231184503018e-05,
      "loss": 0.0784,
      "step": 3267
    },
    {
      "epoch": 6.2158820732287206,
      "grad_norm": 0.3388371467590332,
      "learning_rate": 7.927596062241982e-05,
      "loss": 0.0836,
      "step": 3268
    },
    {
      "epoch": 6.217784117926771,
      "grad_norm": 0.24535390734672546,
      "learning_rate": 7.926960939980947e-05,
      "loss": 0.0799,
      "step": 3269
    },
    {
      "epoch": 6.219686162624821,
      "grad_norm": 0.2897215485572815,
      "learning_rate": 7.926325817719912e-05,
      "loss": 0.0736,
      "step": 3270
    },
    {
      "epoch": 6.221588207322872,
      "grad_norm": 0.3779149353504181,
      "learning_rate": 7.925690695458876e-05,
      "loss": 0.0846,
      "step": 3271
    },
    {
      "epoch": 6.223490252020922,
      "grad_norm": 0.29152947664260864,
      "learning_rate": 7.925055573197841e-05,
      "loss": 0.0868,
      "step": 3272
    },
    {
      "epoch": 6.225392296718973,
      "grad_norm": 0.3078393042087555,
      "learning_rate": 7.924420450936806e-05,
      "loss": 0.1035,
      "step": 3273
    },
    {
      "epoch": 6.227294341417023,
      "grad_norm": 0.21019497513771057,
      "learning_rate": 7.92378532867577e-05,
      "loss": 0.0668,
      "step": 3274
    },
    {
      "epoch": 6.229196386115074,
      "grad_norm": 0.27748751640319824,
      "learning_rate": 7.923150206414735e-05,
      "loss": 0.0835,
      "step": 3275
    },
    {
      "epoch": 6.231098430813124,
      "grad_norm": 0.24059148132801056,
      "learning_rate": 7.9225150841537e-05,
      "loss": 0.0613,
      "step": 3276
    },
    {
      "epoch": 6.233000475511174,
      "grad_norm": 0.2641827166080475,
      "learning_rate": 7.921879961892664e-05,
      "loss": 0.0821,
      "step": 3277
    },
    {
      "epoch": 6.234902520209225,
      "grad_norm": 0.3308301866054535,
      "learning_rate": 7.921244839631629e-05,
      "loss": 0.0858,
      "step": 3278
    },
    {
      "epoch": 6.236804564907275,
      "grad_norm": 0.3596150875091553,
      "learning_rate": 7.920609717370594e-05,
      "loss": 0.0991,
      "step": 3279
    },
    {
      "epoch": 6.238706609605326,
      "grad_norm": 0.23002928495407104,
      "learning_rate": 7.91997459510956e-05,
      "loss": 0.0692,
      "step": 3280
    },
    {
      "epoch": 6.240608654303376,
      "grad_norm": 0.38528430461883545,
      "learning_rate": 7.919339472848523e-05,
      "loss": 0.112,
      "step": 3281
    },
    {
      "epoch": 6.242510699001427,
      "grad_norm": 0.3260602355003357,
      "learning_rate": 7.918704350587489e-05,
      "loss": 0.087,
      "step": 3282
    },
    {
      "epoch": 6.244412743699477,
      "grad_norm": 0.2542354464530945,
      "learning_rate": 7.918069228326454e-05,
      "loss": 0.0701,
      "step": 3283
    },
    {
      "epoch": 6.246314788397528,
      "grad_norm": 0.3322354257106781,
      "learning_rate": 7.917434106065418e-05,
      "loss": 0.0704,
      "step": 3284
    },
    {
      "epoch": 6.248216833095578,
      "grad_norm": 0.31040525436401367,
      "learning_rate": 7.916798983804383e-05,
      "loss": 0.0925,
      "step": 3285
    },
    {
      "epoch": 6.2501188777936285,
      "grad_norm": 0.23275534808635712,
      "learning_rate": 7.916163861543348e-05,
      "loss": 0.0662,
      "step": 3286
    },
    {
      "epoch": 6.2520209224916785,
      "grad_norm": 0.2919691801071167,
      "learning_rate": 7.915528739282312e-05,
      "loss": 0.0799,
      "step": 3287
    },
    {
      "epoch": 6.2539229671897285,
      "grad_norm": 0.2902929186820984,
      "learning_rate": 7.914893617021277e-05,
      "loss": 0.0904,
      "step": 3288
    },
    {
      "epoch": 6.255825011887779,
      "grad_norm": 0.3168974220752716,
      "learning_rate": 7.914258494760242e-05,
      "loss": 0.0925,
      "step": 3289
    },
    {
      "epoch": 6.257727056585829,
      "grad_norm": 0.49860697984695435,
      "learning_rate": 7.913623372499206e-05,
      "loss": 0.1138,
      "step": 3290
    },
    {
      "epoch": 6.25962910128388,
      "grad_norm": 0.35882389545440674,
      "learning_rate": 7.912988250238171e-05,
      "loss": 0.1377,
      "step": 3291
    },
    {
      "epoch": 6.26153114598193,
      "grad_norm": 0.40341517329216003,
      "learning_rate": 7.912353127977135e-05,
      "loss": 0.0884,
      "step": 3292
    },
    {
      "epoch": 6.263433190679981,
      "grad_norm": 0.18433208763599396,
      "learning_rate": 7.911718005716102e-05,
      "loss": 0.055,
      "step": 3293
    },
    {
      "epoch": 6.265335235378031,
      "grad_norm": 0.39911824464797974,
      "learning_rate": 7.911082883455065e-05,
      "loss": 0.0915,
      "step": 3294
    },
    {
      "epoch": 6.267237280076082,
      "grad_norm": 0.32022351026535034,
      "learning_rate": 7.910447761194029e-05,
      "loss": 0.1122,
      "step": 3295
    },
    {
      "epoch": 6.269139324774132,
      "grad_norm": 0.30021417140960693,
      "learning_rate": 7.909812638932996e-05,
      "loss": 0.0918,
      "step": 3296
    },
    {
      "epoch": 6.271041369472183,
      "grad_norm": 0.3369525074958801,
      "learning_rate": 7.90917751667196e-05,
      "loss": 0.0921,
      "step": 3297
    },
    {
      "epoch": 6.272943414170233,
      "grad_norm": 0.28983327746391296,
      "learning_rate": 7.908542394410925e-05,
      "loss": 0.0581,
      "step": 3298
    },
    {
      "epoch": 6.274845458868284,
      "grad_norm": 0.30698540806770325,
      "learning_rate": 7.907907272149889e-05,
      "loss": 0.092,
      "step": 3299
    },
    {
      "epoch": 6.276747503566334,
      "grad_norm": 0.2095540314912796,
      "learning_rate": 7.907272149888854e-05,
      "loss": 0.0724,
      "step": 3300
    },
    {
      "epoch": 6.278649548264384,
      "grad_norm": 0.2522066831588745,
      "learning_rate": 7.906637027627819e-05,
      "loss": 0.0857,
      "step": 3301
    },
    {
      "epoch": 6.280551592962435,
      "grad_norm": 0.23340962827205658,
      "learning_rate": 7.906001905366783e-05,
      "loss": 0.0811,
      "step": 3302
    },
    {
      "epoch": 6.282453637660485,
      "grad_norm": 0.3235514760017395,
      "learning_rate": 7.90536678310575e-05,
      "loss": 0.1103,
      "step": 3303
    },
    {
      "epoch": 6.284355682358536,
      "grad_norm": 0.2366112768650055,
      "learning_rate": 7.904731660844713e-05,
      "loss": 0.0849,
      "step": 3304
    },
    {
      "epoch": 6.286257727056586,
      "grad_norm": 0.35640236735343933,
      "learning_rate": 7.904096538583677e-05,
      "loss": 0.0973,
      "step": 3305
    },
    {
      "epoch": 6.2881597717546365,
      "grad_norm": 0.2726953625679016,
      "learning_rate": 7.903461416322642e-05,
      "loss": 0.0867,
      "step": 3306
    },
    {
      "epoch": 6.2900618164526865,
      "grad_norm": 0.2399393916130066,
      "learning_rate": 7.902826294061607e-05,
      "loss": 0.0804,
      "step": 3307
    },
    {
      "epoch": 6.291963861150737,
      "grad_norm": 0.3313315510749817,
      "learning_rate": 7.902191171800571e-05,
      "loss": 0.1023,
      "step": 3308
    },
    {
      "epoch": 6.293865905848787,
      "grad_norm": 0.30253294110298157,
      "learning_rate": 7.901556049539536e-05,
      "loss": 0.0954,
      "step": 3309
    },
    {
      "epoch": 6.295767950546838,
      "grad_norm": 0.25532180070877075,
      "learning_rate": 7.900920927278502e-05,
      "loss": 0.0708,
      "step": 3310
    },
    {
      "epoch": 6.297669995244888,
      "grad_norm": 0.2780250012874603,
      "learning_rate": 7.900285805017467e-05,
      "loss": 0.0719,
      "step": 3311
    },
    {
      "epoch": 6.299572039942939,
      "grad_norm": 0.31086266040802,
      "learning_rate": 7.89965068275643e-05,
      "loss": 0.0848,
      "step": 3312
    },
    {
      "epoch": 6.301474084640989,
      "grad_norm": 0.2522285580635071,
      "learning_rate": 7.899015560495396e-05,
      "loss": 0.1076,
      "step": 3313
    },
    {
      "epoch": 6.303376129339039,
      "grad_norm": 0.3242189288139343,
      "learning_rate": 7.898380438234361e-05,
      "loss": 0.0824,
      "step": 3314
    },
    {
      "epoch": 6.30527817403709,
      "grad_norm": 0.19826237857341766,
      "learning_rate": 7.897745315973325e-05,
      "loss": 0.0506,
      "step": 3315
    },
    {
      "epoch": 6.30718021873514,
      "grad_norm": 0.295257568359375,
      "learning_rate": 7.89711019371229e-05,
      "loss": 0.0662,
      "step": 3316
    },
    {
      "epoch": 6.309082263433191,
      "grad_norm": 0.2735994756221771,
      "learning_rate": 7.896475071451255e-05,
      "loss": 0.0918,
      "step": 3317
    },
    {
      "epoch": 6.310984308131241,
      "grad_norm": 0.32766851782798767,
      "learning_rate": 7.895839949190219e-05,
      "loss": 0.087,
      "step": 3318
    },
    {
      "epoch": 6.312886352829292,
      "grad_norm": 0.3935610353946686,
      "learning_rate": 7.895204826929184e-05,
      "loss": 0.1329,
      "step": 3319
    },
    {
      "epoch": 6.314788397527342,
      "grad_norm": 0.2454293966293335,
      "learning_rate": 7.89456970466815e-05,
      "loss": 0.0798,
      "step": 3320
    },
    {
      "epoch": 6.316690442225393,
      "grad_norm": 0.20663952827453613,
      "learning_rate": 7.893934582407115e-05,
      "loss": 0.0635,
      "step": 3321
    },
    {
      "epoch": 6.318592486923443,
      "grad_norm": 0.3088918924331665,
      "learning_rate": 7.893299460146078e-05,
      "loss": 0.1099,
      "step": 3322
    },
    {
      "epoch": 6.3204945316214936,
      "grad_norm": 0.3906470239162445,
      "learning_rate": 7.892664337885042e-05,
      "loss": 0.0799,
      "step": 3323
    },
    {
      "epoch": 6.3223965763195435,
      "grad_norm": 0.2616283893585205,
      "learning_rate": 7.892029215624009e-05,
      "loss": 0.0622,
      "step": 3324
    },
    {
      "epoch": 6.3242986210175935,
      "grad_norm": 0.2966347932815552,
      "learning_rate": 7.891394093362973e-05,
      "loss": 0.1009,
      "step": 3325
    },
    {
      "epoch": 6.326200665715644,
      "grad_norm": 0.347552090883255,
      "learning_rate": 7.890758971101936e-05,
      "loss": 0.0819,
      "step": 3326
    },
    {
      "epoch": 6.328102710413694,
      "grad_norm": 0.30803078413009644,
      "learning_rate": 7.890123848840903e-05,
      "loss": 0.0717,
      "step": 3327
    },
    {
      "epoch": 6.330004755111745,
      "grad_norm": 0.4039364755153656,
      "learning_rate": 7.889488726579867e-05,
      "loss": 0.0984,
      "step": 3328
    },
    {
      "epoch": 6.331906799809795,
      "grad_norm": 0.39296120405197144,
      "learning_rate": 7.888853604318832e-05,
      "loss": 0.1066,
      "step": 3329
    },
    {
      "epoch": 6.333808844507846,
      "grad_norm": 0.25951987504959106,
      "learning_rate": 7.888218482057796e-05,
      "loss": 0.0894,
      "step": 3330
    },
    {
      "epoch": 6.335710889205896,
      "grad_norm": 0.3936977982521057,
      "learning_rate": 7.887583359796761e-05,
      "loss": 0.0938,
      "step": 3331
    },
    {
      "epoch": 6.337612933903947,
      "grad_norm": 0.17758914828300476,
      "learning_rate": 7.886948237535726e-05,
      "loss": 0.0618,
      "step": 3332
    },
    {
      "epoch": 6.339514978601997,
      "grad_norm": 0.27204954624176025,
      "learning_rate": 7.88631311527469e-05,
      "loss": 0.0906,
      "step": 3333
    },
    {
      "epoch": 6.341417023300048,
      "grad_norm": 0.3029746115207672,
      "learning_rate": 7.885677993013657e-05,
      "loss": 0.0834,
      "step": 3334
    },
    {
      "epoch": 6.343319067998098,
      "grad_norm": 0.33268237113952637,
      "learning_rate": 7.88504287075262e-05,
      "loss": 0.0922,
      "step": 3335
    },
    {
      "epoch": 6.345221112696148,
      "grad_norm": 0.37622734904289246,
      "learning_rate": 7.884407748491584e-05,
      "loss": 0.096,
      "step": 3336
    },
    {
      "epoch": 6.347123157394199,
      "grad_norm": 0.24578069150447845,
      "learning_rate": 7.88377262623055e-05,
      "loss": 0.0699,
      "step": 3337
    },
    {
      "epoch": 6.349025202092249,
      "grad_norm": 0.33049342036247253,
      "learning_rate": 7.883137503969515e-05,
      "loss": 0.0881,
      "step": 3338
    },
    {
      "epoch": 6.3509272467903,
      "grad_norm": 0.2978206276893616,
      "learning_rate": 7.88250238170848e-05,
      "loss": 0.0876,
      "step": 3339
    },
    {
      "epoch": 6.35282929148835,
      "grad_norm": 0.21608392894268036,
      "learning_rate": 7.881867259447444e-05,
      "loss": 0.0643,
      "step": 3340
    },
    {
      "epoch": 6.354731336186401,
      "grad_norm": 0.28561514616012573,
      "learning_rate": 7.881232137186409e-05,
      "loss": 0.0827,
      "step": 3341
    },
    {
      "epoch": 6.356633380884451,
      "grad_norm": 0.2658670246601105,
      "learning_rate": 7.880597014925374e-05,
      "loss": 0.0918,
      "step": 3342
    },
    {
      "epoch": 6.3585354255825015,
      "grad_norm": 0.31455472111701965,
      "learning_rate": 7.879961892664338e-05,
      "loss": 0.1127,
      "step": 3343
    },
    {
      "epoch": 6.3604374702805515,
      "grad_norm": 0.2520543038845062,
      "learning_rate": 7.879326770403303e-05,
      "loss": 0.074,
      "step": 3344
    },
    {
      "epoch": 6.362339514978602,
      "grad_norm": 0.3586767911911011,
      "learning_rate": 7.878691648142268e-05,
      "loss": 0.1097,
      "step": 3345
    },
    {
      "epoch": 6.364241559676652,
      "grad_norm": 0.23053278028964996,
      "learning_rate": 7.878056525881232e-05,
      "loss": 0.0923,
      "step": 3346
    },
    {
      "epoch": 6.366143604374702,
      "grad_norm": 0.31203171610832214,
      "learning_rate": 7.877421403620197e-05,
      "loss": 0.1194,
      "step": 3347
    },
    {
      "epoch": 6.368045649072753,
      "grad_norm": 0.22267888486385345,
      "learning_rate": 7.876786281359162e-05,
      "loss": 0.0686,
      "step": 3348
    },
    {
      "epoch": 6.369947693770803,
      "grad_norm": 0.30080437660217285,
      "learning_rate": 7.876151159098126e-05,
      "loss": 0.0883,
      "step": 3349
    },
    {
      "epoch": 6.371849738468854,
      "grad_norm": 0.3382682502269745,
      "learning_rate": 7.875516036837091e-05,
      "loss": 0.089,
      "step": 3350
    },
    {
      "epoch": 6.373751783166904,
      "grad_norm": 0.23919892311096191,
      "learning_rate": 7.874880914576057e-05,
      "loss": 0.0693,
      "step": 3351
    },
    {
      "epoch": 6.375653827864955,
      "grad_norm": 0.25442051887512207,
      "learning_rate": 7.874245792315022e-05,
      "loss": 0.0783,
      "step": 3352
    },
    {
      "epoch": 6.377555872563005,
      "grad_norm": 0.24807670712471008,
      "learning_rate": 7.873610670053986e-05,
      "loss": 0.0805,
      "step": 3353
    },
    {
      "epoch": 6.379457917261056,
      "grad_norm": 0.3315785229206085,
      "learning_rate": 7.872975547792951e-05,
      "loss": 0.0812,
      "step": 3354
    },
    {
      "epoch": 6.381359961959106,
      "grad_norm": 0.281994104385376,
      "learning_rate": 7.872340425531916e-05,
      "loss": 0.0945,
      "step": 3355
    },
    {
      "epoch": 6.383262006657157,
      "grad_norm": 0.4207977056503296,
      "learning_rate": 7.87170530327088e-05,
      "loss": 0.1041,
      "step": 3356
    },
    {
      "epoch": 6.385164051355207,
      "grad_norm": 0.2093304991722107,
      "learning_rate": 7.871070181009845e-05,
      "loss": 0.0682,
      "step": 3357
    },
    {
      "epoch": 6.387066096053257,
      "grad_norm": 0.29875820875167847,
      "learning_rate": 7.87043505874881e-05,
      "loss": 0.089,
      "step": 3358
    },
    {
      "epoch": 6.388968140751308,
      "grad_norm": 0.331050843000412,
      "learning_rate": 7.869799936487774e-05,
      "loss": 0.076,
      "step": 3359
    },
    {
      "epoch": 6.390870185449358,
      "grad_norm": 0.32891416549682617,
      "learning_rate": 7.869164814226739e-05,
      "loss": 0.0821,
      "step": 3360
    },
    {
      "epoch": 6.392772230147409,
      "grad_norm": 0.23887665569782257,
      "learning_rate": 7.868529691965704e-05,
      "loss": 0.0703,
      "step": 3361
    },
    {
      "epoch": 6.394674274845459,
      "grad_norm": 0.3243681490421295,
      "learning_rate": 7.867894569704668e-05,
      "loss": 0.0893,
      "step": 3362
    },
    {
      "epoch": 6.3965763195435095,
      "grad_norm": 0.3317023813724518,
      "learning_rate": 7.867259447443633e-05,
      "loss": 0.0913,
      "step": 3363
    },
    {
      "epoch": 6.3984783642415595,
      "grad_norm": 0.4032711982727051,
      "learning_rate": 7.866624325182597e-05,
      "loss": 0.1019,
      "step": 3364
    },
    {
      "epoch": 6.40038040893961,
      "grad_norm": 0.30233386158943176,
      "learning_rate": 7.865989202921564e-05,
      "loss": 0.0817,
      "step": 3365
    },
    {
      "epoch": 6.40228245363766,
      "grad_norm": 0.25192874670028687,
      "learning_rate": 7.865354080660528e-05,
      "loss": 0.0863,
      "step": 3366
    },
    {
      "epoch": 6.404184498335711,
      "grad_norm": 0.3983643651008606,
      "learning_rate": 7.864718958399491e-05,
      "loss": 0.1088,
      "step": 3367
    },
    {
      "epoch": 6.406086543033761,
      "grad_norm": 0.3095570206642151,
      "learning_rate": 7.864083836138457e-05,
      "loss": 0.0808,
      "step": 3368
    },
    {
      "epoch": 6.407988587731811,
      "grad_norm": 0.2907126843929291,
      "learning_rate": 7.863448713877422e-05,
      "loss": 0.07,
      "step": 3369
    },
    {
      "epoch": 6.409890632429862,
      "grad_norm": 0.2748839557170868,
      "learning_rate": 7.862813591616387e-05,
      "loss": 0.0884,
      "step": 3370
    },
    {
      "epoch": 6.411792677127912,
      "grad_norm": 0.23985274136066437,
      "learning_rate": 7.862178469355351e-05,
      "loss": 0.0732,
      "step": 3371
    },
    {
      "epoch": 6.413694721825963,
      "grad_norm": 0.2567084729671478,
      "learning_rate": 7.861543347094316e-05,
      "loss": 0.0873,
      "step": 3372
    },
    {
      "epoch": 6.415596766524013,
      "grad_norm": 0.3082403838634491,
      "learning_rate": 7.860908224833281e-05,
      "loss": 0.0741,
      "step": 3373
    },
    {
      "epoch": 6.417498811222064,
      "grad_norm": 0.1999639868736267,
      "learning_rate": 7.860273102572245e-05,
      "loss": 0.0628,
      "step": 3374
    },
    {
      "epoch": 6.419400855920114,
      "grad_norm": 0.34851139783859253,
      "learning_rate": 7.85963798031121e-05,
      "loss": 0.0915,
      "step": 3375
    },
    {
      "epoch": 6.421302900618165,
      "grad_norm": 0.2502918541431427,
      "learning_rate": 7.859002858050175e-05,
      "loss": 0.0695,
      "step": 3376
    },
    {
      "epoch": 6.423204945316215,
      "grad_norm": 0.2936602532863617,
      "learning_rate": 7.858367735789139e-05,
      "loss": 0.0759,
      "step": 3377
    },
    {
      "epoch": 6.425106990014266,
      "grad_norm": 0.3128640353679657,
      "learning_rate": 7.857732613528104e-05,
      "loss": 0.0733,
      "step": 3378
    },
    {
      "epoch": 6.427009034712316,
      "grad_norm": 0.26009759306907654,
      "learning_rate": 7.85709749126707e-05,
      "loss": 0.0962,
      "step": 3379
    },
    {
      "epoch": 6.4289110794103665,
      "grad_norm": 0.28206828236579895,
      "learning_rate": 7.856462369006033e-05,
      "loss": 0.0901,
      "step": 3380
    },
    {
      "epoch": 6.4308131241084165,
      "grad_norm": 0.2728763818740845,
      "learning_rate": 7.855827246744999e-05,
      "loss": 0.0657,
      "step": 3381
    },
    {
      "epoch": 6.4327151688064665,
      "grad_norm": 0.2516337037086487,
      "learning_rate": 7.855192124483964e-05,
      "loss": 0.0774,
      "step": 3382
    },
    {
      "epoch": 6.434617213504517,
      "grad_norm": 0.310434490442276,
      "learning_rate": 7.854557002222929e-05,
      "loss": 0.0999,
      "step": 3383
    },
    {
      "epoch": 6.436519258202567,
      "grad_norm": 0.2987268567085266,
      "learning_rate": 7.853921879961893e-05,
      "loss": 0.0902,
      "step": 3384
    },
    {
      "epoch": 6.438421302900618,
      "grad_norm": 0.2936643362045288,
      "learning_rate": 7.853286757700858e-05,
      "loss": 0.0761,
      "step": 3385
    },
    {
      "epoch": 6.440323347598668,
      "grad_norm": 0.292655348777771,
      "learning_rate": 7.852651635439823e-05,
      "loss": 0.0755,
      "step": 3386
    },
    {
      "epoch": 6.442225392296719,
      "grad_norm": 0.24866874516010284,
      "learning_rate": 7.852016513178787e-05,
      "loss": 0.0846,
      "step": 3387
    },
    {
      "epoch": 6.444127436994769,
      "grad_norm": 0.29063111543655396,
      "learning_rate": 7.851381390917752e-05,
      "loss": 0.0897,
      "step": 3388
    },
    {
      "epoch": 6.44602948169282,
      "grad_norm": 0.41345977783203125,
      "learning_rate": 7.850746268656717e-05,
      "loss": 0.0922,
      "step": 3389
    },
    {
      "epoch": 6.44793152639087,
      "grad_norm": 0.2677401006221771,
      "learning_rate": 7.850111146395681e-05,
      "loss": 0.0764,
      "step": 3390
    },
    {
      "epoch": 6.449833571088921,
      "grad_norm": 0.2827543318271637,
      "learning_rate": 7.849476024134646e-05,
      "loss": 0.0838,
      "step": 3391
    },
    {
      "epoch": 6.451735615786971,
      "grad_norm": 0.241709366440773,
      "learning_rate": 7.848840901873611e-05,
      "loss": 0.0747,
      "step": 3392
    },
    {
      "epoch": 6.453637660485022,
      "grad_norm": 0.3448995053768158,
      "learning_rate": 7.848205779612577e-05,
      "loss": 0.0882,
      "step": 3393
    },
    {
      "epoch": 6.455539705183072,
      "grad_norm": 0.2734489440917969,
      "learning_rate": 7.84757065735154e-05,
      "loss": 0.0764,
      "step": 3394
    },
    {
      "epoch": 6.457441749881122,
      "grad_norm": 0.21918603777885437,
      "learning_rate": 7.846935535090504e-05,
      "loss": 0.0809,
      "step": 3395
    },
    {
      "epoch": 6.459343794579173,
      "grad_norm": 0.2477739155292511,
      "learning_rate": 7.846300412829471e-05,
      "loss": 0.0807,
      "step": 3396
    },
    {
      "epoch": 6.461245839277223,
      "grad_norm": 0.21536344289779663,
      "learning_rate": 7.845665290568435e-05,
      "loss": 0.0632,
      "step": 3397
    },
    {
      "epoch": 6.463147883975274,
      "grad_norm": 0.2905837297439575,
      "learning_rate": 7.845030168307399e-05,
      "loss": 0.1073,
      "step": 3398
    },
    {
      "epoch": 6.465049928673324,
      "grad_norm": 0.29049018025398254,
      "learning_rate": 7.844395046046365e-05,
      "loss": 0.0915,
      "step": 3399
    },
    {
      "epoch": 6.4669519733713745,
      "grad_norm": 0.21640491485595703,
      "learning_rate": 7.843759923785329e-05,
      "loss": 0.0597,
      "step": 3400
    },
    {
      "epoch": 6.4688540180694245,
      "grad_norm": 0.255434513092041,
      "learning_rate": 7.843124801524294e-05,
      "loss": 0.0772,
      "step": 3401
    },
    {
      "epoch": 6.470756062767475,
      "grad_norm": 0.1592421680688858,
      "learning_rate": 7.842489679263258e-05,
      "loss": 0.0605,
      "step": 3402
    },
    {
      "epoch": 6.472658107465525,
      "grad_norm": 0.285553902387619,
      "learning_rate": 7.841854557002223e-05,
      "loss": 0.0683,
      "step": 3403
    },
    {
      "epoch": 6.474560152163576,
      "grad_norm": 0.20020133256912231,
      "learning_rate": 7.841219434741188e-05,
      "loss": 0.1138,
      "step": 3404
    },
    {
      "epoch": 6.476462196861626,
      "grad_norm": 0.3204385042190552,
      "learning_rate": 7.840584312480152e-05,
      "loss": 0.0766,
      "step": 3405
    },
    {
      "epoch": 6.478364241559676,
      "grad_norm": 0.28798025846481323,
      "learning_rate": 7.839949190219119e-05,
      "loss": 0.0883,
      "step": 3406
    },
    {
      "epoch": 6.480266286257727,
      "grad_norm": 0.3688175082206726,
      "learning_rate": 7.839314067958082e-05,
      "loss": 0.1036,
      "step": 3407
    },
    {
      "epoch": 6.482168330955777,
      "grad_norm": 0.35191601514816284,
      "learning_rate": 7.838678945697046e-05,
      "loss": 0.0864,
      "step": 3408
    },
    {
      "epoch": 6.484070375653828,
      "grad_norm": 0.2684312164783478,
      "learning_rate": 7.838043823436011e-05,
      "loss": 0.0712,
      "step": 3409
    },
    {
      "epoch": 6.485972420351878,
      "grad_norm": 0.39929521083831787,
      "learning_rate": 7.837408701174977e-05,
      "loss": 0.0861,
      "step": 3410
    },
    {
      "epoch": 6.487874465049929,
      "grad_norm": 0.26154014468193054,
      "learning_rate": 7.836773578913942e-05,
      "loss": 0.0704,
      "step": 3411
    },
    {
      "epoch": 6.489776509747979,
      "grad_norm": 0.2874952554702759,
      "learning_rate": 7.836138456652906e-05,
      "loss": 0.0804,
      "step": 3412
    },
    {
      "epoch": 6.49167855444603,
      "grad_norm": 0.2434820979833603,
      "learning_rate": 7.835503334391871e-05,
      "loss": 0.0797,
      "step": 3413
    },
    {
      "epoch": 6.49358059914408,
      "grad_norm": 0.23986493051052094,
      "learning_rate": 7.834868212130836e-05,
      "loss": 0.0598,
      "step": 3414
    },
    {
      "epoch": 6.495482643842131,
      "grad_norm": 0.2969447672367096,
      "learning_rate": 7.8342330898698e-05,
      "loss": 0.0827,
      "step": 3415
    },
    {
      "epoch": 6.497384688540181,
      "grad_norm": 0.27279725670814514,
      "learning_rate": 7.833597967608765e-05,
      "loss": 0.094,
      "step": 3416
    },
    {
      "epoch": 6.499286733238231,
      "grad_norm": 0.29611873626708984,
      "learning_rate": 7.83296284534773e-05,
      "loss": 0.0643,
      "step": 3417
    },
    {
      "epoch": 6.501188777936282,
      "grad_norm": 0.28763914108276367,
      "learning_rate": 7.832327723086694e-05,
      "loss": 0.0707,
      "step": 3418
    },
    {
      "epoch": 6.503090822634332,
      "grad_norm": 0.23590637743473053,
      "learning_rate": 7.831692600825659e-05,
      "loss": 0.0741,
      "step": 3419
    },
    {
      "epoch": 6.5049928673323825,
      "grad_norm": 0.3088235557079315,
      "learning_rate": 7.831057478564624e-05,
      "loss": 0.0766,
      "step": 3420
    },
    {
      "epoch": 6.5068949120304325,
      "grad_norm": 0.24730335175991058,
      "learning_rate": 7.830422356303588e-05,
      "loss": 0.0712,
      "step": 3421
    },
    {
      "epoch": 6.508796956728483,
      "grad_norm": 0.29086196422576904,
      "learning_rate": 7.829787234042553e-05,
      "loss": 0.0842,
      "step": 3422
    },
    {
      "epoch": 6.510699001426533,
      "grad_norm": 0.27357545495033264,
      "learning_rate": 7.829152111781519e-05,
      "loss": 0.0693,
      "step": 3423
    },
    {
      "epoch": 6.512601046124584,
      "grad_norm": 0.2787666618824005,
      "learning_rate": 7.828516989520484e-05,
      "loss": 0.0762,
      "step": 3424
    },
    {
      "epoch": 6.514503090822634,
      "grad_norm": 0.27191832661628723,
      "learning_rate": 7.827881867259448e-05,
      "loss": 0.0744,
      "step": 3425
    },
    {
      "epoch": 6.516405135520685,
      "grad_norm": 0.3606361448764801,
      "learning_rate": 7.827246744998411e-05,
      "loss": 0.0976,
      "step": 3426
    },
    {
      "epoch": 6.518307180218735,
      "grad_norm": 0.25823327898979187,
      "learning_rate": 7.826611622737378e-05,
      "loss": 0.0733,
      "step": 3427
    },
    {
      "epoch": 6.520209224916785,
      "grad_norm": 0.38170698285102844,
      "learning_rate": 7.825976500476342e-05,
      "loss": 0.0961,
      "step": 3428
    },
    {
      "epoch": 6.522111269614836,
      "grad_norm": 0.27328750491142273,
      "learning_rate": 7.825341378215307e-05,
      "loss": 0.0809,
      "step": 3429
    },
    {
      "epoch": 6.524013314312886,
      "grad_norm": 0.2610822916030884,
      "learning_rate": 7.824706255954272e-05,
      "loss": 0.0631,
      "step": 3430
    },
    {
      "epoch": 6.525915359010937,
      "grad_norm": 0.21520115435123444,
      "learning_rate": 7.824071133693236e-05,
      "loss": 0.0815,
      "step": 3431
    },
    {
      "epoch": 6.527817403708987,
      "grad_norm": 0.36525583267211914,
      "learning_rate": 7.823436011432201e-05,
      "loss": 0.0979,
      "step": 3432
    },
    {
      "epoch": 6.529719448407038,
      "grad_norm": 0.31419306993484497,
      "learning_rate": 7.822800889171165e-05,
      "loss": 0.1211,
      "step": 3433
    },
    {
      "epoch": 6.531621493105088,
      "grad_norm": 0.2525794804096222,
      "learning_rate": 7.82216576691013e-05,
      "loss": 0.0918,
      "step": 3434
    },
    {
      "epoch": 6.533523537803139,
      "grad_norm": 0.22929687798023224,
      "learning_rate": 7.821530644649095e-05,
      "loss": 0.058,
      "step": 3435
    },
    {
      "epoch": 6.535425582501189,
      "grad_norm": 0.36891230940818787,
      "learning_rate": 7.820895522388059e-05,
      "loss": 0.1044,
      "step": 3436
    },
    {
      "epoch": 6.5373276271992395,
      "grad_norm": 0.2763042151927948,
      "learning_rate": 7.820260400127026e-05,
      "loss": 0.1427,
      "step": 3437
    },
    {
      "epoch": 6.5392296718972895,
      "grad_norm": 0.34488481283187866,
      "learning_rate": 7.81962527786599e-05,
      "loss": 0.0815,
      "step": 3438
    },
    {
      "epoch": 6.5411317165953395,
      "grad_norm": 0.28253301978111267,
      "learning_rate": 7.818990155604953e-05,
      "loss": 0.0802,
      "step": 3439
    },
    {
      "epoch": 6.54303376129339,
      "grad_norm": 0.24336297810077667,
      "learning_rate": 7.818355033343919e-05,
      "loss": 0.0762,
      "step": 3440
    },
    {
      "epoch": 6.54493580599144,
      "grad_norm": 0.2575397491455078,
      "learning_rate": 7.817719911082884e-05,
      "loss": 0.0927,
      "step": 3441
    },
    {
      "epoch": 6.546837850689491,
      "grad_norm": 0.3071526885032654,
      "learning_rate": 7.817084788821849e-05,
      "loss": 0.0873,
      "step": 3442
    },
    {
      "epoch": 6.548739895387541,
      "grad_norm": 0.3540725111961365,
      "learning_rate": 7.816449666560813e-05,
      "loss": 0.0962,
      "step": 3443
    },
    {
      "epoch": 6.550641940085592,
      "grad_norm": 0.20820970833301544,
      "learning_rate": 7.815814544299778e-05,
      "loss": 0.0623,
      "step": 3444
    },
    {
      "epoch": 6.552543984783642,
      "grad_norm": 0.24989038705825806,
      "learning_rate": 7.815179422038743e-05,
      "loss": 0.0983,
      "step": 3445
    },
    {
      "epoch": 6.554446029481693,
      "grad_norm": 0.2294490784406662,
      "learning_rate": 7.814544299777707e-05,
      "loss": 0.0822,
      "step": 3446
    },
    {
      "epoch": 6.556348074179743,
      "grad_norm": 0.25477972626686096,
      "learning_rate": 7.813909177516672e-05,
      "loss": 0.0552,
      "step": 3447
    },
    {
      "epoch": 6.558250118877794,
      "grad_norm": 0.2907105088233948,
      "learning_rate": 7.813274055255637e-05,
      "loss": 0.0881,
      "step": 3448
    },
    {
      "epoch": 6.560152163575844,
      "grad_norm": 0.3188037574291229,
      "learning_rate": 7.812638932994601e-05,
      "loss": 0.0877,
      "step": 3449
    },
    {
      "epoch": 6.562054208273894,
      "grad_norm": 0.3072590231895447,
      "learning_rate": 7.812003810733566e-05,
      "loss": 0.0989,
      "step": 3450
    },
    {
      "epoch": 6.563956252971945,
      "grad_norm": 0.44314712285995483,
      "learning_rate": 7.811368688472532e-05,
      "loss": 0.0977,
      "step": 3451
    },
    {
      "epoch": 6.565858297669996,
      "grad_norm": 0.2874956727027893,
      "learning_rate": 7.810733566211495e-05,
      "loss": 0.0909,
      "step": 3452
    },
    {
      "epoch": 6.567760342368046,
      "grad_norm": 0.32697319984436035,
      "learning_rate": 7.81009844395046e-05,
      "loss": 0.1043,
      "step": 3453
    },
    {
      "epoch": 6.569662387066096,
      "grad_norm": 0.28816571831703186,
      "learning_rate": 7.809463321689426e-05,
      "loss": 0.0944,
      "step": 3454
    },
    {
      "epoch": 6.571564431764147,
      "grad_norm": 0.4285229742527008,
      "learning_rate": 7.808828199428391e-05,
      "loss": 0.1251,
      "step": 3455
    },
    {
      "epoch": 6.573466476462197,
      "grad_norm": 0.30767038464546204,
      "learning_rate": 7.808193077167355e-05,
      "loss": 0.0864,
      "step": 3456
    },
    {
      "epoch": 6.5753685211602475,
      "grad_norm": 0.3375704884529114,
      "learning_rate": 7.80755795490632e-05,
      "loss": 0.0861,
      "step": 3457
    },
    {
      "epoch": 6.5772705658582975,
      "grad_norm": 0.31623944640159607,
      "learning_rate": 7.806922832645285e-05,
      "loss": 0.0706,
      "step": 3458
    },
    {
      "epoch": 6.579172610556348,
      "grad_norm": 0.3341452181339264,
      "learning_rate": 7.806287710384249e-05,
      "loss": 0.0946,
      "step": 3459
    },
    {
      "epoch": 6.581074655254398,
      "grad_norm": 0.32458797097206116,
      "learning_rate": 7.805652588123214e-05,
      "loss": 0.0976,
      "step": 3460
    },
    {
      "epoch": 6.582976699952448,
      "grad_norm": 0.2971061170101166,
      "learning_rate": 7.80501746586218e-05,
      "loss": 0.0838,
      "step": 3461
    },
    {
      "epoch": 6.584878744650499,
      "grad_norm": 0.29806768894195557,
      "learning_rate": 7.804382343601143e-05,
      "loss": 0.0719,
      "step": 3462
    },
    {
      "epoch": 6.58678078934855,
      "grad_norm": 0.268477201461792,
      "learning_rate": 7.803747221340108e-05,
      "loss": 0.0819,
      "step": 3463
    },
    {
      "epoch": 6.5886828340466,
      "grad_norm": 0.3286268413066864,
      "learning_rate": 7.803112099079074e-05,
      "loss": 0.0919,
      "step": 3464
    },
    {
      "epoch": 6.59058487874465,
      "grad_norm": 0.30549558997154236,
      "learning_rate": 7.802476976818039e-05,
      "loss": 0.0784,
      "step": 3465
    },
    {
      "epoch": 6.592486923442701,
      "grad_norm": 0.3088127374649048,
      "learning_rate": 7.801841854557003e-05,
      "loss": 0.0894,
      "step": 3466
    },
    {
      "epoch": 6.594388968140751,
      "grad_norm": 0.29529303312301636,
      "learning_rate": 7.801206732295966e-05,
      "loss": 0.0945,
      "step": 3467
    },
    {
      "epoch": 6.596291012838802,
      "grad_norm": 0.33194759488105774,
      "learning_rate": 7.800571610034933e-05,
      "loss": 0.0999,
      "step": 3468
    },
    {
      "epoch": 6.598193057536852,
      "grad_norm": 0.3072373867034912,
      "learning_rate": 7.799936487773897e-05,
      "loss": 0.0819,
      "step": 3469
    },
    {
      "epoch": 6.600095102234903,
      "grad_norm": 0.3042803704738617,
      "learning_rate": 7.79930136551286e-05,
      "loss": 0.0919,
      "step": 3470
    },
    {
      "epoch": 6.601997146932953,
      "grad_norm": 0.3292810618877411,
      "learning_rate": 7.798666243251827e-05,
      "loss": 0.0915,
      "step": 3471
    },
    {
      "epoch": 6.603899191631004,
      "grad_norm": 0.36252084374427795,
      "learning_rate": 7.798031120990791e-05,
      "loss": 0.0707,
      "step": 3472
    },
    {
      "epoch": 6.605801236329054,
      "grad_norm": 0.23440448939800262,
      "learning_rate": 7.797395998729756e-05,
      "loss": 0.0692,
      "step": 3473
    },
    {
      "epoch": 6.607703281027105,
      "grad_norm": 0.3150375187397003,
      "learning_rate": 7.79676087646872e-05,
      "loss": 0.0979,
      "step": 3474
    },
    {
      "epoch": 6.609605325725155,
      "grad_norm": 0.2594108581542969,
      "learning_rate": 7.796125754207685e-05,
      "loss": 0.1073,
      "step": 3475
    },
    {
      "epoch": 6.611507370423205,
      "grad_norm": 0.25529977679252625,
      "learning_rate": 7.79549063194665e-05,
      "loss": 0.0907,
      "step": 3476
    },
    {
      "epoch": 6.6134094151212555,
      "grad_norm": 0.2744307219982147,
      "learning_rate": 7.794855509685614e-05,
      "loss": 0.066,
      "step": 3477
    },
    {
      "epoch": 6.6153114598193055,
      "grad_norm": 0.26056280732154846,
      "learning_rate": 7.79422038742458e-05,
      "loss": 0.0883,
      "step": 3478
    },
    {
      "epoch": 6.617213504517356,
      "grad_norm": 0.27727293968200684,
      "learning_rate": 7.793585265163545e-05,
      "loss": 0.1024,
      "step": 3479
    },
    {
      "epoch": 6.619115549215406,
      "grad_norm": 0.31950679421424866,
      "learning_rate": 7.792950142902508e-05,
      "loss": 0.0799,
      "step": 3480
    },
    {
      "epoch": 6.621017593913457,
      "grad_norm": 0.301362007856369,
      "learning_rate": 7.792315020641474e-05,
      "loss": 0.0885,
      "step": 3481
    },
    {
      "epoch": 6.622919638611507,
      "grad_norm": 0.3070707321166992,
      "learning_rate": 7.791679898380439e-05,
      "loss": 0.0851,
      "step": 3482
    },
    {
      "epoch": 6.624821683309558,
      "grad_norm": 0.39499345421791077,
      "learning_rate": 7.791044776119404e-05,
      "loss": 0.1236,
      "step": 3483
    },
    {
      "epoch": 6.626723728007608,
      "grad_norm": 0.30052968859672546,
      "learning_rate": 7.790409653858368e-05,
      "loss": 0.0761,
      "step": 3484
    },
    {
      "epoch": 6.628625772705659,
      "grad_norm": 0.2883196175098419,
      "learning_rate": 7.789774531597333e-05,
      "loss": 0.0718,
      "step": 3485
    },
    {
      "epoch": 6.630527817403709,
      "grad_norm": 0.24781768023967743,
      "learning_rate": 7.789139409336298e-05,
      "loss": 0.0866,
      "step": 3486
    },
    {
      "epoch": 6.632429862101759,
      "grad_norm": 0.3658462464809418,
      "learning_rate": 7.788504287075262e-05,
      "loss": 0.0763,
      "step": 3487
    },
    {
      "epoch": 6.63433190679981,
      "grad_norm": 0.24517424404621124,
      "learning_rate": 7.787869164814227e-05,
      "loss": 0.0734,
      "step": 3488
    },
    {
      "epoch": 6.63623395149786,
      "grad_norm": 0.36587032675743103,
      "learning_rate": 7.787234042553192e-05,
      "loss": 0.1221,
      "step": 3489
    },
    {
      "epoch": 6.638135996195911,
      "grad_norm": 0.3201904892921448,
      "learning_rate": 7.786598920292156e-05,
      "loss": 0.0748,
      "step": 3490
    },
    {
      "epoch": 6.640038040893961,
      "grad_norm": 0.3514649569988251,
      "learning_rate": 7.785963798031121e-05,
      "loss": 0.0971,
      "step": 3491
    },
    {
      "epoch": 6.641940085592012,
      "grad_norm": 0.25201430916786194,
      "learning_rate": 7.785328675770087e-05,
      "loss": 0.071,
      "step": 3492
    },
    {
      "epoch": 6.643842130290062,
      "grad_norm": 0.21259263157844543,
      "learning_rate": 7.78469355350905e-05,
      "loss": 0.0692,
      "step": 3493
    },
    {
      "epoch": 6.6457441749881125,
      "grad_norm": 0.32301944494247437,
      "learning_rate": 7.784058431248016e-05,
      "loss": 0.0798,
      "step": 3494
    },
    {
      "epoch": 6.6476462196861625,
      "grad_norm": 0.32042115926742554,
      "learning_rate": 7.783423308986981e-05,
      "loss": 0.0846,
      "step": 3495
    },
    {
      "epoch": 6.649548264384213,
      "grad_norm": 0.294508695602417,
      "learning_rate": 7.782788186725946e-05,
      "loss": 0.0849,
      "step": 3496
    },
    {
      "epoch": 6.651450309082263,
      "grad_norm": 0.3282386064529419,
      "learning_rate": 7.78215306446491e-05,
      "loss": 0.0823,
      "step": 3497
    },
    {
      "epoch": 6.653352353780313,
      "grad_norm": 0.3124896287918091,
      "learning_rate": 7.781517942203874e-05,
      "loss": 0.0835,
      "step": 3498
    },
    {
      "epoch": 6.655254398478364,
      "grad_norm": 0.38605186343193054,
      "learning_rate": 7.78088281994284e-05,
      "loss": 0.0977,
      "step": 3499
    },
    {
      "epoch": 6.657156443176414,
      "grad_norm": 0.2927373945713043,
      "learning_rate": 7.780247697681804e-05,
      "loss": 0.0835,
      "step": 3500
    },
    {
      "epoch": 6.659058487874465,
      "grad_norm": 0.2956550121307373,
      "learning_rate": 7.779612575420769e-05,
      "loss": 0.0895,
      "step": 3501
    },
    {
      "epoch": 6.660960532572515,
      "grad_norm": 0.3334302008152008,
      "learning_rate": 7.778977453159734e-05,
      "loss": 0.0699,
      "step": 3502
    },
    {
      "epoch": 6.662862577270566,
      "grad_norm": 0.2574530839920044,
      "learning_rate": 7.778342330898698e-05,
      "loss": 0.0815,
      "step": 3503
    },
    {
      "epoch": 6.664764621968616,
      "grad_norm": 0.2664775252342224,
      "learning_rate": 7.777707208637663e-05,
      "loss": 0.0837,
      "step": 3504
    },
    {
      "epoch": 6.666666666666667,
      "grad_norm": 0.25906041264533997,
      "learning_rate": 7.777072086376627e-05,
      "loss": 0.0719,
      "step": 3505
    },
    {
      "epoch": 6.668568711364717,
      "grad_norm": 0.373308390378952,
      "learning_rate": 7.776436964115592e-05,
      "loss": 0.1048,
      "step": 3506
    },
    {
      "epoch": 6.670470756062768,
      "grad_norm": 0.2534428536891937,
      "learning_rate": 7.775801841854558e-05,
      "loss": 0.0784,
      "step": 3507
    },
    {
      "epoch": 6.672372800760818,
      "grad_norm": 0.33284425735473633,
      "learning_rate": 7.775166719593521e-05,
      "loss": 0.0953,
      "step": 3508
    },
    {
      "epoch": 6.674274845458868,
      "grad_norm": 0.3962637484073639,
      "learning_rate": 7.774531597332488e-05,
      "loss": 0.0963,
      "step": 3509
    },
    {
      "epoch": 6.676176890156919,
      "grad_norm": 0.2589205503463745,
      "learning_rate": 7.773896475071452e-05,
      "loss": 0.0747,
      "step": 3510
    },
    {
      "epoch": 6.678078934854969,
      "grad_norm": 0.3683101534843445,
      "learning_rate": 7.773261352810416e-05,
      "loss": 0.1112,
      "step": 3511
    },
    {
      "epoch": 6.67998097955302,
      "grad_norm": 0.3012533187866211,
      "learning_rate": 7.772626230549381e-05,
      "loss": 0.0823,
      "step": 3512
    },
    {
      "epoch": 6.68188302425107,
      "grad_norm": 0.23834310472011566,
      "learning_rate": 7.771991108288346e-05,
      "loss": 0.0791,
      "step": 3513
    },
    {
      "epoch": 6.6837850689491205,
      "grad_norm": 0.28065311908721924,
      "learning_rate": 7.771355986027311e-05,
      "loss": 0.0816,
      "step": 3514
    },
    {
      "epoch": 6.6856871136471705,
      "grad_norm": 0.3004380166530609,
      "learning_rate": 7.770720863766275e-05,
      "loss": 0.0958,
      "step": 3515
    },
    {
      "epoch": 6.687589158345221,
      "grad_norm": 0.32143503427505493,
      "learning_rate": 7.77008574150524e-05,
      "loss": 0.0897,
      "step": 3516
    },
    {
      "epoch": 6.689491203043271,
      "grad_norm": 0.3012734055519104,
      "learning_rate": 7.769450619244205e-05,
      "loss": 0.0783,
      "step": 3517
    },
    {
      "epoch": 6.691393247741322,
      "grad_norm": 0.2757457494735718,
      "learning_rate": 7.768815496983169e-05,
      "loss": 0.0877,
      "step": 3518
    },
    {
      "epoch": 6.693295292439372,
      "grad_norm": 0.3059716820716858,
      "learning_rate": 7.768180374722134e-05,
      "loss": 0.0805,
      "step": 3519
    },
    {
      "epoch": 6.695197337137422,
      "grad_norm": 0.26920875906944275,
      "learning_rate": 7.7675452524611e-05,
      "loss": 0.0753,
      "step": 3520
    },
    {
      "epoch": 6.697099381835473,
      "grad_norm": 0.30633947253227234,
      "learning_rate": 7.766910130200063e-05,
      "loss": 0.071,
      "step": 3521
    },
    {
      "epoch": 6.699001426533523,
      "grad_norm": 0.39395442605018616,
      "learning_rate": 7.766275007939029e-05,
      "loss": 0.1016,
      "step": 3522
    },
    {
      "epoch": 6.700903471231574,
      "grad_norm": 0.40755128860473633,
      "learning_rate": 7.765639885677994e-05,
      "loss": 0.0955,
      "step": 3523
    },
    {
      "epoch": 6.702805515929624,
      "grad_norm": 0.26808515191078186,
      "learning_rate": 7.765004763416958e-05,
      "loss": 0.0732,
      "step": 3524
    },
    {
      "epoch": 6.704707560627675,
      "grad_norm": 0.30976927280426025,
      "learning_rate": 7.764369641155923e-05,
      "loss": 0.0895,
      "step": 3525
    },
    {
      "epoch": 6.706609605325725,
      "grad_norm": 0.3644852936267853,
      "learning_rate": 7.763734518894888e-05,
      "loss": 0.1035,
      "step": 3526
    },
    {
      "epoch": 6.708511650023776,
      "grad_norm": 0.37868261337280273,
      "learning_rate": 7.763099396633853e-05,
      "loss": 0.1074,
      "step": 3527
    },
    {
      "epoch": 6.710413694721826,
      "grad_norm": 0.3462066948413849,
      "learning_rate": 7.762464274372817e-05,
      "loss": 0.1004,
      "step": 3528
    },
    {
      "epoch": 6.712315739419877,
      "grad_norm": 0.4035661518573761,
      "learning_rate": 7.761829152111782e-05,
      "loss": 0.0891,
      "step": 3529
    },
    {
      "epoch": 6.714217784117927,
      "grad_norm": 0.26760900020599365,
      "learning_rate": 7.761194029850747e-05,
      "loss": 0.0834,
      "step": 3530
    },
    {
      "epoch": 6.716119828815977,
      "grad_norm": 0.3930485248565674,
      "learning_rate": 7.760558907589711e-05,
      "loss": 0.0976,
      "step": 3531
    },
    {
      "epoch": 6.718021873514028,
      "grad_norm": 0.24880844354629517,
      "learning_rate": 7.759923785328676e-05,
      "loss": 0.0886,
      "step": 3532
    },
    {
      "epoch": 6.7199239182120785,
      "grad_norm": 0.3307024836540222,
      "learning_rate": 7.759288663067641e-05,
      "loss": 0.0947,
      "step": 3533
    },
    {
      "epoch": 6.7218259629101285,
      "grad_norm": 0.34674742817878723,
      "learning_rate": 7.758653540806605e-05,
      "loss": 0.0987,
      "step": 3534
    },
    {
      "epoch": 6.7237280076081785,
      "grad_norm": 0.2853250801563263,
      "learning_rate": 7.75801841854557e-05,
      "loss": 0.0748,
      "step": 3535
    },
    {
      "epoch": 6.725630052306229,
      "grad_norm": 0.23212608695030212,
      "learning_rate": 7.757383296284534e-05,
      "loss": 0.0716,
      "step": 3536
    },
    {
      "epoch": 6.727532097004279,
      "grad_norm": 0.3743533194065094,
      "learning_rate": 7.756748174023501e-05,
      "loss": 0.0903,
      "step": 3537
    },
    {
      "epoch": 6.72943414170233,
      "grad_norm": 0.2562249004840851,
      "learning_rate": 7.756113051762465e-05,
      "loss": 0.0719,
      "step": 3538
    },
    {
      "epoch": 6.73133618640038,
      "grad_norm": 0.29872533679008484,
      "learning_rate": 7.755477929501429e-05,
      "loss": 0.0983,
      "step": 3539
    },
    {
      "epoch": 6.733238231098431,
      "grad_norm": 0.46953633427619934,
      "learning_rate": 7.754842807240395e-05,
      "loss": 0.1268,
      "step": 3540
    },
    {
      "epoch": 6.735140275796481,
      "grad_norm": 0.3572865128517151,
      "learning_rate": 7.754207684979359e-05,
      "loss": 0.1016,
      "step": 3541
    },
    {
      "epoch": 6.737042320494532,
      "grad_norm": 0.27556324005126953,
      "learning_rate": 7.753572562718323e-05,
      "loss": 0.0873,
      "step": 3542
    },
    {
      "epoch": 6.738944365192582,
      "grad_norm": 0.3115139305591583,
      "learning_rate": 7.752937440457288e-05,
      "loss": 0.0879,
      "step": 3543
    },
    {
      "epoch": 6.740846409890633,
      "grad_norm": 0.2544573247432709,
      "learning_rate": 7.752302318196253e-05,
      "loss": 0.0824,
      "step": 3544
    },
    {
      "epoch": 6.742748454588683,
      "grad_norm": 0.3366973400115967,
      "learning_rate": 7.751667195935218e-05,
      "loss": 0.0912,
      "step": 3545
    },
    {
      "epoch": 6.744650499286733,
      "grad_norm": 0.23436006903648376,
      "learning_rate": 7.751032073674182e-05,
      "loss": 0.0882,
      "step": 3546
    },
    {
      "epoch": 6.746552543984784,
      "grad_norm": 0.41576170921325684,
      "learning_rate": 7.750396951413147e-05,
      "loss": 0.098,
      "step": 3547
    },
    {
      "epoch": 6.748454588682834,
      "grad_norm": 0.23372437059879303,
      "learning_rate": 7.749761829152112e-05,
      "loss": 0.0604,
      "step": 3548
    },
    {
      "epoch": 6.750356633380885,
      "grad_norm": 0.42242592573165894,
      "learning_rate": 7.749126706891076e-05,
      "loss": 0.1247,
      "step": 3549
    },
    {
      "epoch": 6.752258678078935,
      "grad_norm": 0.19408640265464783,
      "learning_rate": 7.748491584630041e-05,
      "loss": 0.0514,
      "step": 3550
    },
    {
      "epoch": 6.7541607227769855,
      "grad_norm": 0.30913442373275757,
      "learning_rate": 7.747856462369007e-05,
      "loss": 0.0785,
      "step": 3551
    },
    {
      "epoch": 6.7560627674750355,
      "grad_norm": 0.3456955850124359,
      "learning_rate": 7.74722134010797e-05,
      "loss": 0.0922,
      "step": 3552
    },
    {
      "epoch": 6.757964812173086,
      "grad_norm": 0.2908725142478943,
      "learning_rate": 7.746586217846936e-05,
      "loss": 0.0837,
      "step": 3553
    },
    {
      "epoch": 6.759866856871136,
      "grad_norm": 0.32404351234436035,
      "learning_rate": 7.745951095585901e-05,
      "loss": 0.0813,
      "step": 3554
    },
    {
      "epoch": 6.761768901569187,
      "grad_norm": 0.3528265357017517,
      "learning_rate": 7.745315973324866e-05,
      "loss": 0.0952,
      "step": 3555
    },
    {
      "epoch": 6.763670946267237,
      "grad_norm": 0.3190256655216217,
      "learning_rate": 7.74468085106383e-05,
      "loss": 0.0753,
      "step": 3556
    },
    {
      "epoch": 6.765572990965287,
      "grad_norm": 0.36666131019592285,
      "learning_rate": 7.744045728802795e-05,
      "loss": 0.0986,
      "step": 3557
    },
    {
      "epoch": 6.767475035663338,
      "grad_norm": 0.3192061185836792,
      "learning_rate": 7.74341060654176e-05,
      "loss": 0.0659,
      "step": 3558
    },
    {
      "epoch": 6.769377080361388,
      "grad_norm": 0.3112170696258545,
      "learning_rate": 7.742775484280724e-05,
      "loss": 0.0783,
      "step": 3559
    },
    {
      "epoch": 6.771279125059439,
      "grad_norm": 0.30128562450408936,
      "learning_rate": 7.742140362019689e-05,
      "loss": 0.078,
      "step": 3560
    },
    {
      "epoch": 6.773181169757489,
      "grad_norm": 0.31869345903396606,
      "learning_rate": 7.741505239758654e-05,
      "loss": 0.0948,
      "step": 3561
    },
    {
      "epoch": 6.77508321445554,
      "grad_norm": 0.23667073249816895,
      "learning_rate": 7.740870117497618e-05,
      "loss": 0.0778,
      "step": 3562
    },
    {
      "epoch": 6.77698525915359,
      "grad_norm": 0.26983892917633057,
      "learning_rate": 7.740234995236583e-05,
      "loss": 0.0779,
      "step": 3563
    },
    {
      "epoch": 6.778887303851641,
      "grad_norm": 0.3361709415912628,
      "learning_rate": 7.739599872975549e-05,
      "loss": 0.0911,
      "step": 3564
    },
    {
      "epoch": 6.780789348549691,
      "grad_norm": 0.25900062918663025,
      "learning_rate": 7.738964750714512e-05,
      "loss": 0.0728,
      "step": 3565
    },
    {
      "epoch": 6.782691393247742,
      "grad_norm": 0.2520677447319031,
      "learning_rate": 7.738329628453478e-05,
      "loss": 0.0704,
      "step": 3566
    },
    {
      "epoch": 6.784593437945792,
      "grad_norm": 0.29848986864089966,
      "learning_rate": 7.737694506192443e-05,
      "loss": 0.0829,
      "step": 3567
    },
    {
      "epoch": 6.786495482643842,
      "grad_norm": 0.27131152153015137,
      "learning_rate": 7.737059383931408e-05,
      "loss": 0.0651,
      "step": 3568
    },
    {
      "epoch": 6.788397527341893,
      "grad_norm": 0.2900139093399048,
      "learning_rate": 7.736424261670372e-05,
      "loss": 0.0684,
      "step": 3569
    },
    {
      "epoch": 6.790299572039943,
      "grad_norm": 0.29935070872306824,
      "learning_rate": 7.735789139409336e-05,
      "loss": 0.0741,
      "step": 3570
    },
    {
      "epoch": 6.7922016167379935,
      "grad_norm": 0.3142547905445099,
      "learning_rate": 7.735154017148302e-05,
      "loss": 0.0993,
      "step": 3571
    },
    {
      "epoch": 6.7941036614360435,
      "grad_norm": 0.2870892286300659,
      "learning_rate": 7.734518894887266e-05,
      "loss": 0.0725,
      "step": 3572
    },
    {
      "epoch": 6.796005706134094,
      "grad_norm": 0.4500356912612915,
      "learning_rate": 7.733883772626231e-05,
      "loss": 0.1104,
      "step": 3573
    },
    {
      "epoch": 6.797907750832144,
      "grad_norm": 0.4293663501739502,
      "learning_rate": 7.733248650365196e-05,
      "loss": 0.0867,
      "step": 3574
    },
    {
      "epoch": 6.799809795530195,
      "grad_norm": 0.33508118987083435,
      "learning_rate": 7.73261352810416e-05,
      "loss": 0.0853,
      "step": 3575
    },
    {
      "epoch": 6.801711840228245,
      "grad_norm": 0.30892953276634216,
      "learning_rate": 7.731978405843125e-05,
      "loss": 0.092,
      "step": 3576
    },
    {
      "epoch": 6.803613884926296,
      "grad_norm": 0.3197793960571289,
      "learning_rate": 7.731343283582089e-05,
      "loss": 0.0783,
      "step": 3577
    },
    {
      "epoch": 6.805515929624346,
      "grad_norm": 0.29358264803886414,
      "learning_rate": 7.730708161321054e-05,
      "loss": 0.0805,
      "step": 3578
    },
    {
      "epoch": 6.807417974322396,
      "grad_norm": 0.3208252489566803,
      "learning_rate": 7.73007303906002e-05,
      "loss": 0.0774,
      "step": 3579
    },
    {
      "epoch": 6.809320019020447,
      "grad_norm": 0.3243962824344635,
      "learning_rate": 7.729437916798983e-05,
      "loss": 0.0807,
      "step": 3580
    },
    {
      "epoch": 6.811222063718497,
      "grad_norm": 0.3556419909000397,
      "learning_rate": 7.72880279453795e-05,
      "loss": 0.0796,
      "step": 3581
    },
    {
      "epoch": 6.813124108416548,
      "grad_norm": 0.2571510076522827,
      "learning_rate": 7.728167672276914e-05,
      "loss": 0.0754,
      "step": 3582
    },
    {
      "epoch": 6.815026153114598,
      "grad_norm": 0.29990503191947937,
      "learning_rate": 7.727532550015878e-05,
      "loss": 0.0907,
      "step": 3583
    },
    {
      "epoch": 6.816928197812649,
      "grad_norm": 0.3055148720741272,
      "learning_rate": 7.726897427754843e-05,
      "loss": 0.0645,
      "step": 3584
    },
    {
      "epoch": 6.818830242510699,
      "grad_norm": 0.30374523997306824,
      "learning_rate": 7.726262305493808e-05,
      "loss": 0.0777,
      "step": 3585
    },
    {
      "epoch": 6.82073228720875,
      "grad_norm": 0.3372064530849457,
      "learning_rate": 7.725627183232773e-05,
      "loss": 0.0998,
      "step": 3586
    },
    {
      "epoch": 6.8226343319068,
      "grad_norm": 0.3002014756202698,
      "learning_rate": 7.724992060971737e-05,
      "loss": 0.069,
      "step": 3587
    },
    {
      "epoch": 6.824536376604851,
      "grad_norm": 0.26137155294418335,
      "learning_rate": 7.724356938710702e-05,
      "loss": 0.1057,
      "step": 3588
    },
    {
      "epoch": 6.826438421302901,
      "grad_norm": 0.3956218361854553,
      "learning_rate": 7.723721816449667e-05,
      "loss": 0.0795,
      "step": 3589
    },
    {
      "epoch": 6.828340466000951,
      "grad_norm": 0.2980736792087555,
      "learning_rate": 7.723086694188631e-05,
      "loss": 0.0934,
      "step": 3590
    },
    {
      "epoch": 6.8302425106990015,
      "grad_norm": 0.37110257148742676,
      "learning_rate": 7.722451571927596e-05,
      "loss": 0.0931,
      "step": 3591
    },
    {
      "epoch": 6.8321445553970515,
      "grad_norm": 0.2767345905303955,
      "learning_rate": 7.721816449666562e-05,
      "loss": 0.0736,
      "step": 3592
    },
    {
      "epoch": 6.834046600095102,
      "grad_norm": 0.31996482610702515,
      "learning_rate": 7.721181327405525e-05,
      "loss": 0.0842,
      "step": 3593
    },
    {
      "epoch": 6.835948644793152,
      "grad_norm": 0.303419291973114,
      "learning_rate": 7.72054620514449e-05,
      "loss": 0.1011,
      "step": 3594
    },
    {
      "epoch": 6.837850689491203,
      "grad_norm": 0.2984336316585541,
      "learning_rate": 7.719911082883456e-05,
      "loss": 0.0876,
      "step": 3595
    },
    {
      "epoch": 6.839752734189253,
      "grad_norm": 0.32384997606277466,
      "learning_rate": 7.71927596062242e-05,
      "loss": 0.0788,
      "step": 3596
    },
    {
      "epoch": 6.841654778887304,
      "grad_norm": 0.23781158030033112,
      "learning_rate": 7.718640838361385e-05,
      "loss": 0.0861,
      "step": 3597
    },
    {
      "epoch": 6.843556823585354,
      "grad_norm": 0.22147323191165924,
      "learning_rate": 7.71800571610035e-05,
      "loss": 0.0614,
      "step": 3598
    },
    {
      "epoch": 6.845458868283405,
      "grad_norm": 0.3343324363231659,
      "learning_rate": 7.717370593839315e-05,
      "loss": 0.0906,
      "step": 3599
    },
    {
      "epoch": 6.847360912981455,
      "grad_norm": 0.4509906470775604,
      "learning_rate": 7.716735471578279e-05,
      "loss": 0.0859,
      "step": 3600
    },
    {
      "epoch": 6.849262957679505,
      "grad_norm": 0.27889499068260193,
      "learning_rate": 7.716100349317243e-05,
      "loss": 0.0832,
      "step": 3601
    },
    {
      "epoch": 6.851165002377556,
      "grad_norm": 0.24554187059402466,
      "learning_rate": 7.71546522705621e-05,
      "loss": 0.061,
      "step": 3602
    },
    {
      "epoch": 6.853067047075607,
      "grad_norm": 0.307670533657074,
      "learning_rate": 7.714830104795173e-05,
      "loss": 0.0853,
      "step": 3603
    },
    {
      "epoch": 6.854969091773657,
      "grad_norm": 0.29913416504859924,
      "learning_rate": 7.714194982534138e-05,
      "loss": 0.064,
      "step": 3604
    },
    {
      "epoch": 6.856871136471707,
      "grad_norm": 0.2952573001384735,
      "learning_rate": 7.713559860273104e-05,
      "loss": 0.0974,
      "step": 3605
    },
    {
      "epoch": 6.858773181169758,
      "grad_norm": 0.3110623359680176,
      "learning_rate": 7.712924738012067e-05,
      "loss": 0.1036,
      "step": 3606
    },
    {
      "epoch": 6.860675225867808,
      "grad_norm": 0.25095778703689575,
      "learning_rate": 7.712289615751033e-05,
      "loss": 0.0867,
      "step": 3607
    },
    {
      "epoch": 6.8625772705658585,
      "grad_norm": 0.23761174082756042,
      "learning_rate": 7.711654493489996e-05,
      "loss": 0.0719,
      "step": 3608
    },
    {
      "epoch": 6.8644793152639085,
      "grad_norm": 0.39075493812561035,
      "learning_rate": 7.711019371228963e-05,
      "loss": 0.0942,
      "step": 3609
    },
    {
      "epoch": 6.866381359961959,
      "grad_norm": 0.20142102241516113,
      "learning_rate": 7.710384248967927e-05,
      "loss": 0.0753,
      "step": 3610
    },
    {
      "epoch": 6.868283404660009,
      "grad_norm": 0.3020033836364746,
      "learning_rate": 7.70974912670689e-05,
      "loss": 0.1073,
      "step": 3611
    },
    {
      "epoch": 6.870185449358059,
      "grad_norm": 0.5035987496376038,
      "learning_rate": 7.709114004445857e-05,
      "loss": 0.0936,
      "step": 3612
    },
    {
      "epoch": 6.87208749405611,
      "grad_norm": 0.25016555190086365,
      "learning_rate": 7.708478882184821e-05,
      "loss": 0.0853,
      "step": 3613
    },
    {
      "epoch": 6.873989538754161,
      "grad_norm": 0.3226104974746704,
      "learning_rate": 7.707843759923785e-05,
      "loss": 0.0984,
      "step": 3614
    },
    {
      "epoch": 6.875891583452211,
      "grad_norm": 0.3413734436035156,
      "learning_rate": 7.70720863766275e-05,
      "loss": 0.0967,
      "step": 3615
    },
    {
      "epoch": 6.877793628150261,
      "grad_norm": 0.2394518107175827,
      "learning_rate": 7.706573515401715e-05,
      "loss": 0.0699,
      "step": 3616
    },
    {
      "epoch": 6.879695672848312,
      "grad_norm": 0.3109191358089447,
      "learning_rate": 7.70593839314068e-05,
      "loss": 0.0941,
      "step": 3617
    },
    {
      "epoch": 6.881597717546362,
      "grad_norm": 0.24442525207996368,
      "learning_rate": 7.705303270879644e-05,
      "loss": 0.0993,
      "step": 3618
    },
    {
      "epoch": 6.883499762244413,
      "grad_norm": 0.350276380777359,
      "learning_rate": 7.70466814861861e-05,
      "loss": 0.1212,
      "step": 3619
    },
    {
      "epoch": 6.885401806942463,
      "grad_norm": 0.3216733932495117,
      "learning_rate": 7.704033026357575e-05,
      "loss": 0.1079,
      "step": 3620
    },
    {
      "epoch": 6.887303851640514,
      "grad_norm": 0.5144888162612915,
      "learning_rate": 7.703397904096538e-05,
      "loss": 0.1056,
      "step": 3621
    },
    {
      "epoch": 6.889205896338564,
      "grad_norm": 0.2998292148113251,
      "learning_rate": 7.702762781835504e-05,
      "loss": 0.0891,
      "step": 3622
    },
    {
      "epoch": 6.891107941036615,
      "grad_norm": 0.39780697226524353,
      "learning_rate": 7.702127659574469e-05,
      "loss": 0.1039,
      "step": 3623
    },
    {
      "epoch": 6.893009985734665,
      "grad_norm": 0.30607423186302185,
      "learning_rate": 7.701492537313433e-05,
      "loss": 0.0864,
      "step": 3624
    },
    {
      "epoch": 6.894912030432716,
      "grad_norm": 0.2678467631340027,
      "learning_rate": 7.700857415052398e-05,
      "loss": 0.0956,
      "step": 3625
    },
    {
      "epoch": 6.896814075130766,
      "grad_norm": 0.2922561764717102,
      "learning_rate": 7.700222292791363e-05,
      "loss": 0.0904,
      "step": 3626
    },
    {
      "epoch": 6.898716119828816,
      "grad_norm": 0.34059253334999084,
      "learning_rate": 7.699587170530328e-05,
      "loss": 0.1142,
      "step": 3627
    },
    {
      "epoch": 6.9006181645268665,
      "grad_norm": 0.34268996119499207,
      "learning_rate": 7.698952048269292e-05,
      "loss": 0.0999,
      "step": 3628
    },
    {
      "epoch": 6.9025202092249165,
      "grad_norm": 0.2850574851036072,
      "learning_rate": 7.698316926008257e-05,
      "loss": 0.0827,
      "step": 3629
    },
    {
      "epoch": 6.904422253922967,
      "grad_norm": 0.44454216957092285,
      "learning_rate": 7.697681803747222e-05,
      "loss": 0.1169,
      "step": 3630
    },
    {
      "epoch": 6.906324298621017,
      "grad_norm": 0.3208524286746979,
      "learning_rate": 7.697046681486186e-05,
      "loss": 0.0989,
      "step": 3631
    },
    {
      "epoch": 6.908226343319068,
      "grad_norm": 0.36460134387016296,
      "learning_rate": 7.696411559225151e-05,
      "loss": 0.0775,
      "step": 3632
    },
    {
      "epoch": 6.910128388017118,
      "grad_norm": 0.34505489468574524,
      "learning_rate": 7.695776436964117e-05,
      "loss": 0.0959,
      "step": 3633
    },
    {
      "epoch": 6.912030432715169,
      "grad_norm": 0.28789353370666504,
      "learning_rate": 7.69514131470308e-05,
      "loss": 0.0862,
      "step": 3634
    },
    {
      "epoch": 6.913932477413219,
      "grad_norm": 0.3089391589164734,
      "learning_rate": 7.694506192442046e-05,
      "loss": 0.0838,
      "step": 3635
    },
    {
      "epoch": 6.91583452211127,
      "grad_norm": 0.38007378578186035,
      "learning_rate": 7.693871070181011e-05,
      "loss": 0.0811,
      "step": 3636
    },
    {
      "epoch": 6.91773656680932,
      "grad_norm": 0.24695605039596558,
      "learning_rate": 7.693235947919975e-05,
      "loss": 0.0886,
      "step": 3637
    },
    {
      "epoch": 6.91963861150737,
      "grad_norm": 0.31701305508613586,
      "learning_rate": 7.69260082565894e-05,
      "loss": 0.0814,
      "step": 3638
    },
    {
      "epoch": 6.921540656205421,
      "grad_norm": 0.2539271116256714,
      "learning_rate": 7.691965703397905e-05,
      "loss": 0.0948,
      "step": 3639
    },
    {
      "epoch": 6.923442700903471,
      "grad_norm": 0.3465598523616791,
      "learning_rate": 7.69133058113687e-05,
      "loss": 0.0983,
      "step": 3640
    },
    {
      "epoch": 6.925344745601522,
      "grad_norm": 0.30124831199645996,
      "learning_rate": 7.690695458875834e-05,
      "loss": 0.0687,
      "step": 3641
    },
    {
      "epoch": 6.927246790299572,
      "grad_norm": 0.3128988444805145,
      "learning_rate": 7.690060336614798e-05,
      "loss": 0.091,
      "step": 3642
    },
    {
      "epoch": 6.929148834997623,
      "grad_norm": 0.30787718296051025,
      "learning_rate": 7.689425214353764e-05,
      "loss": 0.08,
      "step": 3643
    },
    {
      "epoch": 6.931050879695673,
      "grad_norm": 0.26531967520713806,
      "learning_rate": 7.688790092092728e-05,
      "loss": 0.0903,
      "step": 3644
    },
    {
      "epoch": 6.932952924393724,
      "grad_norm": 0.28860077261924744,
      "learning_rate": 7.688154969831693e-05,
      "loss": 0.1021,
      "step": 3645
    },
    {
      "epoch": 6.934854969091774,
      "grad_norm": 0.3609332740306854,
      "learning_rate": 7.687519847570657e-05,
      "loss": 0.1108,
      "step": 3646
    },
    {
      "epoch": 6.9367570137898245,
      "grad_norm": 0.2961665987968445,
      "learning_rate": 7.686884725309622e-05,
      "loss": 0.0732,
      "step": 3647
    },
    {
      "epoch": 6.9386590584878745,
      "grad_norm": 0.38632073998451233,
      "learning_rate": 7.686249603048587e-05,
      "loss": 0.0999,
      "step": 3648
    },
    {
      "epoch": 6.9405611031859245,
      "grad_norm": 0.26912546157836914,
      "learning_rate": 7.685614480787551e-05,
      "loss": 0.0803,
      "step": 3649
    },
    {
      "epoch": 6.942463147883975,
      "grad_norm": 0.505782961845398,
      "learning_rate": 7.684979358526517e-05,
      "loss": 0.1164,
      "step": 3650
    },
    {
      "epoch": 6.944365192582025,
      "grad_norm": 0.202500119805336,
      "learning_rate": 7.684344236265482e-05,
      "loss": 0.0714,
      "step": 3651
    },
    {
      "epoch": 6.946267237280076,
      "grad_norm": 0.3188091814517975,
      "learning_rate": 7.683709114004446e-05,
      "loss": 0.0758,
      "step": 3652
    },
    {
      "epoch": 6.948169281978126,
      "grad_norm": 0.34093478322029114,
      "learning_rate": 7.683073991743411e-05,
      "loss": 0.0787,
      "step": 3653
    },
    {
      "epoch": 6.950071326676177,
      "grad_norm": 0.4178665280342102,
      "learning_rate": 7.682438869482376e-05,
      "loss": 0.1045,
      "step": 3654
    },
    {
      "epoch": 6.951973371374227,
      "grad_norm": 0.29486292600631714,
      "learning_rate": 7.68180374722134e-05,
      "loss": 0.081,
      "step": 3655
    },
    {
      "epoch": 6.953875416072278,
      "grad_norm": 0.23391938209533691,
      "learning_rate": 7.681168624960305e-05,
      "loss": 0.0653,
      "step": 3656
    },
    {
      "epoch": 6.955777460770328,
      "grad_norm": 0.26563167572021484,
      "learning_rate": 7.68053350269927e-05,
      "loss": 0.0815,
      "step": 3657
    },
    {
      "epoch": 6.957679505468379,
      "grad_norm": 0.25844621658325195,
      "learning_rate": 7.679898380438235e-05,
      "loss": 0.0657,
      "step": 3658
    },
    {
      "epoch": 6.959581550166429,
      "grad_norm": 0.2926645874977112,
      "learning_rate": 7.679263258177199e-05,
      "loss": 0.0807,
      "step": 3659
    },
    {
      "epoch": 6.961483594864479,
      "grad_norm": 0.2784111499786377,
      "learning_rate": 7.678628135916164e-05,
      "loss": 0.0973,
      "step": 3660
    },
    {
      "epoch": 6.96338563956253,
      "grad_norm": 0.3077068030834198,
      "learning_rate": 7.67799301365513e-05,
      "loss": 0.0902,
      "step": 3661
    },
    {
      "epoch": 6.96528768426058,
      "grad_norm": 0.3007901906967163,
      "learning_rate": 7.677357891394093e-05,
      "loss": 0.093,
      "step": 3662
    },
    {
      "epoch": 6.967189728958631,
      "grad_norm": 0.3273557126522064,
      "learning_rate": 7.676722769133058e-05,
      "loss": 0.0958,
      "step": 3663
    },
    {
      "epoch": 6.969091773656681,
      "grad_norm": 0.2853504419326782,
      "learning_rate": 7.676087646872024e-05,
      "loss": 0.0949,
      "step": 3664
    },
    {
      "epoch": 6.9709938183547315,
      "grad_norm": 0.30276769399642944,
      "learning_rate": 7.675452524610987e-05,
      "loss": 0.0615,
      "step": 3665
    },
    {
      "epoch": 6.9728958630527815,
      "grad_norm": 0.23220020532608032,
      "learning_rate": 7.674817402349953e-05,
      "loss": 0.0605,
      "step": 3666
    },
    {
      "epoch": 6.974797907750832,
      "grad_norm": 0.26779958605766296,
      "learning_rate": 7.674182280088918e-05,
      "loss": 0.0844,
      "step": 3667
    },
    {
      "epoch": 6.976699952448882,
      "grad_norm": 0.32445698976516724,
      "learning_rate": 7.673547157827882e-05,
      "loss": 0.0944,
      "step": 3668
    },
    {
      "epoch": 6.978601997146933,
      "grad_norm": 0.36110180616378784,
      "learning_rate": 7.672912035566847e-05,
      "loss": 0.083,
      "step": 3669
    },
    {
      "epoch": 6.980504041844983,
      "grad_norm": 0.3307937681674957,
      "learning_rate": 7.672276913305812e-05,
      "loss": 0.0843,
      "step": 3670
    },
    {
      "epoch": 6.982406086543033,
      "grad_norm": 0.32733941078186035,
      "learning_rate": 7.671641791044777e-05,
      "loss": 0.0837,
      "step": 3671
    },
    {
      "epoch": 6.984308131241084,
      "grad_norm": 0.8110985159873962,
      "learning_rate": 7.671006668783741e-05,
      "loss": 0.1877,
      "step": 3672
    },
    {
      "epoch": 6.986210175939134,
      "grad_norm": 0.41781920194625854,
      "learning_rate": 7.670371546522705e-05,
      "loss": 0.1121,
      "step": 3673
    },
    {
      "epoch": 6.988112220637185,
      "grad_norm": 0.2781015932559967,
      "learning_rate": 7.669736424261671e-05,
      "loss": 0.0735,
      "step": 3674
    },
    {
      "epoch": 6.990014265335235,
      "grad_norm": 0.38082799315452576,
      "learning_rate": 7.669101302000635e-05,
      "loss": 0.0951,
      "step": 3675
    },
    {
      "epoch": 6.991916310033286,
      "grad_norm": 0.3210826814174652,
      "learning_rate": 7.6684661797396e-05,
      "loss": 0.0947,
      "step": 3676
    },
    {
      "epoch": 6.993818354731336,
      "grad_norm": 0.2684693932533264,
      "learning_rate": 7.667831057478566e-05,
      "loss": 0.1023,
      "step": 3677
    },
    {
      "epoch": 6.995720399429387,
      "grad_norm": 0.4197976589202881,
      "learning_rate": 7.66719593521753e-05,
      "loss": 0.0941,
      "step": 3678
    },
    {
      "epoch": 6.997622444127437,
      "grad_norm": 0.3351534605026245,
      "learning_rate": 7.666560812956495e-05,
      "loss": 0.0748,
      "step": 3679
    },
    {
      "epoch": 6.999524488825488,
      "grad_norm": 0.3510059118270874,
      "learning_rate": 7.665925690695458e-05,
      "loss": 0.0993,
      "step": 3680
    },
    {
      "epoch": 7.001426533523538,
      "grad_norm": 0.2770400643348694,
      "learning_rate": 7.665290568434425e-05,
      "loss": 0.0962,
      "step": 3681
    },
    {
      "epoch": 7.003328578221589,
      "grad_norm": 0.21430467069149017,
      "learning_rate": 7.664655446173389e-05,
      "loss": 0.0689,
      "step": 3682
    },
    {
      "epoch": 7.005230622919639,
      "grad_norm": 0.3444157540798187,
      "learning_rate": 7.664020323912353e-05,
      "loss": 0.061,
      "step": 3683
    },
    {
      "epoch": 7.007132667617689,
      "grad_norm": 0.3022604286670685,
      "learning_rate": 7.663385201651319e-05,
      "loss": 0.0792,
      "step": 3684
    },
    {
      "epoch": 7.0090347123157395,
      "grad_norm": 0.233037069439888,
      "learning_rate": 7.662750079390283e-05,
      "loss": 0.0821,
      "step": 3685
    },
    {
      "epoch": 7.0109367570137895,
      "grad_norm": 0.254637211561203,
      "learning_rate": 7.662114957129247e-05,
      "loss": 0.0957,
      "step": 3686
    },
    {
      "epoch": 7.01283880171184,
      "grad_norm": 0.30084213614463806,
      "learning_rate": 7.661479834868212e-05,
      "loss": 0.0813,
      "step": 3687
    },
    {
      "epoch": 7.01474084640989,
      "grad_norm": 0.30689504742622375,
      "learning_rate": 7.660844712607177e-05,
      "loss": 0.0863,
      "step": 3688
    },
    {
      "epoch": 7.016642891107941,
      "grad_norm": 0.24477942287921906,
      "learning_rate": 7.660209590346142e-05,
      "loss": 0.071,
      "step": 3689
    },
    {
      "epoch": 7.018544935805991,
      "grad_norm": 0.25171712040901184,
      "learning_rate": 7.659574468085106e-05,
      "loss": 0.0641,
      "step": 3690
    },
    {
      "epoch": 7.020446980504042,
      "grad_norm": 0.21781741082668304,
      "learning_rate": 7.658939345824071e-05,
      "loss": 0.0676,
      "step": 3691
    },
    {
      "epoch": 7.022349025202092,
      "grad_norm": 0.23356683552265167,
      "learning_rate": 7.658304223563037e-05,
      "loss": 0.0654,
      "step": 3692
    },
    {
      "epoch": 7.024251069900143,
      "grad_norm": 0.2565113604068756,
      "learning_rate": 7.657669101302e-05,
      "loss": 0.0753,
      "step": 3693
    },
    {
      "epoch": 7.026153114598193,
      "grad_norm": 0.3282324969768524,
      "learning_rate": 7.657033979040966e-05,
      "loss": 0.0725,
      "step": 3694
    },
    {
      "epoch": 7.028055159296243,
      "grad_norm": 0.2607407867908478,
      "learning_rate": 7.656398856779931e-05,
      "loss": 0.0962,
      "step": 3695
    },
    {
      "epoch": 7.029957203994294,
      "grad_norm": 0.18478083610534668,
      "learning_rate": 7.655763734518895e-05,
      "loss": 0.0694,
      "step": 3696
    },
    {
      "epoch": 7.031859248692344,
      "grad_norm": 0.21331697702407837,
      "learning_rate": 7.65512861225786e-05,
      "loss": 0.059,
      "step": 3697
    },
    {
      "epoch": 7.033761293390395,
      "grad_norm": 0.20701389014720917,
      "learning_rate": 7.654493489996825e-05,
      "loss": 0.0745,
      "step": 3698
    },
    {
      "epoch": 7.035663338088445,
      "grad_norm": 0.35444629192352295,
      "learning_rate": 7.65385836773579e-05,
      "loss": 0.1039,
      "step": 3699
    },
    {
      "epoch": 7.037565382786496,
      "grad_norm": 0.24495595693588257,
      "learning_rate": 7.653223245474754e-05,
      "loss": 0.0695,
      "step": 3700
    },
    {
      "epoch": 7.039467427484546,
      "grad_norm": 0.24500828981399536,
      "learning_rate": 7.652588123213719e-05,
      "loss": 0.0663,
      "step": 3701
    },
    {
      "epoch": 7.041369472182597,
      "grad_norm": 0.2540830075740814,
      "learning_rate": 7.651953000952684e-05,
      "loss": 0.0546,
      "step": 3702
    },
    {
      "epoch": 7.043271516880647,
      "grad_norm": 0.3355015814304352,
      "learning_rate": 7.651317878691648e-05,
      "loss": 0.0744,
      "step": 3703
    },
    {
      "epoch": 7.0451735615786975,
      "grad_norm": 0.2634127736091614,
      "learning_rate": 7.650682756430612e-05,
      "loss": 0.0752,
      "step": 3704
    },
    {
      "epoch": 7.0470756062767475,
      "grad_norm": 0.21350815892219543,
      "learning_rate": 7.650047634169579e-05,
      "loss": 0.0743,
      "step": 3705
    },
    {
      "epoch": 7.048977650974798,
      "grad_norm": 0.3560217320919037,
      "learning_rate": 7.649412511908542e-05,
      "loss": 0.0894,
      "step": 3706
    },
    {
      "epoch": 7.050879695672848,
      "grad_norm": 0.2929129898548126,
      "learning_rate": 7.648777389647508e-05,
      "loss": 0.0632,
      "step": 3707
    },
    {
      "epoch": 7.052781740370898,
      "grad_norm": 0.34552210569381714,
      "learning_rate": 7.648142267386473e-05,
      "loss": 0.0909,
      "step": 3708
    },
    {
      "epoch": 7.054683785068949,
      "grad_norm": 0.21399307250976562,
      "learning_rate": 7.647507145125437e-05,
      "loss": 0.0687,
      "step": 3709
    },
    {
      "epoch": 7.056585829766999,
      "grad_norm": 0.22795552015304565,
      "learning_rate": 7.646872022864402e-05,
      "loss": 0.0694,
      "step": 3710
    },
    {
      "epoch": 7.05848787446505,
      "grad_norm": 0.4743165075778961,
      "learning_rate": 7.646236900603366e-05,
      "loss": 0.1167,
      "step": 3711
    },
    {
      "epoch": 7.0603899191631,
      "grad_norm": 0.25579768419265747,
      "learning_rate": 7.645601778342332e-05,
      "loss": 0.052,
      "step": 3712
    },
    {
      "epoch": 7.062291963861151,
      "grad_norm": 0.41287311911582947,
      "learning_rate": 7.644966656081296e-05,
      "loss": 0.0891,
      "step": 3713
    },
    {
      "epoch": 7.064194008559201,
      "grad_norm": 0.4381440281867981,
      "learning_rate": 7.64433153382026e-05,
      "loss": 0.1165,
      "step": 3714
    },
    {
      "epoch": 7.066096053257252,
      "grad_norm": 0.4422372281551361,
      "learning_rate": 7.643696411559226e-05,
      "loss": 0.0979,
      "step": 3715
    },
    {
      "epoch": 7.067998097955302,
      "grad_norm": 0.3064451813697815,
      "learning_rate": 7.64306128929819e-05,
      "loss": 0.0781,
      "step": 3716
    },
    {
      "epoch": 7.069900142653353,
      "grad_norm": 0.30070748925209045,
      "learning_rate": 7.642426167037155e-05,
      "loss": 0.0737,
      "step": 3717
    },
    {
      "epoch": 7.071802187351403,
      "grad_norm": 0.34751248359680176,
      "learning_rate": 7.641791044776119e-05,
      "loss": 0.0682,
      "step": 3718
    },
    {
      "epoch": 7.073704232049453,
      "grad_norm": 0.33590537309646606,
      "learning_rate": 7.641155922515084e-05,
      "loss": 0.0944,
      "step": 3719
    },
    {
      "epoch": 7.075606276747504,
      "grad_norm": 0.4045461416244507,
      "learning_rate": 7.64052080025405e-05,
      "loss": 0.0846,
      "step": 3720
    },
    {
      "epoch": 7.077508321445554,
      "grad_norm": 0.189460888504982,
      "learning_rate": 7.639885677993013e-05,
      "loss": 0.0715,
      "step": 3721
    },
    {
      "epoch": 7.0794103661436045,
      "grad_norm": 0.2401493638753891,
      "learning_rate": 7.639250555731979e-05,
      "loss": 0.0558,
      "step": 3722
    },
    {
      "epoch": 7.0813124108416545,
      "grad_norm": 0.265419602394104,
      "learning_rate": 7.638615433470944e-05,
      "loss": 0.0761,
      "step": 3723
    },
    {
      "epoch": 7.083214455539705,
      "grad_norm": 0.22794272005558014,
      "learning_rate": 7.637980311209908e-05,
      "loss": 0.0594,
      "step": 3724
    },
    {
      "epoch": 7.085116500237755,
      "grad_norm": 0.2641700804233551,
      "learning_rate": 7.637345188948873e-05,
      "loss": 0.0789,
      "step": 3725
    },
    {
      "epoch": 7.087018544935806,
      "grad_norm": 0.29633039236068726,
      "learning_rate": 7.636710066687838e-05,
      "loss": 0.0788,
      "step": 3726
    },
    {
      "epoch": 7.088920589633856,
      "grad_norm": 0.3040051758289337,
      "learning_rate": 7.636074944426802e-05,
      "loss": 0.0853,
      "step": 3727
    },
    {
      "epoch": 7.090822634331907,
      "grad_norm": 0.24118536710739136,
      "learning_rate": 7.635439822165767e-05,
      "loss": 0.0745,
      "step": 3728
    },
    {
      "epoch": 7.092724679029957,
      "grad_norm": 0.26473772525787354,
      "learning_rate": 7.634804699904732e-05,
      "loss": 0.0654,
      "step": 3729
    },
    {
      "epoch": 7.094626723728007,
      "grad_norm": 0.3334522247314453,
      "learning_rate": 7.634169577643697e-05,
      "loss": 0.0896,
      "step": 3730
    },
    {
      "epoch": 7.096528768426058,
      "grad_norm": 0.3315863311290741,
      "learning_rate": 7.633534455382661e-05,
      "loss": 0.0878,
      "step": 3731
    },
    {
      "epoch": 7.098430813124108,
      "grad_norm": 0.2381017655134201,
      "learning_rate": 7.632899333121626e-05,
      "loss": 0.07,
      "step": 3732
    },
    {
      "epoch": 7.100332857822159,
      "grad_norm": 0.25926443934440613,
      "learning_rate": 7.632264210860592e-05,
      "loss": 0.0786,
      "step": 3733
    },
    {
      "epoch": 7.102234902520209,
      "grad_norm": 0.38049688935279846,
      "learning_rate": 7.631629088599555e-05,
      "loss": 0.0953,
      "step": 3734
    },
    {
      "epoch": 7.10413694721826,
      "grad_norm": 0.36096832156181335,
      "learning_rate": 7.63099396633852e-05,
      "loss": 0.0894,
      "step": 3735
    },
    {
      "epoch": 7.10603899191631,
      "grad_norm": 0.2273632436990738,
      "learning_rate": 7.630358844077486e-05,
      "loss": 0.071,
      "step": 3736
    },
    {
      "epoch": 7.107941036614361,
      "grad_norm": 0.24754773080348969,
      "learning_rate": 7.62972372181645e-05,
      "loss": 0.0881,
      "step": 3737
    },
    {
      "epoch": 7.109843081312411,
      "grad_norm": 0.3349929749965668,
      "learning_rate": 7.629088599555415e-05,
      "loss": 0.0845,
      "step": 3738
    },
    {
      "epoch": 7.111745126010462,
      "grad_norm": 0.21691037714481354,
      "learning_rate": 7.62845347729438e-05,
      "loss": 0.0733,
      "step": 3739
    },
    {
      "epoch": 7.113647170708512,
      "grad_norm": 0.2048906683921814,
      "learning_rate": 7.627818355033344e-05,
      "loss": 0.0492,
      "step": 3740
    },
    {
      "epoch": 7.1155492154065625,
      "grad_norm": 0.2179352343082428,
      "learning_rate": 7.627183232772309e-05,
      "loss": 0.0759,
      "step": 3741
    },
    {
      "epoch": 7.1174512601046125,
      "grad_norm": 0.24795210361480713,
      "learning_rate": 7.626548110511274e-05,
      "loss": 0.0731,
      "step": 3742
    },
    {
      "epoch": 7.1193533048026625,
      "grad_norm": 0.28693392872810364,
      "learning_rate": 7.62591298825024e-05,
      "loss": 0.0865,
      "step": 3743
    },
    {
      "epoch": 7.121255349500713,
      "grad_norm": 0.2696365416049957,
      "learning_rate": 7.625277865989203e-05,
      "loss": 0.0723,
      "step": 3744
    },
    {
      "epoch": 7.123157394198763,
      "grad_norm": 0.3165305256843567,
      "learning_rate": 7.624642743728167e-05,
      "loss": 0.0955,
      "step": 3745
    },
    {
      "epoch": 7.125059438896814,
      "grad_norm": 0.2257310301065445,
      "learning_rate": 7.624007621467134e-05,
      "loss": 0.0699,
      "step": 3746
    },
    {
      "epoch": 7.126961483594864,
      "grad_norm": 0.14401155710220337,
      "learning_rate": 7.623372499206097e-05,
      "loss": 0.0582,
      "step": 3747
    },
    {
      "epoch": 7.128863528292915,
      "grad_norm": 0.40427732467651367,
      "learning_rate": 7.622737376945063e-05,
      "loss": 0.1004,
      "step": 3748
    },
    {
      "epoch": 7.130765572990965,
      "grad_norm": 0.3528768718242645,
      "learning_rate": 7.622102254684028e-05,
      "loss": 0.0902,
      "step": 3749
    },
    {
      "epoch": 7.132667617689016,
      "grad_norm": 0.2220061868429184,
      "learning_rate": 7.621467132422992e-05,
      "loss": 0.0517,
      "step": 3750
    },
    {
      "epoch": 7.134569662387066,
      "grad_norm": 0.3189872205257416,
      "learning_rate": 7.620832010161957e-05,
      "loss": 0.0837,
      "step": 3751
    },
    {
      "epoch": 7.136471707085117,
      "grad_norm": 0.21646323800086975,
      "learning_rate": 7.62019688790092e-05,
      "loss": 0.0546,
      "step": 3752
    },
    {
      "epoch": 7.138373751783167,
      "grad_norm": 0.3528130054473877,
      "learning_rate": 7.619561765639887e-05,
      "loss": 0.0781,
      "step": 3753
    },
    {
      "epoch": 7.140275796481217,
      "grad_norm": 0.2020796537399292,
      "learning_rate": 7.618926643378851e-05,
      "loss": 0.0652,
      "step": 3754
    },
    {
      "epoch": 7.142177841179268,
      "grad_norm": 0.24950513243675232,
      "learning_rate": 7.618291521117815e-05,
      "loss": 0.0741,
      "step": 3755
    },
    {
      "epoch": 7.144079885877318,
      "grad_norm": 0.24914821982383728,
      "learning_rate": 7.61765639885678e-05,
      "loss": 0.0734,
      "step": 3756
    },
    {
      "epoch": 7.145981930575369,
      "grad_norm": 0.2292976677417755,
      "learning_rate": 7.617021276595745e-05,
      "loss": 0.0754,
      "step": 3757
    },
    {
      "epoch": 7.147883975273419,
      "grad_norm": 0.3274495005607605,
      "learning_rate": 7.616386154334709e-05,
      "loss": 0.1309,
      "step": 3758
    },
    {
      "epoch": 7.14978601997147,
      "grad_norm": 0.28548160195350647,
      "learning_rate": 7.615751032073674e-05,
      "loss": 0.0801,
      "step": 3759
    },
    {
      "epoch": 7.15168806466952,
      "grad_norm": 0.20921234786510468,
      "learning_rate": 7.61511590981264e-05,
      "loss": 0.0686,
      "step": 3760
    },
    {
      "epoch": 7.1535901093675704,
      "grad_norm": 0.2667579650878906,
      "learning_rate": 7.614480787551605e-05,
      "loss": 0.0831,
      "step": 3761
    },
    {
      "epoch": 7.1554921540656204,
      "grad_norm": 0.4038238227367401,
      "learning_rate": 7.613845665290568e-05,
      "loss": 0.0965,
      "step": 3762
    },
    {
      "epoch": 7.157394198763671,
      "grad_norm": 0.2721848785877228,
      "learning_rate": 7.613210543029534e-05,
      "loss": 0.0752,
      "step": 3763
    },
    {
      "epoch": 7.159296243461721,
      "grad_norm": 0.2838439345359802,
      "learning_rate": 7.612575420768499e-05,
      "loss": 0.0758,
      "step": 3764
    },
    {
      "epoch": 7.161198288159771,
      "grad_norm": 0.3026263117790222,
      "learning_rate": 7.611940298507463e-05,
      "loss": 0.0913,
      "step": 3765
    },
    {
      "epoch": 7.163100332857822,
      "grad_norm": 0.7450515031814575,
      "learning_rate": 7.611305176246428e-05,
      "loss": 0.0912,
      "step": 3766
    },
    {
      "epoch": 7.165002377555872,
      "grad_norm": 0.2942456007003784,
      "learning_rate": 7.610670053985393e-05,
      "loss": 0.0644,
      "step": 3767
    },
    {
      "epoch": 7.166904422253923,
      "grad_norm": 0.24996860325336456,
      "learning_rate": 7.610034931724357e-05,
      "loss": 0.0906,
      "step": 3768
    },
    {
      "epoch": 7.168806466951973,
      "grad_norm": 0.23108260333538055,
      "learning_rate": 7.609399809463322e-05,
      "loss": 0.069,
      "step": 3769
    },
    {
      "epoch": 7.170708511650024,
      "grad_norm": 0.28415048122406006,
      "learning_rate": 7.608764687202287e-05,
      "loss": 0.0727,
      "step": 3770
    },
    {
      "epoch": 7.172610556348074,
      "grad_norm": 0.2941918969154358,
      "learning_rate": 7.608129564941252e-05,
      "loss": 0.0749,
      "step": 3771
    },
    {
      "epoch": 7.174512601046125,
      "grad_norm": 0.2570197284221649,
      "learning_rate": 7.607494442680216e-05,
      "loss": 0.0798,
      "step": 3772
    },
    {
      "epoch": 7.176414645744175,
      "grad_norm": 0.20420850813388824,
      "learning_rate": 7.606859320419181e-05,
      "loss": 0.061,
      "step": 3773
    },
    {
      "epoch": 7.178316690442226,
      "grad_norm": 0.31141209602355957,
      "learning_rate": 7.606224198158146e-05,
      "loss": 0.0708,
      "step": 3774
    },
    {
      "epoch": 7.180218735140276,
      "grad_norm": 0.2804230749607086,
      "learning_rate": 7.60558907589711e-05,
      "loss": 0.1223,
      "step": 3775
    },
    {
      "epoch": 7.182120779838327,
      "grad_norm": 0.28173887729644775,
      "learning_rate": 7.604953953636074e-05,
      "loss": 0.0847,
      "step": 3776
    },
    {
      "epoch": 7.184022824536377,
      "grad_norm": 0.2961934208869934,
      "learning_rate": 7.604318831375041e-05,
      "loss": 0.0675,
      "step": 3777
    },
    {
      "epoch": 7.185924869234427,
      "grad_norm": 0.3397238850593567,
      "learning_rate": 7.603683709114005e-05,
      "loss": 0.0896,
      "step": 3778
    },
    {
      "epoch": 7.1878269139324775,
      "grad_norm": 0.3696925938129425,
      "learning_rate": 7.60304858685297e-05,
      "loss": 0.1039,
      "step": 3779
    },
    {
      "epoch": 7.1897289586305275,
      "grad_norm": 0.2568362355232239,
      "learning_rate": 7.602413464591935e-05,
      "loss": 0.0764,
      "step": 3780
    },
    {
      "epoch": 7.191631003328578,
      "grad_norm": 0.23331722617149353,
      "learning_rate": 7.601778342330899e-05,
      "loss": 0.0799,
      "step": 3781
    },
    {
      "epoch": 7.193533048026628,
      "grad_norm": 0.2694757878780365,
      "learning_rate": 7.601143220069864e-05,
      "loss": 0.0986,
      "step": 3782
    },
    {
      "epoch": 7.195435092724679,
      "grad_norm": 0.3996799886226654,
      "learning_rate": 7.600508097808828e-05,
      "loss": 0.0824,
      "step": 3783
    },
    {
      "epoch": 7.197337137422729,
      "grad_norm": 0.341553270816803,
      "learning_rate": 7.599872975547794e-05,
      "loss": 0.0935,
      "step": 3784
    },
    {
      "epoch": 7.19923918212078,
      "grad_norm": 0.41450145840644836,
      "learning_rate": 7.599237853286758e-05,
      "loss": 0.0944,
      "step": 3785
    },
    {
      "epoch": 7.20114122681883,
      "grad_norm": 0.30726349353790283,
      "learning_rate": 7.598602731025722e-05,
      "loss": 0.0811,
      "step": 3786
    },
    {
      "epoch": 7.203043271516881,
      "grad_norm": 0.29750320315361023,
      "learning_rate": 7.597967608764688e-05,
      "loss": 0.0706,
      "step": 3787
    },
    {
      "epoch": 7.204945316214931,
      "grad_norm": 0.37351077795028687,
      "learning_rate": 7.597332486503652e-05,
      "loss": 0.1056,
      "step": 3788
    },
    {
      "epoch": 7.206847360912981,
      "grad_norm": 0.2162342220544815,
      "learning_rate": 7.596697364242617e-05,
      "loss": 0.0732,
      "step": 3789
    },
    {
      "epoch": 7.208749405611032,
      "grad_norm": 0.2836562693119049,
      "learning_rate": 7.596062241981581e-05,
      "loss": 0.0779,
      "step": 3790
    },
    {
      "epoch": 7.210651450309082,
      "grad_norm": 0.23386134207248688,
      "learning_rate": 7.595427119720546e-05,
      "loss": 0.0741,
      "step": 3791
    },
    {
      "epoch": 7.212553495007133,
      "grad_norm": 0.2197110950946808,
      "learning_rate": 7.594791997459512e-05,
      "loss": 0.0628,
      "step": 3792
    },
    {
      "epoch": 7.214455539705183,
      "grad_norm": 0.2751940190792084,
      "learning_rate": 7.594156875198476e-05,
      "loss": 0.0873,
      "step": 3793
    },
    {
      "epoch": 7.216357584403234,
      "grad_norm": 0.2839178740978241,
      "learning_rate": 7.593521752937441e-05,
      "loss": 0.0861,
      "step": 3794
    },
    {
      "epoch": 7.218259629101284,
      "grad_norm": 0.22591270506381989,
      "learning_rate": 7.592886630676406e-05,
      "loss": 0.083,
      "step": 3795
    },
    {
      "epoch": 7.220161673799335,
      "grad_norm": 0.20070408284664154,
      "learning_rate": 7.59225150841537e-05,
      "loss": 0.0663,
      "step": 3796
    },
    {
      "epoch": 7.222063718497385,
      "grad_norm": 0.2874523401260376,
      "learning_rate": 7.591616386154335e-05,
      "loss": 0.0729,
      "step": 3797
    },
    {
      "epoch": 7.2239657631954355,
      "grad_norm": 0.2688714563846588,
      "learning_rate": 7.5909812638933e-05,
      "loss": 0.0757,
      "step": 3798
    },
    {
      "epoch": 7.2258678078934855,
      "grad_norm": 0.19655251502990723,
      "learning_rate": 7.590346141632264e-05,
      "loss": 0.0566,
      "step": 3799
    },
    {
      "epoch": 7.2277698525915355,
      "grad_norm": 0.27301642298698425,
      "learning_rate": 7.589711019371229e-05,
      "loss": 0.0846,
      "step": 3800
    },
    {
      "epoch": 7.229671897289586,
      "grad_norm": 0.24339912831783295,
      "learning_rate": 7.589075897110194e-05,
      "loss": 0.0689,
      "step": 3801
    },
    {
      "epoch": 7.231573941987636,
      "grad_norm": 0.22319412231445312,
      "learning_rate": 7.58844077484916e-05,
      "loss": 0.0671,
      "step": 3802
    },
    {
      "epoch": 7.233475986685687,
      "grad_norm": 0.2316451221704483,
      "learning_rate": 7.587805652588123e-05,
      "loss": 0.0924,
      "step": 3803
    },
    {
      "epoch": 7.235378031383737,
      "grad_norm": 0.24457992613315582,
      "learning_rate": 7.587170530327088e-05,
      "loss": 0.0694,
      "step": 3804
    },
    {
      "epoch": 7.237280076081788,
      "grad_norm": 0.2121913731098175,
      "learning_rate": 7.586535408066054e-05,
      "loss": 0.1148,
      "step": 3805
    },
    {
      "epoch": 7.239182120779838,
      "grad_norm": 0.19238021969795227,
      "learning_rate": 7.585900285805017e-05,
      "loss": 0.0518,
      "step": 3806
    },
    {
      "epoch": 7.241084165477889,
      "grad_norm": 0.2916553318500519,
      "learning_rate": 7.585265163543983e-05,
      "loss": 0.0851,
      "step": 3807
    },
    {
      "epoch": 7.242986210175939,
      "grad_norm": 0.263918399810791,
      "learning_rate": 7.584630041282948e-05,
      "loss": 0.0828,
      "step": 3808
    },
    {
      "epoch": 7.24488825487399,
      "grad_norm": 0.2717205286026001,
      "learning_rate": 7.583994919021912e-05,
      "loss": 0.0808,
      "step": 3809
    },
    {
      "epoch": 7.24679029957204,
      "grad_norm": 0.37340980768203735,
      "learning_rate": 7.583359796760877e-05,
      "loss": 0.1054,
      "step": 3810
    },
    {
      "epoch": 7.24869234427009,
      "grad_norm": 0.3086722195148468,
      "learning_rate": 7.582724674499842e-05,
      "loss": 0.0744,
      "step": 3811
    },
    {
      "epoch": 7.250594388968141,
      "grad_norm": 0.3170717656612396,
      "learning_rate": 7.582089552238806e-05,
      "loss": 0.0709,
      "step": 3812
    },
    {
      "epoch": 7.252496433666191,
      "grad_norm": 0.3427761495113373,
      "learning_rate": 7.581454429977771e-05,
      "loss": 0.0853,
      "step": 3813
    },
    {
      "epoch": 7.254398478364242,
      "grad_norm": 0.36829718947410583,
      "learning_rate": 7.580819307716735e-05,
      "loss": 0.0911,
      "step": 3814
    },
    {
      "epoch": 7.256300523062292,
      "grad_norm": 0.2933104634284973,
      "learning_rate": 7.580184185455701e-05,
      "loss": 0.0788,
      "step": 3815
    },
    {
      "epoch": 7.258202567760343,
      "grad_norm": 0.30918237566947937,
      "learning_rate": 7.579549063194665e-05,
      "loss": 0.0897,
      "step": 3816
    },
    {
      "epoch": 7.260104612458393,
      "grad_norm": 0.27912604808807373,
      "learning_rate": 7.578913940933629e-05,
      "loss": 0.0781,
      "step": 3817
    },
    {
      "epoch": 7.2620066571564434,
      "grad_norm": 0.3042118549346924,
      "learning_rate": 7.578278818672596e-05,
      "loss": 0.0768,
      "step": 3818
    },
    {
      "epoch": 7.263908701854493,
      "grad_norm": 0.34230148792266846,
      "learning_rate": 7.57764369641156e-05,
      "loss": 0.0985,
      "step": 3819
    },
    {
      "epoch": 7.265810746552544,
      "grad_norm": 0.3558116853237152,
      "learning_rate": 7.577008574150525e-05,
      "loss": 0.0907,
      "step": 3820
    },
    {
      "epoch": 7.267712791250594,
      "grad_norm": 0.25006282329559326,
      "learning_rate": 7.576373451889488e-05,
      "loss": 0.0704,
      "step": 3821
    },
    {
      "epoch": 7.269614835948644,
      "grad_norm": 0.2956601679325104,
      "learning_rate": 7.575738329628454e-05,
      "loss": 0.1006,
      "step": 3822
    },
    {
      "epoch": 7.271516880646695,
      "grad_norm": 0.23409853875637054,
      "learning_rate": 7.575103207367419e-05,
      "loss": 0.0847,
      "step": 3823
    },
    {
      "epoch": 7.273418925344745,
      "grad_norm": 0.3059806227684021,
      "learning_rate": 7.574468085106383e-05,
      "loss": 0.0726,
      "step": 3824
    },
    {
      "epoch": 7.275320970042796,
      "grad_norm": 0.22190988063812256,
      "learning_rate": 7.573832962845349e-05,
      "loss": 0.083,
      "step": 3825
    },
    {
      "epoch": 7.277223014740846,
      "grad_norm": 0.25523841381073,
      "learning_rate": 7.573197840584313e-05,
      "loss": 0.1011,
      "step": 3826
    },
    {
      "epoch": 7.279125059438897,
      "grad_norm": 0.2877829372882843,
      "learning_rate": 7.572562718323277e-05,
      "loss": 0.1145,
      "step": 3827
    },
    {
      "epoch": 7.281027104136947,
      "grad_norm": 0.28623905777931213,
      "learning_rate": 7.571927596062242e-05,
      "loss": 0.0988,
      "step": 3828
    },
    {
      "epoch": 7.282929148834998,
      "grad_norm": 0.26607346534729004,
      "learning_rate": 7.571292473801207e-05,
      "loss": 0.0683,
      "step": 3829
    },
    {
      "epoch": 7.284831193533048,
      "grad_norm": 0.26464006304740906,
      "learning_rate": 7.570657351540171e-05,
      "loss": 0.0813,
      "step": 3830
    },
    {
      "epoch": 7.286733238231099,
      "grad_norm": 0.37109822034835815,
      "learning_rate": 7.570022229279136e-05,
      "loss": 0.0709,
      "step": 3831
    },
    {
      "epoch": 7.288635282929149,
      "grad_norm": 0.2537643313407898,
      "learning_rate": 7.569387107018101e-05,
      "loss": 0.0907,
      "step": 3832
    },
    {
      "epoch": 7.2905373276272,
      "grad_norm": 0.2658579647541046,
      "learning_rate": 7.568751984757067e-05,
      "loss": 0.0712,
      "step": 3833
    },
    {
      "epoch": 7.29243937232525,
      "grad_norm": 0.29602423310279846,
      "learning_rate": 7.56811686249603e-05,
      "loss": 0.0741,
      "step": 3834
    },
    {
      "epoch": 7.2943414170233,
      "grad_norm": 0.25278082489967346,
      "learning_rate": 7.567481740234996e-05,
      "loss": 0.0975,
      "step": 3835
    },
    {
      "epoch": 7.2962434617213505,
      "grad_norm": 0.2224166989326477,
      "learning_rate": 7.566846617973961e-05,
      "loss": 0.0606,
      "step": 3836
    },
    {
      "epoch": 7.2981455064194005,
      "grad_norm": 0.2645518481731415,
      "learning_rate": 7.566211495712925e-05,
      "loss": 0.0691,
      "step": 3837
    },
    {
      "epoch": 7.300047551117451,
      "grad_norm": 0.2529650628566742,
      "learning_rate": 7.56557637345189e-05,
      "loss": 0.0823,
      "step": 3838
    },
    {
      "epoch": 7.301949595815501,
      "grad_norm": 0.3536492586135864,
      "learning_rate": 7.564941251190855e-05,
      "loss": 0.0881,
      "step": 3839
    },
    {
      "epoch": 7.303851640513552,
      "grad_norm": 0.270697683095932,
      "learning_rate": 7.564306128929819e-05,
      "loss": 0.0852,
      "step": 3840
    },
    {
      "epoch": 7.305753685211602,
      "grad_norm": 0.18432219326496124,
      "learning_rate": 7.563671006668784e-05,
      "loss": 0.0491,
      "step": 3841
    },
    {
      "epoch": 7.307655729909653,
      "grad_norm": 0.2726588547229767,
      "learning_rate": 7.563035884407749e-05,
      "loss": 0.0813,
      "step": 3842
    },
    {
      "epoch": 7.309557774607703,
      "grad_norm": 0.2138698250055313,
      "learning_rate": 7.562400762146714e-05,
      "loss": 0.0522,
      "step": 3843
    },
    {
      "epoch": 7.311459819305754,
      "grad_norm": 0.24019719660282135,
      "learning_rate": 7.561765639885678e-05,
      "loss": 0.0901,
      "step": 3844
    },
    {
      "epoch": 7.313361864003804,
      "grad_norm": 0.4069961905479431,
      "learning_rate": 7.561130517624643e-05,
      "loss": 0.0882,
      "step": 3845
    },
    {
      "epoch": 7.315263908701855,
      "grad_norm": 0.29907310009002686,
      "learning_rate": 7.560495395363609e-05,
      "loss": 0.0733,
      "step": 3846
    },
    {
      "epoch": 7.317165953399905,
      "grad_norm": 0.29490938782691956,
      "learning_rate": 7.559860273102572e-05,
      "loss": 0.0554,
      "step": 3847
    },
    {
      "epoch": 7.319067998097955,
      "grad_norm": 0.2691021263599396,
      "learning_rate": 7.559225150841536e-05,
      "loss": 0.0824,
      "step": 3848
    },
    {
      "epoch": 7.320970042796006,
      "grad_norm": 0.2889997959136963,
      "learning_rate": 7.558590028580503e-05,
      "loss": 0.0751,
      "step": 3849
    },
    {
      "epoch": 7.322872087494056,
      "grad_norm": 0.27042216062545776,
      "learning_rate": 7.557954906319467e-05,
      "loss": 0.0683,
      "step": 3850
    },
    {
      "epoch": 7.324774132192107,
      "grad_norm": 0.2516496181488037,
      "learning_rate": 7.557319784058432e-05,
      "loss": 0.0821,
      "step": 3851
    },
    {
      "epoch": 7.326676176890157,
      "grad_norm": 0.25301793217658997,
      "learning_rate": 7.556684661797397e-05,
      "loss": 0.0684,
      "step": 3852
    },
    {
      "epoch": 7.328578221588208,
      "grad_norm": 0.20801399648189545,
      "learning_rate": 7.556049539536361e-05,
      "loss": 0.0594,
      "step": 3853
    },
    {
      "epoch": 7.330480266286258,
      "grad_norm": 0.2855316400527954,
      "learning_rate": 7.555414417275326e-05,
      "loss": 0.1444,
      "step": 3854
    },
    {
      "epoch": 7.3323823109843085,
      "grad_norm": 0.23078607022762299,
      "learning_rate": 7.55477929501429e-05,
      "loss": 0.0847,
      "step": 3855
    },
    {
      "epoch": 7.3342843556823585,
      "grad_norm": 0.29791387915611267,
      "learning_rate": 7.554144172753256e-05,
      "loss": 0.0811,
      "step": 3856
    },
    {
      "epoch": 7.336186400380409,
      "grad_norm": 0.3947182297706604,
      "learning_rate": 7.55350905049222e-05,
      "loss": 0.0792,
      "step": 3857
    },
    {
      "epoch": 7.338088445078459,
      "grad_norm": 0.29231664538383484,
      "learning_rate": 7.552873928231184e-05,
      "loss": 0.0867,
      "step": 3858
    },
    {
      "epoch": 7.339990489776509,
      "grad_norm": 0.27032309770584106,
      "learning_rate": 7.55223880597015e-05,
      "loss": 0.0728,
      "step": 3859
    },
    {
      "epoch": 7.34189253447456,
      "grad_norm": 0.27698999643325806,
      "learning_rate": 7.551603683709114e-05,
      "loss": 0.0661,
      "step": 3860
    },
    {
      "epoch": 7.34379457917261,
      "grad_norm": 0.2770868241786957,
      "learning_rate": 7.55096856144808e-05,
      "loss": 0.0598,
      "step": 3861
    },
    {
      "epoch": 7.345696623870661,
      "grad_norm": 0.26074889302253723,
      "learning_rate": 7.550333439187043e-05,
      "loss": 0.0703,
      "step": 3862
    },
    {
      "epoch": 7.347598668568711,
      "grad_norm": 0.1882147341966629,
      "learning_rate": 7.549698316926009e-05,
      "loss": 0.0684,
      "step": 3863
    },
    {
      "epoch": 7.349500713266762,
      "grad_norm": 0.3019726276397705,
      "learning_rate": 7.549063194664974e-05,
      "loss": 0.0762,
      "step": 3864
    },
    {
      "epoch": 7.351402757964812,
      "grad_norm": 0.32880455255508423,
      "learning_rate": 7.548428072403938e-05,
      "loss": 0.0786,
      "step": 3865
    },
    {
      "epoch": 7.353304802662863,
      "grad_norm": 0.280349463224411,
      "learning_rate": 7.547792950142903e-05,
      "loss": 0.0682,
      "step": 3866
    },
    {
      "epoch": 7.355206847360913,
      "grad_norm": 0.3259483277797699,
      "learning_rate": 7.547157827881868e-05,
      "loss": 0.0867,
      "step": 3867
    },
    {
      "epoch": 7.357108892058964,
      "grad_norm": 0.2592945992946625,
      "learning_rate": 7.546522705620832e-05,
      "loss": 0.0587,
      "step": 3868
    },
    {
      "epoch": 7.359010936757014,
      "grad_norm": 0.2538476288318634,
      "learning_rate": 7.545887583359797e-05,
      "loss": 0.0635,
      "step": 3869
    },
    {
      "epoch": 7.360912981455064,
      "grad_norm": 0.38723552227020264,
      "learning_rate": 7.545252461098762e-05,
      "loss": 0.0843,
      "step": 3870
    },
    {
      "epoch": 7.362815026153115,
      "grad_norm": 0.20521502196788788,
      "learning_rate": 7.544617338837726e-05,
      "loss": 0.0608,
      "step": 3871
    },
    {
      "epoch": 7.364717070851165,
      "grad_norm": 0.26786184310913086,
      "learning_rate": 7.543982216576691e-05,
      "loss": 0.0798,
      "step": 3872
    },
    {
      "epoch": 7.366619115549216,
      "grad_norm": 0.30265524983406067,
      "learning_rate": 7.543347094315656e-05,
      "loss": 0.0761,
      "step": 3873
    },
    {
      "epoch": 7.3685211602472656,
      "grad_norm": 0.383024126291275,
      "learning_rate": 7.542711972054622e-05,
      "loss": 0.1,
      "step": 3874
    },
    {
      "epoch": 7.370423204945316,
      "grad_norm": 0.3774031400680542,
      "learning_rate": 7.542076849793585e-05,
      "loss": 0.1151,
      "step": 3875
    },
    {
      "epoch": 7.372325249643366,
      "grad_norm": 0.1787770837545395,
      "learning_rate": 7.54144172753255e-05,
      "loss": 0.0774,
      "step": 3876
    },
    {
      "epoch": 7.374227294341417,
      "grad_norm": 0.3548794686794281,
      "learning_rate": 7.540806605271516e-05,
      "loss": 0.0982,
      "step": 3877
    },
    {
      "epoch": 7.376129339039467,
      "grad_norm": 0.4059542119503021,
      "learning_rate": 7.54017148301048e-05,
      "loss": 0.0912,
      "step": 3878
    },
    {
      "epoch": 7.378031383737518,
      "grad_norm": 0.5270293951034546,
      "learning_rate": 7.539536360749445e-05,
      "loss": 0.1042,
      "step": 3879
    },
    {
      "epoch": 7.379933428435568,
      "grad_norm": 0.3405616581439972,
      "learning_rate": 7.53890123848841e-05,
      "loss": 0.0899,
      "step": 3880
    },
    {
      "epoch": 7.381835473133618,
      "grad_norm": 0.3558886647224426,
      "learning_rate": 7.538266116227374e-05,
      "loss": 0.091,
      "step": 3881
    },
    {
      "epoch": 7.383737517831669,
      "grad_norm": 0.28821122646331787,
      "learning_rate": 7.537630993966339e-05,
      "loss": 0.0819,
      "step": 3882
    },
    {
      "epoch": 7.385639562529719,
      "grad_norm": 0.3417015075683594,
      "learning_rate": 7.536995871705304e-05,
      "loss": 0.078,
      "step": 3883
    },
    {
      "epoch": 7.38754160722777,
      "grad_norm": 0.32220759987831116,
      "learning_rate": 7.536360749444268e-05,
      "loss": 0.0865,
      "step": 3884
    },
    {
      "epoch": 7.38944365192582,
      "grad_norm": 0.3623081147670746,
      "learning_rate": 7.535725627183233e-05,
      "loss": 0.0844,
      "step": 3885
    },
    {
      "epoch": 7.391345696623871,
      "grad_norm": 0.2569948732852936,
      "learning_rate": 7.535090504922197e-05,
      "loss": 0.0932,
      "step": 3886
    },
    {
      "epoch": 7.393247741321921,
      "grad_norm": 0.33517327904701233,
      "learning_rate": 7.534455382661164e-05,
      "loss": 0.102,
      "step": 3887
    },
    {
      "epoch": 7.395149786019972,
      "grad_norm": 0.3194878101348877,
      "learning_rate": 7.533820260400127e-05,
      "loss": 0.0942,
      "step": 3888
    },
    {
      "epoch": 7.397051830718022,
      "grad_norm": 0.28731876611709595,
      "learning_rate": 7.533185138139091e-05,
      "loss": 0.0881,
      "step": 3889
    },
    {
      "epoch": 7.398953875416073,
      "grad_norm": 0.2113313227891922,
      "learning_rate": 7.532550015878058e-05,
      "loss": 0.0554,
      "step": 3890
    },
    {
      "epoch": 7.400855920114123,
      "grad_norm": 0.30572062730789185,
      "learning_rate": 7.531914893617022e-05,
      "loss": 0.0849,
      "step": 3891
    },
    {
      "epoch": 7.402757964812173,
      "grad_norm": 0.17145153880119324,
      "learning_rate": 7.531279771355987e-05,
      "loss": 0.07,
      "step": 3892
    },
    {
      "epoch": 7.4046600095102235,
      "grad_norm": 0.30596452951431274,
      "learning_rate": 7.53064464909495e-05,
      "loss": 0.0752,
      "step": 3893
    },
    {
      "epoch": 7.4065620542082735,
      "grad_norm": 0.2836361825466156,
      "learning_rate": 7.530009526833916e-05,
      "loss": 0.0864,
      "step": 3894
    },
    {
      "epoch": 7.408464098906324,
      "grad_norm": 0.37935930490493774,
      "learning_rate": 7.529374404572881e-05,
      "loss": 0.097,
      "step": 3895
    },
    {
      "epoch": 7.410366143604374,
      "grad_norm": 0.24081386625766754,
      "learning_rate": 7.528739282311845e-05,
      "loss": 0.0643,
      "step": 3896
    },
    {
      "epoch": 7.412268188302425,
      "grad_norm": 0.324744313955307,
      "learning_rate": 7.528104160050811e-05,
      "loss": 0.0892,
      "step": 3897
    },
    {
      "epoch": 7.414170233000475,
      "grad_norm": 0.29037338495254517,
      "learning_rate": 7.527469037789775e-05,
      "loss": 0.0792,
      "step": 3898
    },
    {
      "epoch": 7.416072277698526,
      "grad_norm": 0.20849809050559998,
      "learning_rate": 7.526833915528739e-05,
      "loss": 0.0688,
      "step": 3899
    },
    {
      "epoch": 7.417974322396576,
      "grad_norm": 0.4752673804759979,
      "learning_rate": 7.526198793267704e-05,
      "loss": 0.0848,
      "step": 3900
    },
    {
      "epoch": 7.419876367094627,
      "grad_norm": 0.20970579981803894,
      "learning_rate": 7.525563671006669e-05,
      "loss": 0.0685,
      "step": 3901
    },
    {
      "epoch": 7.421778411792677,
      "grad_norm": 0.2748827338218689,
      "learning_rate": 7.524928548745633e-05,
      "loss": 0.0869,
      "step": 3902
    },
    {
      "epoch": 7.423680456490728,
      "grad_norm": 0.4105650782585144,
      "learning_rate": 7.524293426484598e-05,
      "loss": 0.0907,
      "step": 3903
    },
    {
      "epoch": 7.425582501188778,
      "grad_norm": 0.4483408033847809,
      "learning_rate": 7.523658304223564e-05,
      "loss": 0.0933,
      "step": 3904
    },
    {
      "epoch": 7.427484545886828,
      "grad_norm": 0.2609012722969055,
      "learning_rate": 7.523023181962529e-05,
      "loss": 0.0711,
      "step": 3905
    },
    {
      "epoch": 7.429386590584879,
      "grad_norm": 0.2896825075149536,
      "learning_rate": 7.522388059701493e-05,
      "loss": 0.0669,
      "step": 3906
    },
    {
      "epoch": 7.431288635282929,
      "grad_norm": 0.3085344731807709,
      "learning_rate": 7.521752937440458e-05,
      "loss": 0.0827,
      "step": 3907
    },
    {
      "epoch": 7.43319067998098,
      "grad_norm": 0.3237287104129791,
      "learning_rate": 7.521117815179423e-05,
      "loss": 0.0827,
      "step": 3908
    },
    {
      "epoch": 7.43509272467903,
      "grad_norm": 0.27414318919181824,
      "learning_rate": 7.520482692918387e-05,
      "loss": 0.0728,
      "step": 3909
    },
    {
      "epoch": 7.436994769377081,
      "grad_norm": 0.23370763659477234,
      "learning_rate": 7.519847570657352e-05,
      "loss": 0.067,
      "step": 3910
    },
    {
      "epoch": 7.438896814075131,
      "grad_norm": 0.20903317630290985,
      "learning_rate": 7.519212448396317e-05,
      "loss": 0.057,
      "step": 3911
    },
    {
      "epoch": 7.4407988587731815,
      "grad_norm": 0.2630857527256012,
      "learning_rate": 7.518577326135281e-05,
      "loss": 0.0713,
      "step": 3912
    },
    {
      "epoch": 7.4427009034712315,
      "grad_norm": 0.27731242775917053,
      "learning_rate": 7.517942203874246e-05,
      "loss": 0.0804,
      "step": 3913
    },
    {
      "epoch": 7.444602948169282,
      "grad_norm": 0.2399328500032425,
      "learning_rate": 7.517307081613211e-05,
      "loss": 0.0696,
      "step": 3914
    },
    {
      "epoch": 7.446504992867332,
      "grad_norm": 0.3606988787651062,
      "learning_rate": 7.516671959352176e-05,
      "loss": 0.0662,
      "step": 3915
    },
    {
      "epoch": 7.448407037565383,
      "grad_norm": 0.30132564902305603,
      "learning_rate": 7.51603683709114e-05,
      "loss": 0.0925,
      "step": 3916
    },
    {
      "epoch": 7.450309082263433,
      "grad_norm": 0.36007988452911377,
      "learning_rate": 7.515401714830104e-05,
      "loss": 0.1102,
      "step": 3917
    },
    {
      "epoch": 7.452211126961483,
      "grad_norm": 0.2463337630033493,
      "learning_rate": 7.51476659256907e-05,
      "loss": 0.0769,
      "step": 3918
    },
    {
      "epoch": 7.454113171659534,
      "grad_norm": 0.21934674680233002,
      "learning_rate": 7.514131470308034e-05,
      "loss": 0.0603,
      "step": 3919
    },
    {
      "epoch": 7.456015216357584,
      "grad_norm": 0.3735393285751343,
      "learning_rate": 7.513496348046998e-05,
      "loss": 0.088,
      "step": 3920
    },
    {
      "epoch": 7.457917261055635,
      "grad_norm": 0.3590213656425476,
      "learning_rate": 7.512861225785965e-05,
      "loss": 0.0892,
      "step": 3921
    },
    {
      "epoch": 7.459819305753685,
      "grad_norm": 0.2909318506717682,
      "learning_rate": 7.512226103524929e-05,
      "loss": 0.0743,
      "step": 3922
    },
    {
      "epoch": 7.461721350451736,
      "grad_norm": 0.2760285437107086,
      "learning_rate": 7.511590981263894e-05,
      "loss": 0.0743,
      "step": 3923
    },
    {
      "epoch": 7.463623395149786,
      "grad_norm": 0.20594924688339233,
      "learning_rate": 7.510955859002858e-05,
      "loss": 0.0712,
      "step": 3924
    },
    {
      "epoch": 7.465525439847837,
      "grad_norm": 0.36622461676597595,
      "learning_rate": 7.510320736741823e-05,
      "loss": 0.0796,
      "step": 3925
    },
    {
      "epoch": 7.467427484545887,
      "grad_norm": 0.47549140453338623,
      "learning_rate": 7.509685614480788e-05,
      "loss": 0.1338,
      "step": 3926
    },
    {
      "epoch": 7.469329529243938,
      "grad_norm": 0.22030964493751526,
      "learning_rate": 7.509050492219752e-05,
      "loss": 0.0696,
      "step": 3927
    },
    {
      "epoch": 7.471231573941988,
      "grad_norm": 0.3158624470233917,
      "learning_rate": 7.508415369958718e-05,
      "loss": 0.0804,
      "step": 3928
    },
    {
      "epoch": 7.473133618640038,
      "grad_norm": 0.28002792596817017,
      "learning_rate": 7.507780247697682e-05,
      "loss": 0.0859,
      "step": 3929
    },
    {
      "epoch": 7.4750356633380886,
      "grad_norm": 0.42471325397491455,
      "learning_rate": 7.507145125436646e-05,
      "loss": 0.0943,
      "step": 3930
    },
    {
      "epoch": 7.4769377080361386,
      "grad_norm": 0.25244924426078796,
      "learning_rate": 7.506510003175611e-05,
      "loss": 0.0683,
      "step": 3931
    },
    {
      "epoch": 7.478839752734189,
      "grad_norm": 0.22861167788505554,
      "learning_rate": 7.505874880914576e-05,
      "loss": 0.0601,
      "step": 3932
    },
    {
      "epoch": 7.480741797432239,
      "grad_norm": 0.2393944412469864,
      "learning_rate": 7.505239758653542e-05,
      "loss": 0.0565,
      "step": 3933
    },
    {
      "epoch": 7.48264384213029,
      "grad_norm": 0.22203993797302246,
      "learning_rate": 7.504604636392505e-05,
      "loss": 0.078,
      "step": 3934
    },
    {
      "epoch": 7.48454588682834,
      "grad_norm": 0.22644877433776855,
      "learning_rate": 7.50396951413147e-05,
      "loss": 0.0793,
      "step": 3935
    },
    {
      "epoch": 7.486447931526391,
      "grad_norm": 0.2989737093448639,
      "learning_rate": 7.503334391870436e-05,
      "loss": 0.0893,
      "step": 3936
    },
    {
      "epoch": 7.488349976224441,
      "grad_norm": 0.2532472014427185,
      "learning_rate": 7.5026992696094e-05,
      "loss": 0.0892,
      "step": 3937
    },
    {
      "epoch": 7.490252020922492,
      "grad_norm": 0.29805564880371094,
      "learning_rate": 7.502064147348365e-05,
      "loss": 0.0744,
      "step": 3938
    },
    {
      "epoch": 7.492154065620542,
      "grad_norm": 0.2866828441619873,
      "learning_rate": 7.50142902508733e-05,
      "loss": 0.0854,
      "step": 3939
    },
    {
      "epoch": 7.494056110318592,
      "grad_norm": 0.4060494005680084,
      "learning_rate": 7.500793902826294e-05,
      "loss": 0.1008,
      "step": 3940
    },
    {
      "epoch": 7.495958155016643,
      "grad_norm": 0.24941779673099518,
      "learning_rate": 7.500158780565259e-05,
      "loss": 0.07,
      "step": 3941
    },
    {
      "epoch": 7.497860199714693,
      "grad_norm": 0.260382741689682,
      "learning_rate": 7.499523658304224e-05,
      "loss": 0.0799,
      "step": 3942
    },
    {
      "epoch": 7.499762244412744,
      "grad_norm": 0.24394255876541138,
      "learning_rate": 7.498888536043188e-05,
      "loss": 0.0607,
      "step": 3943
    },
    {
      "epoch": 7.501664289110794,
      "grad_norm": 0.3588791489601135,
      "learning_rate": 7.498253413782153e-05,
      "loss": 0.0862,
      "step": 3944
    },
    {
      "epoch": 7.503566333808845,
      "grad_norm": 0.29977500438690186,
      "learning_rate": 7.497618291521118e-05,
      "loss": 0.0885,
      "step": 3945
    },
    {
      "epoch": 7.505468378506895,
      "grad_norm": 0.37303635478019714,
      "learning_rate": 7.496983169260084e-05,
      "loss": 0.0879,
      "step": 3946
    },
    {
      "epoch": 7.507370423204946,
      "grad_norm": 0.2976795434951782,
      "learning_rate": 7.496348046999047e-05,
      "loss": 0.0796,
      "step": 3947
    },
    {
      "epoch": 7.509272467902996,
      "grad_norm": 0.2982828617095947,
      "learning_rate": 7.495712924738013e-05,
      "loss": 0.0813,
      "step": 3948
    },
    {
      "epoch": 7.5111745126010465,
      "grad_norm": 0.2461419552564621,
      "learning_rate": 7.495077802476978e-05,
      "loss": 0.0843,
      "step": 3949
    },
    {
      "epoch": 7.5130765572990965,
      "grad_norm": 0.3808850646018982,
      "learning_rate": 7.494442680215942e-05,
      "loss": 0.0855,
      "step": 3950
    },
    {
      "epoch": 7.5149786019971465,
      "grad_norm": 0.5055833458900452,
      "learning_rate": 7.493807557954907e-05,
      "loss": 0.1101,
      "step": 3951
    },
    {
      "epoch": 7.516880646695197,
      "grad_norm": 0.22496581077575684,
      "learning_rate": 7.493172435693872e-05,
      "loss": 0.124,
      "step": 3952
    },
    {
      "epoch": 7.518782691393247,
      "grad_norm": 0.29180511832237244,
      "learning_rate": 7.492537313432836e-05,
      "loss": 0.0865,
      "step": 3953
    },
    {
      "epoch": 7.520684736091298,
      "grad_norm": 0.35880160331726074,
      "learning_rate": 7.491902191171801e-05,
      "loss": 0.1123,
      "step": 3954
    },
    {
      "epoch": 7.522586780789348,
      "grad_norm": 0.36466071009635925,
      "learning_rate": 7.491267068910766e-05,
      "loss": 0.1021,
      "step": 3955
    },
    {
      "epoch": 7.524488825487399,
      "grad_norm": 0.3240335285663605,
      "learning_rate": 7.49063194664973e-05,
      "loss": 0.0757,
      "step": 3956
    },
    {
      "epoch": 7.526390870185449,
      "grad_norm": 0.28873270750045776,
      "learning_rate": 7.489996824388695e-05,
      "loss": 0.1093,
      "step": 3957
    },
    {
      "epoch": 7.5282929148835,
      "grad_norm": 0.3459244668483734,
      "learning_rate": 7.489361702127659e-05,
      "loss": 0.0997,
      "step": 3958
    },
    {
      "epoch": 7.53019495958155,
      "grad_norm": 0.3284059762954712,
      "learning_rate": 7.488726579866626e-05,
      "loss": 0.0948,
      "step": 3959
    },
    {
      "epoch": 7.532097004279601,
      "grad_norm": 0.3020002245903015,
      "learning_rate": 7.48809145760559e-05,
      "loss": 0.0847,
      "step": 3960
    },
    {
      "epoch": 7.533999048977651,
      "grad_norm": 0.2214907556772232,
      "learning_rate": 7.487456335344553e-05,
      "loss": 0.0719,
      "step": 3961
    },
    {
      "epoch": 7.535901093675701,
      "grad_norm": 0.19691269099712372,
      "learning_rate": 7.48682121308352e-05,
      "loss": 0.0491,
      "step": 3962
    },
    {
      "epoch": 7.537803138373752,
      "grad_norm": 0.2436630129814148,
      "learning_rate": 7.486186090822484e-05,
      "loss": 0.0696,
      "step": 3963
    },
    {
      "epoch": 7.539705183071802,
      "grad_norm": 0.3241187632083893,
      "learning_rate": 7.485550968561449e-05,
      "loss": 0.1074,
      "step": 3964
    },
    {
      "epoch": 7.541607227769853,
      "grad_norm": 0.2550383508205414,
      "learning_rate": 7.484915846300413e-05,
      "loss": 0.0729,
      "step": 3965
    },
    {
      "epoch": 7.543509272467903,
      "grad_norm": 0.28291863203048706,
      "learning_rate": 7.484280724039378e-05,
      "loss": 0.0704,
      "step": 3966
    },
    {
      "epoch": 7.545411317165954,
      "grad_norm": 0.3558233380317688,
      "learning_rate": 7.483645601778343e-05,
      "loss": 0.0924,
      "step": 3967
    },
    {
      "epoch": 7.547313361864004,
      "grad_norm": 0.29661667346954346,
      "learning_rate": 7.483010479517307e-05,
      "loss": 0.0755,
      "step": 3968
    },
    {
      "epoch": 7.5492154065620545,
      "grad_norm": 0.2704700231552124,
      "learning_rate": 7.482375357256272e-05,
      "loss": 0.0869,
      "step": 3969
    },
    {
      "epoch": 7.5511174512601045,
      "grad_norm": 0.23262625932693481,
      "learning_rate": 7.481740234995237e-05,
      "loss": 0.06,
      "step": 3970
    },
    {
      "epoch": 7.553019495958155,
      "grad_norm": 0.24703703820705414,
      "learning_rate": 7.481105112734201e-05,
      "loss": 0.0617,
      "step": 3971
    },
    {
      "epoch": 7.554921540656205,
      "grad_norm": 0.3128037750720978,
      "learning_rate": 7.480469990473166e-05,
      "loss": 0.085,
      "step": 3972
    },
    {
      "epoch": 7.556823585354255,
      "grad_norm": 0.27781766653060913,
      "learning_rate": 7.479834868212131e-05,
      "loss": 0.0731,
      "step": 3973
    },
    {
      "epoch": 7.558725630052306,
      "grad_norm": 0.21794871985912323,
      "learning_rate": 7.479199745951095e-05,
      "loss": 0.064,
      "step": 3974
    },
    {
      "epoch": 7.560627674750357,
      "grad_norm": 0.332007497549057,
      "learning_rate": 7.47856462369006e-05,
      "loss": 0.1261,
      "step": 3975
    },
    {
      "epoch": 7.562529719448407,
      "grad_norm": 0.36492758989334106,
      "learning_rate": 7.477929501429026e-05,
      "loss": 0.0869,
      "step": 3976
    },
    {
      "epoch": 7.564431764146457,
      "grad_norm": 0.36482545733451843,
      "learning_rate": 7.477294379167991e-05,
      "loss": 0.0956,
      "step": 3977
    },
    {
      "epoch": 7.566333808844508,
      "grad_norm": 0.3136618137359619,
      "learning_rate": 7.476659256906955e-05,
      "loss": 0.0724,
      "step": 3978
    },
    {
      "epoch": 7.568235853542558,
      "grad_norm": 0.1828143447637558,
      "learning_rate": 7.47602413464592e-05,
      "loss": 0.0656,
      "step": 3979
    },
    {
      "epoch": 7.570137898240609,
      "grad_norm": 0.2850959599018097,
      "learning_rate": 7.475389012384885e-05,
      "loss": 0.0807,
      "step": 3980
    },
    {
      "epoch": 7.572039942938659,
      "grad_norm": 0.23250633478164673,
      "learning_rate": 7.474753890123849e-05,
      "loss": 0.0879,
      "step": 3981
    },
    {
      "epoch": 7.57394198763671,
      "grad_norm": 0.22114038467407227,
      "learning_rate": 7.474118767862814e-05,
      "loss": 0.0881,
      "step": 3982
    },
    {
      "epoch": 7.57584403233476,
      "grad_norm": 0.304195374250412,
      "learning_rate": 7.473483645601779e-05,
      "loss": 0.0854,
      "step": 3983
    },
    {
      "epoch": 7.57774607703281,
      "grad_norm": 0.3512796461582184,
      "learning_rate": 7.472848523340743e-05,
      "loss": 0.0984,
      "step": 3984
    },
    {
      "epoch": 7.579648121730861,
      "grad_norm": 0.19406934082508087,
      "learning_rate": 7.472213401079708e-05,
      "loss": 0.0527,
      "step": 3985
    },
    {
      "epoch": 7.5815501664289116,
      "grad_norm": 0.35052844882011414,
      "learning_rate": 7.471578278818673e-05,
      "loss": 0.093,
      "step": 3986
    },
    {
      "epoch": 7.5834522111269616,
      "grad_norm": 0.3175174593925476,
      "learning_rate": 7.470943156557639e-05,
      "loss": 0.0765,
      "step": 3987
    },
    {
      "epoch": 7.5853542558250115,
      "grad_norm": 0.2843180298805237,
      "learning_rate": 7.470308034296602e-05,
      "loss": 0.0794,
      "step": 3988
    },
    {
      "epoch": 7.587256300523062,
      "grad_norm": 0.25421711802482605,
      "learning_rate": 7.469672912035566e-05,
      "loss": 0.0559,
      "step": 3989
    },
    {
      "epoch": 7.589158345221112,
      "grad_norm": 0.26305899024009705,
      "learning_rate": 7.469037789774533e-05,
      "loss": 0.083,
      "step": 3990
    },
    {
      "epoch": 7.591060389919163,
      "grad_norm": 0.2685282528400421,
      "learning_rate": 7.468402667513497e-05,
      "loss": 0.0735,
      "step": 3991
    },
    {
      "epoch": 7.592962434617213,
      "grad_norm": 0.2922079265117645,
      "learning_rate": 7.46776754525246e-05,
      "loss": 0.0752,
      "step": 3992
    },
    {
      "epoch": 7.594864479315264,
      "grad_norm": 0.4285946190357208,
      "learning_rate": 7.467132422991427e-05,
      "loss": 0.0875,
      "step": 3993
    },
    {
      "epoch": 7.596766524013314,
      "grad_norm": 0.3181219696998596,
      "learning_rate": 7.466497300730391e-05,
      "loss": 0.0664,
      "step": 3994
    },
    {
      "epoch": 7.598668568711365,
      "grad_norm": 0.2952035069465637,
      "learning_rate": 7.465862178469356e-05,
      "loss": 0.0814,
      "step": 3995
    },
    {
      "epoch": 7.600570613409415,
      "grad_norm": 0.2553558647632599,
      "learning_rate": 7.46522705620832e-05,
      "loss": 0.066,
      "step": 3996
    },
    {
      "epoch": 7.602472658107466,
      "grad_norm": 0.23761653900146484,
      "learning_rate": 7.464591933947285e-05,
      "loss": 0.0642,
      "step": 3997
    },
    {
      "epoch": 7.604374702805516,
      "grad_norm": 0.2953890562057495,
      "learning_rate": 7.46395681168625e-05,
      "loss": 0.0888,
      "step": 3998
    },
    {
      "epoch": 7.606276747503566,
      "grad_norm": 0.29672572016716003,
      "learning_rate": 7.463321689425214e-05,
      "loss": 0.0856,
      "step": 3999
    },
    {
      "epoch": 7.608178792201617,
      "grad_norm": 0.48846060037612915,
      "learning_rate": 7.46268656716418e-05,
      "loss": 0.1094,
      "step": 4000
    },
    {
      "epoch": 7.610080836899667,
      "grad_norm": 0.27442026138305664,
      "learning_rate": 7.462051444903144e-05,
      "loss": 0.0699,
      "step": 4001
    },
    {
      "epoch": 7.611982881597718,
      "grad_norm": 0.27723994851112366,
      "learning_rate": 7.461416322642108e-05,
      "loss": 0.0826,
      "step": 4002
    },
    {
      "epoch": 7.613884926295768,
      "grad_norm": 0.19016140699386597,
      "learning_rate": 7.460781200381073e-05,
      "loss": 0.0863,
      "step": 4003
    },
    {
      "epoch": 7.615786970993819,
      "grad_norm": 0.2775978744029999,
      "learning_rate": 7.460146078120039e-05,
      "loss": 0.0775,
      "step": 4004
    },
    {
      "epoch": 7.617689015691869,
      "grad_norm": 0.23393043875694275,
      "learning_rate": 7.459510955859004e-05,
      "loss": 0.0774,
      "step": 4005
    },
    {
      "epoch": 7.6195910603899195,
      "grad_norm": 0.3676578104496002,
      "learning_rate": 7.458875833597968e-05,
      "loss": 0.0849,
      "step": 4006
    },
    {
      "epoch": 7.6214931050879695,
      "grad_norm": 0.22967612743377686,
      "learning_rate": 7.458240711336933e-05,
      "loss": 0.0585,
      "step": 4007
    },
    {
      "epoch": 7.62339514978602,
      "grad_norm": 0.20074373483657837,
      "learning_rate": 7.457605589075898e-05,
      "loss": 0.0756,
      "step": 4008
    },
    {
      "epoch": 7.62529719448407,
      "grad_norm": 0.32885944843292236,
      "learning_rate": 7.456970466814862e-05,
      "loss": 0.094,
      "step": 4009
    },
    {
      "epoch": 7.62719923918212,
      "grad_norm": 0.34579211473464966,
      "learning_rate": 7.456335344553827e-05,
      "loss": 0.0667,
      "step": 4010
    },
    {
      "epoch": 7.629101283880171,
      "grad_norm": 0.26802295446395874,
      "learning_rate": 7.455700222292792e-05,
      "loss": 0.073,
      "step": 4011
    },
    {
      "epoch": 7.631003328578221,
      "grad_norm": 0.315384179353714,
      "learning_rate": 7.455065100031756e-05,
      "loss": 0.0792,
      "step": 4012
    },
    {
      "epoch": 7.632905373276272,
      "grad_norm": 0.35064810514450073,
      "learning_rate": 7.454429977770721e-05,
      "loss": 0.0859,
      "step": 4013
    },
    {
      "epoch": 7.634807417974322,
      "grad_norm": 0.3940977156162262,
      "learning_rate": 7.453794855509686e-05,
      "loss": 0.0899,
      "step": 4014
    },
    {
      "epoch": 7.636709462672373,
      "grad_norm": 0.28424572944641113,
      "learning_rate": 7.45315973324865e-05,
      "loss": 0.0826,
      "step": 4015
    },
    {
      "epoch": 7.638611507370423,
      "grad_norm": 0.2842091917991638,
      "learning_rate": 7.452524610987615e-05,
      "loss": 0.0727,
      "step": 4016
    },
    {
      "epoch": 7.640513552068474,
      "grad_norm": 0.2675850987434387,
      "learning_rate": 7.45188948872658e-05,
      "loss": 0.085,
      "step": 4017
    },
    {
      "epoch": 7.642415596766524,
      "grad_norm": 0.20844288170337677,
      "learning_rate": 7.451254366465546e-05,
      "loss": 0.0662,
      "step": 4018
    },
    {
      "epoch": 7.644317641464575,
      "grad_norm": 0.1705649346113205,
      "learning_rate": 7.45061924420451e-05,
      "loss": 0.0673,
      "step": 4019
    },
    {
      "epoch": 7.646219686162625,
      "grad_norm": 0.22716879844665527,
      "learning_rate": 7.449984121943475e-05,
      "loss": 0.0737,
      "step": 4020
    },
    {
      "epoch": 7.648121730860675,
      "grad_norm": 0.22440075874328613,
      "learning_rate": 7.44934899968244e-05,
      "loss": 0.082,
      "step": 4021
    },
    {
      "epoch": 7.650023775558726,
      "grad_norm": 0.18389202654361725,
      "learning_rate": 7.448713877421404e-05,
      "loss": 0.056,
      "step": 4022
    },
    {
      "epoch": 7.651925820256776,
      "grad_norm": 0.32997074723243713,
      "learning_rate": 7.448078755160369e-05,
      "loss": 0.0852,
      "step": 4023
    },
    {
      "epoch": 7.653827864954827,
      "grad_norm": 0.2446168065071106,
      "learning_rate": 7.447443632899334e-05,
      "loss": 0.0612,
      "step": 4024
    },
    {
      "epoch": 7.655729909652877,
      "grad_norm": 0.2455371469259262,
      "learning_rate": 7.446808510638298e-05,
      "loss": 0.0595,
      "step": 4025
    },
    {
      "epoch": 7.6576319543509275,
      "grad_norm": 0.3117963373661041,
      "learning_rate": 7.446173388377263e-05,
      "loss": 0.0869,
      "step": 4026
    },
    {
      "epoch": 7.6595339990489775,
      "grad_norm": 0.2107362449169159,
      "learning_rate": 7.445538266116227e-05,
      "loss": 0.0726,
      "step": 4027
    },
    {
      "epoch": 7.661436043747028,
      "grad_norm": 0.21624886989593506,
      "learning_rate": 7.444903143855192e-05,
      "loss": 0.0721,
      "step": 4028
    },
    {
      "epoch": 7.663338088445078,
      "grad_norm": 0.30223172903060913,
      "learning_rate": 7.444268021594157e-05,
      "loss": 0.1137,
      "step": 4029
    },
    {
      "epoch": 7.665240133143129,
      "grad_norm": 0.3354337513446808,
      "learning_rate": 7.443632899333121e-05,
      "loss": 0.0718,
      "step": 4030
    },
    {
      "epoch": 7.667142177841179,
      "grad_norm": 0.20831426978111267,
      "learning_rate": 7.442997777072088e-05,
      "loss": 0.0696,
      "step": 4031
    },
    {
      "epoch": 7.669044222539229,
      "grad_norm": 0.222997784614563,
      "learning_rate": 7.442362654811052e-05,
      "loss": 0.0716,
      "step": 4032
    },
    {
      "epoch": 7.67094626723728,
      "grad_norm": 0.29635244607925415,
      "learning_rate": 7.441727532550015e-05,
      "loss": 0.1017,
      "step": 4033
    },
    {
      "epoch": 7.67284831193533,
      "grad_norm": 0.24566298723220825,
      "learning_rate": 7.44109241028898e-05,
      "loss": 0.0757,
      "step": 4034
    },
    {
      "epoch": 7.674750356633381,
      "grad_norm": 0.21996475756168365,
      "learning_rate": 7.440457288027946e-05,
      "loss": 0.0899,
      "step": 4035
    },
    {
      "epoch": 7.676652401331431,
      "grad_norm": 0.36381396651268005,
      "learning_rate": 7.439822165766911e-05,
      "loss": 0.0827,
      "step": 4036
    },
    {
      "epoch": 7.678554446029482,
      "grad_norm": 0.2817314863204956,
      "learning_rate": 7.439187043505875e-05,
      "loss": 0.071,
      "step": 4037
    },
    {
      "epoch": 7.680456490727532,
      "grad_norm": 0.24392902851104736,
      "learning_rate": 7.43855192124484e-05,
      "loss": 0.0575,
      "step": 4038
    },
    {
      "epoch": 7.682358535425583,
      "grad_norm": 0.22329150140285492,
      "learning_rate": 7.437916798983805e-05,
      "loss": 0.0659,
      "step": 4039
    },
    {
      "epoch": 7.684260580123633,
      "grad_norm": 0.302490770816803,
      "learning_rate": 7.437281676722769e-05,
      "loss": 0.0863,
      "step": 4040
    },
    {
      "epoch": 7.686162624821684,
      "grad_norm": 0.2640133202075958,
      "learning_rate": 7.436646554461734e-05,
      "loss": 0.0977,
      "step": 4041
    },
    {
      "epoch": 7.688064669519734,
      "grad_norm": 0.26939862966537476,
      "learning_rate": 7.436011432200699e-05,
      "loss": 0.0791,
      "step": 4042
    },
    {
      "epoch": 7.689966714217784,
      "grad_norm": 0.2610327899456024,
      "learning_rate": 7.435376309939663e-05,
      "loss": 0.0539,
      "step": 4043
    },
    {
      "epoch": 7.6918687589158345,
      "grad_norm": 0.2487347424030304,
      "learning_rate": 7.434741187678628e-05,
      "loss": 0.0661,
      "step": 4044
    },
    {
      "epoch": 7.6937708036138845,
      "grad_norm": 0.34148916602134705,
      "learning_rate": 7.434106065417593e-05,
      "loss": 0.083,
      "step": 4045
    },
    {
      "epoch": 7.695672848311935,
      "grad_norm": 0.28194597363471985,
      "learning_rate": 7.433470943156557e-05,
      "loss": 0.0681,
      "step": 4046
    },
    {
      "epoch": 7.697574893009985,
      "grad_norm": 0.2983434200286865,
      "learning_rate": 7.432835820895523e-05,
      "loss": 0.1014,
      "step": 4047
    },
    {
      "epoch": 7.699476937708036,
      "grad_norm": 0.27884411811828613,
      "learning_rate": 7.432200698634488e-05,
      "loss": 0.0711,
      "step": 4048
    },
    {
      "epoch": 7.701378982406086,
      "grad_norm": 0.27753472328186035,
      "learning_rate": 7.431565576373453e-05,
      "loss": 0.0815,
      "step": 4049
    },
    {
      "epoch": 7.703281027104137,
      "grad_norm": 0.33670538663864136,
      "learning_rate": 7.430930454112417e-05,
      "loss": 0.0829,
      "step": 4050
    },
    {
      "epoch": 7.705183071802187,
      "grad_norm": 0.38544511795043945,
      "learning_rate": 7.430295331851382e-05,
      "loss": 0.0764,
      "step": 4051
    },
    {
      "epoch": 7.707085116500238,
      "grad_norm": 0.2954690158367157,
      "learning_rate": 7.429660209590347e-05,
      "loss": 0.0779,
      "step": 4052
    },
    {
      "epoch": 7.708987161198288,
      "grad_norm": 0.2750054597854614,
      "learning_rate": 7.429025087329311e-05,
      "loss": 0.0668,
      "step": 4053
    },
    {
      "epoch": 7.710889205896338,
      "grad_norm": 0.27135616540908813,
      "learning_rate": 7.428389965068276e-05,
      "loss": 0.0936,
      "step": 4054
    },
    {
      "epoch": 7.712791250594389,
      "grad_norm": 0.25627055764198303,
      "learning_rate": 7.427754842807241e-05,
      "loss": 0.0585,
      "step": 4055
    },
    {
      "epoch": 7.71469329529244,
      "grad_norm": 0.3443787395954132,
      "learning_rate": 7.427119720546205e-05,
      "loss": 0.0825,
      "step": 4056
    },
    {
      "epoch": 7.71659533999049,
      "grad_norm": 0.23502713441848755,
      "learning_rate": 7.42648459828517e-05,
      "loss": 0.0772,
      "step": 4057
    },
    {
      "epoch": 7.71849738468854,
      "grad_norm": 0.28659313917160034,
      "learning_rate": 7.425849476024135e-05,
      "loss": 0.1007,
      "step": 4058
    },
    {
      "epoch": 7.720399429386591,
      "grad_norm": 0.2622101902961731,
      "learning_rate": 7.4252143537631e-05,
      "loss": 0.0751,
      "step": 4059
    },
    {
      "epoch": 7.722301474084641,
      "grad_norm": 0.2769562304019928,
      "learning_rate": 7.424579231502064e-05,
      "loss": 0.0744,
      "step": 4060
    },
    {
      "epoch": 7.724203518782692,
      "grad_norm": 0.29820317029953003,
      "learning_rate": 7.423944109241028e-05,
      "loss": 0.0968,
      "step": 4061
    },
    {
      "epoch": 7.726105563480742,
      "grad_norm": 0.1930849850177765,
      "learning_rate": 7.423308986979995e-05,
      "loss": 0.0622,
      "step": 4062
    },
    {
      "epoch": 7.7280076081787925,
      "grad_norm": 0.3451099395751953,
      "learning_rate": 7.422673864718959e-05,
      "loss": 0.1056,
      "step": 4063
    },
    {
      "epoch": 7.7299096528768425,
      "grad_norm": 0.4290725588798523,
      "learning_rate": 7.422038742457923e-05,
      "loss": 0.0813,
      "step": 4064
    },
    {
      "epoch": 7.731811697574893,
      "grad_norm": 0.25179606676101685,
      "learning_rate": 7.421403620196889e-05,
      "loss": 0.0713,
      "step": 4065
    },
    {
      "epoch": 7.733713742272943,
      "grad_norm": 0.30341050028800964,
      "learning_rate": 7.420768497935853e-05,
      "loss": 0.0797,
      "step": 4066
    },
    {
      "epoch": 7.735615786970994,
      "grad_norm": 0.2818777561187744,
      "learning_rate": 7.420133375674818e-05,
      "loss": 0.0844,
      "step": 4067
    },
    {
      "epoch": 7.737517831669044,
      "grad_norm": 0.2710643410682678,
      "learning_rate": 7.419498253413782e-05,
      "loss": 0.085,
      "step": 4068
    },
    {
      "epoch": 7.739419876367094,
      "grad_norm": 0.29590675234794617,
      "learning_rate": 7.418863131152747e-05,
      "loss": 0.0844,
      "step": 4069
    },
    {
      "epoch": 7.741321921065145,
      "grad_norm": 0.26093828678131104,
      "learning_rate": 7.418228008891712e-05,
      "loss": 0.0927,
      "step": 4070
    },
    {
      "epoch": 7.743223965763195,
      "grad_norm": 0.20662081241607666,
      "learning_rate": 7.417592886630676e-05,
      "loss": 0.0842,
      "step": 4071
    },
    {
      "epoch": 7.745126010461246,
      "grad_norm": 0.2244255244731903,
      "learning_rate": 7.416957764369643e-05,
      "loss": 0.0691,
      "step": 4072
    },
    {
      "epoch": 7.747028055159296,
      "grad_norm": 0.1915290802717209,
      "learning_rate": 7.416322642108606e-05,
      "loss": 0.0765,
      "step": 4073
    },
    {
      "epoch": 7.748930099857347,
      "grad_norm": 0.34651103615760803,
      "learning_rate": 7.41568751984757e-05,
      "loss": 0.0809,
      "step": 4074
    },
    {
      "epoch": 7.750832144555397,
      "grad_norm": 0.2221180945634842,
      "learning_rate": 7.415052397586535e-05,
      "loss": 0.0618,
      "step": 4075
    },
    {
      "epoch": 7.752734189253448,
      "grad_norm": 0.2723255455493927,
      "learning_rate": 7.4144172753255e-05,
      "loss": 0.0921,
      "step": 4076
    },
    {
      "epoch": 7.754636233951498,
      "grad_norm": 0.2689955234527588,
      "learning_rate": 7.413782153064466e-05,
      "loss": 0.06,
      "step": 4077
    },
    {
      "epoch": 7.756538278649549,
      "grad_norm": 0.2710064649581909,
      "learning_rate": 7.41314703080343e-05,
      "loss": 0.0786,
      "step": 4078
    },
    {
      "epoch": 7.758440323347599,
      "grad_norm": 0.2953267991542816,
      "learning_rate": 7.412511908542395e-05,
      "loss": 0.0662,
      "step": 4079
    },
    {
      "epoch": 7.760342368045649,
      "grad_norm": 0.40185266733169556,
      "learning_rate": 7.41187678628136e-05,
      "loss": 0.0758,
      "step": 4080
    },
    {
      "epoch": 7.7622444127437,
      "grad_norm": 0.3208361268043518,
      "learning_rate": 7.411241664020324e-05,
      "loss": 0.091,
      "step": 4081
    },
    {
      "epoch": 7.76414645744175,
      "grad_norm": 0.20456193387508392,
      "learning_rate": 7.410606541759289e-05,
      "loss": 0.062,
      "step": 4082
    },
    {
      "epoch": 7.7660485021398005,
      "grad_norm": 0.38072583079338074,
      "learning_rate": 7.409971419498254e-05,
      "loss": 0.0859,
      "step": 4083
    },
    {
      "epoch": 7.7679505468378505,
      "grad_norm": 0.19965219497680664,
      "learning_rate": 7.409336297237218e-05,
      "loss": 0.0632,
      "step": 4084
    },
    {
      "epoch": 7.769852591535901,
      "grad_norm": 0.3247738778591156,
      "learning_rate": 7.408701174976183e-05,
      "loss": 0.0902,
      "step": 4085
    },
    {
      "epoch": 7.771754636233951,
      "grad_norm": 0.3265827000141144,
      "learning_rate": 7.408066052715148e-05,
      "loss": 0.0849,
      "step": 4086
    },
    {
      "epoch": 7.773656680932002,
      "grad_norm": 0.254866361618042,
      "learning_rate": 7.407430930454112e-05,
      "loss": 0.0721,
      "step": 4087
    },
    {
      "epoch": 7.775558725630052,
      "grad_norm": 0.24622495472431183,
      "learning_rate": 7.406795808193077e-05,
      "loss": 0.0744,
      "step": 4088
    },
    {
      "epoch": 7.777460770328103,
      "grad_norm": 0.3360063135623932,
      "learning_rate": 7.406160685932043e-05,
      "loss": 0.0869,
      "step": 4089
    },
    {
      "epoch": 7.779362815026153,
      "grad_norm": 0.23800595104694366,
      "learning_rate": 7.405525563671008e-05,
      "loss": 0.0662,
      "step": 4090
    },
    {
      "epoch": 7.781264859724203,
      "grad_norm": 0.2821212112903595,
      "learning_rate": 7.404890441409972e-05,
      "loss": 0.0739,
      "step": 4091
    },
    {
      "epoch": 7.783166904422254,
      "grad_norm": 0.293625146150589,
      "learning_rate": 7.404255319148935e-05,
      "loss": 0.0781,
      "step": 4092
    },
    {
      "epoch": 7.785068949120304,
      "grad_norm": 0.26291313767433167,
      "learning_rate": 7.403620196887902e-05,
      "loss": 0.0708,
      "step": 4093
    },
    {
      "epoch": 7.786970993818355,
      "grad_norm": 0.20752035081386566,
      "learning_rate": 7.402985074626866e-05,
      "loss": 0.0638,
      "step": 4094
    },
    {
      "epoch": 7.788873038516405,
      "grad_norm": 0.36044710874557495,
      "learning_rate": 7.402349952365831e-05,
      "loss": 0.079,
      "step": 4095
    },
    {
      "epoch": 7.790775083214456,
      "grad_norm": 0.32189956307411194,
      "learning_rate": 7.401714830104796e-05,
      "loss": 0.0754,
      "step": 4096
    },
    {
      "epoch": 7.792677127912506,
      "grad_norm": 0.259915828704834,
      "learning_rate": 7.40107970784376e-05,
      "loss": 0.083,
      "step": 4097
    },
    {
      "epoch": 7.794579172610557,
      "grad_norm": 0.2856104373931885,
      "learning_rate": 7.400444585582725e-05,
      "loss": 0.0742,
      "step": 4098
    },
    {
      "epoch": 7.796481217308607,
      "grad_norm": 0.27823686599731445,
      "learning_rate": 7.399809463321689e-05,
      "loss": 0.0697,
      "step": 4099
    },
    {
      "epoch": 7.7983832620066575,
      "grad_norm": 0.291198194026947,
      "learning_rate": 7.399174341060654e-05,
      "loss": 0.0771,
      "step": 4100
    },
    {
      "epoch": 7.8002853067047075,
      "grad_norm": 0.322517067193985,
      "learning_rate": 7.39853921879962e-05,
      "loss": 0.0585,
      "step": 4101
    },
    {
      "epoch": 7.8021873514027575,
      "grad_norm": 0.3800067603588104,
      "learning_rate": 7.397904096538583e-05,
      "loss": 0.0941,
      "step": 4102
    },
    {
      "epoch": 7.804089396100808,
      "grad_norm": 0.33139878511428833,
      "learning_rate": 7.39726897427755e-05,
      "loss": 0.0879,
      "step": 4103
    },
    {
      "epoch": 7.805991440798858,
      "grad_norm": 0.3596518039703369,
      "learning_rate": 7.396633852016514e-05,
      "loss": 0.1046,
      "step": 4104
    },
    {
      "epoch": 7.807893485496909,
      "grad_norm": 0.24730372428894043,
      "learning_rate": 7.395998729755477e-05,
      "loss": 0.0727,
      "step": 4105
    },
    {
      "epoch": 7.809795530194959,
      "grad_norm": 0.24898457527160645,
      "learning_rate": 7.395363607494443e-05,
      "loss": 0.0787,
      "step": 4106
    },
    {
      "epoch": 7.81169757489301,
      "grad_norm": 0.27778348326683044,
      "learning_rate": 7.394728485233408e-05,
      "loss": 0.0789,
      "step": 4107
    },
    {
      "epoch": 7.81359961959106,
      "grad_norm": 0.4483128488063812,
      "learning_rate": 7.394093362972373e-05,
      "loss": 0.0903,
      "step": 4108
    },
    {
      "epoch": 7.815501664289111,
      "grad_norm": 0.25830453634262085,
      "learning_rate": 7.393458240711337e-05,
      "loss": 0.0776,
      "step": 4109
    },
    {
      "epoch": 7.817403708987161,
      "grad_norm": 0.26678788661956787,
      "learning_rate": 7.392823118450302e-05,
      "loss": 0.0689,
      "step": 4110
    },
    {
      "epoch": 7.819305753685212,
      "grad_norm": 0.3530676066875458,
      "learning_rate": 7.392187996189267e-05,
      "loss": 0.0877,
      "step": 4111
    },
    {
      "epoch": 7.821207798383262,
      "grad_norm": 0.27258220314979553,
      "learning_rate": 7.391552873928231e-05,
      "loss": 0.0849,
      "step": 4112
    },
    {
      "epoch": 7.823109843081312,
      "grad_norm": 0.3042837679386139,
      "learning_rate": 7.390917751667196e-05,
      "loss": 0.0764,
      "step": 4113
    },
    {
      "epoch": 7.825011887779363,
      "grad_norm": 0.26109039783477783,
      "learning_rate": 7.390282629406161e-05,
      "loss": 0.0687,
      "step": 4114
    },
    {
      "epoch": 7.826913932477413,
      "grad_norm": 0.36038345098495483,
      "learning_rate": 7.389647507145125e-05,
      "loss": 0.1176,
      "step": 4115
    },
    {
      "epoch": 7.828815977175464,
      "grad_norm": 0.24119070172309875,
      "learning_rate": 7.38901238488409e-05,
      "loss": 0.0803,
      "step": 4116
    },
    {
      "epoch": 7.830718021873514,
      "grad_norm": 0.279041588306427,
      "learning_rate": 7.388377262623056e-05,
      "loss": 0.0748,
      "step": 4117
    },
    {
      "epoch": 7.832620066571565,
      "grad_norm": 0.24732699990272522,
      "learning_rate": 7.38774214036202e-05,
      "loss": 0.0749,
      "step": 4118
    },
    {
      "epoch": 7.834522111269615,
      "grad_norm": 0.19942957162857056,
      "learning_rate": 7.387107018100985e-05,
      "loss": 0.058,
      "step": 4119
    },
    {
      "epoch": 7.8364241559676655,
      "grad_norm": 0.3800085484981537,
      "learning_rate": 7.38647189583995e-05,
      "loss": 0.0959,
      "step": 4120
    },
    {
      "epoch": 7.8383262006657155,
      "grad_norm": 0.24655404686927795,
      "learning_rate": 7.385836773578915e-05,
      "loss": 0.0704,
      "step": 4121
    },
    {
      "epoch": 7.840228245363766,
      "grad_norm": 0.29220011830329895,
      "learning_rate": 7.385201651317879e-05,
      "loss": 0.0686,
      "step": 4122
    },
    {
      "epoch": 7.842130290061816,
      "grad_norm": 0.29641228914260864,
      "learning_rate": 7.384566529056844e-05,
      "loss": 0.1011,
      "step": 4123
    },
    {
      "epoch": 7.844032334759866,
      "grad_norm": 0.3333776593208313,
      "learning_rate": 7.383931406795809e-05,
      "loss": 0.0656,
      "step": 4124
    },
    {
      "epoch": 7.845934379457917,
      "grad_norm": 0.3494468331336975,
      "learning_rate": 7.383296284534773e-05,
      "loss": 0.0884,
      "step": 4125
    },
    {
      "epoch": 7.847836424155968,
      "grad_norm": 0.2233295738697052,
      "learning_rate": 7.382661162273738e-05,
      "loss": 0.0835,
      "step": 4126
    },
    {
      "epoch": 7.849738468854018,
      "grad_norm": 0.26915204524993896,
      "learning_rate": 7.382026040012703e-05,
      "loss": 0.0748,
      "step": 4127
    },
    {
      "epoch": 7.851640513552068,
      "grad_norm": 0.3454749286174774,
      "learning_rate": 7.381390917751667e-05,
      "loss": 0.0839,
      "step": 4128
    },
    {
      "epoch": 7.853542558250119,
      "grad_norm": 0.23676830530166626,
      "learning_rate": 7.380755795490632e-05,
      "loss": 0.0802,
      "step": 4129
    },
    {
      "epoch": 7.855444602948169,
      "grad_norm": 0.38189131021499634,
      "learning_rate": 7.380120673229598e-05,
      "loss": 0.0968,
      "step": 4130
    },
    {
      "epoch": 7.85734664764622,
      "grad_norm": 0.40328049659729004,
      "learning_rate": 7.379485550968563e-05,
      "loss": 0.0985,
      "step": 4131
    },
    {
      "epoch": 7.85924869234427,
      "grad_norm": 0.2955927848815918,
      "learning_rate": 7.378850428707527e-05,
      "loss": 0.0761,
      "step": 4132
    },
    {
      "epoch": 7.861150737042321,
      "grad_norm": 0.34690070152282715,
      "learning_rate": 7.37821530644649e-05,
      "loss": 0.0956,
      "step": 4133
    },
    {
      "epoch": 7.863052781740371,
      "grad_norm": 0.18155311048030853,
      "learning_rate": 7.377580184185457e-05,
      "loss": 0.0389,
      "step": 4134
    },
    {
      "epoch": 7.864954826438421,
      "grad_norm": 0.23770956695079803,
      "learning_rate": 7.376945061924421e-05,
      "loss": 0.0744,
      "step": 4135
    },
    {
      "epoch": 7.866856871136472,
      "grad_norm": 0.32734113931655884,
      "learning_rate": 7.376309939663385e-05,
      "loss": 0.0786,
      "step": 4136
    },
    {
      "epoch": 7.868758915834523,
      "grad_norm": 0.24840562045574188,
      "learning_rate": 7.37567481740235e-05,
      "loss": 0.0702,
      "step": 4137
    },
    {
      "epoch": 7.870660960532573,
      "grad_norm": 0.3209112882614136,
      "learning_rate": 7.375039695141315e-05,
      "loss": 0.084,
      "step": 4138
    },
    {
      "epoch": 7.872563005230623,
      "grad_norm": 0.34609726071357727,
      "learning_rate": 7.37440457288028e-05,
      "loss": 0.086,
      "step": 4139
    },
    {
      "epoch": 7.8744650499286735,
      "grad_norm": 0.24158047139644623,
      "learning_rate": 7.373769450619244e-05,
      "loss": 0.0799,
      "step": 4140
    },
    {
      "epoch": 7.8763670946267235,
      "grad_norm": 0.2257915884256363,
      "learning_rate": 7.373134328358209e-05,
      "loss": 0.0644,
      "step": 4141
    },
    {
      "epoch": 7.878269139324774,
      "grad_norm": 0.2671010196208954,
      "learning_rate": 7.372499206097174e-05,
      "loss": 0.0786,
      "step": 4142
    },
    {
      "epoch": 7.880171184022824,
      "grad_norm": 0.4526705741882324,
      "learning_rate": 7.371864083836138e-05,
      "loss": 0.1118,
      "step": 4143
    },
    {
      "epoch": 7.882073228720875,
      "grad_norm": 0.2693113684654236,
      "learning_rate": 7.371228961575103e-05,
      "loss": 0.0588,
      "step": 4144
    },
    {
      "epoch": 7.883975273418925,
      "grad_norm": 0.20709961652755737,
      "learning_rate": 7.370593839314069e-05,
      "loss": 0.0651,
      "step": 4145
    },
    {
      "epoch": 7.885877318116976,
      "grad_norm": 0.21224233508110046,
      "learning_rate": 7.369958717053032e-05,
      "loss": 0.0664,
      "step": 4146
    },
    {
      "epoch": 7.887779362815026,
      "grad_norm": 0.3241201639175415,
      "learning_rate": 7.369323594791998e-05,
      "loss": 0.0787,
      "step": 4147
    },
    {
      "epoch": 7.889681407513077,
      "grad_norm": 0.24492232501506805,
      "learning_rate": 7.368688472530963e-05,
      "loss": 0.0791,
      "step": 4148
    },
    {
      "epoch": 7.891583452211127,
      "grad_norm": 0.33030468225479126,
      "learning_rate": 7.368053350269928e-05,
      "loss": 0.0738,
      "step": 4149
    },
    {
      "epoch": 7.893485496909177,
      "grad_norm": 0.2864938974380493,
      "learning_rate": 7.367418228008892e-05,
      "loss": 0.0818,
      "step": 4150
    },
    {
      "epoch": 7.895387541607228,
      "grad_norm": 0.2391495555639267,
      "learning_rate": 7.366783105747857e-05,
      "loss": 0.0678,
      "step": 4151
    },
    {
      "epoch": 7.897289586305278,
      "grad_norm": 0.3550390303134918,
      "learning_rate": 7.366147983486822e-05,
      "loss": 0.0835,
      "step": 4152
    },
    {
      "epoch": 7.899191631003329,
      "grad_norm": 0.26836395263671875,
      "learning_rate": 7.365512861225786e-05,
      "loss": 0.0757,
      "step": 4153
    },
    {
      "epoch": 7.901093675701379,
      "grad_norm": 0.2554214894771576,
      "learning_rate": 7.364877738964751e-05,
      "loss": 0.0888,
      "step": 4154
    },
    {
      "epoch": 7.90299572039943,
      "grad_norm": 0.2723173499107361,
      "learning_rate": 7.364242616703716e-05,
      "loss": 0.0765,
      "step": 4155
    },
    {
      "epoch": 7.90489776509748,
      "grad_norm": 0.3314184844493866,
      "learning_rate": 7.36360749444268e-05,
      "loss": 0.1036,
      "step": 4156
    },
    {
      "epoch": 7.9067998097955305,
      "grad_norm": 0.2639399766921997,
      "learning_rate": 7.362972372181645e-05,
      "loss": 0.0573,
      "step": 4157
    },
    {
      "epoch": 7.9087018544935805,
      "grad_norm": 0.229736328125,
      "learning_rate": 7.36233724992061e-05,
      "loss": 0.0736,
      "step": 4158
    },
    {
      "epoch": 7.910603899191631,
      "grad_norm": 0.34588804841041565,
      "learning_rate": 7.361702127659574e-05,
      "loss": 0.0606,
      "step": 4159
    },
    {
      "epoch": 7.912505943889681,
      "grad_norm": 0.37485557794570923,
      "learning_rate": 7.36106700539854e-05,
      "loss": 0.0861,
      "step": 4160
    },
    {
      "epoch": 7.914407988587731,
      "grad_norm": 0.23467861115932465,
      "learning_rate": 7.360431883137505e-05,
      "loss": 0.0909,
      "step": 4161
    },
    {
      "epoch": 7.916310033285782,
      "grad_norm": 0.2061249315738678,
      "learning_rate": 7.35979676087647e-05,
      "loss": 0.0689,
      "step": 4162
    },
    {
      "epoch": 7.918212077983832,
      "grad_norm": 0.31142982840538025,
      "learning_rate": 7.359161638615434e-05,
      "loss": 0.0897,
      "step": 4163
    },
    {
      "epoch": 7.920114122681883,
      "grad_norm": 0.21099460124969482,
      "learning_rate": 7.358526516354398e-05,
      "loss": 0.0689,
      "step": 4164
    },
    {
      "epoch": 7.922016167379933,
      "grad_norm": 0.2100033015012741,
      "learning_rate": 7.357891394093364e-05,
      "loss": 0.0565,
      "step": 4165
    },
    {
      "epoch": 7.923918212077984,
      "grad_norm": 0.26403024792671204,
      "learning_rate": 7.357256271832328e-05,
      "loss": 0.0674,
      "step": 4166
    },
    {
      "epoch": 7.925820256776034,
      "grad_norm": 0.26634496450424194,
      "learning_rate": 7.356621149571293e-05,
      "loss": 0.0657,
      "step": 4167
    },
    {
      "epoch": 7.927722301474085,
      "grad_norm": 0.1982484757900238,
      "learning_rate": 7.355986027310258e-05,
      "loss": 0.0554,
      "step": 4168
    },
    {
      "epoch": 7.929624346172135,
      "grad_norm": 0.24268779158592224,
      "learning_rate": 7.355350905049222e-05,
      "loss": 0.0669,
      "step": 4169
    },
    {
      "epoch": 7.931526390870186,
      "grad_norm": 0.4950437545776367,
      "learning_rate": 7.354715782788187e-05,
      "loss": 0.101,
      "step": 4170
    },
    {
      "epoch": 7.933428435568236,
      "grad_norm": 0.39911749958992004,
      "learning_rate": 7.354080660527151e-05,
      "loss": 0.0784,
      "step": 4171
    },
    {
      "epoch": 7.935330480266286,
      "grad_norm": 0.3811649680137634,
      "learning_rate": 7.353445538266116e-05,
      "loss": 0.0884,
      "step": 4172
    },
    {
      "epoch": 7.937232524964337,
      "grad_norm": 0.20468108355998993,
      "learning_rate": 7.352810416005081e-05,
      "loss": 0.0578,
      "step": 4173
    },
    {
      "epoch": 7.939134569662387,
      "grad_norm": 0.34287476539611816,
      "learning_rate": 7.352175293744045e-05,
      "loss": 0.0692,
      "step": 4174
    },
    {
      "epoch": 7.941036614360438,
      "grad_norm": 0.23349088430404663,
      "learning_rate": 7.351540171483012e-05,
      "loss": 0.0706,
      "step": 4175
    },
    {
      "epoch": 7.942938659058488,
      "grad_norm": 0.33331581950187683,
      "learning_rate": 7.350905049221976e-05,
      "loss": 0.0788,
      "step": 4176
    },
    {
      "epoch": 7.9448407037565385,
      "grad_norm": 0.2772316038608551,
      "learning_rate": 7.35026992696094e-05,
      "loss": 0.0766,
      "step": 4177
    },
    {
      "epoch": 7.9467427484545885,
      "grad_norm": 0.23243772983551025,
      "learning_rate": 7.349634804699905e-05,
      "loss": 0.0588,
      "step": 4178
    },
    {
      "epoch": 7.948644793152639,
      "grad_norm": 0.2223922461271286,
      "learning_rate": 7.34899968243887e-05,
      "loss": 0.051,
      "step": 4179
    },
    {
      "epoch": 7.950546837850689,
      "grad_norm": 0.29967790842056274,
      "learning_rate": 7.348364560177835e-05,
      "loss": 0.0754,
      "step": 4180
    },
    {
      "epoch": 7.95244888254874,
      "grad_norm": 0.317216694355011,
      "learning_rate": 7.347729437916799e-05,
      "loss": 0.0873,
      "step": 4181
    },
    {
      "epoch": 7.95435092724679,
      "grad_norm": 0.28003180027008057,
      "learning_rate": 7.347094315655764e-05,
      "loss": 0.0795,
      "step": 4182
    },
    {
      "epoch": 7.95625297194484,
      "grad_norm": 0.38882380723953247,
      "learning_rate": 7.346459193394729e-05,
      "loss": 0.0874,
      "step": 4183
    },
    {
      "epoch": 7.958155016642891,
      "grad_norm": 0.21491193771362305,
      "learning_rate": 7.345824071133693e-05,
      "loss": 0.0829,
      "step": 4184
    },
    {
      "epoch": 7.960057061340941,
      "grad_norm": 0.2690449655056,
      "learning_rate": 7.345188948872658e-05,
      "loss": 0.0613,
      "step": 4185
    },
    {
      "epoch": 7.961959106038992,
      "grad_norm": 0.27885550260543823,
      "learning_rate": 7.344553826611623e-05,
      "loss": 0.0692,
      "step": 4186
    },
    {
      "epoch": 7.963861150737042,
      "grad_norm": 0.20747528970241547,
      "learning_rate": 7.343918704350587e-05,
      "loss": 0.069,
      "step": 4187
    },
    {
      "epoch": 7.965763195435093,
      "grad_norm": 0.33041733503341675,
      "learning_rate": 7.343283582089552e-05,
      "loss": 0.099,
      "step": 4188
    },
    {
      "epoch": 7.967665240133143,
      "grad_norm": 0.2511996626853943,
      "learning_rate": 7.342648459828518e-05,
      "loss": 0.0775,
      "step": 4189
    },
    {
      "epoch": 7.969567284831194,
      "grad_norm": 0.2743570804595947,
      "learning_rate": 7.342013337567481e-05,
      "loss": 0.0877,
      "step": 4190
    },
    {
      "epoch": 7.971469329529244,
      "grad_norm": 0.2678762972354889,
      "learning_rate": 7.341378215306447e-05,
      "loss": 0.0848,
      "step": 4191
    },
    {
      "epoch": 7.973371374227295,
      "grad_norm": 0.23124247789382935,
      "learning_rate": 7.340743093045412e-05,
      "loss": 0.0725,
      "step": 4192
    },
    {
      "epoch": 7.975273418925345,
      "grad_norm": 0.2571721374988556,
      "learning_rate": 7.340107970784377e-05,
      "loss": 0.0806,
      "step": 4193
    },
    {
      "epoch": 7.977175463623395,
      "grad_norm": 0.2708531618118286,
      "learning_rate": 7.339472848523341e-05,
      "loss": 0.073,
      "step": 4194
    },
    {
      "epoch": 7.979077508321446,
      "grad_norm": 0.17527925968170166,
      "learning_rate": 7.338837726262305e-05,
      "loss": 0.0977,
      "step": 4195
    },
    {
      "epoch": 7.980979553019496,
      "grad_norm": 0.23408342897891998,
      "learning_rate": 7.338202604001271e-05,
      "loss": 0.0769,
      "step": 4196
    },
    {
      "epoch": 7.9828815977175465,
      "grad_norm": 0.27631494402885437,
      "learning_rate": 7.337567481740235e-05,
      "loss": 0.0878,
      "step": 4197
    },
    {
      "epoch": 7.9847836424155965,
      "grad_norm": 0.28769686818122864,
      "learning_rate": 7.3369323594792e-05,
      "loss": 0.0979,
      "step": 4198
    },
    {
      "epoch": 7.986685687113647,
      "grad_norm": 0.2594468593597412,
      "learning_rate": 7.336297237218165e-05,
      "loss": 0.0829,
      "step": 4199
    },
    {
      "epoch": 7.988587731811697,
      "grad_norm": 0.46053510904312134,
      "learning_rate": 7.335662114957129e-05,
      "loss": 0.0974,
      "step": 4200
    },
    {
      "epoch": 7.990489776509748,
      "grad_norm": 0.29801467061042786,
      "learning_rate": 7.335026992696094e-05,
      "loss": 0.0948,
      "step": 4201
    },
    {
      "epoch": 7.992391821207798,
      "grad_norm": 0.374401330947876,
      "learning_rate": 7.334391870435058e-05,
      "loss": 0.1119,
      "step": 4202
    },
    {
      "epoch": 7.994293865905849,
      "grad_norm": 0.3322446644306183,
      "learning_rate": 7.333756748174025e-05,
      "loss": 0.0921,
      "step": 4203
    },
    {
      "epoch": 7.996195910603899,
      "grad_norm": 0.3914404511451721,
      "learning_rate": 7.333121625912989e-05,
      "loss": 0.0879,
      "step": 4204
    },
    {
      "epoch": 7.998097955301949,
      "grad_norm": 0.3230751156806946,
      "learning_rate": 7.332486503651952e-05,
      "loss": 0.0785,
      "step": 4205
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.3116169571876526,
      "learning_rate": 7.331851381390919e-05,
      "loss": 0.1309,
      "step": 4206
    },
    {
      "epoch": 8.00190204469805,
      "grad_norm": 0.1682499498128891,
      "learning_rate": 7.331216259129883e-05,
      "loss": 0.0519,
      "step": 4207
    },
    {
      "epoch": 8.0038040893961,
      "grad_norm": 0.31888794898986816,
      "learning_rate": 7.330581136868847e-05,
      "loss": 0.0693,
      "step": 4208
    },
    {
      "epoch": 8.00570613409415,
      "grad_norm": 0.1675415188074112,
      "learning_rate": 7.329946014607812e-05,
      "loss": 0.049,
      "step": 4209
    },
    {
      "epoch": 8.007608178792202,
      "grad_norm": 0.2735571265220642,
      "learning_rate": 7.329310892346777e-05,
      "loss": 0.0682,
      "step": 4210
    },
    {
      "epoch": 8.009510223490253,
      "grad_norm": 0.2755778133869171,
      "learning_rate": 7.328675770085742e-05,
      "loss": 0.072,
      "step": 4211
    },
    {
      "epoch": 8.011412268188302,
      "grad_norm": 0.2477390468120575,
      "learning_rate": 7.328040647824706e-05,
      "loss": 0.0613,
      "step": 4212
    },
    {
      "epoch": 8.013314312886353,
      "grad_norm": 0.21356599032878876,
      "learning_rate": 7.327405525563671e-05,
      "loss": 0.0637,
      "step": 4213
    },
    {
      "epoch": 8.015216357584404,
      "grad_norm": 0.2848338782787323,
      "learning_rate": 7.326770403302636e-05,
      "loss": 0.0742,
      "step": 4214
    },
    {
      "epoch": 8.017118402282454,
      "grad_norm": 0.30361059308052063,
      "learning_rate": 7.3261352810416e-05,
      "loss": 0.0722,
      "step": 4215
    },
    {
      "epoch": 8.019020446980504,
      "grad_norm": 0.2148815244436264,
      "learning_rate": 7.325500158780565e-05,
      "loss": 0.072,
      "step": 4216
    },
    {
      "epoch": 8.020922491678554,
      "grad_norm": 0.3315644860267639,
      "learning_rate": 7.32486503651953e-05,
      "loss": 0.1286,
      "step": 4217
    },
    {
      "epoch": 8.022824536376605,
      "grad_norm": 0.30502501130104065,
      "learning_rate": 7.324229914258494e-05,
      "loss": 0.0857,
      "step": 4218
    },
    {
      "epoch": 8.024726581074654,
      "grad_norm": 0.32918721437454224,
      "learning_rate": 7.32359479199746e-05,
      "loss": 0.0695,
      "step": 4219
    },
    {
      "epoch": 8.026628625772705,
      "grad_norm": 0.19223280251026154,
      "learning_rate": 7.322959669736425e-05,
      "loss": 0.0594,
      "step": 4220
    },
    {
      "epoch": 8.028530670470756,
      "grad_norm": 0.3204280138015747,
      "learning_rate": 7.32232454747539e-05,
      "loss": 0.0752,
      "step": 4221
    },
    {
      "epoch": 8.030432715168807,
      "grad_norm": 0.27245232462882996,
      "learning_rate": 7.321689425214354e-05,
      "loss": 0.0617,
      "step": 4222
    },
    {
      "epoch": 8.032334759866856,
      "grad_norm": 0.26873305439949036,
      "learning_rate": 7.321054302953319e-05,
      "loss": 0.1057,
      "step": 4223
    },
    {
      "epoch": 8.034236804564907,
      "grad_norm": 0.36406102776527405,
      "learning_rate": 7.320419180692284e-05,
      "loss": 0.0583,
      "step": 4224
    },
    {
      "epoch": 8.036138849262958,
      "grad_norm": 0.25883886218070984,
      "learning_rate": 7.319784058431248e-05,
      "loss": 0.0634,
      "step": 4225
    },
    {
      "epoch": 8.038040893961009,
      "grad_norm": 0.30319416522979736,
      "learning_rate": 7.319148936170213e-05,
      "loss": 0.0706,
      "step": 4226
    },
    {
      "epoch": 8.039942938659058,
      "grad_norm": 0.3315434753894806,
      "learning_rate": 7.318513813909178e-05,
      "loss": 0.0611,
      "step": 4227
    },
    {
      "epoch": 8.041844983357109,
      "grad_norm": 0.22568808495998383,
      "learning_rate": 7.317878691648142e-05,
      "loss": 0.0847,
      "step": 4228
    },
    {
      "epoch": 8.04374702805516,
      "grad_norm": 0.3340080976486206,
      "learning_rate": 7.317243569387107e-05,
      "loss": 0.0767,
      "step": 4229
    },
    {
      "epoch": 8.045649072753209,
      "grad_norm": 0.20436561107635498,
      "learning_rate": 7.316608447126073e-05,
      "loss": 0.0849,
      "step": 4230
    },
    {
      "epoch": 8.04755111745126,
      "grad_norm": 0.2909284830093384,
      "learning_rate": 7.315973324865036e-05,
      "loss": 0.055,
      "step": 4231
    },
    {
      "epoch": 8.04945316214931,
      "grad_norm": 0.24360747635364532,
      "learning_rate": 7.315338202604002e-05,
      "loss": 0.0852,
      "step": 4232
    },
    {
      "epoch": 8.051355206847362,
      "grad_norm": 0.26607367396354675,
      "learning_rate": 7.314703080342967e-05,
      "loss": 0.0739,
      "step": 4233
    },
    {
      "epoch": 8.05325725154541,
      "grad_norm": 0.3718374967575073,
      "learning_rate": 7.314067958081932e-05,
      "loss": 0.074,
      "step": 4234
    },
    {
      "epoch": 8.055159296243461,
      "grad_norm": 0.2629009485244751,
      "learning_rate": 7.313432835820896e-05,
      "loss": 0.0966,
      "step": 4235
    },
    {
      "epoch": 8.057061340941512,
      "grad_norm": 0.24625834822654724,
      "learning_rate": 7.31279771355986e-05,
      "loss": 0.0733,
      "step": 4236
    },
    {
      "epoch": 8.058963385639563,
      "grad_norm": 0.2897358238697052,
      "learning_rate": 7.312162591298826e-05,
      "loss": 0.0497,
      "step": 4237
    },
    {
      "epoch": 8.060865430337612,
      "grad_norm": 0.23193830251693726,
      "learning_rate": 7.31152746903779e-05,
      "loss": 0.0612,
      "step": 4238
    },
    {
      "epoch": 8.062767475035663,
      "grad_norm": 0.22554795444011688,
      "learning_rate": 7.310892346776755e-05,
      "loss": 0.0819,
      "step": 4239
    },
    {
      "epoch": 8.064669519733714,
      "grad_norm": 0.2712651491165161,
      "learning_rate": 7.31025722451572e-05,
      "loss": 0.079,
      "step": 4240
    },
    {
      "epoch": 8.066571564431765,
      "grad_norm": 0.2655814290046692,
      "learning_rate": 7.309622102254684e-05,
      "loss": 0.0824,
      "step": 4241
    },
    {
      "epoch": 8.068473609129814,
      "grad_norm": 0.29547134041786194,
      "learning_rate": 7.30898697999365e-05,
      "loss": 0.0776,
      "step": 4242
    },
    {
      "epoch": 8.070375653827865,
      "grad_norm": 0.19004866480827332,
      "learning_rate": 7.308351857732613e-05,
      "loss": 0.0631,
      "step": 4243
    },
    {
      "epoch": 8.072277698525916,
      "grad_norm": 0.2522698640823364,
      "learning_rate": 7.307716735471578e-05,
      "loss": 0.0882,
      "step": 4244
    },
    {
      "epoch": 8.074179743223965,
      "grad_norm": 0.1891438066959381,
      "learning_rate": 7.307081613210544e-05,
      "loss": 0.0744,
      "step": 4245
    },
    {
      "epoch": 8.076081787922016,
      "grad_norm": 0.18746241927146912,
      "learning_rate": 7.306446490949507e-05,
      "loss": 0.0534,
      "step": 4246
    },
    {
      "epoch": 8.077983832620067,
      "grad_norm": 0.2318631261587143,
      "learning_rate": 7.305811368688473e-05,
      "loss": 0.0749,
      "step": 4247
    },
    {
      "epoch": 8.079885877318118,
      "grad_norm": 0.2980051040649414,
      "learning_rate": 7.305176246427438e-05,
      "loss": 0.0996,
      "step": 4248
    },
    {
      "epoch": 8.081787922016167,
      "grad_norm": 0.26681259274482727,
      "learning_rate": 7.304541124166402e-05,
      "loss": 0.0759,
      "step": 4249
    },
    {
      "epoch": 8.083689966714218,
      "grad_norm": 0.27438244223594666,
      "learning_rate": 7.303906001905367e-05,
      "loss": 0.1157,
      "step": 4250
    },
    {
      "epoch": 8.085592011412269,
      "grad_norm": 0.3032076060771942,
      "learning_rate": 7.303270879644332e-05,
      "loss": 0.0696,
      "step": 4251
    },
    {
      "epoch": 8.08749405611032,
      "grad_norm": 0.23937378823757172,
      "learning_rate": 7.302635757383297e-05,
      "loss": 0.0784,
      "step": 4252
    },
    {
      "epoch": 8.089396100808369,
      "grad_norm": 0.24228748679161072,
      "learning_rate": 7.302000635122261e-05,
      "loss": 0.0677,
      "step": 4253
    },
    {
      "epoch": 8.09129814550642,
      "grad_norm": 0.14060363173484802,
      "learning_rate": 7.301365512861226e-05,
      "loss": 0.0618,
      "step": 4254
    },
    {
      "epoch": 8.09320019020447,
      "grad_norm": 0.23839721083641052,
      "learning_rate": 7.300730390600191e-05,
      "loss": 0.0686,
      "step": 4255
    },
    {
      "epoch": 8.09510223490252,
      "grad_norm": 0.41366299986839294,
      "learning_rate": 7.300095268339155e-05,
      "loss": 0.0925,
      "step": 4256
    },
    {
      "epoch": 8.09700427960057,
      "grad_norm": 0.2656678855419159,
      "learning_rate": 7.29946014607812e-05,
      "loss": 0.0579,
      "step": 4257
    },
    {
      "epoch": 8.098906324298621,
      "grad_norm": 0.25097447633743286,
      "learning_rate": 7.298825023817086e-05,
      "loss": 0.0733,
      "step": 4258
    },
    {
      "epoch": 8.100808368996672,
      "grad_norm": 0.30444836616516113,
      "learning_rate": 7.29818990155605e-05,
      "loss": 0.0576,
      "step": 4259
    },
    {
      "epoch": 8.102710413694721,
      "grad_norm": 0.2674120366573334,
      "learning_rate": 7.297554779295015e-05,
      "loss": 0.0662,
      "step": 4260
    },
    {
      "epoch": 8.104612458392772,
      "grad_norm": 0.30446937680244446,
      "learning_rate": 7.29691965703398e-05,
      "loss": 0.064,
      "step": 4261
    },
    {
      "epoch": 8.106514503090823,
      "grad_norm": 0.20771685242652893,
      "learning_rate": 7.296284534772944e-05,
      "loss": 0.0689,
      "step": 4262
    },
    {
      "epoch": 8.108416547788874,
      "grad_norm": 0.30618518590927124,
      "learning_rate": 7.295649412511909e-05,
      "loss": 0.0701,
      "step": 4263
    },
    {
      "epoch": 8.110318592486923,
      "grad_norm": 0.22746066749095917,
      "learning_rate": 7.295014290250874e-05,
      "loss": 0.0888,
      "step": 4264
    },
    {
      "epoch": 8.112220637184974,
      "grad_norm": 0.19552601873874664,
      "learning_rate": 7.294379167989839e-05,
      "loss": 0.0638,
      "step": 4265
    },
    {
      "epoch": 8.114122681883025,
      "grad_norm": 0.2784650921821594,
      "learning_rate": 7.293744045728803e-05,
      "loss": 0.0715,
      "step": 4266
    },
    {
      "epoch": 8.116024726581074,
      "grad_norm": 0.28357699513435364,
      "learning_rate": 7.293108923467767e-05,
      "loss": 0.0779,
      "step": 4267
    },
    {
      "epoch": 8.117926771279125,
      "grad_norm": 0.3271869421005249,
      "learning_rate": 7.292473801206733e-05,
      "loss": 0.0788,
      "step": 4268
    },
    {
      "epoch": 8.119828815977176,
      "grad_norm": 0.29517510533332825,
      "learning_rate": 7.291838678945697e-05,
      "loss": 0.0851,
      "step": 4269
    },
    {
      "epoch": 8.121730860675227,
      "grad_norm": 0.1606910526752472,
      "learning_rate": 7.291203556684662e-05,
      "loss": 0.0703,
      "step": 4270
    },
    {
      "epoch": 8.123632905373276,
      "grad_norm": 0.2820865213871002,
      "learning_rate": 7.290568434423628e-05,
      "loss": 0.0785,
      "step": 4271
    },
    {
      "epoch": 8.125534950071327,
      "grad_norm": 0.2265879064798355,
      "learning_rate": 7.289933312162591e-05,
      "loss": 0.0593,
      "step": 4272
    },
    {
      "epoch": 8.127436994769377,
      "grad_norm": 0.25888168811798096,
      "learning_rate": 7.289298189901557e-05,
      "loss": 0.08,
      "step": 4273
    },
    {
      "epoch": 8.129339039467428,
      "grad_norm": 0.3094843924045563,
      "learning_rate": 7.28866306764052e-05,
      "loss": 0.0724,
      "step": 4274
    },
    {
      "epoch": 8.131241084165477,
      "grad_norm": 0.29158642888069153,
      "learning_rate": 7.288027945379487e-05,
      "loss": 0.0704,
      "step": 4275
    },
    {
      "epoch": 8.133143128863528,
      "grad_norm": 0.4067235291004181,
      "learning_rate": 7.287392823118451e-05,
      "loss": 0.0848,
      "step": 4276
    },
    {
      "epoch": 8.13504517356158,
      "grad_norm": 0.31271129846572876,
      "learning_rate": 7.286757700857415e-05,
      "loss": 0.0753,
      "step": 4277
    },
    {
      "epoch": 8.136947218259628,
      "grad_norm": 0.2223985642194748,
      "learning_rate": 7.286122578596381e-05,
      "loss": 0.0619,
      "step": 4278
    },
    {
      "epoch": 8.13884926295768,
      "grad_norm": 0.19151456654071808,
      "learning_rate": 7.285487456335345e-05,
      "loss": 0.0774,
      "step": 4279
    },
    {
      "epoch": 8.14075130765573,
      "grad_norm": 0.2343282848596573,
      "learning_rate": 7.284852334074309e-05,
      "loss": 0.0662,
      "step": 4280
    },
    {
      "epoch": 8.142653352353781,
      "grad_norm": 0.22527043521404266,
      "learning_rate": 7.284217211813274e-05,
      "loss": 0.0779,
      "step": 4281
    },
    {
      "epoch": 8.14455539705183,
      "grad_norm": 0.13312441110610962,
      "learning_rate": 7.283582089552239e-05,
      "loss": 0.0523,
      "step": 4282
    },
    {
      "epoch": 8.146457441749881,
      "grad_norm": 0.18979710340499878,
      "learning_rate": 7.282946967291204e-05,
      "loss": 0.0811,
      "step": 4283
    },
    {
      "epoch": 8.148359486447932,
      "grad_norm": 0.31543537974357605,
      "learning_rate": 7.282311845030168e-05,
      "loss": 0.0756,
      "step": 4284
    },
    {
      "epoch": 8.150261531145983,
      "grad_norm": 0.23982515931129456,
      "learning_rate": 7.281676722769133e-05,
      "loss": 0.0726,
      "step": 4285
    },
    {
      "epoch": 8.152163575844032,
      "grad_norm": 0.26044005155563354,
      "learning_rate": 7.281041600508099e-05,
      "loss": 0.0571,
      "step": 4286
    },
    {
      "epoch": 8.154065620542083,
      "grad_norm": 0.26153257489204407,
      "learning_rate": 7.280406478247062e-05,
      "loss": 0.0729,
      "step": 4287
    },
    {
      "epoch": 8.155967665240134,
      "grad_norm": 0.3593063950538635,
      "learning_rate": 7.279771355986028e-05,
      "loss": 0.1001,
      "step": 4288
    },
    {
      "epoch": 8.157869709938183,
      "grad_norm": 0.2667100429534912,
      "learning_rate": 7.279136233724993e-05,
      "loss": 0.0642,
      "step": 4289
    },
    {
      "epoch": 8.159771754636234,
      "grad_norm": 0.2793238162994385,
      "learning_rate": 7.278501111463957e-05,
      "loss": 0.0629,
      "step": 4290
    },
    {
      "epoch": 8.161673799334284,
      "grad_norm": 0.20465384423732758,
      "learning_rate": 7.277865989202922e-05,
      "loss": 0.0547,
      "step": 4291
    },
    {
      "epoch": 8.163575844032335,
      "grad_norm": 0.19889406859874725,
      "learning_rate": 7.277230866941887e-05,
      "loss": 0.0595,
      "step": 4292
    },
    {
      "epoch": 8.165477888730384,
      "grad_norm": 0.3476722836494446,
      "learning_rate": 7.276595744680852e-05,
      "loss": 0.0899,
      "step": 4293
    },
    {
      "epoch": 8.167379933428435,
      "grad_norm": 0.3568250834941864,
      "learning_rate": 7.275960622419816e-05,
      "loss": 0.0834,
      "step": 4294
    },
    {
      "epoch": 8.169281978126486,
      "grad_norm": 0.37351176142692566,
      "learning_rate": 7.275325500158781e-05,
      "loss": 0.0731,
      "step": 4295
    },
    {
      "epoch": 8.171184022824537,
      "grad_norm": 0.2659788429737091,
      "learning_rate": 7.274690377897746e-05,
      "loss": 0.0664,
      "step": 4296
    },
    {
      "epoch": 8.173086067522586,
      "grad_norm": 0.29493045806884766,
      "learning_rate": 7.27405525563671e-05,
      "loss": 0.0903,
      "step": 4297
    },
    {
      "epoch": 8.174988112220637,
      "grad_norm": 0.23891212046146393,
      "learning_rate": 7.273420133375674e-05,
      "loss": 0.0754,
      "step": 4298
    },
    {
      "epoch": 8.176890156918688,
      "grad_norm": 0.3760624825954437,
      "learning_rate": 7.27278501111464e-05,
      "loss": 0.0863,
      "step": 4299
    },
    {
      "epoch": 8.178792201616737,
      "grad_norm": 0.23999914526939392,
      "learning_rate": 7.272149888853604e-05,
      "loss": 0.0622,
      "step": 4300
    },
    {
      "epoch": 8.180694246314788,
      "grad_norm": 0.3227231204509735,
      "learning_rate": 7.27151476659257e-05,
      "loss": 0.0695,
      "step": 4301
    },
    {
      "epoch": 8.182596291012839,
      "grad_norm": 0.20113304257392883,
      "learning_rate": 7.270879644331535e-05,
      "loss": 0.0687,
      "step": 4302
    },
    {
      "epoch": 8.18449833571089,
      "grad_norm": 0.31149378418922424,
      "learning_rate": 7.270244522070499e-05,
      "loss": 0.069,
      "step": 4303
    },
    {
      "epoch": 8.186400380408939,
      "grad_norm": 0.2091590166091919,
      "learning_rate": 7.269609399809464e-05,
      "loss": 0.0623,
      "step": 4304
    },
    {
      "epoch": 8.18830242510699,
      "grad_norm": 0.1678960770368576,
      "learning_rate": 7.268974277548428e-05,
      "loss": 0.0634,
      "step": 4305
    },
    {
      "epoch": 8.19020446980504,
      "grad_norm": 0.1816246211528778,
      "learning_rate": 7.268339155287394e-05,
      "loss": 0.0619,
      "step": 4306
    },
    {
      "epoch": 8.192106514503092,
      "grad_norm": 0.21957528591156006,
      "learning_rate": 7.267704033026358e-05,
      "loss": 0.0648,
      "step": 4307
    },
    {
      "epoch": 8.19400855920114,
      "grad_norm": 0.24338406324386597,
      "learning_rate": 7.267068910765322e-05,
      "loss": 0.0674,
      "step": 4308
    },
    {
      "epoch": 8.195910603899192,
      "grad_norm": 0.29942211508750916,
      "learning_rate": 7.266433788504288e-05,
      "loss": 0.0993,
      "step": 4309
    },
    {
      "epoch": 8.197812648597242,
      "grad_norm": 0.3957151770591736,
      "learning_rate": 7.265798666243252e-05,
      "loss": 0.0981,
      "step": 4310
    },
    {
      "epoch": 8.199714693295292,
      "grad_norm": 0.5199161767959595,
      "learning_rate": 7.265163543982217e-05,
      "loss": 0.0858,
      "step": 4311
    },
    {
      "epoch": 8.201616737993342,
      "grad_norm": 0.3623242974281311,
      "learning_rate": 7.264528421721181e-05,
      "loss": 0.0801,
      "step": 4312
    },
    {
      "epoch": 8.203518782691393,
      "grad_norm": 0.29007983207702637,
      "learning_rate": 7.263893299460146e-05,
      "loss": 0.0754,
      "step": 4313
    },
    {
      "epoch": 8.205420827389444,
      "grad_norm": 0.21117328107357025,
      "learning_rate": 7.263258177199111e-05,
      "loss": 0.0665,
      "step": 4314
    },
    {
      "epoch": 8.207322872087493,
      "grad_norm": 0.3226154148578644,
      "learning_rate": 7.262623054938075e-05,
      "loss": 0.0776,
      "step": 4315
    },
    {
      "epoch": 8.209224916785544,
      "grad_norm": 0.23427355289459229,
      "learning_rate": 7.26198793267704e-05,
      "loss": 0.0711,
      "step": 4316
    },
    {
      "epoch": 8.211126961483595,
      "grad_norm": 0.26657065749168396,
      "learning_rate": 7.261352810416006e-05,
      "loss": 0.0693,
      "step": 4317
    },
    {
      "epoch": 8.213029006181646,
      "grad_norm": 0.2999359965324402,
      "learning_rate": 7.26071768815497e-05,
      "loss": 0.0762,
      "step": 4318
    },
    {
      "epoch": 8.214931050879695,
      "grad_norm": 0.2279568612575531,
      "learning_rate": 7.260082565893935e-05,
      "loss": 0.0802,
      "step": 4319
    },
    {
      "epoch": 8.216833095577746,
      "grad_norm": 0.17314577102661133,
      "learning_rate": 7.2594474436329e-05,
      "loss": 0.0658,
      "step": 4320
    },
    {
      "epoch": 8.218735140275797,
      "grad_norm": 0.3194338381290436,
      "learning_rate": 7.258812321371864e-05,
      "loss": 0.0891,
      "step": 4321
    },
    {
      "epoch": 8.220637184973846,
      "grad_norm": 0.32515963912010193,
      "learning_rate": 7.258177199110829e-05,
      "loss": 0.0813,
      "step": 4322
    },
    {
      "epoch": 8.222539229671897,
      "grad_norm": 0.25820595026016235,
      "learning_rate": 7.257542076849794e-05,
      "loss": 0.0851,
      "step": 4323
    },
    {
      "epoch": 8.224441274369948,
      "grad_norm": 0.27671653032302856,
      "learning_rate": 7.256906954588759e-05,
      "loss": 0.075,
      "step": 4324
    },
    {
      "epoch": 8.226343319067999,
      "grad_norm": 0.3091162145137787,
      "learning_rate": 7.256271832327723e-05,
      "loss": 0.0876,
      "step": 4325
    },
    {
      "epoch": 8.228245363766048,
      "grad_norm": 0.2991148829460144,
      "learning_rate": 7.255636710066688e-05,
      "loss": 0.0826,
      "step": 4326
    },
    {
      "epoch": 8.230147408464099,
      "grad_norm": 0.2437266856431961,
      "learning_rate": 7.255001587805653e-05,
      "loss": 0.0801,
      "step": 4327
    },
    {
      "epoch": 8.23204945316215,
      "grad_norm": 0.3508904278278351,
      "learning_rate": 7.254366465544617e-05,
      "loss": 0.1039,
      "step": 4328
    },
    {
      "epoch": 8.2339514978602,
      "grad_norm": 0.21462726593017578,
      "learning_rate": 7.253731343283582e-05,
      "loss": 0.0528,
      "step": 4329
    },
    {
      "epoch": 8.23585354255825,
      "grad_norm": 0.24533016979694366,
      "learning_rate": 7.253096221022548e-05,
      "loss": 0.1122,
      "step": 4330
    },
    {
      "epoch": 8.2377555872563,
      "grad_norm": 0.36849719285964966,
      "learning_rate": 7.252461098761511e-05,
      "loss": 0.1083,
      "step": 4331
    },
    {
      "epoch": 8.239657631954351,
      "grad_norm": 0.1650083214044571,
      "learning_rate": 7.251825976500477e-05,
      "loss": 0.0499,
      "step": 4332
    },
    {
      "epoch": 8.241559676652402,
      "grad_norm": 0.27161771059036255,
      "learning_rate": 7.251190854239442e-05,
      "loss": 0.0839,
      "step": 4333
    },
    {
      "epoch": 8.243461721350451,
      "grad_norm": 0.43029820919036865,
      "learning_rate": 7.250555731978406e-05,
      "loss": 0.099,
      "step": 4334
    },
    {
      "epoch": 8.245363766048502,
      "grad_norm": 0.19337327778339386,
      "learning_rate": 7.249920609717371e-05,
      "loss": 0.0653,
      "step": 4335
    },
    {
      "epoch": 8.247265810746553,
      "grad_norm": 0.24718555808067322,
      "learning_rate": 7.249285487456336e-05,
      "loss": 0.0796,
      "step": 4336
    },
    {
      "epoch": 8.249167855444602,
      "grad_norm": 0.20396992564201355,
      "learning_rate": 7.248650365195301e-05,
      "loss": 0.0651,
      "step": 4337
    },
    {
      "epoch": 8.251069900142653,
      "grad_norm": 0.3806256055831909,
      "learning_rate": 7.248015242934265e-05,
      "loss": 0.0906,
      "step": 4338
    },
    {
      "epoch": 8.252971944840704,
      "grad_norm": 0.2927926778793335,
      "learning_rate": 7.247380120673229e-05,
      "loss": 0.0719,
      "step": 4339
    },
    {
      "epoch": 8.254873989538755,
      "grad_norm": 0.2693699300289154,
      "learning_rate": 7.246744998412195e-05,
      "loss": 0.0835,
      "step": 4340
    },
    {
      "epoch": 8.256776034236804,
      "grad_norm": 0.22693417966365814,
      "learning_rate": 7.246109876151159e-05,
      "loss": 0.0642,
      "step": 4341
    },
    {
      "epoch": 8.258678078934855,
      "grad_norm": 0.25865957140922546,
      "learning_rate": 7.245474753890124e-05,
      "loss": 0.0855,
      "step": 4342
    },
    {
      "epoch": 8.260580123632906,
      "grad_norm": 0.21432361006736755,
      "learning_rate": 7.24483963162909e-05,
      "loss": 0.0497,
      "step": 4343
    },
    {
      "epoch": 8.262482168330957,
      "grad_norm": 0.21584494411945343,
      "learning_rate": 7.244204509368053e-05,
      "loss": 0.0741,
      "step": 4344
    },
    {
      "epoch": 8.264384213029006,
      "grad_norm": 0.26091018319129944,
      "learning_rate": 7.243569387107019e-05,
      "loss": 0.0681,
      "step": 4345
    },
    {
      "epoch": 8.266286257727057,
      "grad_norm": 0.1850460171699524,
      "learning_rate": 7.242934264845982e-05,
      "loss": 0.075,
      "step": 4346
    },
    {
      "epoch": 8.268188302425107,
      "grad_norm": 0.21847312152385712,
      "learning_rate": 7.242299142584949e-05,
      "loss": 0.0798,
      "step": 4347
    },
    {
      "epoch": 8.270090347123157,
      "grad_norm": 0.1917494237422943,
      "learning_rate": 7.241664020323913e-05,
      "loss": 0.0556,
      "step": 4348
    },
    {
      "epoch": 8.271992391821207,
      "grad_norm": 0.320363849401474,
      "learning_rate": 7.241028898062877e-05,
      "loss": 0.0649,
      "step": 4349
    },
    {
      "epoch": 8.273894436519258,
      "grad_norm": 0.2682509422302246,
      "learning_rate": 7.240393775801843e-05,
      "loss": 0.0865,
      "step": 4350
    },
    {
      "epoch": 8.27579648121731,
      "grad_norm": 0.2643677294254303,
      "learning_rate": 7.239758653540807e-05,
      "loss": 0.0794,
      "step": 4351
    },
    {
      "epoch": 8.277698525915358,
      "grad_norm": 0.21975865960121155,
      "learning_rate": 7.239123531279771e-05,
      "loss": 0.0756,
      "step": 4352
    },
    {
      "epoch": 8.27960057061341,
      "grad_norm": 0.4391152560710907,
      "learning_rate": 7.238488409018736e-05,
      "loss": 0.0836,
      "step": 4353
    },
    {
      "epoch": 8.28150261531146,
      "grad_norm": 0.2395821362733841,
      "learning_rate": 7.237853286757701e-05,
      "loss": 0.0619,
      "step": 4354
    },
    {
      "epoch": 8.283404660009511,
      "grad_norm": 0.24098101258277893,
      "learning_rate": 7.237218164496666e-05,
      "loss": 0.0827,
      "step": 4355
    },
    {
      "epoch": 8.28530670470756,
      "grad_norm": 0.2938069701194763,
      "learning_rate": 7.23658304223563e-05,
      "loss": 0.0726,
      "step": 4356
    },
    {
      "epoch": 8.287208749405611,
      "grad_norm": 0.24898219108581543,
      "learning_rate": 7.235947919974595e-05,
      "loss": 0.0709,
      "step": 4357
    },
    {
      "epoch": 8.289110794103662,
      "grad_norm": 0.361019104719162,
      "learning_rate": 7.23531279771356e-05,
      "loss": 0.0702,
      "step": 4358
    },
    {
      "epoch": 8.291012838801711,
      "grad_norm": 0.2318553328514099,
      "learning_rate": 7.234677675452524e-05,
      "loss": 0.0546,
      "step": 4359
    },
    {
      "epoch": 8.292914883499762,
      "grad_norm": 0.23865951597690582,
      "learning_rate": 7.23404255319149e-05,
      "loss": 0.0737,
      "step": 4360
    },
    {
      "epoch": 8.294816928197813,
      "grad_norm": 0.21517156064510345,
      "learning_rate": 7.233407430930455e-05,
      "loss": 0.0627,
      "step": 4361
    },
    {
      "epoch": 8.296718972895864,
      "grad_norm": 0.30916860699653625,
      "learning_rate": 7.232772308669419e-05,
      "loss": 0.0637,
      "step": 4362
    },
    {
      "epoch": 8.298621017593913,
      "grad_norm": 0.16915148496627808,
      "learning_rate": 7.232137186408384e-05,
      "loss": 0.061,
      "step": 4363
    },
    {
      "epoch": 8.300523062291964,
      "grad_norm": 0.19052615761756897,
      "learning_rate": 7.231502064147349e-05,
      "loss": 0.0504,
      "step": 4364
    },
    {
      "epoch": 8.302425106990015,
      "grad_norm": 0.35525116324424744,
      "learning_rate": 7.230866941886314e-05,
      "loss": 0.0802,
      "step": 4365
    },
    {
      "epoch": 8.304327151688065,
      "grad_norm": 0.2397274225950241,
      "learning_rate": 7.230231819625278e-05,
      "loss": 0.088,
      "step": 4366
    },
    {
      "epoch": 8.306229196386115,
      "grad_norm": 0.30947449803352356,
      "learning_rate": 7.229596697364243e-05,
      "loss": 0.0997,
      "step": 4367
    },
    {
      "epoch": 8.308131241084165,
      "grad_norm": 0.3109225630760193,
      "learning_rate": 7.228961575103208e-05,
      "loss": 0.0631,
      "step": 4368
    },
    {
      "epoch": 8.310033285782216,
      "grad_norm": 0.27324262261390686,
      "learning_rate": 7.228326452842172e-05,
      "loss": 0.0741,
      "step": 4369
    },
    {
      "epoch": 8.311935330480265,
      "grad_norm": 0.3617977201938629,
      "learning_rate": 7.227691330581136e-05,
      "loss": 0.0787,
      "step": 4370
    },
    {
      "epoch": 8.313837375178316,
      "grad_norm": 0.2727469205856323,
      "learning_rate": 7.227056208320103e-05,
      "loss": 0.0584,
      "step": 4371
    },
    {
      "epoch": 8.315739419876367,
      "grad_norm": 0.24662989377975464,
      "learning_rate": 7.226421086059066e-05,
      "loss": 0.0556,
      "step": 4372
    },
    {
      "epoch": 8.317641464574418,
      "grad_norm": 0.264032244682312,
      "learning_rate": 7.225785963798032e-05,
      "loss": 0.067,
      "step": 4373
    },
    {
      "epoch": 8.319543509272467,
      "grad_norm": 0.19273340702056885,
      "learning_rate": 7.225150841536997e-05,
      "loss": 0.0626,
      "step": 4374
    },
    {
      "epoch": 8.321445553970518,
      "grad_norm": 0.2576651871204376,
      "learning_rate": 7.22451571927596e-05,
      "loss": 0.0613,
      "step": 4375
    },
    {
      "epoch": 8.323347598668569,
      "grad_norm": 0.40740740299224854,
      "learning_rate": 7.223880597014926e-05,
      "loss": 0.0928,
      "step": 4376
    },
    {
      "epoch": 8.32524964336662,
      "grad_norm": 0.30207934975624084,
      "learning_rate": 7.22324547475389e-05,
      "loss": 0.0837,
      "step": 4377
    },
    {
      "epoch": 8.327151688064669,
      "grad_norm": 0.3514809012413025,
      "learning_rate": 7.222610352492856e-05,
      "loss": 0.1216,
      "step": 4378
    },
    {
      "epoch": 8.32905373276272,
      "grad_norm": 0.3978719711303711,
      "learning_rate": 7.22197523023182e-05,
      "loss": 0.0879,
      "step": 4379
    },
    {
      "epoch": 8.33095577746077,
      "grad_norm": 0.23223471641540527,
      "learning_rate": 7.221340107970784e-05,
      "loss": 0.0664,
      "step": 4380
    },
    {
      "epoch": 8.332857822158822,
      "grad_norm": 0.31868988275527954,
      "learning_rate": 7.22070498570975e-05,
      "loss": 0.0947,
      "step": 4381
    },
    {
      "epoch": 8.33475986685687,
      "grad_norm": 0.23482367396354675,
      "learning_rate": 7.220069863448714e-05,
      "loss": 0.0642,
      "step": 4382
    },
    {
      "epoch": 8.336661911554922,
      "grad_norm": 0.2527036964893341,
      "learning_rate": 7.21943474118768e-05,
      "loss": 0.0639,
      "step": 4383
    },
    {
      "epoch": 8.338563956252973,
      "grad_norm": 0.3074730932712555,
      "learning_rate": 7.218799618926643e-05,
      "loss": 0.0575,
      "step": 4384
    },
    {
      "epoch": 8.340466000951022,
      "grad_norm": 0.253987193107605,
      "learning_rate": 7.218164496665608e-05,
      "loss": 0.0787,
      "step": 4385
    },
    {
      "epoch": 8.342368045649073,
      "grad_norm": 0.19843977689743042,
      "learning_rate": 7.217529374404574e-05,
      "loss": 0.0733,
      "step": 4386
    },
    {
      "epoch": 8.344270090347123,
      "grad_norm": 0.3194156587123871,
      "learning_rate": 7.216894252143537e-05,
      "loss": 0.0811,
      "step": 4387
    },
    {
      "epoch": 8.346172135045174,
      "grad_norm": 0.3143181800842285,
      "learning_rate": 7.216259129882503e-05,
      "loss": 0.0753,
      "step": 4388
    },
    {
      "epoch": 8.348074179743223,
      "grad_norm": 0.24854514002799988,
      "learning_rate": 7.215624007621468e-05,
      "loss": 0.0545,
      "step": 4389
    },
    {
      "epoch": 8.349976224441274,
      "grad_norm": 0.30429551005363464,
      "learning_rate": 7.214988885360432e-05,
      "loss": 0.0629,
      "step": 4390
    },
    {
      "epoch": 8.351878269139325,
      "grad_norm": 0.23647554218769073,
      "learning_rate": 7.214353763099397e-05,
      "loss": 0.0614,
      "step": 4391
    },
    {
      "epoch": 8.353780313837376,
      "grad_norm": 0.2253941148519516,
      "learning_rate": 7.213718640838362e-05,
      "loss": 0.0715,
      "step": 4392
    },
    {
      "epoch": 8.355682358535425,
      "grad_norm": 0.24400204420089722,
      "learning_rate": 7.213083518577326e-05,
      "loss": 0.074,
      "step": 4393
    },
    {
      "epoch": 8.357584403233476,
      "grad_norm": 0.27485939860343933,
      "learning_rate": 7.212448396316291e-05,
      "loss": 0.075,
      "step": 4394
    },
    {
      "epoch": 8.359486447931527,
      "grad_norm": 0.25887802243232727,
      "learning_rate": 7.211813274055256e-05,
      "loss": 0.073,
      "step": 4395
    },
    {
      "epoch": 8.361388492629576,
      "grad_norm": 0.25531402230262756,
      "learning_rate": 7.211178151794221e-05,
      "loss": 0.0914,
      "step": 4396
    },
    {
      "epoch": 8.363290537327627,
      "grad_norm": 0.3148556649684906,
      "learning_rate": 7.210543029533185e-05,
      "loss": 0.0757,
      "step": 4397
    },
    {
      "epoch": 8.365192582025678,
      "grad_norm": 0.19946114718914032,
      "learning_rate": 7.20990790727215e-05,
      "loss": 0.073,
      "step": 4398
    },
    {
      "epoch": 8.367094626723729,
      "grad_norm": 0.30881115794181824,
      "learning_rate": 7.209272785011116e-05,
      "loss": 0.0966,
      "step": 4399
    },
    {
      "epoch": 8.368996671421778,
      "grad_norm": 0.20002616941928864,
      "learning_rate": 7.20863766275008e-05,
      "loss": 0.0611,
      "step": 4400
    },
    {
      "epoch": 8.370898716119829,
      "grad_norm": 0.34020841121673584,
      "learning_rate": 7.208002540489045e-05,
      "loss": 0.0885,
      "step": 4401
    },
    {
      "epoch": 8.37280076081788,
      "grad_norm": 0.36568552255630493,
      "learning_rate": 7.20736741822801e-05,
      "loss": 0.0942,
      "step": 4402
    },
    {
      "epoch": 8.37470280551593,
      "grad_norm": 0.25258252024650574,
      "learning_rate": 7.206732295966974e-05,
      "loss": 0.0684,
      "step": 4403
    },
    {
      "epoch": 8.37660485021398,
      "grad_norm": 0.34854474663734436,
      "learning_rate": 7.206097173705939e-05,
      "loss": 0.0822,
      "step": 4404
    },
    {
      "epoch": 8.37850689491203,
      "grad_norm": 0.2970679700374603,
      "learning_rate": 7.205462051444904e-05,
      "loss": 0.0773,
      "step": 4405
    },
    {
      "epoch": 8.380408939610081,
      "grad_norm": 0.2980916500091553,
      "learning_rate": 7.204826929183868e-05,
      "loss": 0.0983,
      "step": 4406
    },
    {
      "epoch": 8.38231098430813,
      "grad_norm": 0.21473757922649384,
      "learning_rate": 7.204191806922833e-05,
      "loss": 0.0624,
      "step": 4407
    },
    {
      "epoch": 8.384213029006181,
      "grad_norm": 0.41438958048820496,
      "learning_rate": 7.203556684661797e-05,
      "loss": 0.0941,
      "step": 4408
    },
    {
      "epoch": 8.386115073704232,
      "grad_norm": 0.30517882108688354,
      "learning_rate": 7.202921562400763e-05,
      "loss": 0.0812,
      "step": 4409
    },
    {
      "epoch": 8.388017118402283,
      "grad_norm": 0.2721097469329834,
      "learning_rate": 7.202286440139727e-05,
      "loss": 0.0696,
      "step": 4410
    },
    {
      "epoch": 8.389919163100332,
      "grad_norm": 0.1893528401851654,
      "learning_rate": 7.201651317878691e-05,
      "loss": 0.0823,
      "step": 4411
    },
    {
      "epoch": 8.391821207798383,
      "grad_norm": 0.33133113384246826,
      "learning_rate": 7.201016195617658e-05,
      "loss": 0.0812,
      "step": 4412
    },
    {
      "epoch": 8.393723252496434,
      "grad_norm": 0.35746121406555176,
      "learning_rate": 7.200381073356621e-05,
      "loss": 0.093,
      "step": 4413
    },
    {
      "epoch": 8.395625297194485,
      "grad_norm": 0.20589201152324677,
      "learning_rate": 7.199745951095587e-05,
      "loss": 0.0832,
      "step": 4414
    },
    {
      "epoch": 8.397527341892534,
      "grad_norm": 0.29606151580810547,
      "learning_rate": 7.19911082883455e-05,
      "loss": 0.0726,
      "step": 4415
    },
    {
      "epoch": 8.399429386590585,
      "grad_norm": 0.17773225903511047,
      "learning_rate": 7.198475706573516e-05,
      "loss": 0.0609,
      "step": 4416
    },
    {
      "epoch": 8.401331431288636,
      "grad_norm": 0.2160770446062088,
      "learning_rate": 7.197840584312481e-05,
      "loss": 0.093,
      "step": 4417
    },
    {
      "epoch": 8.403233475986685,
      "grad_norm": 0.2273487150669098,
      "learning_rate": 7.197205462051445e-05,
      "loss": 0.0617,
      "step": 4418
    },
    {
      "epoch": 8.405135520684736,
      "grad_norm": 0.30558666586875916,
      "learning_rate": 7.196570339790411e-05,
      "loss": 0.0875,
      "step": 4419
    },
    {
      "epoch": 8.407037565382787,
      "grad_norm": 0.208307147026062,
      "learning_rate": 7.195935217529375e-05,
      "loss": 0.0832,
      "step": 4420
    },
    {
      "epoch": 8.408939610080838,
      "grad_norm": 0.2679693400859833,
      "learning_rate": 7.195300095268339e-05,
      "loss": 0.0654,
      "step": 4421
    },
    {
      "epoch": 8.410841654778887,
      "grad_norm": 0.2670406997203827,
      "learning_rate": 7.194664973007304e-05,
      "loss": 0.0612,
      "step": 4422
    },
    {
      "epoch": 8.412743699476938,
      "grad_norm": 0.21855837106704712,
      "learning_rate": 7.194029850746269e-05,
      "loss": 0.0712,
      "step": 4423
    },
    {
      "epoch": 8.414645744174988,
      "grad_norm": 0.2817660868167877,
      "learning_rate": 7.193394728485233e-05,
      "loss": 0.0824,
      "step": 4424
    },
    {
      "epoch": 8.41654778887304,
      "grad_norm": 0.26271113753318787,
      "learning_rate": 7.192759606224198e-05,
      "loss": 0.067,
      "step": 4425
    },
    {
      "epoch": 8.418449833571088,
      "grad_norm": 0.23378713428974152,
      "learning_rate": 7.192124483963163e-05,
      "loss": 0.0664,
      "step": 4426
    },
    {
      "epoch": 8.42035187826914,
      "grad_norm": 0.3081713616847992,
      "learning_rate": 7.191489361702129e-05,
      "loss": 0.0784,
      "step": 4427
    },
    {
      "epoch": 8.42225392296719,
      "grad_norm": 0.3694870173931122,
      "learning_rate": 7.190854239441092e-05,
      "loss": 0.0782,
      "step": 4428
    },
    {
      "epoch": 8.42415596766524,
      "grad_norm": 0.273703396320343,
      "learning_rate": 7.190219117180058e-05,
      "loss": 0.0975,
      "step": 4429
    },
    {
      "epoch": 8.42605801236329,
      "grad_norm": 0.21073737740516663,
      "learning_rate": 7.189583994919023e-05,
      "loss": 0.0639,
      "step": 4430
    },
    {
      "epoch": 8.427960057061341,
      "grad_norm": 0.273165225982666,
      "learning_rate": 7.188948872657987e-05,
      "loss": 0.0653,
      "step": 4431
    },
    {
      "epoch": 8.429862101759392,
      "grad_norm": 0.23675204813480377,
      "learning_rate": 7.188313750396952e-05,
      "loss": 0.0555,
      "step": 4432
    },
    {
      "epoch": 8.431764146457441,
      "grad_norm": 0.35497403144836426,
      "learning_rate": 7.187678628135917e-05,
      "loss": 0.0789,
      "step": 4433
    },
    {
      "epoch": 8.433666191155492,
      "grad_norm": 0.24616917967796326,
      "learning_rate": 7.187043505874881e-05,
      "loss": 0.0532,
      "step": 4434
    },
    {
      "epoch": 8.435568235853543,
      "grad_norm": 0.2543208599090576,
      "learning_rate": 7.186408383613846e-05,
      "loss": 0.0732,
      "step": 4435
    },
    {
      "epoch": 8.437470280551594,
      "grad_norm": 0.378913551568985,
      "learning_rate": 7.185773261352811e-05,
      "loss": 0.0916,
      "step": 4436
    },
    {
      "epoch": 8.439372325249643,
      "grad_norm": 0.19589996337890625,
      "learning_rate": 7.185138139091776e-05,
      "loss": 0.0573,
      "step": 4437
    },
    {
      "epoch": 8.441274369947694,
      "grad_norm": 0.27976134419441223,
      "learning_rate": 7.18450301683074e-05,
      "loss": 0.0723,
      "step": 4438
    },
    {
      "epoch": 8.443176414645745,
      "grad_norm": 0.1991090476512909,
      "learning_rate": 7.183867894569705e-05,
      "loss": 0.0618,
      "step": 4439
    },
    {
      "epoch": 8.445078459343794,
      "grad_norm": 0.18157024681568146,
      "learning_rate": 7.18323277230867e-05,
      "loss": 0.0744,
      "step": 4440
    },
    {
      "epoch": 8.446980504041845,
      "grad_norm": 0.2461085170507431,
      "learning_rate": 7.182597650047634e-05,
      "loss": 0.0622,
      "step": 4441
    },
    {
      "epoch": 8.448882548739896,
      "grad_norm": 0.35755425691604614,
      "learning_rate": 7.181962527786598e-05,
      "loss": 0.0822,
      "step": 4442
    },
    {
      "epoch": 8.450784593437946,
      "grad_norm": 0.2072787582874298,
      "learning_rate": 7.181327405525565e-05,
      "loss": 0.0664,
      "step": 4443
    },
    {
      "epoch": 8.452686638135996,
      "grad_norm": 0.3196927607059479,
      "learning_rate": 7.180692283264528e-05,
      "loss": 0.0717,
      "step": 4444
    },
    {
      "epoch": 8.454588682834046,
      "grad_norm": 0.23601149022579193,
      "learning_rate": 7.180057161003494e-05,
      "loss": 0.1213,
      "step": 4445
    },
    {
      "epoch": 8.456490727532097,
      "grad_norm": 0.18101097643375397,
      "learning_rate": 7.179422038742459e-05,
      "loss": 0.0645,
      "step": 4446
    },
    {
      "epoch": 8.458392772230148,
      "grad_norm": 0.29692402482032776,
      "learning_rate": 7.178786916481423e-05,
      "loss": 0.0667,
      "step": 4447
    },
    {
      "epoch": 8.460294816928197,
      "grad_norm": 0.32833683490753174,
      "learning_rate": 7.178151794220388e-05,
      "loss": 0.1109,
      "step": 4448
    },
    {
      "epoch": 8.462196861626248,
      "grad_norm": 0.4002103805541992,
      "learning_rate": 7.177516671959352e-05,
      "loss": 0.0804,
      "step": 4449
    },
    {
      "epoch": 8.464098906324299,
      "grad_norm": 0.40316230058670044,
      "learning_rate": 7.176881549698318e-05,
      "loss": 0.1148,
      "step": 4450
    },
    {
      "epoch": 8.466000951022348,
      "grad_norm": 0.24910256266593933,
      "learning_rate": 7.176246427437282e-05,
      "loss": 0.089,
      "step": 4451
    },
    {
      "epoch": 8.467902995720399,
      "grad_norm": 0.31366637349128723,
      "learning_rate": 7.175611305176246e-05,
      "loss": 0.0822,
      "step": 4452
    },
    {
      "epoch": 8.46980504041845,
      "grad_norm": 0.21720853447914124,
      "learning_rate": 7.174976182915212e-05,
      "loss": 0.0842,
      "step": 4453
    },
    {
      "epoch": 8.4717070851165,
      "grad_norm": 0.4722314476966858,
      "learning_rate": 7.174341060654176e-05,
      "loss": 0.0882,
      "step": 4454
    },
    {
      "epoch": 8.47360912981455,
      "grad_norm": 0.2937127649784088,
      "learning_rate": 7.173705938393141e-05,
      "loss": 0.0752,
      "step": 4455
    },
    {
      "epoch": 8.4755111745126,
      "grad_norm": 0.3475329577922821,
      "learning_rate": 7.173070816132105e-05,
      "loss": 0.0652,
      "step": 4456
    },
    {
      "epoch": 8.477413219210652,
      "grad_norm": 0.32552871108055115,
      "learning_rate": 7.17243569387107e-05,
      "loss": 0.0824,
      "step": 4457
    },
    {
      "epoch": 8.479315263908703,
      "grad_norm": 0.3245328664779663,
      "learning_rate": 7.171800571610036e-05,
      "loss": 0.078,
      "step": 4458
    },
    {
      "epoch": 8.481217308606752,
      "grad_norm": 0.3682286739349365,
      "learning_rate": 7.171165449349e-05,
      "loss": 0.0672,
      "step": 4459
    },
    {
      "epoch": 8.483119353304803,
      "grad_norm": 0.37250223755836487,
      "learning_rate": 7.170530327087965e-05,
      "loss": 0.073,
      "step": 4460
    },
    {
      "epoch": 8.485021398002853,
      "grad_norm": 0.3498341739177704,
      "learning_rate": 7.16989520482693e-05,
      "loss": 0.0817,
      "step": 4461
    },
    {
      "epoch": 8.486923442700903,
      "grad_norm": 0.2839718759059906,
      "learning_rate": 7.169260082565894e-05,
      "loss": 0.0652,
      "step": 4462
    },
    {
      "epoch": 8.488825487398953,
      "grad_norm": 0.2696433663368225,
      "learning_rate": 7.168624960304859e-05,
      "loss": 0.0693,
      "step": 4463
    },
    {
      "epoch": 8.490727532097004,
      "grad_norm": 0.3266785144805908,
      "learning_rate": 7.167989838043824e-05,
      "loss": 0.0927,
      "step": 4464
    },
    {
      "epoch": 8.492629576795055,
      "grad_norm": 0.275698184967041,
      "learning_rate": 7.167354715782788e-05,
      "loss": 0.068,
      "step": 4465
    },
    {
      "epoch": 8.494531621493104,
      "grad_norm": 0.24140353500843048,
      "learning_rate": 7.166719593521753e-05,
      "loss": 0.0543,
      "step": 4466
    },
    {
      "epoch": 8.496433666191155,
      "grad_norm": 0.2588673233985901,
      "learning_rate": 7.166084471260718e-05,
      "loss": 0.0983,
      "step": 4467
    },
    {
      "epoch": 8.498335710889206,
      "grad_norm": 0.2226792275905609,
      "learning_rate": 7.165449348999683e-05,
      "loss": 0.0728,
      "step": 4468
    },
    {
      "epoch": 8.500237755587257,
      "grad_norm": 0.2721460461616516,
      "learning_rate": 7.164814226738647e-05,
      "loss": 0.0807,
      "step": 4469
    },
    {
      "epoch": 8.502139800285306,
      "grad_norm": 0.2585650384426117,
      "learning_rate": 7.164179104477612e-05,
      "loss": 0.0973,
      "step": 4470
    },
    {
      "epoch": 8.504041844983357,
      "grad_norm": 0.20774824917316437,
      "learning_rate": 7.163543982216578e-05,
      "loss": 0.0598,
      "step": 4471
    },
    {
      "epoch": 8.505943889681408,
      "grad_norm": 0.22531108558177948,
      "learning_rate": 7.162908859955541e-05,
      "loss": 0.0658,
      "step": 4472
    },
    {
      "epoch": 8.507845934379457,
      "grad_norm": 0.23569051921367645,
      "learning_rate": 7.162273737694507e-05,
      "loss": 0.0778,
      "step": 4473
    },
    {
      "epoch": 8.509747979077508,
      "grad_norm": 0.28449130058288574,
      "learning_rate": 7.161638615433472e-05,
      "loss": 0.0868,
      "step": 4474
    },
    {
      "epoch": 8.511650023775559,
      "grad_norm": 0.2436315268278122,
      "learning_rate": 7.161003493172436e-05,
      "loss": 0.071,
      "step": 4475
    },
    {
      "epoch": 8.51355206847361,
      "grad_norm": 0.2228497564792633,
      "learning_rate": 7.160368370911401e-05,
      "loss": 0.0742,
      "step": 4476
    },
    {
      "epoch": 8.515454113171659,
      "grad_norm": 0.2390798032283783,
      "learning_rate": 7.159733248650366e-05,
      "loss": 0.0694,
      "step": 4477
    },
    {
      "epoch": 8.51735615786971,
      "grad_norm": 0.15328125655651093,
      "learning_rate": 7.15909812638933e-05,
      "loss": 0.0558,
      "step": 4478
    },
    {
      "epoch": 8.51925820256776,
      "grad_norm": 0.1791938692331314,
      "learning_rate": 7.158463004128295e-05,
      "loss": 0.0662,
      "step": 4479
    },
    {
      "epoch": 8.521160247265811,
      "grad_norm": 0.20867963135242462,
      "learning_rate": 7.157827881867259e-05,
      "loss": 0.0658,
      "step": 4480
    },
    {
      "epoch": 8.52306229196386,
      "grad_norm": 0.21594001352787018,
      "learning_rate": 7.157192759606225e-05,
      "loss": 0.053,
      "step": 4481
    },
    {
      "epoch": 8.524964336661911,
      "grad_norm": 0.18117885291576385,
      "learning_rate": 7.156557637345189e-05,
      "loss": 0.0497,
      "step": 4482
    },
    {
      "epoch": 8.526866381359962,
      "grad_norm": 0.3606829345226288,
      "learning_rate": 7.155922515084153e-05,
      "loss": 0.0831,
      "step": 4483
    },
    {
      "epoch": 8.528768426058011,
      "grad_norm": 0.26378872990608215,
      "learning_rate": 7.15528739282312e-05,
      "loss": 0.081,
      "step": 4484
    },
    {
      "epoch": 8.530670470756062,
      "grad_norm": 0.38574284315109253,
      "learning_rate": 7.154652270562083e-05,
      "loss": 0.0995,
      "step": 4485
    },
    {
      "epoch": 8.532572515454113,
      "grad_norm": 0.241069957613945,
      "learning_rate": 7.154017148301049e-05,
      "loss": 0.0672,
      "step": 4486
    },
    {
      "epoch": 8.534474560152164,
      "grad_norm": 0.32855287194252014,
      "learning_rate": 7.153382026040012e-05,
      "loss": 0.0887,
      "step": 4487
    },
    {
      "epoch": 8.536376604850213,
      "grad_norm": 0.2607368230819702,
      "learning_rate": 7.152746903778978e-05,
      "loss": 0.0835,
      "step": 4488
    },
    {
      "epoch": 8.538278649548264,
      "grad_norm": 0.23832017183303833,
      "learning_rate": 7.152111781517943e-05,
      "loss": 0.07,
      "step": 4489
    },
    {
      "epoch": 8.540180694246315,
      "grad_norm": 0.3213230073451996,
      "learning_rate": 7.151476659256907e-05,
      "loss": 0.0777,
      "step": 4490
    },
    {
      "epoch": 8.542082738944366,
      "grad_norm": 0.36647650599479675,
      "learning_rate": 7.150841536995873e-05,
      "loss": 0.0909,
      "step": 4491
    },
    {
      "epoch": 8.543984783642415,
      "grad_norm": 0.20226097106933594,
      "learning_rate": 7.150206414734837e-05,
      "loss": 0.0679,
      "step": 4492
    },
    {
      "epoch": 8.545886828340466,
      "grad_norm": 0.2921244502067566,
      "learning_rate": 7.149571292473801e-05,
      "loss": 0.0736,
      "step": 4493
    },
    {
      "epoch": 8.547788873038517,
      "grad_norm": 0.24473552405834198,
      "learning_rate": 7.148936170212766e-05,
      "loss": 0.0706,
      "step": 4494
    },
    {
      "epoch": 8.549690917736568,
      "grad_norm": 0.5814130306243896,
      "learning_rate": 7.148301047951731e-05,
      "loss": 0.1231,
      "step": 4495
    },
    {
      "epoch": 8.551592962434617,
      "grad_norm": 0.33063530921936035,
      "learning_rate": 7.147665925690695e-05,
      "loss": 0.0891,
      "step": 4496
    },
    {
      "epoch": 8.553495007132668,
      "grad_norm": 0.3366851508617401,
      "learning_rate": 7.14703080342966e-05,
      "loss": 0.0909,
      "step": 4497
    },
    {
      "epoch": 8.555397051830719,
      "grad_norm": 0.2970556914806366,
      "learning_rate": 7.146395681168625e-05,
      "loss": 0.0826,
      "step": 4498
    },
    {
      "epoch": 8.557299096528768,
      "grad_norm": 0.2414432168006897,
      "learning_rate": 7.14576055890759e-05,
      "loss": 0.0778,
      "step": 4499
    },
    {
      "epoch": 8.559201141226819,
      "grad_norm": 0.22268299758434296,
      "learning_rate": 7.145125436646554e-05,
      "loss": 0.0621,
      "step": 4500
    },
    {
      "epoch": 8.56110318592487,
      "grad_norm": 0.2911908030509949,
      "learning_rate": 7.14449031438552e-05,
      "loss": 0.058,
      "step": 4501
    },
    {
      "epoch": 8.56300523062292,
      "grad_norm": 0.17397868633270264,
      "learning_rate": 7.143855192124485e-05,
      "loss": 0.0554,
      "step": 4502
    },
    {
      "epoch": 8.56490727532097,
      "grad_norm": 0.2673262357711792,
      "learning_rate": 7.143220069863449e-05,
      "loss": 0.0591,
      "step": 4503
    },
    {
      "epoch": 8.56680932001902,
      "grad_norm": 0.29203251004219055,
      "learning_rate": 7.142584947602414e-05,
      "loss": 0.0919,
      "step": 4504
    },
    {
      "epoch": 8.568711364717071,
      "grad_norm": 0.22142189741134644,
      "learning_rate": 7.141949825341379e-05,
      "loss": 0.0674,
      "step": 4505
    },
    {
      "epoch": 8.570613409415122,
      "grad_norm": 0.34359076619148254,
      "learning_rate": 7.141314703080343e-05,
      "loss": 0.0677,
      "step": 4506
    },
    {
      "epoch": 8.572515454113171,
      "grad_norm": 0.23393453657627106,
      "learning_rate": 7.140679580819308e-05,
      "loss": 0.0712,
      "step": 4507
    },
    {
      "epoch": 8.574417498811222,
      "grad_norm": 0.2184855043888092,
      "learning_rate": 7.140044458558273e-05,
      "loss": 0.0737,
      "step": 4508
    },
    {
      "epoch": 8.576319543509273,
      "grad_norm": 0.26712971925735474,
      "learning_rate": 7.139409336297238e-05,
      "loss": 0.0863,
      "step": 4509
    },
    {
      "epoch": 8.578221588207322,
      "grad_norm": 0.22099913656711578,
      "learning_rate": 7.138774214036202e-05,
      "loss": 0.0679,
      "step": 4510
    },
    {
      "epoch": 8.580123632905373,
      "grad_norm": 0.389970600605011,
      "learning_rate": 7.138139091775167e-05,
      "loss": 0.0776,
      "step": 4511
    },
    {
      "epoch": 8.582025677603424,
      "grad_norm": 0.2818027138710022,
      "learning_rate": 7.137503969514133e-05,
      "loss": 0.0751,
      "step": 4512
    },
    {
      "epoch": 8.583927722301475,
      "grad_norm": 0.2537659704685211,
      "learning_rate": 7.136868847253096e-05,
      "loss": 0.0708,
      "step": 4513
    },
    {
      "epoch": 8.585829766999524,
      "grad_norm": 0.2738887369632721,
      "learning_rate": 7.13623372499206e-05,
      "loss": 0.0544,
      "step": 4514
    },
    {
      "epoch": 8.587731811697575,
      "grad_norm": 0.21738064289093018,
      "learning_rate": 7.135598602731027e-05,
      "loss": 0.0663,
      "step": 4515
    },
    {
      "epoch": 8.589633856395626,
      "grad_norm": 0.2945701479911804,
      "learning_rate": 7.13496348046999e-05,
      "loss": 0.0558,
      "step": 4516
    },
    {
      "epoch": 8.591535901093676,
      "grad_norm": 0.2870883345603943,
      "learning_rate": 7.134328358208956e-05,
      "loss": 0.0695,
      "step": 4517
    },
    {
      "epoch": 8.593437945791726,
      "grad_norm": 0.28586509823799133,
      "learning_rate": 7.13369323594792e-05,
      "loss": 0.0715,
      "step": 4518
    },
    {
      "epoch": 8.595339990489776,
      "grad_norm": 0.25800830125808716,
      "learning_rate": 7.133058113686885e-05,
      "loss": 0.0791,
      "step": 4519
    },
    {
      "epoch": 8.597242035187827,
      "grad_norm": 0.23650942742824554,
      "learning_rate": 7.13242299142585e-05,
      "loss": 0.0628,
      "step": 4520
    },
    {
      "epoch": 8.599144079885878,
      "grad_norm": 0.5239331126213074,
      "learning_rate": 7.131787869164814e-05,
      "loss": 0.0821,
      "step": 4521
    },
    {
      "epoch": 8.601046124583927,
      "grad_norm": 0.2804233431816101,
      "learning_rate": 7.13115274690378e-05,
      "loss": 0.087,
      "step": 4522
    },
    {
      "epoch": 8.602948169281978,
      "grad_norm": 0.29622703790664673,
      "learning_rate": 7.130517624642744e-05,
      "loss": 0.0746,
      "step": 4523
    },
    {
      "epoch": 8.60485021398003,
      "grad_norm": 0.2862629294395447,
      "learning_rate": 7.129882502381708e-05,
      "loss": 0.0854,
      "step": 4524
    },
    {
      "epoch": 8.606752258678078,
      "grad_norm": 0.23067869246006012,
      "learning_rate": 7.129247380120673e-05,
      "loss": 0.0726,
      "step": 4525
    },
    {
      "epoch": 8.60865430337613,
      "grad_norm": 0.2480565309524536,
      "learning_rate": 7.128612257859638e-05,
      "loss": 0.0594,
      "step": 4526
    },
    {
      "epoch": 8.61055634807418,
      "grad_norm": 0.22137805819511414,
      "learning_rate": 7.127977135598604e-05,
      "loss": 0.06,
      "step": 4527
    },
    {
      "epoch": 8.612458392772231,
      "grad_norm": 0.1964171975851059,
      "learning_rate": 7.127342013337567e-05,
      "loss": 0.0529,
      "step": 4528
    },
    {
      "epoch": 8.61436043747028,
      "grad_norm": 0.2646423280239105,
      "learning_rate": 7.126706891076533e-05,
      "loss": 0.0757,
      "step": 4529
    },
    {
      "epoch": 8.616262482168331,
      "grad_norm": 0.25427189469337463,
      "learning_rate": 7.126071768815498e-05,
      "loss": 0.0652,
      "step": 4530
    },
    {
      "epoch": 8.618164526866382,
      "grad_norm": 0.22125300765037537,
      "learning_rate": 7.125436646554462e-05,
      "loss": 0.0838,
      "step": 4531
    },
    {
      "epoch": 8.620066571564433,
      "grad_norm": 0.33113163709640503,
      "learning_rate": 7.124801524293427e-05,
      "loss": 0.0735,
      "step": 4532
    },
    {
      "epoch": 8.621968616262482,
      "grad_norm": 0.21249188482761383,
      "learning_rate": 7.124166402032392e-05,
      "loss": 0.0655,
      "step": 4533
    },
    {
      "epoch": 8.623870660960533,
      "grad_norm": 0.3268553912639618,
      "learning_rate": 7.123531279771356e-05,
      "loss": 0.0619,
      "step": 4534
    },
    {
      "epoch": 8.625772705658584,
      "grad_norm": 0.3507033586502075,
      "learning_rate": 7.122896157510321e-05,
      "loss": 0.0734,
      "step": 4535
    },
    {
      "epoch": 8.627674750356633,
      "grad_norm": 0.20632556080818176,
      "learning_rate": 7.122261035249286e-05,
      "loss": 0.0614,
      "step": 4536
    },
    {
      "epoch": 8.629576795054684,
      "grad_norm": 0.23562835156917572,
      "learning_rate": 7.12162591298825e-05,
      "loss": 0.0592,
      "step": 4537
    },
    {
      "epoch": 8.631478839752734,
      "grad_norm": 0.2938970625400543,
      "learning_rate": 7.120990790727215e-05,
      "loss": 0.0793,
      "step": 4538
    },
    {
      "epoch": 8.633380884450785,
      "grad_norm": 0.25379982590675354,
      "learning_rate": 7.12035566846618e-05,
      "loss": 0.0658,
      "step": 4539
    },
    {
      "epoch": 8.635282929148834,
      "grad_norm": 0.32854923605918884,
      "learning_rate": 7.119720546205146e-05,
      "loss": 0.0785,
      "step": 4540
    },
    {
      "epoch": 8.637184973846885,
      "grad_norm": 0.25158101320266724,
      "learning_rate": 7.11908542394411e-05,
      "loss": 0.0885,
      "step": 4541
    },
    {
      "epoch": 8.639087018544936,
      "grad_norm": 0.3338216543197632,
      "learning_rate": 7.118450301683075e-05,
      "loss": 0.103,
      "step": 4542
    },
    {
      "epoch": 8.640989063242987,
      "grad_norm": 0.24056802690029144,
      "learning_rate": 7.11781517942204e-05,
      "loss": 0.0738,
      "step": 4543
    },
    {
      "epoch": 8.642891107941036,
      "grad_norm": 0.27864179015159607,
      "learning_rate": 7.117180057161004e-05,
      "loss": 0.0782,
      "step": 4544
    },
    {
      "epoch": 8.644793152639087,
      "grad_norm": 0.33596450090408325,
      "learning_rate": 7.116544934899969e-05,
      "loss": 0.0737,
      "step": 4545
    },
    {
      "epoch": 8.646695197337138,
      "grad_norm": 0.3385752737522125,
      "learning_rate": 7.115909812638934e-05,
      "loss": 0.0962,
      "step": 4546
    },
    {
      "epoch": 8.648597242035187,
      "grad_norm": 0.24388425052165985,
      "learning_rate": 7.115274690377898e-05,
      "loss": 0.0548,
      "step": 4547
    },
    {
      "epoch": 8.650499286733238,
      "grad_norm": 0.22161011397838593,
      "learning_rate": 7.114639568116863e-05,
      "loss": 0.0614,
      "step": 4548
    },
    {
      "epoch": 8.652401331431289,
      "grad_norm": 0.26100924611091614,
      "learning_rate": 7.114004445855828e-05,
      "loss": 0.0645,
      "step": 4549
    },
    {
      "epoch": 8.65430337612934,
      "grad_norm": 0.26137739419937134,
      "learning_rate": 7.113369323594792e-05,
      "loss": 0.0793,
      "step": 4550
    },
    {
      "epoch": 8.656205420827389,
      "grad_norm": 0.40450796484947205,
      "learning_rate": 7.112734201333757e-05,
      "loss": 0.08,
      "step": 4551
    },
    {
      "epoch": 8.65810746552544,
      "grad_norm": 0.3062434196472168,
      "learning_rate": 7.112099079072721e-05,
      "loss": 0.0827,
      "step": 4552
    },
    {
      "epoch": 8.66000951022349,
      "grad_norm": 0.3404581844806671,
      "learning_rate": 7.111463956811687e-05,
      "loss": 0.0839,
      "step": 4553
    },
    {
      "epoch": 8.661911554921542,
      "grad_norm": 0.24370938539505005,
      "learning_rate": 7.110828834550651e-05,
      "loss": 0.0686,
      "step": 4554
    },
    {
      "epoch": 8.66381359961959,
      "grad_norm": 0.24574245512485504,
      "learning_rate": 7.110193712289615e-05,
      "loss": 0.0696,
      "step": 4555
    },
    {
      "epoch": 8.665715644317642,
      "grad_norm": 0.26258182525634766,
      "learning_rate": 7.109558590028582e-05,
      "loss": 0.0828,
      "step": 4556
    },
    {
      "epoch": 8.667617689015692,
      "grad_norm": 0.3371654748916626,
      "learning_rate": 7.108923467767546e-05,
      "loss": 0.0709,
      "step": 4557
    },
    {
      "epoch": 8.669519733713742,
      "grad_norm": 0.2799132168292999,
      "learning_rate": 7.108288345506511e-05,
      "loss": 0.0739,
      "step": 4558
    },
    {
      "epoch": 8.671421778411792,
      "grad_norm": 0.6267777681350708,
      "learning_rate": 7.107653223245475e-05,
      "loss": 0.0934,
      "step": 4559
    },
    {
      "epoch": 8.673323823109843,
      "grad_norm": 0.28459432721138,
      "learning_rate": 7.10701810098444e-05,
      "loss": 0.0906,
      "step": 4560
    },
    {
      "epoch": 8.675225867807894,
      "grad_norm": 0.29621008038520813,
      "learning_rate": 7.106382978723405e-05,
      "loss": 0.0561,
      "step": 4561
    },
    {
      "epoch": 8.677127912505943,
      "grad_norm": 0.20961788296699524,
      "learning_rate": 7.105747856462369e-05,
      "loss": 0.067,
      "step": 4562
    },
    {
      "epoch": 8.679029957203994,
      "grad_norm": 0.37413978576660156,
      "learning_rate": 7.105112734201335e-05,
      "loss": 0.0912,
      "step": 4563
    },
    {
      "epoch": 8.680932001902045,
      "grad_norm": 0.2561793029308319,
      "learning_rate": 7.104477611940299e-05,
      "loss": 0.0665,
      "step": 4564
    },
    {
      "epoch": 8.682834046600096,
      "grad_norm": 0.3220752477645874,
      "learning_rate": 7.103842489679263e-05,
      "loss": 0.0621,
      "step": 4565
    },
    {
      "epoch": 8.684736091298145,
      "grad_norm": 0.18698498606681824,
      "learning_rate": 7.103207367418228e-05,
      "loss": 0.0548,
      "step": 4566
    },
    {
      "epoch": 8.686638135996196,
      "grad_norm": 0.3139233887195587,
      "learning_rate": 7.102572245157193e-05,
      "loss": 0.0971,
      "step": 4567
    },
    {
      "epoch": 8.688540180694247,
      "grad_norm": 0.2640330195426941,
      "learning_rate": 7.101937122896157e-05,
      "loss": 0.08,
      "step": 4568
    },
    {
      "epoch": 8.690442225392296,
      "grad_norm": 0.2758345901966095,
      "learning_rate": 7.101302000635122e-05,
      "loss": 0.0663,
      "step": 4569
    },
    {
      "epoch": 8.692344270090347,
      "grad_norm": 0.23548677563667297,
      "learning_rate": 7.100666878374087e-05,
      "loss": 0.0753,
      "step": 4570
    },
    {
      "epoch": 8.694246314788398,
      "grad_norm": 0.4258490204811096,
      "learning_rate": 7.100031756113053e-05,
      "loss": 0.1119,
      "step": 4571
    },
    {
      "epoch": 8.696148359486449,
      "grad_norm": 0.25525277853012085,
      "learning_rate": 7.099396633852017e-05,
      "loss": 0.0762,
      "step": 4572
    },
    {
      "epoch": 8.698050404184498,
      "grad_norm": 0.34199386835098267,
      "learning_rate": 7.098761511590982e-05,
      "loss": 0.0828,
      "step": 4573
    },
    {
      "epoch": 8.699952448882549,
      "grad_norm": 0.31173181533813477,
      "learning_rate": 7.098126389329947e-05,
      "loss": 0.0855,
      "step": 4574
    },
    {
      "epoch": 8.7018544935806,
      "grad_norm": 0.20065343379974365,
      "learning_rate": 7.097491267068911e-05,
      "loss": 0.0639,
      "step": 4575
    },
    {
      "epoch": 8.70375653827865,
      "grad_norm": 0.32433775067329407,
      "learning_rate": 7.096856144807876e-05,
      "loss": 0.0799,
      "step": 4576
    },
    {
      "epoch": 8.7056585829767,
      "grad_norm": 0.43522873520851135,
      "learning_rate": 7.096221022546841e-05,
      "loss": 0.0882,
      "step": 4577
    },
    {
      "epoch": 8.70756062767475,
      "grad_norm": 0.33370131254196167,
      "learning_rate": 7.095585900285805e-05,
      "loss": 0.0996,
      "step": 4578
    },
    {
      "epoch": 8.709462672372801,
      "grad_norm": 0.3209904730319977,
      "learning_rate": 7.09495077802477e-05,
      "loss": 0.0722,
      "step": 4579
    },
    {
      "epoch": 8.71136471707085,
      "grad_norm": 0.2941846549510956,
      "learning_rate": 7.094315655763735e-05,
      "loss": 0.0746,
      "step": 4580
    },
    {
      "epoch": 8.713266761768901,
      "grad_norm": 0.27437055110931396,
      "learning_rate": 7.0936805335027e-05,
      "loss": 0.0772,
      "step": 4581
    },
    {
      "epoch": 8.715168806466952,
      "grad_norm": 0.19886012375354767,
      "learning_rate": 7.093045411241664e-05,
      "loss": 0.0569,
      "step": 4582
    },
    {
      "epoch": 8.717070851165003,
      "grad_norm": 0.2602776885032654,
      "learning_rate": 7.092410288980628e-05,
      "loss": 0.0843,
      "step": 4583
    },
    {
      "epoch": 8.718972895863052,
      "grad_norm": 0.20333616435527802,
      "learning_rate": 7.091775166719595e-05,
      "loss": 0.1018,
      "step": 4584
    },
    {
      "epoch": 8.720874940561103,
      "grad_norm": 0.2698846757411957,
      "learning_rate": 7.091140044458558e-05,
      "loss": 0.0777,
      "step": 4585
    },
    {
      "epoch": 8.722776985259154,
      "grad_norm": 0.18672236800193787,
      "learning_rate": 7.090504922197522e-05,
      "loss": 0.0608,
      "step": 4586
    },
    {
      "epoch": 8.724679029957205,
      "grad_norm": 0.2933334410190582,
      "learning_rate": 7.089869799936489e-05,
      "loss": 0.0783,
      "step": 4587
    },
    {
      "epoch": 8.726581074655254,
      "grad_norm": 0.2911440432071686,
      "learning_rate": 7.089234677675453e-05,
      "loss": 0.0637,
      "step": 4588
    },
    {
      "epoch": 8.728483119353305,
      "grad_norm": 0.19904588162899017,
      "learning_rate": 7.088599555414418e-05,
      "loss": 0.0629,
      "step": 4589
    },
    {
      "epoch": 8.730385164051356,
      "grad_norm": 0.22173476219177246,
      "learning_rate": 7.087964433153382e-05,
      "loss": 0.0611,
      "step": 4590
    },
    {
      "epoch": 8.732287208749405,
      "grad_norm": 0.3003878891468048,
      "learning_rate": 7.087329310892347e-05,
      "loss": 0.0939,
      "step": 4591
    },
    {
      "epoch": 8.734189253447456,
      "grad_norm": 0.27259764075279236,
      "learning_rate": 7.086694188631312e-05,
      "loss": 0.0798,
      "step": 4592
    },
    {
      "epoch": 8.736091298145507,
      "grad_norm": 0.25377964973449707,
      "learning_rate": 7.086059066370276e-05,
      "loss": 0.0756,
      "step": 4593
    },
    {
      "epoch": 8.737993342843557,
      "grad_norm": 0.2532641887664795,
      "learning_rate": 7.085423944109242e-05,
      "loss": 0.0767,
      "step": 4594
    },
    {
      "epoch": 8.739895387541607,
      "grad_norm": 0.33837416768074036,
      "learning_rate": 7.084788821848206e-05,
      "loss": 0.0906,
      "step": 4595
    },
    {
      "epoch": 8.741797432239657,
      "grad_norm": 0.20276691019535065,
      "learning_rate": 7.08415369958717e-05,
      "loss": 0.0744,
      "step": 4596
    },
    {
      "epoch": 8.743699476937708,
      "grad_norm": 0.2975214123725891,
      "learning_rate": 7.083518577326135e-05,
      "loss": 0.0771,
      "step": 4597
    },
    {
      "epoch": 8.74560152163576,
      "grad_norm": 0.3634226322174072,
      "learning_rate": 7.0828834550651e-05,
      "loss": 0.0946,
      "step": 4598
    },
    {
      "epoch": 8.747503566333808,
      "grad_norm": 0.1982811689376831,
      "learning_rate": 7.082248332804066e-05,
      "loss": 0.057,
      "step": 4599
    },
    {
      "epoch": 8.74940561103186,
      "grad_norm": 0.2674190104007721,
      "learning_rate": 7.08161321054303e-05,
      "loss": 0.0695,
      "step": 4600
    },
    {
      "epoch": 8.75130765572991,
      "grad_norm": 0.40828996896743774,
      "learning_rate": 7.080978088281995e-05,
      "loss": 0.0814,
      "step": 4601
    },
    {
      "epoch": 8.75320970042796,
      "grad_norm": 0.2784786522388458,
      "learning_rate": 7.08034296602096e-05,
      "loss": 0.0791,
      "step": 4602
    },
    {
      "epoch": 8.75511174512601,
      "grad_norm": 0.26131200790405273,
      "learning_rate": 7.079707843759924e-05,
      "loss": 0.0668,
      "step": 4603
    },
    {
      "epoch": 8.757013789824061,
      "grad_norm": 0.34598174691200256,
      "learning_rate": 7.079072721498889e-05,
      "loss": 0.0671,
      "step": 4604
    },
    {
      "epoch": 8.758915834522112,
      "grad_norm": 0.28290489315986633,
      "learning_rate": 7.078437599237854e-05,
      "loss": 0.0901,
      "step": 4605
    },
    {
      "epoch": 8.760817879220161,
      "grad_norm": 0.4226842522621155,
      "learning_rate": 7.077802476976818e-05,
      "loss": 0.0848,
      "step": 4606
    },
    {
      "epoch": 8.762719923918212,
      "grad_norm": 0.23531925678253174,
      "learning_rate": 7.077167354715783e-05,
      "loss": 0.0738,
      "step": 4607
    },
    {
      "epoch": 8.764621968616263,
      "grad_norm": 0.3046768605709076,
      "learning_rate": 7.076532232454748e-05,
      "loss": 0.0723,
      "step": 4608
    },
    {
      "epoch": 8.766524013314314,
      "grad_norm": 0.22470355033874512,
      "learning_rate": 7.075897110193712e-05,
      "loss": 0.0493,
      "step": 4609
    },
    {
      "epoch": 8.768426058012363,
      "grad_norm": 0.23630782961845398,
      "learning_rate": 7.075261987932677e-05,
      "loss": 0.0627,
      "step": 4610
    },
    {
      "epoch": 8.770328102710414,
      "grad_norm": 0.38506048917770386,
      "learning_rate": 7.074626865671642e-05,
      "loss": 0.1092,
      "step": 4611
    },
    {
      "epoch": 8.772230147408465,
      "grad_norm": 0.23271790146827698,
      "learning_rate": 7.073991743410608e-05,
      "loss": 0.0726,
      "step": 4612
    },
    {
      "epoch": 8.774132192106514,
      "grad_norm": 0.22639422118663788,
      "learning_rate": 7.073356621149571e-05,
      "loss": 0.064,
      "step": 4613
    },
    {
      "epoch": 8.776034236804565,
      "grad_norm": 0.3787291347980499,
      "learning_rate": 7.072721498888537e-05,
      "loss": 0.0871,
      "step": 4614
    },
    {
      "epoch": 8.777936281502615,
      "grad_norm": 0.23703598976135254,
      "learning_rate": 7.072086376627502e-05,
      "loss": 0.0677,
      "step": 4615
    },
    {
      "epoch": 8.779838326200666,
      "grad_norm": 0.3365970551967621,
      "learning_rate": 7.071451254366466e-05,
      "loss": 0.0949,
      "step": 4616
    },
    {
      "epoch": 8.781740370898715,
      "grad_norm": 0.25305354595184326,
      "learning_rate": 7.070816132105431e-05,
      "loss": 0.0612,
      "step": 4617
    },
    {
      "epoch": 8.783642415596766,
      "grad_norm": 0.23190952837467194,
      "learning_rate": 7.070181009844396e-05,
      "loss": 0.0649,
      "step": 4618
    },
    {
      "epoch": 8.785544460294817,
      "grad_norm": 0.2549265921115875,
      "learning_rate": 7.06954588758336e-05,
      "loss": 0.0695,
      "step": 4619
    },
    {
      "epoch": 8.787446504992868,
      "grad_norm": 0.26542478799819946,
      "learning_rate": 7.068910765322325e-05,
      "loss": 0.0776,
      "step": 4620
    },
    {
      "epoch": 8.789348549690917,
      "grad_norm": 0.27121585607528687,
      "learning_rate": 7.06827564306129e-05,
      "loss": 0.1015,
      "step": 4621
    },
    {
      "epoch": 8.791250594388968,
      "grad_norm": 0.1988026648759842,
      "learning_rate": 7.067640520800254e-05,
      "loss": 0.058,
      "step": 4622
    },
    {
      "epoch": 8.793152639087019,
      "grad_norm": 0.386599063873291,
      "learning_rate": 7.067005398539219e-05,
      "loss": 0.1015,
      "step": 4623
    },
    {
      "epoch": 8.795054683785068,
      "grad_norm": 0.28242504596710205,
      "learning_rate": 7.066370276278183e-05,
      "loss": 0.0683,
      "step": 4624
    },
    {
      "epoch": 8.796956728483119,
      "grad_norm": 0.2033800631761551,
      "learning_rate": 7.06573515401715e-05,
      "loss": 0.0734,
      "step": 4625
    },
    {
      "epoch": 8.79885877318117,
      "grad_norm": 0.25961750745773315,
      "learning_rate": 7.065100031756113e-05,
      "loss": 0.0564,
      "step": 4626
    },
    {
      "epoch": 8.80076081787922,
      "grad_norm": 0.26695361733436584,
      "learning_rate": 7.064464909495077e-05,
      "loss": 0.0815,
      "step": 4627
    },
    {
      "epoch": 8.80266286257727,
      "grad_norm": 0.3340332508087158,
      "learning_rate": 7.063829787234042e-05,
      "loss": 0.0598,
      "step": 4628
    },
    {
      "epoch": 8.80456490727532,
      "grad_norm": 0.3151671290397644,
      "learning_rate": 7.063194664973008e-05,
      "loss": 0.1035,
      "step": 4629
    },
    {
      "epoch": 8.806466951973372,
      "grad_norm": 0.31824958324432373,
      "learning_rate": 7.062559542711973e-05,
      "loss": 0.0666,
      "step": 4630
    },
    {
      "epoch": 8.808368996671422,
      "grad_norm": 0.24082911014556885,
      "learning_rate": 7.061924420450937e-05,
      "loss": 0.065,
      "step": 4631
    },
    {
      "epoch": 8.810271041369472,
      "grad_norm": 0.2710180878639221,
      "learning_rate": 7.061289298189902e-05,
      "loss": 0.074,
      "step": 4632
    },
    {
      "epoch": 8.812173086067522,
      "grad_norm": 0.31997546553611755,
      "learning_rate": 7.060654175928867e-05,
      "loss": 0.0841,
      "step": 4633
    },
    {
      "epoch": 8.814075130765573,
      "grad_norm": 0.40407681465148926,
      "learning_rate": 7.060019053667831e-05,
      "loss": 0.0784,
      "step": 4634
    },
    {
      "epoch": 8.815977175463622,
      "grad_norm": 0.2711647152900696,
      "learning_rate": 7.059383931406796e-05,
      "loss": 0.0959,
      "step": 4635
    },
    {
      "epoch": 8.817879220161673,
      "grad_norm": 0.25653815269470215,
      "learning_rate": 7.058748809145761e-05,
      "loss": 0.0697,
      "step": 4636
    },
    {
      "epoch": 8.819781264859724,
      "grad_norm": 0.17264167964458466,
      "learning_rate": 7.058113686884725e-05,
      "loss": 0.0484,
      "step": 4637
    },
    {
      "epoch": 8.821683309557775,
      "grad_norm": 0.37043115496635437,
      "learning_rate": 7.05747856462369e-05,
      "loss": 0.0829,
      "step": 4638
    },
    {
      "epoch": 8.823585354255824,
      "grad_norm": 0.24396714568138123,
      "learning_rate": 7.056843442362655e-05,
      "loss": 0.0732,
      "step": 4639
    },
    {
      "epoch": 8.825487398953875,
      "grad_norm": 0.33010244369506836,
      "learning_rate": 7.056208320101619e-05,
      "loss": 0.0742,
      "step": 4640
    },
    {
      "epoch": 8.827389443651926,
      "grad_norm": 0.2644335925579071,
      "learning_rate": 7.055573197840584e-05,
      "loss": 0.0699,
      "step": 4641
    },
    {
      "epoch": 8.829291488349977,
      "grad_norm": 0.2645286023616791,
      "learning_rate": 7.05493807557955e-05,
      "loss": 0.0773,
      "step": 4642
    },
    {
      "epoch": 8.831193533048026,
      "grad_norm": 0.21194998919963837,
      "learning_rate": 7.054302953318515e-05,
      "loss": 0.1088,
      "step": 4643
    },
    {
      "epoch": 8.833095577746077,
      "grad_norm": 0.36791402101516724,
      "learning_rate": 7.053667831057479e-05,
      "loss": 0.0897,
      "step": 4644
    },
    {
      "epoch": 8.834997622444128,
      "grad_norm": 0.44055166840553284,
      "learning_rate": 7.053032708796444e-05,
      "loss": 0.099,
      "step": 4645
    },
    {
      "epoch": 8.836899667142179,
      "grad_norm": 0.2763393819332123,
      "learning_rate": 7.052397586535409e-05,
      "loss": 0.0786,
      "step": 4646
    },
    {
      "epoch": 8.838801711840228,
      "grad_norm": 0.2285812348127365,
      "learning_rate": 7.051762464274373e-05,
      "loss": 0.0465,
      "step": 4647
    },
    {
      "epoch": 8.840703756538279,
      "grad_norm": 0.266569048166275,
      "learning_rate": 7.051127342013338e-05,
      "loss": 0.0778,
      "step": 4648
    },
    {
      "epoch": 8.84260580123633,
      "grad_norm": 0.29792171716690063,
      "learning_rate": 7.050492219752303e-05,
      "loss": 0.1043,
      "step": 4649
    },
    {
      "epoch": 8.844507845934379,
      "grad_norm": 0.27587541937828064,
      "learning_rate": 7.049857097491267e-05,
      "loss": 0.0901,
      "step": 4650
    },
    {
      "epoch": 8.84640989063243,
      "grad_norm": 0.2589787542819977,
      "learning_rate": 7.049221975230232e-05,
      "loss": 0.0597,
      "step": 4651
    },
    {
      "epoch": 8.84831193533048,
      "grad_norm": 0.1944485306739807,
      "learning_rate": 7.048586852969197e-05,
      "loss": 0.0744,
      "step": 4652
    },
    {
      "epoch": 8.850213980028531,
      "grad_norm": 0.27121320366859436,
      "learning_rate": 7.047951730708163e-05,
      "loss": 0.0826,
      "step": 4653
    },
    {
      "epoch": 8.85211602472658,
      "grad_norm": 0.28086745738983154,
      "learning_rate": 7.047316608447126e-05,
      "loss": 0.0855,
      "step": 4654
    },
    {
      "epoch": 8.854018069424631,
      "grad_norm": 0.2198602259159088,
      "learning_rate": 7.04668148618609e-05,
      "loss": 0.0641,
      "step": 4655
    },
    {
      "epoch": 8.855920114122682,
      "grad_norm": 0.3515179455280304,
      "learning_rate": 7.046046363925057e-05,
      "loss": 0.0929,
      "step": 4656
    },
    {
      "epoch": 8.857822158820733,
      "grad_norm": 0.294048935174942,
      "learning_rate": 7.04541124166402e-05,
      "loss": 0.0705,
      "step": 4657
    },
    {
      "epoch": 8.859724203518782,
      "grad_norm": 0.25119510293006897,
      "learning_rate": 7.044776119402984e-05,
      "loss": 0.0622,
      "step": 4658
    },
    {
      "epoch": 8.861626248216833,
      "grad_norm": 0.2803022265434265,
      "learning_rate": 7.044140997141951e-05,
      "loss": 0.0543,
      "step": 4659
    },
    {
      "epoch": 8.863528292914884,
      "grad_norm": 0.25134971737861633,
      "learning_rate": 7.043505874880915e-05,
      "loss": 0.0848,
      "step": 4660
    },
    {
      "epoch": 8.865430337612933,
      "grad_norm": 0.23869898915290833,
      "learning_rate": 7.04287075261988e-05,
      "loss": 0.0786,
      "step": 4661
    },
    {
      "epoch": 8.867332382310984,
      "grad_norm": 0.19293729960918427,
      "learning_rate": 7.042235630358844e-05,
      "loss": 0.0615,
      "step": 4662
    },
    {
      "epoch": 8.869234427009035,
      "grad_norm": 0.23141486942768097,
      "learning_rate": 7.041600508097809e-05,
      "loss": 0.0871,
      "step": 4663
    },
    {
      "epoch": 8.871136471707086,
      "grad_norm": 0.28251227736473083,
      "learning_rate": 7.040965385836774e-05,
      "loss": 0.0707,
      "step": 4664
    },
    {
      "epoch": 8.873038516405135,
      "grad_norm": 0.1875603348016739,
      "learning_rate": 7.040330263575738e-05,
      "loss": 0.0693,
      "step": 4665
    },
    {
      "epoch": 8.874940561103186,
      "grad_norm": 0.24016056954860687,
      "learning_rate": 7.039695141314705e-05,
      "loss": 0.0656,
      "step": 4666
    },
    {
      "epoch": 8.876842605801237,
      "grad_norm": 0.4740009605884552,
      "learning_rate": 7.039060019053668e-05,
      "loss": 0.0807,
      "step": 4667
    },
    {
      "epoch": 8.878744650499288,
      "grad_norm": 0.24484196305274963,
      "learning_rate": 7.038424896792632e-05,
      "loss": 0.0679,
      "step": 4668
    },
    {
      "epoch": 8.880646695197337,
      "grad_norm": 0.46116209030151367,
      "learning_rate": 7.037789774531597e-05,
      "loss": 0.0982,
      "step": 4669
    },
    {
      "epoch": 8.882548739895388,
      "grad_norm": 0.3259490728378296,
      "learning_rate": 7.037154652270563e-05,
      "loss": 0.1054,
      "step": 4670
    },
    {
      "epoch": 8.884450784593438,
      "grad_norm": 0.27219080924987793,
      "learning_rate": 7.036519530009528e-05,
      "loss": 0.0712,
      "step": 4671
    },
    {
      "epoch": 8.88635282929149,
      "grad_norm": 0.2472408562898636,
      "learning_rate": 7.035884407748492e-05,
      "loss": 0.0564,
      "step": 4672
    },
    {
      "epoch": 8.888254873989538,
      "grad_norm": 0.26915243268013,
      "learning_rate": 7.035249285487457e-05,
      "loss": 0.0743,
      "step": 4673
    },
    {
      "epoch": 8.89015691868759,
      "grad_norm": 0.27465927600860596,
      "learning_rate": 7.034614163226422e-05,
      "loss": 0.0679,
      "step": 4674
    },
    {
      "epoch": 8.89205896338564,
      "grad_norm": 0.2622419595718384,
      "learning_rate": 7.033979040965386e-05,
      "loss": 0.0762,
      "step": 4675
    },
    {
      "epoch": 8.89396100808369,
      "grad_norm": 0.32345646619796753,
      "learning_rate": 7.033343918704351e-05,
      "loss": 0.0816,
      "step": 4676
    },
    {
      "epoch": 8.89586305278174,
      "grad_norm": 0.2678157389163971,
      "learning_rate": 7.032708796443316e-05,
      "loss": 0.0672,
      "step": 4677
    },
    {
      "epoch": 8.897765097479791,
      "grad_norm": 0.3485300838947296,
      "learning_rate": 7.03207367418228e-05,
      "loss": 0.0953,
      "step": 4678
    },
    {
      "epoch": 8.899667142177842,
      "grad_norm": 0.2810485363006592,
      "learning_rate": 7.031438551921245e-05,
      "loss": 0.069,
      "step": 4679
    },
    {
      "epoch": 8.901569186875891,
      "grad_norm": 0.3006494343280792,
      "learning_rate": 7.03080342966021e-05,
      "loss": 0.0717,
      "step": 4680
    },
    {
      "epoch": 8.903471231573942,
      "grad_norm": 0.29095882177352905,
      "learning_rate": 7.030168307399174e-05,
      "loss": 0.0762,
      "step": 4681
    },
    {
      "epoch": 8.905373276271993,
      "grad_norm": 0.21266767382621765,
      "learning_rate": 7.02953318513814e-05,
      "loss": 0.0771,
      "step": 4682
    },
    {
      "epoch": 8.907275320970044,
      "grad_norm": 0.23947753012180328,
      "learning_rate": 7.028898062877105e-05,
      "loss": 0.0641,
      "step": 4683
    },
    {
      "epoch": 8.909177365668093,
      "grad_norm": 0.24538327753543854,
      "learning_rate": 7.02826294061607e-05,
      "loss": 0.0538,
      "step": 4684
    },
    {
      "epoch": 8.911079410366144,
      "grad_norm": 0.45514872670173645,
      "learning_rate": 7.027627818355034e-05,
      "loss": 0.0759,
      "step": 4685
    },
    {
      "epoch": 8.912981455064195,
      "grad_norm": 0.31402572989463806,
      "learning_rate": 7.026992696093997e-05,
      "loss": 0.0654,
      "step": 4686
    },
    {
      "epoch": 8.914883499762244,
      "grad_norm": 0.2629324793815613,
      "learning_rate": 7.026357573832964e-05,
      "loss": 0.0729,
      "step": 4687
    },
    {
      "epoch": 8.916785544460295,
      "grad_norm": 0.34424546360969543,
      "learning_rate": 7.025722451571928e-05,
      "loss": 0.0643,
      "step": 4688
    },
    {
      "epoch": 8.918687589158345,
      "grad_norm": 0.45988425612449646,
      "learning_rate": 7.025087329310893e-05,
      "loss": 0.0942,
      "step": 4689
    },
    {
      "epoch": 8.920589633856396,
      "grad_norm": 0.2842889428138733,
      "learning_rate": 7.024452207049858e-05,
      "loss": 0.0679,
      "step": 4690
    },
    {
      "epoch": 8.922491678554445,
      "grad_norm": 0.3553121089935303,
      "learning_rate": 7.023817084788822e-05,
      "loss": 0.0778,
      "step": 4691
    },
    {
      "epoch": 8.924393723252496,
      "grad_norm": 0.2820618152618408,
      "learning_rate": 7.023181962527787e-05,
      "loss": 0.072,
      "step": 4692
    },
    {
      "epoch": 8.926295767950547,
      "grad_norm": 0.23294129967689514,
      "learning_rate": 7.022546840266751e-05,
      "loss": 0.0628,
      "step": 4693
    },
    {
      "epoch": 8.928197812648598,
      "grad_norm": 0.3666904866695404,
      "learning_rate": 7.021911718005716e-05,
      "loss": 0.0983,
      "step": 4694
    },
    {
      "epoch": 8.930099857346647,
      "grad_norm": 0.22570650279521942,
      "learning_rate": 7.021276595744681e-05,
      "loss": 0.0984,
      "step": 4695
    },
    {
      "epoch": 8.932001902044698,
      "grad_norm": 0.26434043049812317,
      "learning_rate": 7.020641473483645e-05,
      "loss": 0.0852,
      "step": 4696
    },
    {
      "epoch": 8.933903946742749,
      "grad_norm": 0.32865387201309204,
      "learning_rate": 7.020006351222612e-05,
      "loss": 0.0922,
      "step": 4697
    },
    {
      "epoch": 8.935805991440798,
      "grad_norm": 0.36958640813827515,
      "learning_rate": 7.019371228961576e-05,
      "loss": 0.0797,
      "step": 4698
    },
    {
      "epoch": 8.937708036138849,
      "grad_norm": 0.3567565679550171,
      "learning_rate": 7.01873610670054e-05,
      "loss": 0.1245,
      "step": 4699
    },
    {
      "epoch": 8.9396100808369,
      "grad_norm": 0.2572322189807892,
      "learning_rate": 7.018100984439505e-05,
      "loss": 0.0662,
      "step": 4700
    },
    {
      "epoch": 8.94151212553495,
      "grad_norm": 0.30103451013565063,
      "learning_rate": 7.01746586217847e-05,
      "loss": 0.0751,
      "step": 4701
    },
    {
      "epoch": 8.943414170233,
      "grad_norm": 0.2606084644794464,
      "learning_rate": 7.016830739917435e-05,
      "loss": 0.0679,
      "step": 4702
    },
    {
      "epoch": 8.94531621493105,
      "grad_norm": 0.2592700719833374,
      "learning_rate": 7.016195617656399e-05,
      "loss": 0.057,
      "step": 4703
    },
    {
      "epoch": 8.947218259629102,
      "grad_norm": 0.3322167694568634,
      "learning_rate": 7.015560495395364e-05,
      "loss": 0.0798,
      "step": 4704
    },
    {
      "epoch": 8.949120304327153,
      "grad_norm": 0.2863213121891022,
      "learning_rate": 7.014925373134329e-05,
      "loss": 0.0744,
      "step": 4705
    },
    {
      "epoch": 8.951022349025202,
      "grad_norm": 0.3357941806316376,
      "learning_rate": 7.014290250873293e-05,
      "loss": 0.0797,
      "step": 4706
    },
    {
      "epoch": 8.952924393723253,
      "grad_norm": 0.21185575425624847,
      "learning_rate": 7.013655128612258e-05,
      "loss": 0.0689,
      "step": 4707
    },
    {
      "epoch": 8.954826438421303,
      "grad_norm": 0.414275199174881,
      "learning_rate": 7.013020006351223e-05,
      "loss": 0.0863,
      "step": 4708
    },
    {
      "epoch": 8.956728483119353,
      "grad_norm": 0.3176036477088928,
      "learning_rate": 7.012384884090187e-05,
      "loss": 0.0641,
      "step": 4709
    },
    {
      "epoch": 8.958630527817403,
      "grad_norm": 0.2829712927341461,
      "learning_rate": 7.011749761829152e-05,
      "loss": 0.0722,
      "step": 4710
    },
    {
      "epoch": 8.960532572515454,
      "grad_norm": 0.2736256718635559,
      "learning_rate": 7.011114639568117e-05,
      "loss": 0.1003,
      "step": 4711
    },
    {
      "epoch": 8.962434617213505,
      "grad_norm": 0.2921677827835083,
      "learning_rate": 7.010479517307081e-05,
      "loss": 0.0662,
      "step": 4712
    },
    {
      "epoch": 8.964336661911554,
      "grad_norm": 0.3244280219078064,
      "learning_rate": 7.009844395046046e-05,
      "loss": 0.0774,
      "step": 4713
    },
    {
      "epoch": 8.966238706609605,
      "grad_norm": 0.3001691997051239,
      "learning_rate": 7.009209272785012e-05,
      "loss": 0.0896,
      "step": 4714
    },
    {
      "epoch": 8.968140751307656,
      "grad_norm": 0.2865767776966095,
      "learning_rate": 7.008574150523977e-05,
      "loss": 0.0636,
      "step": 4715
    },
    {
      "epoch": 8.970042796005707,
      "grad_norm": 0.28869691491127014,
      "learning_rate": 7.007939028262941e-05,
      "loss": 0.077,
      "step": 4716
    },
    {
      "epoch": 8.971944840703756,
      "grad_norm": 0.22912059724330902,
      "learning_rate": 7.007303906001906e-05,
      "loss": 0.0701,
      "step": 4717
    },
    {
      "epoch": 8.973846885401807,
      "grad_norm": 0.2385643869638443,
      "learning_rate": 7.006668783740871e-05,
      "loss": 0.0729,
      "step": 4718
    },
    {
      "epoch": 8.975748930099858,
      "grad_norm": 0.26056259870529175,
      "learning_rate": 7.006033661479835e-05,
      "loss": 0.0752,
      "step": 4719
    },
    {
      "epoch": 8.977650974797907,
      "grad_norm": 0.21698705852031708,
      "learning_rate": 7.0053985392188e-05,
      "loss": 0.0684,
      "step": 4720
    },
    {
      "epoch": 8.979553019495958,
      "grad_norm": 0.30771854519844055,
      "learning_rate": 7.004763416957765e-05,
      "loss": 0.0815,
      "step": 4721
    },
    {
      "epoch": 8.981455064194009,
      "grad_norm": 0.2765178978443146,
      "learning_rate": 7.004128294696729e-05,
      "loss": 0.064,
      "step": 4722
    },
    {
      "epoch": 8.98335710889206,
      "grad_norm": 0.31932932138442993,
      "learning_rate": 7.003493172435694e-05,
      "loss": 0.085,
      "step": 4723
    },
    {
      "epoch": 8.985259153590109,
      "grad_norm": 0.23978936672210693,
      "learning_rate": 7.00285805017466e-05,
      "loss": 0.0767,
      "step": 4724
    },
    {
      "epoch": 8.98716119828816,
      "grad_norm": 0.3429613411426544,
      "learning_rate": 7.002222927913625e-05,
      "loss": 0.0924,
      "step": 4725
    },
    {
      "epoch": 8.98906324298621,
      "grad_norm": 0.2052287459373474,
      "learning_rate": 7.001587805652588e-05,
      "loss": 0.0552,
      "step": 4726
    },
    {
      "epoch": 8.990965287684261,
      "grad_norm": 0.31074854731559753,
      "learning_rate": 7.000952683391552e-05,
      "loss": 0.0966,
      "step": 4727
    },
    {
      "epoch": 8.99286733238231,
      "grad_norm": 0.274406760931015,
      "learning_rate": 7.000317561130519e-05,
      "loss": 0.066,
      "step": 4728
    },
    {
      "epoch": 8.994769377080361,
      "grad_norm": 0.3344108462333679,
      "learning_rate": 6.999682438869483e-05,
      "loss": 0.0894,
      "step": 4729
    },
    {
      "epoch": 8.996671421778412,
      "grad_norm": 0.263041615486145,
      "learning_rate": 6.999047316608446e-05,
      "loss": 0.0786,
      "step": 4730
    },
    {
      "epoch": 8.998573466476461,
      "grad_norm": 0.33452650904655457,
      "learning_rate": 6.998412194347413e-05,
      "loss": 0.0856,
      "step": 4731
    },
    {
      "epoch": 9.000475511174512,
      "grad_norm": 0.3001817464828491,
      "learning_rate": 6.997777072086377e-05,
      "loss": 0.0864,
      "step": 4732
    },
    {
      "epoch": 9.002377555872563,
      "grad_norm": 0.541235089302063,
      "learning_rate": 6.997141949825342e-05,
      "loss": 0.1138,
      "step": 4733
    },
    {
      "epoch": 9.004279600570614,
      "grad_norm": 0.1997150033712387,
      "learning_rate": 6.996506827564306e-05,
      "loss": 0.069,
      "step": 4734
    },
    {
      "epoch": 9.006181645268663,
      "grad_norm": 0.2090476155281067,
      "learning_rate": 6.995871705303271e-05,
      "loss": 0.0624,
      "step": 4735
    },
    {
      "epoch": 9.008083689966714,
      "grad_norm": 0.15848073363304138,
      "learning_rate": 6.995236583042236e-05,
      "loss": 0.0836,
      "step": 4736
    },
    {
      "epoch": 9.009985734664765,
      "grad_norm": 0.15844178199768066,
      "learning_rate": 6.9946014607812e-05,
      "loss": 0.0624,
      "step": 4737
    },
    {
      "epoch": 9.011887779362816,
      "grad_norm": 0.2605695426464081,
      "learning_rate": 6.993966338520165e-05,
      "loss": 0.0645,
      "step": 4738
    },
    {
      "epoch": 9.013789824060865,
      "grad_norm": 0.25842902064323425,
      "learning_rate": 6.99333121625913e-05,
      "loss": 0.0806,
      "step": 4739
    },
    {
      "epoch": 9.015691868758916,
      "grad_norm": 0.18746209144592285,
      "learning_rate": 6.992696093998094e-05,
      "loss": 0.0649,
      "step": 4740
    },
    {
      "epoch": 9.017593913456967,
      "grad_norm": 0.3465903699398041,
      "learning_rate": 6.99206097173706e-05,
      "loss": 0.0652,
      "step": 4741
    },
    {
      "epoch": 9.019495958155016,
      "grad_norm": 0.14225976169109344,
      "learning_rate": 6.991425849476025e-05,
      "loss": 0.05,
      "step": 4742
    },
    {
      "epoch": 9.021398002853067,
      "grad_norm": 0.1438269317150116,
      "learning_rate": 6.99079072721499e-05,
      "loss": 0.0581,
      "step": 4743
    },
    {
      "epoch": 9.023300047551118,
      "grad_norm": 0.2879999279975891,
      "learning_rate": 6.990155604953954e-05,
      "loss": 0.078,
      "step": 4744
    },
    {
      "epoch": 9.025202092249168,
      "grad_norm": 0.25889408588409424,
      "learning_rate": 6.989520482692919e-05,
      "loss": 0.0631,
      "step": 4745
    },
    {
      "epoch": 9.027104136947218,
      "grad_norm": 0.2674861252307892,
      "learning_rate": 6.988885360431884e-05,
      "loss": 0.0626,
      "step": 4746
    },
    {
      "epoch": 9.029006181645268,
      "grad_norm": 0.2341015636920929,
      "learning_rate": 6.988250238170848e-05,
      "loss": 0.072,
      "step": 4747
    },
    {
      "epoch": 9.03090822634332,
      "grad_norm": 0.2379583716392517,
      "learning_rate": 6.987615115909813e-05,
      "loss": 0.059,
      "step": 4748
    },
    {
      "epoch": 9.03281027104137,
      "grad_norm": 0.24688489735126495,
      "learning_rate": 6.986979993648778e-05,
      "loss": 0.0767,
      "step": 4749
    },
    {
      "epoch": 9.03471231573942,
      "grad_norm": 0.33658576011657715,
      "learning_rate": 6.986344871387742e-05,
      "loss": 0.0893,
      "step": 4750
    },
    {
      "epoch": 9.03661436043747,
      "grad_norm": 0.28995007276535034,
      "learning_rate": 6.985709749126707e-05,
      "loss": 0.0602,
      "step": 4751
    },
    {
      "epoch": 9.038516405135521,
      "grad_norm": 0.23219893872737885,
      "learning_rate": 6.985074626865672e-05,
      "loss": 0.0603,
      "step": 4752
    },
    {
      "epoch": 9.04041844983357,
      "grad_norm": 0.23928780853748322,
      "learning_rate": 6.984439504604636e-05,
      "loss": 0.0714,
      "step": 4753
    },
    {
      "epoch": 9.042320494531621,
      "grad_norm": 0.3113102912902832,
      "learning_rate": 6.983804382343601e-05,
      "loss": 0.0856,
      "step": 4754
    },
    {
      "epoch": 9.044222539229672,
      "grad_norm": 0.2090459018945694,
      "learning_rate": 6.983169260082567e-05,
      "loss": 0.0649,
      "step": 4755
    },
    {
      "epoch": 9.046124583927723,
      "grad_norm": 0.31228578090667725,
      "learning_rate": 6.982534137821532e-05,
      "loss": 0.078,
      "step": 4756
    },
    {
      "epoch": 9.048026628625772,
      "grad_norm": 0.26794952154159546,
      "learning_rate": 6.981899015560496e-05,
      "loss": 0.1032,
      "step": 4757
    },
    {
      "epoch": 9.049928673323823,
      "grad_norm": 0.20607098937034607,
      "learning_rate": 6.98126389329946e-05,
      "loss": 0.0652,
      "step": 4758
    },
    {
      "epoch": 9.051830718021874,
      "grad_norm": 0.23113508522510529,
      "learning_rate": 6.980628771038426e-05,
      "loss": 0.0701,
      "step": 4759
    },
    {
      "epoch": 9.053732762719925,
      "grad_norm": 0.35468852519989014,
      "learning_rate": 6.97999364877739e-05,
      "loss": 0.117,
      "step": 4760
    },
    {
      "epoch": 9.055634807417974,
      "grad_norm": 0.271899551153183,
      "learning_rate": 6.979358526516355e-05,
      "loss": 0.0856,
      "step": 4761
    },
    {
      "epoch": 9.057536852116025,
      "grad_norm": 0.19537365436553955,
      "learning_rate": 6.97872340425532e-05,
      "loss": 0.0458,
      "step": 4762
    },
    {
      "epoch": 9.059438896814076,
      "grad_norm": 0.27871695160865784,
      "learning_rate": 6.978088281994284e-05,
      "loss": 0.0541,
      "step": 4763
    },
    {
      "epoch": 9.061340941512125,
      "grad_norm": 0.30046334862709045,
      "learning_rate": 6.977453159733249e-05,
      "loss": 0.0952,
      "step": 4764
    },
    {
      "epoch": 9.063242986210176,
      "grad_norm": 0.2028159648180008,
      "learning_rate": 6.976818037472213e-05,
      "loss": 0.0916,
      "step": 4765
    },
    {
      "epoch": 9.065145030908226,
      "grad_norm": 0.24774155020713806,
      "learning_rate": 6.976182915211178e-05,
      "loss": 0.054,
      "step": 4766
    },
    {
      "epoch": 9.067047075606277,
      "grad_norm": 0.16707736253738403,
      "learning_rate": 6.975547792950143e-05,
      "loss": 0.0524,
      "step": 4767
    },
    {
      "epoch": 9.068949120304326,
      "grad_norm": 0.16940726339817047,
      "learning_rate": 6.974912670689107e-05,
      "loss": 0.0475,
      "step": 4768
    },
    {
      "epoch": 9.070851165002377,
      "grad_norm": 0.25898852944374084,
      "learning_rate": 6.974277548428074e-05,
      "loss": 0.0697,
      "step": 4769
    },
    {
      "epoch": 9.072753209700428,
      "grad_norm": 0.27324533462524414,
      "learning_rate": 6.973642426167038e-05,
      "loss": 0.0858,
      "step": 4770
    },
    {
      "epoch": 9.074655254398479,
      "grad_norm": 0.20153963565826416,
      "learning_rate": 6.973007303906001e-05,
      "loss": 0.0829,
      "step": 4771
    },
    {
      "epoch": 9.076557299096528,
      "grad_norm": 0.2274760901927948,
      "learning_rate": 6.972372181644967e-05,
      "loss": 0.0609,
      "step": 4772
    },
    {
      "epoch": 9.078459343794579,
      "grad_norm": 0.2810436189174652,
      "learning_rate": 6.971737059383932e-05,
      "loss": 0.074,
      "step": 4773
    },
    {
      "epoch": 9.08036138849263,
      "grad_norm": 0.24662359058856964,
      "learning_rate": 6.971101937122897e-05,
      "loss": 0.0644,
      "step": 4774
    },
    {
      "epoch": 9.08226343319068,
      "grad_norm": 0.17531703412532806,
      "learning_rate": 6.970466814861861e-05,
      "loss": 0.0588,
      "step": 4775
    },
    {
      "epoch": 9.08416547788873,
      "grad_norm": 0.3015911877155304,
      "learning_rate": 6.969831692600826e-05,
      "loss": 0.0759,
      "step": 4776
    },
    {
      "epoch": 9.08606752258678,
      "grad_norm": 0.31380707025527954,
      "learning_rate": 6.969196570339791e-05,
      "loss": 0.0656,
      "step": 4777
    },
    {
      "epoch": 9.087969567284832,
      "grad_norm": 0.1681576669216156,
      "learning_rate": 6.968561448078755e-05,
      "loss": 0.0623,
      "step": 4778
    },
    {
      "epoch": 9.08987161198288,
      "grad_norm": 0.23140835762023926,
      "learning_rate": 6.96792632581772e-05,
      "loss": 0.0744,
      "step": 4779
    },
    {
      "epoch": 9.091773656680932,
      "grad_norm": 0.1460658758878708,
      "learning_rate": 6.967291203556685e-05,
      "loss": 0.054,
      "step": 4780
    },
    {
      "epoch": 9.093675701378983,
      "grad_norm": 0.1995638608932495,
      "learning_rate": 6.966656081295649e-05,
      "loss": 0.0617,
      "step": 4781
    },
    {
      "epoch": 9.095577746077034,
      "grad_norm": 0.2675533890724182,
      "learning_rate": 6.966020959034614e-05,
      "loss": 0.0667,
      "step": 4782
    },
    {
      "epoch": 9.097479790775083,
      "grad_norm": 0.3835360109806061,
      "learning_rate": 6.96538583677358e-05,
      "loss": 0.0763,
      "step": 4783
    },
    {
      "epoch": 9.099381835473134,
      "grad_norm": 0.28690552711486816,
      "learning_rate": 6.964750714512543e-05,
      "loss": 0.0822,
      "step": 4784
    },
    {
      "epoch": 9.101283880171184,
      "grad_norm": 0.18676474690437317,
      "learning_rate": 6.964115592251509e-05,
      "loss": 0.0521,
      "step": 4785
    },
    {
      "epoch": 9.103185924869235,
      "grad_norm": 0.28001606464385986,
      "learning_rate": 6.963480469990474e-05,
      "loss": 0.0596,
      "step": 4786
    },
    {
      "epoch": 9.105087969567284,
      "grad_norm": 0.34591495990753174,
      "learning_rate": 6.962845347729439e-05,
      "loss": 0.0739,
      "step": 4787
    },
    {
      "epoch": 9.106990014265335,
      "grad_norm": 0.2598930299282074,
      "learning_rate": 6.962210225468403e-05,
      "loss": 0.0648,
      "step": 4788
    },
    {
      "epoch": 9.108892058963386,
      "grad_norm": 0.17923042178153992,
      "learning_rate": 6.961575103207367e-05,
      "loss": 0.0624,
      "step": 4789
    },
    {
      "epoch": 9.110794103661435,
      "grad_norm": 0.30225229263305664,
      "learning_rate": 6.960939980946333e-05,
      "loss": 0.0735,
      "step": 4790
    },
    {
      "epoch": 9.112696148359486,
      "grad_norm": 0.20305217802524567,
      "learning_rate": 6.960304858685297e-05,
      "loss": 0.0654,
      "step": 4791
    },
    {
      "epoch": 9.114598193057537,
      "grad_norm": 0.24463124573230743,
      "learning_rate": 6.959669736424262e-05,
      "loss": 0.0885,
      "step": 4792
    },
    {
      "epoch": 9.116500237755588,
      "grad_norm": 0.2814115881919861,
      "learning_rate": 6.959034614163227e-05,
      "loss": 0.0716,
      "step": 4793
    },
    {
      "epoch": 9.118402282453637,
      "grad_norm": 0.1762290596961975,
      "learning_rate": 6.958399491902191e-05,
      "loss": 0.0552,
      "step": 4794
    },
    {
      "epoch": 9.120304327151688,
      "grad_norm": 0.30058228969573975,
      "learning_rate": 6.957764369641156e-05,
      "loss": 0.0728,
      "step": 4795
    },
    {
      "epoch": 9.122206371849739,
      "grad_norm": 0.25093328952789307,
      "learning_rate": 6.95712924738012e-05,
      "loss": 0.0598,
      "step": 4796
    },
    {
      "epoch": 9.12410841654779,
      "grad_norm": 0.2654763460159302,
      "learning_rate": 6.956494125119087e-05,
      "loss": 0.0685,
      "step": 4797
    },
    {
      "epoch": 9.126010461245839,
      "grad_norm": 0.22402413189411163,
      "learning_rate": 6.95585900285805e-05,
      "loss": 0.0561,
      "step": 4798
    },
    {
      "epoch": 9.12791250594389,
      "grad_norm": 0.22796908020973206,
      "learning_rate": 6.955223880597014e-05,
      "loss": 0.0689,
      "step": 4799
    },
    {
      "epoch": 9.12981455064194,
      "grad_norm": 0.26183420419692993,
      "learning_rate": 6.954588758335981e-05,
      "loss": 0.0693,
      "step": 4800
    },
    {
      "epoch": 9.13171659533999,
      "grad_norm": 0.21732604503631592,
      "learning_rate": 6.953953636074945e-05,
      "loss": 0.0572,
      "step": 4801
    },
    {
      "epoch": 9.13361864003804,
      "grad_norm": 0.244783416390419,
      "learning_rate": 6.953318513813909e-05,
      "loss": 0.0714,
      "step": 4802
    },
    {
      "epoch": 9.135520684736091,
      "grad_norm": 0.25078529119491577,
      "learning_rate": 6.952683391552874e-05,
      "loss": 0.0651,
      "step": 4803
    },
    {
      "epoch": 9.137422729434142,
      "grad_norm": 0.2012512981891632,
      "learning_rate": 6.952048269291839e-05,
      "loss": 0.0621,
      "step": 4804
    },
    {
      "epoch": 9.139324774132191,
      "grad_norm": 0.26933446526527405,
      "learning_rate": 6.951413147030804e-05,
      "loss": 0.0636,
      "step": 4805
    },
    {
      "epoch": 9.141226818830242,
      "grad_norm": 0.18506479263305664,
      "learning_rate": 6.950778024769768e-05,
      "loss": 0.0628,
      "step": 4806
    },
    {
      "epoch": 9.143128863528293,
      "grad_norm": 0.2639950215816498,
      "learning_rate": 6.950142902508733e-05,
      "loss": 0.0595,
      "step": 4807
    },
    {
      "epoch": 9.145030908226344,
      "grad_norm": 0.2577735483646393,
      "learning_rate": 6.949507780247698e-05,
      "loss": 0.0602,
      "step": 4808
    },
    {
      "epoch": 9.146932952924393,
      "grad_norm": 0.20550990104675293,
      "learning_rate": 6.948872657986662e-05,
      "loss": 0.0731,
      "step": 4809
    },
    {
      "epoch": 9.148834997622444,
      "grad_norm": 0.18981294333934784,
      "learning_rate": 6.948237535725627e-05,
      "loss": 0.0567,
      "step": 4810
    },
    {
      "epoch": 9.150737042320495,
      "grad_norm": 0.2644405961036682,
      "learning_rate": 6.947602413464593e-05,
      "loss": 0.0631,
      "step": 4811
    },
    {
      "epoch": 9.152639087018544,
      "grad_norm": 0.17926043272018433,
      "learning_rate": 6.946967291203556e-05,
      "loss": 0.0567,
      "step": 4812
    },
    {
      "epoch": 9.154541131716595,
      "grad_norm": 0.2050900012254715,
      "learning_rate": 6.946332168942522e-05,
      "loss": 0.0968,
      "step": 4813
    },
    {
      "epoch": 9.156443176414646,
      "grad_norm": 0.29818764328956604,
      "learning_rate": 6.945697046681487e-05,
      "loss": 0.0643,
      "step": 4814
    },
    {
      "epoch": 9.158345221112697,
      "grad_norm": 0.24317260086536407,
      "learning_rate": 6.945061924420452e-05,
      "loss": 0.0799,
      "step": 4815
    },
    {
      "epoch": 9.160247265810746,
      "grad_norm": 0.16913026571273804,
      "learning_rate": 6.944426802159416e-05,
      "loss": 0.0799,
      "step": 4816
    },
    {
      "epoch": 9.162149310508797,
      "grad_norm": 0.14065365493297577,
      "learning_rate": 6.943791679898381e-05,
      "loss": 0.0666,
      "step": 4817
    },
    {
      "epoch": 9.164051355206848,
      "grad_norm": 0.16402816772460938,
      "learning_rate": 6.943156557637346e-05,
      "loss": 0.0589,
      "step": 4818
    },
    {
      "epoch": 9.165953399904899,
      "grad_norm": 0.2175387740135193,
      "learning_rate": 6.94252143537631e-05,
      "loss": 0.0647,
      "step": 4819
    },
    {
      "epoch": 9.167855444602948,
      "grad_norm": 0.3297087848186493,
      "learning_rate": 6.941886313115275e-05,
      "loss": 0.0875,
      "step": 4820
    },
    {
      "epoch": 9.169757489300999,
      "grad_norm": 0.14931192994117737,
      "learning_rate": 6.94125119085424e-05,
      "loss": 0.058,
      "step": 4821
    },
    {
      "epoch": 9.17165953399905,
      "grad_norm": 0.21912738680839539,
      "learning_rate": 6.940616068593204e-05,
      "loss": 0.0648,
      "step": 4822
    },
    {
      "epoch": 9.173561578697099,
      "grad_norm": 0.27302685379981995,
      "learning_rate": 6.939980946332169e-05,
      "loss": 0.0711,
      "step": 4823
    },
    {
      "epoch": 9.17546362339515,
      "grad_norm": 0.23038659989833832,
      "learning_rate": 6.939345824071134e-05,
      "loss": 0.0741,
      "step": 4824
    },
    {
      "epoch": 9.1773656680932,
      "grad_norm": 0.24258142709732056,
      "learning_rate": 6.938710701810098e-05,
      "loss": 0.0857,
      "step": 4825
    },
    {
      "epoch": 9.179267712791251,
      "grad_norm": 0.2215842455625534,
      "learning_rate": 6.938075579549064e-05,
      "loss": 0.0856,
      "step": 4826
    },
    {
      "epoch": 9.1811697574893,
      "grad_norm": 0.2076491415500641,
      "learning_rate": 6.937440457288029e-05,
      "loss": 0.0711,
      "step": 4827
    },
    {
      "epoch": 9.183071802187351,
      "grad_norm": 0.4951212704181671,
      "learning_rate": 6.936805335026994e-05,
      "loss": 0.1213,
      "step": 4828
    },
    {
      "epoch": 9.184973846885402,
      "grad_norm": 0.1944156438112259,
      "learning_rate": 6.936170212765958e-05,
      "loss": 0.0682,
      "step": 4829
    },
    {
      "epoch": 9.186875891583453,
      "grad_norm": 0.17844007909297943,
      "learning_rate": 6.935535090504922e-05,
      "loss": 0.0624,
      "step": 4830
    },
    {
      "epoch": 9.188777936281502,
      "grad_norm": 0.47420743107795715,
      "learning_rate": 6.934899968243888e-05,
      "loss": 0.0875,
      "step": 4831
    },
    {
      "epoch": 9.190679980979553,
      "grad_norm": 0.2568189799785614,
      "learning_rate": 6.934264845982852e-05,
      "loss": 0.0715,
      "step": 4832
    },
    {
      "epoch": 9.192582025677604,
      "grad_norm": 0.16426631808280945,
      "learning_rate": 6.933629723721817e-05,
      "loss": 0.0776,
      "step": 4833
    },
    {
      "epoch": 9.194484070375653,
      "grad_norm": 0.2905197739601135,
      "learning_rate": 6.932994601460782e-05,
      "loss": 0.0875,
      "step": 4834
    },
    {
      "epoch": 9.196386115073704,
      "grad_norm": 0.1947975754737854,
      "learning_rate": 6.932359479199746e-05,
      "loss": 0.0528,
      "step": 4835
    },
    {
      "epoch": 9.198288159771755,
      "grad_norm": 0.24473321437835693,
      "learning_rate": 6.931724356938711e-05,
      "loss": 0.047,
      "step": 4836
    },
    {
      "epoch": 9.200190204469806,
      "grad_norm": 0.2849128544330597,
      "learning_rate": 6.931089234677675e-05,
      "loss": 0.0708,
      "step": 4837
    },
    {
      "epoch": 9.202092249167855,
      "grad_norm": 0.3052026033401489,
      "learning_rate": 6.93045411241664e-05,
      "loss": 0.0761,
      "step": 4838
    },
    {
      "epoch": 9.203994293865906,
      "grad_norm": 0.1367628425359726,
      "learning_rate": 6.929818990155605e-05,
      "loss": 0.0465,
      "step": 4839
    },
    {
      "epoch": 9.205896338563957,
      "grad_norm": 0.2635146975517273,
      "learning_rate": 6.929183867894569e-05,
      "loss": 0.0715,
      "step": 4840
    },
    {
      "epoch": 9.207798383262007,
      "grad_norm": 0.21499952673912048,
      "learning_rate": 6.928548745633536e-05,
      "loss": 0.0593,
      "step": 4841
    },
    {
      "epoch": 9.209700427960057,
      "grad_norm": 0.20185910165309906,
      "learning_rate": 6.9279136233725e-05,
      "loss": 0.0777,
      "step": 4842
    },
    {
      "epoch": 9.211602472658107,
      "grad_norm": 0.20574524998664856,
      "learning_rate": 6.927278501111464e-05,
      "loss": 0.0682,
      "step": 4843
    },
    {
      "epoch": 9.213504517356158,
      "grad_norm": 0.30292728543281555,
      "learning_rate": 6.926643378850429e-05,
      "loss": 0.0913,
      "step": 4844
    },
    {
      "epoch": 9.21540656205421,
      "grad_norm": 0.21002617478370667,
      "learning_rate": 6.926008256589394e-05,
      "loss": 0.0655,
      "step": 4845
    },
    {
      "epoch": 9.217308606752258,
      "grad_norm": 0.30659377574920654,
      "learning_rate": 6.925373134328359e-05,
      "loss": 0.0746,
      "step": 4846
    },
    {
      "epoch": 9.21921065145031,
      "grad_norm": 0.3024582266807556,
      "learning_rate": 6.924738012067323e-05,
      "loss": 0.0754,
      "step": 4847
    },
    {
      "epoch": 9.22111269614836,
      "grad_norm": 0.20387329161167145,
      "learning_rate": 6.924102889806288e-05,
      "loss": 0.0538,
      "step": 4848
    },
    {
      "epoch": 9.22301474084641,
      "grad_norm": 0.19420789182186127,
      "learning_rate": 6.923467767545253e-05,
      "loss": 0.0668,
      "step": 4849
    },
    {
      "epoch": 9.22491678554446,
      "grad_norm": 0.18014901876449585,
      "learning_rate": 6.922832645284217e-05,
      "loss": 0.0643,
      "step": 4850
    },
    {
      "epoch": 9.226818830242511,
      "grad_norm": 0.27916622161865234,
      "learning_rate": 6.922197523023182e-05,
      "loss": 0.097,
      "step": 4851
    },
    {
      "epoch": 9.228720874940562,
      "grad_norm": 0.38052618503570557,
      "learning_rate": 6.921562400762147e-05,
      "loss": 0.0773,
      "step": 4852
    },
    {
      "epoch": 9.230622919638611,
      "grad_norm": 0.2899182438850403,
      "learning_rate": 6.920927278501111e-05,
      "loss": 0.0739,
      "step": 4853
    },
    {
      "epoch": 9.232524964336662,
      "grad_norm": 0.16164492070674896,
      "learning_rate": 6.920292156240076e-05,
      "loss": 0.0683,
      "step": 4854
    },
    {
      "epoch": 9.234427009034713,
      "grad_norm": 0.38230976462364197,
      "learning_rate": 6.919657033979042e-05,
      "loss": 0.0665,
      "step": 4855
    },
    {
      "epoch": 9.236329053732764,
      "grad_norm": 0.21252687275409698,
      "learning_rate": 6.919021911718005e-05,
      "loss": 0.0693,
      "step": 4856
    },
    {
      "epoch": 9.238231098430813,
      "grad_norm": 0.31365564465522766,
      "learning_rate": 6.91838678945697e-05,
      "loss": 0.0898,
      "step": 4857
    },
    {
      "epoch": 9.240133143128864,
      "grad_norm": 0.25702884793281555,
      "learning_rate": 6.917751667195936e-05,
      "loss": 0.0934,
      "step": 4858
    },
    {
      "epoch": 9.242035187826914,
      "grad_norm": 0.17167243361473083,
      "learning_rate": 6.917116544934901e-05,
      "loss": 0.0664,
      "step": 4859
    },
    {
      "epoch": 9.243937232524964,
      "grad_norm": 0.2831684648990631,
      "learning_rate": 6.916481422673865e-05,
      "loss": 0.0836,
      "step": 4860
    },
    {
      "epoch": 9.245839277223014,
      "grad_norm": 0.32651644945144653,
      "learning_rate": 6.915846300412829e-05,
      "loss": 0.0855,
      "step": 4861
    },
    {
      "epoch": 9.247741321921065,
      "grad_norm": 0.5113350749015808,
      "learning_rate": 6.915211178151795e-05,
      "loss": 0.097,
      "step": 4862
    },
    {
      "epoch": 9.249643366619116,
      "grad_norm": 0.23918244242668152,
      "learning_rate": 6.914576055890759e-05,
      "loss": 0.0676,
      "step": 4863
    },
    {
      "epoch": 9.251545411317165,
      "grad_norm": 0.24368615448474884,
      "learning_rate": 6.913940933629724e-05,
      "loss": 0.0679,
      "step": 4864
    },
    {
      "epoch": 9.253447456015216,
      "grad_norm": 0.19113053381443024,
      "learning_rate": 6.91330581136869e-05,
      "loss": 0.0661,
      "step": 4865
    },
    {
      "epoch": 9.255349500713267,
      "grad_norm": 0.3215082585811615,
      "learning_rate": 6.912670689107653e-05,
      "loss": 0.0782,
      "step": 4866
    },
    {
      "epoch": 9.257251545411318,
      "grad_norm": 0.18558810651302338,
      "learning_rate": 6.912035566846618e-05,
      "loss": 0.0771,
      "step": 4867
    },
    {
      "epoch": 9.259153590109367,
      "grad_norm": 0.25669926404953003,
      "learning_rate": 6.911400444585582e-05,
      "loss": 0.0877,
      "step": 4868
    },
    {
      "epoch": 9.261055634807418,
      "grad_norm": 0.37997862696647644,
      "learning_rate": 6.910765322324549e-05,
      "loss": 0.0919,
      "step": 4869
    },
    {
      "epoch": 9.262957679505469,
      "grad_norm": 0.27809038758277893,
      "learning_rate": 6.910130200063513e-05,
      "loss": 0.0598,
      "step": 4870
    },
    {
      "epoch": 9.264859724203518,
      "grad_norm": 0.1689881980419159,
      "learning_rate": 6.909495077802476e-05,
      "loss": 0.0479,
      "step": 4871
    },
    {
      "epoch": 9.266761768901569,
      "grad_norm": 0.14065071940422058,
      "learning_rate": 6.908859955541443e-05,
      "loss": 0.0515,
      "step": 4872
    },
    {
      "epoch": 9.26866381359962,
      "grad_norm": 0.3398038148880005,
      "learning_rate": 6.908224833280407e-05,
      "loss": 0.1203,
      "step": 4873
    },
    {
      "epoch": 9.27056585829767,
      "grad_norm": 0.21026046574115753,
      "learning_rate": 6.90758971101937e-05,
      "loss": 0.0633,
      "step": 4874
    },
    {
      "epoch": 9.27246790299572,
      "grad_norm": 0.16988354921340942,
      "learning_rate": 6.906954588758336e-05,
      "loss": 0.0663,
      "step": 4875
    },
    {
      "epoch": 9.27436994769377,
      "grad_norm": 0.18398746848106384,
      "learning_rate": 6.906319466497301e-05,
      "loss": 0.0744,
      "step": 4876
    },
    {
      "epoch": 9.276271992391822,
      "grad_norm": 0.20699293911457062,
      "learning_rate": 6.905684344236266e-05,
      "loss": 0.0801,
      "step": 4877
    },
    {
      "epoch": 9.278174037089872,
      "grad_norm": 0.2684108316898346,
      "learning_rate": 6.90504922197523e-05,
      "loss": 0.0823,
      "step": 4878
    },
    {
      "epoch": 9.280076081787922,
      "grad_norm": 0.3421522378921509,
      "learning_rate": 6.904414099714195e-05,
      "loss": 0.0975,
      "step": 4879
    },
    {
      "epoch": 9.281978126485972,
      "grad_norm": 0.216671422123909,
      "learning_rate": 6.90377897745316e-05,
      "loss": 0.0657,
      "step": 4880
    },
    {
      "epoch": 9.283880171184023,
      "grad_norm": 0.23740790784358978,
      "learning_rate": 6.903143855192124e-05,
      "loss": 0.0921,
      "step": 4881
    },
    {
      "epoch": 9.285782215882072,
      "grad_norm": 0.2520216107368469,
      "learning_rate": 6.90250873293109e-05,
      "loss": 0.0746,
      "step": 4882
    },
    {
      "epoch": 9.287684260580123,
      "grad_norm": 0.22703856229782104,
      "learning_rate": 6.901873610670055e-05,
      "loss": 0.0669,
      "step": 4883
    },
    {
      "epoch": 9.289586305278174,
      "grad_norm": 0.20447954535484314,
      "learning_rate": 6.901238488409018e-05,
      "loss": 0.0625,
      "step": 4884
    },
    {
      "epoch": 9.291488349976225,
      "grad_norm": 0.2351393699645996,
      "learning_rate": 6.900603366147984e-05,
      "loss": 0.0561,
      "step": 4885
    },
    {
      "epoch": 9.293390394674274,
      "grad_norm": 0.19731055200099945,
      "learning_rate": 6.899968243886949e-05,
      "loss": 0.0675,
      "step": 4886
    },
    {
      "epoch": 9.295292439372325,
      "grad_norm": 0.29714658856391907,
      "learning_rate": 6.899333121625914e-05,
      "loss": 0.0762,
      "step": 4887
    },
    {
      "epoch": 9.297194484070376,
      "grad_norm": 0.37264585494995117,
      "learning_rate": 6.898697999364878e-05,
      "loss": 0.0869,
      "step": 4888
    },
    {
      "epoch": 9.299096528768427,
      "grad_norm": 0.16094455122947693,
      "learning_rate": 6.898062877103843e-05,
      "loss": 0.0754,
      "step": 4889
    },
    {
      "epoch": 9.300998573466476,
      "grad_norm": 0.3849022686481476,
      "learning_rate": 6.897427754842808e-05,
      "loss": 0.1015,
      "step": 4890
    },
    {
      "epoch": 9.302900618164527,
      "grad_norm": 0.2579275667667389,
      "learning_rate": 6.896792632581772e-05,
      "loss": 0.0725,
      "step": 4891
    },
    {
      "epoch": 9.304802662862578,
      "grad_norm": 0.2276908904314041,
      "learning_rate": 6.896157510320737e-05,
      "loss": 0.0586,
      "step": 4892
    },
    {
      "epoch": 9.306704707560627,
      "grad_norm": 0.2620905637741089,
      "learning_rate": 6.895522388059702e-05,
      "loss": 0.0504,
      "step": 4893
    },
    {
      "epoch": 9.308606752258678,
      "grad_norm": 0.21400023996829987,
      "learning_rate": 6.894887265798666e-05,
      "loss": 0.0581,
      "step": 4894
    },
    {
      "epoch": 9.310508796956729,
      "grad_norm": 0.3381659984588623,
      "learning_rate": 6.894252143537631e-05,
      "loss": 0.0745,
      "step": 4895
    },
    {
      "epoch": 9.31241084165478,
      "grad_norm": 0.3350960910320282,
      "learning_rate": 6.893617021276597e-05,
      "loss": 0.0796,
      "step": 4896
    },
    {
      "epoch": 9.314312886352829,
      "grad_norm": 0.14392538368701935,
      "learning_rate": 6.89298189901556e-05,
      "loss": 0.0615,
      "step": 4897
    },
    {
      "epoch": 9.31621493105088,
      "grad_norm": 0.19645485281944275,
      "learning_rate": 6.892346776754526e-05,
      "loss": 0.0698,
      "step": 4898
    },
    {
      "epoch": 9.31811697574893,
      "grad_norm": 0.2611951231956482,
      "learning_rate": 6.89171165449349e-05,
      "loss": 0.0731,
      "step": 4899
    },
    {
      "epoch": 9.320019020446981,
      "grad_norm": 0.20194761455059052,
      "learning_rate": 6.891076532232456e-05,
      "loss": 0.0649,
      "step": 4900
    },
    {
      "epoch": 9.32192106514503,
      "grad_norm": 0.17792201042175293,
      "learning_rate": 6.89044140997142e-05,
      "loss": 0.0809,
      "step": 4901
    },
    {
      "epoch": 9.323823109843081,
      "grad_norm": 0.26867902278900146,
      "learning_rate": 6.889806287710384e-05,
      "loss": 0.0715,
      "step": 4902
    },
    {
      "epoch": 9.325725154541132,
      "grad_norm": 0.2845976948738098,
      "learning_rate": 6.88917116544935e-05,
      "loss": 0.0738,
      "step": 4903
    },
    {
      "epoch": 9.327627199239181,
      "grad_norm": 0.39770546555519104,
      "learning_rate": 6.888536043188314e-05,
      "loss": 0.0905,
      "step": 4904
    },
    {
      "epoch": 9.329529243937232,
      "grad_norm": 0.27722543478012085,
      "learning_rate": 6.887900920927279e-05,
      "loss": 0.063,
      "step": 4905
    },
    {
      "epoch": 9.331431288635283,
      "grad_norm": 0.23620283603668213,
      "learning_rate": 6.887265798666243e-05,
      "loss": 0.0701,
      "step": 4906
    },
    {
      "epoch": 9.333333333333334,
      "grad_norm": 0.1973799765110016,
      "learning_rate": 6.886630676405208e-05,
      "loss": 0.0704,
      "step": 4907
    },
    {
      "epoch": 9.335235378031383,
      "grad_norm": 0.29582053422927856,
      "learning_rate": 6.885995554144173e-05,
      "loss": 0.0702,
      "step": 4908
    },
    {
      "epoch": 9.337137422729434,
      "grad_norm": 0.32822665572166443,
      "learning_rate": 6.885360431883137e-05,
      "loss": 0.0773,
      "step": 4909
    },
    {
      "epoch": 9.339039467427485,
      "grad_norm": 0.2938401401042938,
      "learning_rate": 6.884725309622102e-05,
      "loss": 0.0686,
      "step": 4910
    },
    {
      "epoch": 9.340941512125536,
      "grad_norm": 0.23710989952087402,
      "learning_rate": 6.884090187361068e-05,
      "loss": 0.071,
      "step": 4911
    },
    {
      "epoch": 9.342843556823585,
      "grad_norm": 0.37398549914360046,
      "learning_rate": 6.883455065100031e-05,
      "loss": 0.088,
      "step": 4912
    },
    {
      "epoch": 9.344745601521636,
      "grad_norm": 0.3170340657234192,
      "learning_rate": 6.882819942838997e-05,
      "loss": 0.0736,
      "step": 4913
    },
    {
      "epoch": 9.346647646219687,
      "grad_norm": 0.21240757405757904,
      "learning_rate": 6.882184820577962e-05,
      "loss": 0.0569,
      "step": 4914
    },
    {
      "epoch": 9.348549690917736,
      "grad_norm": 0.2711277902126312,
      "learning_rate": 6.881549698316926e-05,
      "loss": 0.0644,
      "step": 4915
    },
    {
      "epoch": 9.350451735615787,
      "grad_norm": 0.3614808917045593,
      "learning_rate": 6.880914576055891e-05,
      "loss": 0.0888,
      "step": 4916
    },
    {
      "epoch": 9.352353780313837,
      "grad_norm": 0.2747959792613983,
      "learning_rate": 6.880279453794856e-05,
      "loss": 0.0933,
      "step": 4917
    },
    {
      "epoch": 9.354255825011888,
      "grad_norm": 0.20947587490081787,
      "learning_rate": 6.879644331533821e-05,
      "loss": 0.0658,
      "step": 4918
    },
    {
      "epoch": 9.356157869709937,
      "grad_norm": 0.19765761494636536,
      "learning_rate": 6.879009209272785e-05,
      "loss": 0.0739,
      "step": 4919
    },
    {
      "epoch": 9.358059914407988,
      "grad_norm": 0.22963549196720123,
      "learning_rate": 6.87837408701175e-05,
      "loss": 0.0485,
      "step": 4920
    },
    {
      "epoch": 9.35996195910604,
      "grad_norm": 0.22210000455379486,
      "learning_rate": 6.877738964750715e-05,
      "loss": 0.0731,
      "step": 4921
    },
    {
      "epoch": 9.36186400380409,
      "grad_norm": 0.22614774107933044,
      "learning_rate": 6.877103842489679e-05,
      "loss": 0.1153,
      "step": 4922
    },
    {
      "epoch": 9.36376604850214,
      "grad_norm": 0.20579391717910767,
      "learning_rate": 6.876468720228644e-05,
      "loss": 0.0631,
      "step": 4923
    },
    {
      "epoch": 9.36566809320019,
      "grad_norm": 0.25111064314842224,
      "learning_rate": 6.87583359796761e-05,
      "loss": 0.0606,
      "step": 4924
    },
    {
      "epoch": 9.367570137898241,
      "grad_norm": 0.20417334139347076,
      "learning_rate": 6.875198475706573e-05,
      "loss": 0.0904,
      "step": 4925
    },
    {
      "epoch": 9.36947218259629,
      "grad_norm": 0.21653708815574646,
      "learning_rate": 6.874563353445539e-05,
      "loss": 0.0585,
      "step": 4926
    },
    {
      "epoch": 9.371374227294341,
      "grad_norm": 0.28104594349861145,
      "learning_rate": 6.873928231184504e-05,
      "loss": 0.064,
      "step": 4927
    },
    {
      "epoch": 9.373276271992392,
      "grad_norm": 0.275823712348938,
      "learning_rate": 6.873293108923468e-05,
      "loss": 0.0749,
      "step": 4928
    },
    {
      "epoch": 9.375178316690443,
      "grad_norm": 0.16498956084251404,
      "learning_rate": 6.872657986662433e-05,
      "loss": 0.0601,
      "step": 4929
    },
    {
      "epoch": 9.377080361388492,
      "grad_norm": 0.278196781873703,
      "learning_rate": 6.872022864401398e-05,
      "loss": 0.0765,
      "step": 4930
    },
    {
      "epoch": 9.378982406086543,
      "grad_norm": 0.2505214512348175,
      "learning_rate": 6.871387742140363e-05,
      "loss": 0.0612,
      "step": 4931
    },
    {
      "epoch": 9.380884450784594,
      "grad_norm": 0.26770439743995667,
      "learning_rate": 6.870752619879327e-05,
      "loss": 0.0572,
      "step": 4932
    },
    {
      "epoch": 9.382786495482645,
      "grad_norm": 0.24536767601966858,
      "learning_rate": 6.870117497618291e-05,
      "loss": 0.0579,
      "step": 4933
    },
    {
      "epoch": 9.384688540180694,
      "grad_norm": 0.26408347487449646,
      "learning_rate": 6.869482375357257e-05,
      "loss": 0.0767,
      "step": 4934
    },
    {
      "epoch": 9.386590584878745,
      "grad_norm": 0.2711765766143799,
      "learning_rate": 6.868847253096221e-05,
      "loss": 0.0797,
      "step": 4935
    },
    {
      "epoch": 9.388492629576795,
      "grad_norm": 0.2707313299179077,
      "learning_rate": 6.868212130835186e-05,
      "loss": 0.0687,
      "step": 4936
    },
    {
      "epoch": 9.390394674274846,
      "grad_norm": 0.21455048024654388,
      "learning_rate": 6.867577008574152e-05,
      "loss": 0.0737,
      "step": 4937
    },
    {
      "epoch": 9.392296718972895,
      "grad_norm": 0.24911226332187653,
      "learning_rate": 6.866941886313115e-05,
      "loss": 0.0589,
      "step": 4938
    },
    {
      "epoch": 9.394198763670946,
      "grad_norm": 0.17448081076145172,
      "learning_rate": 6.86630676405208e-05,
      "loss": 0.0682,
      "step": 4939
    },
    {
      "epoch": 9.396100808368997,
      "grad_norm": 0.3200829327106476,
      "learning_rate": 6.865671641791044e-05,
      "loss": 0.0846,
      "step": 4940
    },
    {
      "epoch": 9.398002853067046,
      "grad_norm": 0.18561430275440216,
      "learning_rate": 6.865036519530011e-05,
      "loss": 0.0685,
      "step": 4941
    },
    {
      "epoch": 9.399904897765097,
      "grad_norm": 0.2448703497648239,
      "learning_rate": 6.864401397268975e-05,
      "loss": 0.0912,
      "step": 4942
    },
    {
      "epoch": 9.401806942463148,
      "grad_norm": 0.19484390318393707,
      "learning_rate": 6.863766275007939e-05,
      "loss": 0.0543,
      "step": 4943
    },
    {
      "epoch": 9.403708987161199,
      "grad_norm": 0.1991858035326004,
      "learning_rate": 6.863131152746905e-05,
      "loss": 0.0716,
      "step": 4944
    },
    {
      "epoch": 9.405611031859248,
      "grad_norm": 0.2221297323703766,
      "learning_rate": 6.862496030485869e-05,
      "loss": 0.0621,
      "step": 4945
    },
    {
      "epoch": 9.407513076557299,
      "grad_norm": 0.26933857798576355,
      "learning_rate": 6.861860908224833e-05,
      "loss": 0.0525,
      "step": 4946
    },
    {
      "epoch": 9.40941512125535,
      "grad_norm": 0.275558203458786,
      "learning_rate": 6.861225785963798e-05,
      "loss": 0.0717,
      "step": 4947
    },
    {
      "epoch": 9.4113171659534,
      "grad_norm": 0.24430836737155914,
      "learning_rate": 6.860590663702763e-05,
      "loss": 0.0624,
      "step": 4948
    },
    {
      "epoch": 9.41321921065145,
      "grad_norm": 0.19468329846858978,
      "learning_rate": 6.859955541441728e-05,
      "loss": 0.0546,
      "step": 4949
    },
    {
      "epoch": 9.4151212553495,
      "grad_norm": 0.1639413833618164,
      "learning_rate": 6.859320419180692e-05,
      "loss": 0.0617,
      "step": 4950
    },
    {
      "epoch": 9.417023300047552,
      "grad_norm": 0.25814270973205566,
      "learning_rate": 6.858685296919657e-05,
      "loss": 0.0654,
      "step": 4951
    },
    {
      "epoch": 9.4189253447456,
      "grad_norm": 0.18765553832054138,
      "learning_rate": 6.858050174658623e-05,
      "loss": 0.0757,
      "step": 4952
    },
    {
      "epoch": 9.420827389443652,
      "grad_norm": 0.17281563580036163,
      "learning_rate": 6.857415052397586e-05,
      "loss": 0.063,
      "step": 4953
    },
    {
      "epoch": 9.422729434141702,
      "grad_norm": 0.1971518099308014,
      "learning_rate": 6.856779930136552e-05,
      "loss": 0.0695,
      "step": 4954
    },
    {
      "epoch": 9.424631478839753,
      "grad_norm": 0.20317022502422333,
      "learning_rate": 6.856144807875517e-05,
      "loss": 0.0531,
      "step": 4955
    },
    {
      "epoch": 9.426533523537802,
      "grad_norm": 0.2848222851753235,
      "learning_rate": 6.85550968561448e-05,
      "loss": 0.067,
      "step": 4956
    },
    {
      "epoch": 9.428435568235853,
      "grad_norm": 0.20473995804786682,
      "learning_rate": 6.854874563353446e-05,
      "loss": 0.0723,
      "step": 4957
    },
    {
      "epoch": 9.430337612933904,
      "grad_norm": 0.20673026144504547,
      "learning_rate": 6.854239441092411e-05,
      "loss": 0.054,
      "step": 4958
    },
    {
      "epoch": 9.432239657631955,
      "grad_norm": 0.20065629482269287,
      "learning_rate": 6.853604318831376e-05,
      "loss": 0.0723,
      "step": 4959
    },
    {
      "epoch": 9.434141702330004,
      "grad_norm": 0.2596694529056549,
      "learning_rate": 6.85296919657034e-05,
      "loss": 0.045,
      "step": 4960
    },
    {
      "epoch": 9.436043747028055,
      "grad_norm": 0.22267058491706848,
      "learning_rate": 6.852334074309305e-05,
      "loss": 0.0486,
      "step": 4961
    },
    {
      "epoch": 9.437945791726106,
      "grad_norm": 0.18863868713378906,
      "learning_rate": 6.85169895204827e-05,
      "loss": 0.0638,
      "step": 4962
    },
    {
      "epoch": 9.439847836424155,
      "grad_norm": 0.23035591840744019,
      "learning_rate": 6.851063829787234e-05,
      "loss": 0.0731,
      "step": 4963
    },
    {
      "epoch": 9.441749881122206,
      "grad_norm": 0.19516980648040771,
      "learning_rate": 6.850428707526198e-05,
      "loss": 0.0589,
      "step": 4964
    },
    {
      "epoch": 9.443651925820257,
      "grad_norm": 0.22958572208881378,
      "learning_rate": 6.849793585265164e-05,
      "loss": 0.0682,
      "step": 4965
    },
    {
      "epoch": 9.445553970518308,
      "grad_norm": 0.33866357803344727,
      "learning_rate": 6.849158463004128e-05,
      "loss": 0.089,
      "step": 4966
    },
    {
      "epoch": 9.447456015216357,
      "grad_norm": 0.27389445900917053,
      "learning_rate": 6.848523340743093e-05,
      "loss": 0.0699,
      "step": 4967
    },
    {
      "epoch": 9.449358059914408,
      "grad_norm": 0.28872737288475037,
      "learning_rate": 6.847888218482059e-05,
      "loss": 0.0646,
      "step": 4968
    },
    {
      "epoch": 9.451260104612459,
      "grad_norm": 0.22543446719646454,
      "learning_rate": 6.847253096221023e-05,
      "loss": 0.0639,
      "step": 4969
    },
    {
      "epoch": 9.45316214931051,
      "grad_norm": 0.23369933664798737,
      "learning_rate": 6.846617973959988e-05,
      "loss": 0.0697,
      "step": 4970
    },
    {
      "epoch": 9.455064194008559,
      "grad_norm": 0.3129672110080719,
      "learning_rate": 6.845982851698952e-05,
      "loss": 0.0638,
      "step": 4971
    },
    {
      "epoch": 9.45696623870661,
      "grad_norm": 0.5016154646873474,
      "learning_rate": 6.845347729437918e-05,
      "loss": 0.0995,
      "step": 4972
    },
    {
      "epoch": 9.45886828340466,
      "grad_norm": 0.2865123748779297,
      "learning_rate": 6.844712607176882e-05,
      "loss": 0.0482,
      "step": 4973
    },
    {
      "epoch": 9.46077032810271,
      "grad_norm": 0.23946310579776764,
      "learning_rate": 6.844077484915846e-05,
      "loss": 0.0602,
      "step": 4974
    },
    {
      "epoch": 9.46267237280076,
      "grad_norm": 0.3832375705242157,
      "learning_rate": 6.843442362654812e-05,
      "loss": 0.1025,
      "step": 4975
    },
    {
      "epoch": 9.464574417498811,
      "grad_norm": 0.2058791220188141,
      "learning_rate": 6.842807240393776e-05,
      "loss": 0.0605,
      "step": 4976
    },
    {
      "epoch": 9.466476462196862,
      "grad_norm": 0.24849610030651093,
      "learning_rate": 6.842172118132741e-05,
      "loss": 0.0699,
      "step": 4977
    },
    {
      "epoch": 9.468378506894911,
      "grad_norm": 0.2340250462293625,
      "learning_rate": 6.841536995871705e-05,
      "loss": 0.0606,
      "step": 4978
    },
    {
      "epoch": 9.470280551592962,
      "grad_norm": 0.2013719379901886,
      "learning_rate": 6.84090187361067e-05,
      "loss": 0.0658,
      "step": 4979
    },
    {
      "epoch": 9.472182596291013,
      "grad_norm": 0.24580608308315277,
      "learning_rate": 6.840266751349635e-05,
      "loss": 0.0722,
      "step": 4980
    },
    {
      "epoch": 9.474084640989064,
      "grad_norm": 0.23459140956401825,
      "learning_rate": 6.839631629088599e-05,
      "loss": 0.0578,
      "step": 4981
    },
    {
      "epoch": 9.475986685687113,
      "grad_norm": 0.2569440007209778,
      "learning_rate": 6.838996506827564e-05,
      "loss": 0.0775,
      "step": 4982
    },
    {
      "epoch": 9.477888730385164,
      "grad_norm": 0.3572919964790344,
      "learning_rate": 6.83836138456653e-05,
      "loss": 0.0848,
      "step": 4983
    },
    {
      "epoch": 9.479790775083215,
      "grad_norm": 0.252504825592041,
      "learning_rate": 6.837726262305493e-05,
      "loss": 0.0838,
      "step": 4984
    },
    {
      "epoch": 9.481692819781266,
      "grad_norm": 0.2865563929080963,
      "learning_rate": 6.837091140044459e-05,
      "loss": 0.07,
      "step": 4985
    },
    {
      "epoch": 9.483594864479315,
      "grad_norm": 0.32399818301200867,
      "learning_rate": 6.836456017783424e-05,
      "loss": 0.0794,
      "step": 4986
    },
    {
      "epoch": 9.485496909177366,
      "grad_norm": 0.1883072704076767,
      "learning_rate": 6.835820895522388e-05,
      "loss": 0.0694,
      "step": 4987
    },
    {
      "epoch": 9.487398953875417,
      "grad_norm": 0.3224576413631439,
      "learning_rate": 6.835185773261353e-05,
      "loss": 0.0734,
      "step": 4988
    },
    {
      "epoch": 9.489300998573466,
      "grad_norm": 0.2641156315803528,
      "learning_rate": 6.834550651000318e-05,
      "loss": 0.0738,
      "step": 4989
    },
    {
      "epoch": 9.491203043271517,
      "grad_norm": 0.3000323176383972,
      "learning_rate": 6.833915528739283e-05,
      "loss": 0.0646,
      "step": 4990
    },
    {
      "epoch": 9.493105087969568,
      "grad_norm": 0.20200790464878082,
      "learning_rate": 6.833280406478247e-05,
      "loss": 0.0732,
      "step": 4991
    },
    {
      "epoch": 9.495007132667618,
      "grad_norm": 0.13494519889354706,
      "learning_rate": 6.832645284217212e-05,
      "loss": 0.0533,
      "step": 4992
    },
    {
      "epoch": 9.496909177365668,
      "grad_norm": 0.22805655002593994,
      "learning_rate": 6.832010161956177e-05,
      "loss": 0.0506,
      "step": 4993
    },
    {
      "epoch": 9.498811222063718,
      "grad_norm": 0.2714557349681854,
      "learning_rate": 6.831375039695141e-05,
      "loss": 0.0798,
      "step": 4994
    },
    {
      "epoch": 9.50071326676177,
      "grad_norm": 0.2856943607330322,
      "learning_rate": 6.830739917434106e-05,
      "loss": 0.0755,
      "step": 4995
    },
    {
      "epoch": 9.50261531145982,
      "grad_norm": 0.39549219608306885,
      "learning_rate": 6.830104795173072e-05,
      "loss": 0.0854,
      "step": 4996
    },
    {
      "epoch": 9.50451735615787,
      "grad_norm": 0.24781033396720886,
      "learning_rate": 6.829469672912035e-05,
      "loss": 0.0784,
      "step": 4997
    },
    {
      "epoch": 9.50641940085592,
      "grad_norm": 0.2739943861961365,
      "learning_rate": 6.828834550651e-05,
      "loss": 0.0627,
      "step": 4998
    },
    {
      "epoch": 9.508321445553971,
      "grad_norm": 0.3200620710849762,
      "learning_rate": 6.828199428389966e-05,
      "loss": 0.1221,
      "step": 4999
    },
    {
      "epoch": 9.51022349025202,
      "grad_norm": 0.2922022342681885,
      "learning_rate": 6.82756430612893e-05,
      "loss": 0.0765,
      "step": 5000
    },
    {
      "epoch": 9.512125534950071,
      "grad_norm": 0.2247658371925354,
      "learning_rate": 6.826929183867895e-05,
      "loss": 0.0586,
      "step": 5001
    },
    {
      "epoch": 9.514027579648122,
      "grad_norm": 0.18211832642555237,
      "learning_rate": 6.82629406160686e-05,
      "loss": 0.0559,
      "step": 5002
    },
    {
      "epoch": 9.515929624346173,
      "grad_norm": 0.4612724483013153,
      "learning_rate": 6.825658939345825e-05,
      "loss": 0.1092,
      "step": 5003
    },
    {
      "epoch": 9.517831669044222,
      "grad_norm": 0.2759299874305725,
      "learning_rate": 6.825023817084789e-05,
      "loss": 0.0885,
      "step": 5004
    },
    {
      "epoch": 9.519733713742273,
      "grad_norm": 0.28181999921798706,
      "learning_rate": 6.824388694823753e-05,
      "loss": 0.0585,
      "step": 5005
    },
    {
      "epoch": 9.521635758440324,
      "grad_norm": 0.249153271317482,
      "learning_rate": 6.82375357256272e-05,
      "loss": 0.0625,
      "step": 5006
    },
    {
      "epoch": 9.523537803138375,
      "grad_norm": 0.23787033557891846,
      "learning_rate": 6.823118450301683e-05,
      "loss": 0.0599,
      "step": 5007
    },
    {
      "epoch": 9.525439847836424,
      "grad_norm": 0.30787795782089233,
      "learning_rate": 6.822483328040648e-05,
      "loss": 0.0652,
      "step": 5008
    },
    {
      "epoch": 9.527341892534475,
      "grad_norm": 0.21903198957443237,
      "learning_rate": 6.821848205779612e-05,
      "loss": 0.0689,
      "step": 5009
    },
    {
      "epoch": 9.529243937232525,
      "grad_norm": 0.2730168402194977,
      "learning_rate": 6.821213083518577e-05,
      "loss": 0.088,
      "step": 5010
    },
    {
      "epoch": 9.531145981930575,
      "grad_norm": 0.2813446819782257,
      "learning_rate": 6.820577961257543e-05,
      "loss": 0.0734,
      "step": 5011
    },
    {
      "epoch": 9.533048026628625,
      "grad_norm": 0.22539003193378448,
      "learning_rate": 6.819942838996506e-05,
      "loss": 0.0781,
      "step": 5012
    },
    {
      "epoch": 9.534950071326676,
      "grad_norm": 0.24514015018939972,
      "learning_rate": 6.819307716735473e-05,
      "loss": 0.0629,
      "step": 5013
    },
    {
      "epoch": 9.536852116024727,
      "grad_norm": 0.2931705415248871,
      "learning_rate": 6.818672594474437e-05,
      "loss": 0.0696,
      "step": 5014
    },
    {
      "epoch": 9.538754160722776,
      "grad_norm": 0.16477547585964203,
      "learning_rate": 6.8180374722134e-05,
      "loss": 0.0619,
      "step": 5015
    },
    {
      "epoch": 9.540656205420827,
      "grad_norm": 0.27690985798835754,
      "learning_rate": 6.817402349952366e-05,
      "loss": 0.0736,
      "step": 5016
    },
    {
      "epoch": 9.542558250118878,
      "grad_norm": 0.27963119745254517,
      "learning_rate": 6.816767227691331e-05,
      "loss": 0.0947,
      "step": 5017
    },
    {
      "epoch": 9.544460294816929,
      "grad_norm": 0.2638995945453644,
      "learning_rate": 6.816132105430295e-05,
      "loss": 0.0891,
      "step": 5018
    },
    {
      "epoch": 9.546362339514978,
      "grad_norm": 0.2374192625284195,
      "learning_rate": 6.81549698316926e-05,
      "loss": 0.0668,
      "step": 5019
    },
    {
      "epoch": 9.548264384213029,
      "grad_norm": 0.16655726730823517,
      "learning_rate": 6.814861860908225e-05,
      "loss": 0.0516,
      "step": 5020
    },
    {
      "epoch": 9.55016642891108,
      "grad_norm": 0.2931881844997406,
      "learning_rate": 6.81422673864719e-05,
      "loss": 0.08,
      "step": 5021
    },
    {
      "epoch": 9.552068473609129,
      "grad_norm": 0.257124125957489,
      "learning_rate": 6.813591616386154e-05,
      "loss": 0.0693,
      "step": 5022
    },
    {
      "epoch": 9.55397051830718,
      "grad_norm": 0.20207662880420685,
      "learning_rate": 6.81295649412512e-05,
      "loss": 0.0619,
      "step": 5023
    },
    {
      "epoch": 9.55587256300523,
      "grad_norm": 0.21767112612724304,
      "learning_rate": 6.812321371864085e-05,
      "loss": 0.0612,
      "step": 5024
    },
    {
      "epoch": 9.557774607703282,
      "grad_norm": 0.21762673556804657,
      "learning_rate": 6.811686249603048e-05,
      "loss": 0.073,
      "step": 5025
    },
    {
      "epoch": 9.55967665240133,
      "grad_norm": 0.2587988078594208,
      "learning_rate": 6.811051127342014e-05,
      "loss": 0.0646,
      "step": 5026
    },
    {
      "epoch": 9.561578697099382,
      "grad_norm": 0.27404069900512695,
      "learning_rate": 6.810416005080979e-05,
      "loss": 0.0675,
      "step": 5027
    },
    {
      "epoch": 9.563480741797433,
      "grad_norm": 0.2376728504896164,
      "learning_rate": 6.809780882819943e-05,
      "loss": 0.0646,
      "step": 5028
    },
    {
      "epoch": 9.565382786495483,
      "grad_norm": 0.20966622233390808,
      "learning_rate": 6.809145760558908e-05,
      "loss": 0.0772,
      "step": 5029
    },
    {
      "epoch": 9.567284831193533,
      "grad_norm": 0.3153587579727173,
      "learning_rate": 6.808510638297873e-05,
      "loss": 0.0896,
      "step": 5030
    },
    {
      "epoch": 9.569186875891583,
      "grad_norm": 0.4038897156715393,
      "learning_rate": 6.807875516036838e-05,
      "loss": 0.1006,
      "step": 5031
    },
    {
      "epoch": 9.571088920589634,
      "grad_norm": 0.16164061427116394,
      "learning_rate": 6.807240393775802e-05,
      "loss": 0.0567,
      "step": 5032
    },
    {
      "epoch": 9.572990965287683,
      "grad_norm": 0.33663511276245117,
      "learning_rate": 6.806605271514767e-05,
      "loss": 0.0767,
      "step": 5033
    },
    {
      "epoch": 9.574893009985734,
      "grad_norm": 0.2418556958436966,
      "learning_rate": 6.805970149253732e-05,
      "loss": 0.0766,
      "step": 5034
    },
    {
      "epoch": 9.576795054683785,
      "grad_norm": 0.26614710688591003,
      "learning_rate": 6.805335026992696e-05,
      "loss": 0.0691,
      "step": 5035
    },
    {
      "epoch": 9.578697099381836,
      "grad_norm": 0.35526224970817566,
      "learning_rate": 6.80469990473166e-05,
      "loss": 0.0751,
      "step": 5036
    },
    {
      "epoch": 9.580599144079885,
      "grad_norm": 0.16759097576141357,
      "learning_rate": 6.804064782470627e-05,
      "loss": 0.0548,
      "step": 5037
    },
    {
      "epoch": 9.582501188777936,
      "grad_norm": 0.3385069668292999,
      "learning_rate": 6.80342966020959e-05,
      "loss": 0.0734,
      "step": 5038
    },
    {
      "epoch": 9.584403233475987,
      "grad_norm": 0.24936793744564056,
      "learning_rate": 6.802794537948556e-05,
      "loss": 0.0846,
      "step": 5039
    },
    {
      "epoch": 9.586305278174038,
      "grad_norm": 0.22732244431972504,
      "learning_rate": 6.802159415687521e-05,
      "loss": 0.0588,
      "step": 5040
    },
    {
      "epoch": 9.588207322872087,
      "grad_norm": 0.2636946141719818,
      "learning_rate": 6.801524293426485e-05,
      "loss": 0.0652,
      "step": 5041
    },
    {
      "epoch": 9.590109367570138,
      "grad_norm": 0.25417861342430115,
      "learning_rate": 6.80088917116545e-05,
      "loss": 0.0701,
      "step": 5042
    },
    {
      "epoch": 9.592011412268189,
      "grad_norm": 0.2900865375995636,
      "learning_rate": 6.800254048904414e-05,
      "loss": 0.1018,
      "step": 5043
    },
    {
      "epoch": 9.593913456966238,
      "grad_norm": 0.21319755911827087,
      "learning_rate": 6.79961892664338e-05,
      "loss": 0.0673,
      "step": 5044
    },
    {
      "epoch": 9.595815501664289,
      "grad_norm": 0.4287901222705841,
      "learning_rate": 6.798983804382344e-05,
      "loss": 0.1022,
      "step": 5045
    },
    {
      "epoch": 9.59771754636234,
      "grad_norm": 0.22494946420192719,
      "learning_rate": 6.798348682121308e-05,
      "loss": 0.0494,
      "step": 5046
    },
    {
      "epoch": 9.59961959106039,
      "grad_norm": 0.3113909065723419,
      "learning_rate": 6.797713559860274e-05,
      "loss": 0.2004,
      "step": 5047
    },
    {
      "epoch": 9.60152163575844,
      "grad_norm": 0.3399859368801117,
      "learning_rate": 6.797078437599238e-05,
      "loss": 0.0865,
      "step": 5048
    },
    {
      "epoch": 9.60342368045649,
      "grad_norm": 0.3913676142692566,
      "learning_rate": 6.796443315338203e-05,
      "loss": 0.0732,
      "step": 5049
    },
    {
      "epoch": 9.605325725154541,
      "grad_norm": 0.2704809308052063,
      "learning_rate": 6.795808193077167e-05,
      "loss": 0.0718,
      "step": 5050
    },
    {
      "epoch": 9.607227769852592,
      "grad_norm": 0.1767328977584839,
      "learning_rate": 6.795173070816132e-05,
      "loss": 0.0618,
      "step": 5051
    },
    {
      "epoch": 9.609129814550641,
      "grad_norm": 0.3268260061740875,
      "learning_rate": 6.794537948555098e-05,
      "loss": 0.081,
      "step": 5052
    },
    {
      "epoch": 9.611031859248692,
      "grad_norm": 0.24618151783943176,
      "learning_rate": 6.793902826294061e-05,
      "loss": 0.0791,
      "step": 5053
    },
    {
      "epoch": 9.612933903946743,
      "grad_norm": 0.3275710940361023,
      "learning_rate": 6.793267704033027e-05,
      "loss": 0.0838,
      "step": 5054
    },
    {
      "epoch": 9.614835948644792,
      "grad_norm": 0.2791629135608673,
      "learning_rate": 6.792632581771992e-05,
      "loss": 0.0446,
      "step": 5055
    },
    {
      "epoch": 9.616737993342843,
      "grad_norm": 0.2838995158672333,
      "learning_rate": 6.791997459510956e-05,
      "loss": 0.0627,
      "step": 5056
    },
    {
      "epoch": 9.618640038040894,
      "grad_norm": 0.32235971093177795,
      "learning_rate": 6.791362337249921e-05,
      "loss": 0.0832,
      "step": 5057
    },
    {
      "epoch": 9.620542082738945,
      "grad_norm": 0.3361624777317047,
      "learning_rate": 6.790727214988886e-05,
      "loss": 0.0731,
      "step": 5058
    },
    {
      "epoch": 9.622444127436994,
      "grad_norm": 0.31241574883461,
      "learning_rate": 6.79009209272785e-05,
      "loss": 0.0823,
      "step": 5059
    },
    {
      "epoch": 9.624346172135045,
      "grad_norm": 0.5286279916763306,
      "learning_rate": 6.789456970466815e-05,
      "loss": 0.0851,
      "step": 5060
    },
    {
      "epoch": 9.626248216833096,
      "grad_norm": 0.2209668606519699,
      "learning_rate": 6.78882184820578e-05,
      "loss": 0.0851,
      "step": 5061
    },
    {
      "epoch": 9.628150261531147,
      "grad_norm": 0.23965579271316528,
      "learning_rate": 6.788186725944745e-05,
      "loss": 0.0752,
      "step": 5062
    },
    {
      "epoch": 9.630052306229196,
      "grad_norm": 0.320737361907959,
      "learning_rate": 6.787551603683709e-05,
      "loss": 0.0671,
      "step": 5063
    },
    {
      "epoch": 9.631954350927247,
      "grad_norm": 0.23596954345703125,
      "learning_rate": 6.786916481422674e-05,
      "loss": 0.0661,
      "step": 5064
    },
    {
      "epoch": 9.633856395625298,
      "grad_norm": 0.24826346337795258,
      "learning_rate": 6.78628135916164e-05,
      "loss": 0.0673,
      "step": 5065
    },
    {
      "epoch": 9.635758440323347,
      "grad_norm": 0.19784359633922577,
      "learning_rate": 6.785646236900603e-05,
      "loss": 0.0644,
      "step": 5066
    },
    {
      "epoch": 9.637660485021398,
      "grad_norm": 0.2708280682563782,
      "learning_rate": 6.785011114639569e-05,
      "loss": 0.0962,
      "step": 5067
    },
    {
      "epoch": 9.639562529719448,
      "grad_norm": 0.22095151245594025,
      "learning_rate": 6.784375992378534e-05,
      "loss": 0.0792,
      "step": 5068
    },
    {
      "epoch": 9.6414645744175,
      "grad_norm": 0.3527117967605591,
      "learning_rate": 6.783740870117498e-05,
      "loss": 0.0867,
      "step": 5069
    },
    {
      "epoch": 9.643366619115548,
      "grad_norm": 0.27398914098739624,
      "learning_rate": 6.783105747856463e-05,
      "loss": 0.0707,
      "step": 5070
    },
    {
      "epoch": 9.6452686638136,
      "grad_norm": 0.251521497964859,
      "learning_rate": 6.782470625595428e-05,
      "loss": 0.0494,
      "step": 5071
    },
    {
      "epoch": 9.64717070851165,
      "grad_norm": 0.23466455936431885,
      "learning_rate": 6.781835503334392e-05,
      "loss": 0.0762,
      "step": 5072
    },
    {
      "epoch": 9.649072753209701,
      "grad_norm": 0.25730809569358826,
      "learning_rate": 6.781200381073357e-05,
      "loss": 0.0598,
      "step": 5073
    },
    {
      "epoch": 9.65097479790775,
      "grad_norm": 0.23233462870121002,
      "learning_rate": 6.780565258812321e-05,
      "loss": 0.0696,
      "step": 5074
    },
    {
      "epoch": 9.652876842605801,
      "grad_norm": 0.34221500158309937,
      "learning_rate": 6.779930136551287e-05,
      "loss": 0.0859,
      "step": 5075
    },
    {
      "epoch": 9.654778887303852,
      "grad_norm": 0.24130195379257202,
      "learning_rate": 6.779295014290251e-05,
      "loss": 0.0625,
      "step": 5076
    },
    {
      "epoch": 9.656680932001901,
      "grad_norm": 0.27201899886131287,
      "learning_rate": 6.778659892029215e-05,
      "loss": 0.0839,
      "step": 5077
    },
    {
      "epoch": 9.658582976699952,
      "grad_norm": 0.2817753255367279,
      "learning_rate": 6.778024769768181e-05,
      "loss": 0.0709,
      "step": 5078
    },
    {
      "epoch": 9.660485021398003,
      "grad_norm": 0.3388703465461731,
      "learning_rate": 6.777389647507145e-05,
      "loss": 0.0655,
      "step": 5079
    },
    {
      "epoch": 9.662387066096054,
      "grad_norm": 0.3778761625289917,
      "learning_rate": 6.77675452524611e-05,
      "loss": 0.0941,
      "step": 5080
    },
    {
      "epoch": 9.664289110794103,
      "grad_norm": 0.23203331232070923,
      "learning_rate": 6.776119402985074e-05,
      "loss": 0.0725,
      "step": 5081
    },
    {
      "epoch": 9.666191155492154,
      "grad_norm": 0.32718273997306824,
      "learning_rate": 6.77548428072404e-05,
      "loss": 0.061,
      "step": 5082
    },
    {
      "epoch": 9.668093200190205,
      "grad_norm": 0.21871528029441833,
      "learning_rate": 6.774849158463005e-05,
      "loss": 0.0615,
      "step": 5083
    },
    {
      "epoch": 9.669995244888256,
      "grad_norm": 0.20493032038211823,
      "learning_rate": 6.774214036201969e-05,
      "loss": 0.0708,
      "step": 5084
    },
    {
      "epoch": 9.671897289586305,
      "grad_norm": 0.2302158623933792,
      "learning_rate": 6.773578913940935e-05,
      "loss": 0.0857,
      "step": 5085
    },
    {
      "epoch": 9.673799334284356,
      "grad_norm": 0.2125237137079239,
      "learning_rate": 6.772943791679899e-05,
      "loss": 0.0888,
      "step": 5086
    },
    {
      "epoch": 9.675701378982406,
      "grad_norm": 0.2613074779510498,
      "learning_rate": 6.772308669418863e-05,
      "loss": 0.0651,
      "step": 5087
    },
    {
      "epoch": 9.677603423680456,
      "grad_norm": 0.3250272274017334,
      "learning_rate": 6.771673547157828e-05,
      "loss": 0.099,
      "step": 5088
    },
    {
      "epoch": 9.679505468378506,
      "grad_norm": 0.18376415967941284,
      "learning_rate": 6.771038424896793e-05,
      "loss": 0.0609,
      "step": 5089
    },
    {
      "epoch": 9.681407513076557,
      "grad_norm": 0.14176584780216217,
      "learning_rate": 6.770403302635757e-05,
      "loss": 0.0511,
      "step": 5090
    },
    {
      "epoch": 9.683309557774608,
      "grad_norm": 0.21321751177310944,
      "learning_rate": 6.769768180374722e-05,
      "loss": 0.0586,
      "step": 5091
    },
    {
      "epoch": 9.685211602472657,
      "grad_norm": 0.19097277522087097,
      "learning_rate": 6.769133058113687e-05,
      "loss": 0.069,
      "step": 5092
    },
    {
      "epoch": 9.687113647170708,
      "grad_norm": 0.3134055733680725,
      "learning_rate": 6.768497935852652e-05,
      "loss": 0.0799,
      "step": 5093
    },
    {
      "epoch": 9.689015691868759,
      "grad_norm": 0.19542521238327026,
      "learning_rate": 6.767862813591616e-05,
      "loss": 0.0513,
      "step": 5094
    },
    {
      "epoch": 9.69091773656681,
      "grad_norm": 0.2252933830022812,
      "learning_rate": 6.767227691330581e-05,
      "loss": 0.0717,
      "step": 5095
    },
    {
      "epoch": 9.692819781264859,
      "grad_norm": 0.24214401841163635,
      "learning_rate": 6.766592569069547e-05,
      "loss": 0.0713,
      "step": 5096
    },
    {
      "epoch": 9.69472182596291,
      "grad_norm": 0.38027697801589966,
      "learning_rate": 6.76595744680851e-05,
      "loss": 0.0793,
      "step": 5097
    },
    {
      "epoch": 9.69662387066096,
      "grad_norm": 0.2558886706829071,
      "learning_rate": 6.765322324547476e-05,
      "loss": 0.0595,
      "step": 5098
    },
    {
      "epoch": 9.698525915359012,
      "grad_norm": 0.22915878891944885,
      "learning_rate": 6.764687202286441e-05,
      "loss": 0.0804,
      "step": 5099
    },
    {
      "epoch": 9.70042796005706,
      "grad_norm": 0.2833732068538666,
      "learning_rate": 6.764052080025405e-05,
      "loss": 0.0802,
      "step": 5100
    },
    {
      "epoch": 9.702330004755112,
      "grad_norm": 0.20914685726165771,
      "learning_rate": 6.76341695776437e-05,
      "loss": 0.0546,
      "step": 5101
    },
    {
      "epoch": 9.704232049453163,
      "grad_norm": 0.26040971279144287,
      "learning_rate": 6.762781835503335e-05,
      "loss": 0.0718,
      "step": 5102
    },
    {
      "epoch": 9.706134094151212,
      "grad_norm": 0.21980637311935425,
      "learning_rate": 6.7621467132423e-05,
      "loss": 0.0758,
      "step": 5103
    },
    {
      "epoch": 9.708036138849263,
      "grad_norm": 0.24792368710041046,
      "learning_rate": 6.761511590981264e-05,
      "loss": 0.0799,
      "step": 5104
    },
    {
      "epoch": 9.709938183547314,
      "grad_norm": 0.23755225539207458,
      "learning_rate": 6.760876468720229e-05,
      "loss": 0.0606,
      "step": 5105
    },
    {
      "epoch": 9.711840228245364,
      "grad_norm": 0.14159350097179413,
      "learning_rate": 6.760241346459194e-05,
      "loss": 0.0666,
      "step": 5106
    },
    {
      "epoch": 9.713742272943414,
      "grad_norm": 0.29056262969970703,
      "learning_rate": 6.759606224198158e-05,
      "loss": 0.0782,
      "step": 5107
    },
    {
      "epoch": 9.715644317641464,
      "grad_norm": 0.23274298012256622,
      "learning_rate": 6.758971101937122e-05,
      "loss": 0.0672,
      "step": 5108
    },
    {
      "epoch": 9.717546362339515,
      "grad_norm": 0.37022560834884644,
      "learning_rate": 6.758335979676089e-05,
      "loss": 0.0746,
      "step": 5109
    },
    {
      "epoch": 9.719448407037566,
      "grad_norm": 0.18731588125228882,
      "learning_rate": 6.757700857415052e-05,
      "loss": 0.0537,
      "step": 5110
    },
    {
      "epoch": 9.721350451735615,
      "grad_norm": 0.33824440836906433,
      "learning_rate": 6.757065735154018e-05,
      "loss": 0.0741,
      "step": 5111
    },
    {
      "epoch": 9.723252496433666,
      "grad_norm": 0.3501187860965729,
      "learning_rate": 6.756430612892983e-05,
      "loss": 0.0853,
      "step": 5112
    },
    {
      "epoch": 9.725154541131717,
      "grad_norm": 0.25903284549713135,
      "learning_rate": 6.755795490631947e-05,
      "loss": 0.0572,
      "step": 5113
    },
    {
      "epoch": 9.727056585829766,
      "grad_norm": 0.19478189945220947,
      "learning_rate": 6.755160368370912e-05,
      "loss": 0.0643,
      "step": 5114
    },
    {
      "epoch": 9.728958630527817,
      "grad_norm": 0.249579519033432,
      "learning_rate": 6.754525246109876e-05,
      "loss": 0.0737,
      "step": 5115
    },
    {
      "epoch": 9.730860675225868,
      "grad_norm": 0.2832339107990265,
      "learning_rate": 6.753890123848842e-05,
      "loss": 0.093,
      "step": 5116
    },
    {
      "epoch": 9.732762719923919,
      "grad_norm": 0.22517451643943787,
      "learning_rate": 6.753255001587806e-05,
      "loss": 0.0696,
      "step": 5117
    },
    {
      "epoch": 9.734664764621968,
      "grad_norm": 0.28813955187797546,
      "learning_rate": 6.75261987932677e-05,
      "loss": 0.0872,
      "step": 5118
    },
    {
      "epoch": 9.736566809320019,
      "grad_norm": 0.36511409282684326,
      "learning_rate": 6.751984757065735e-05,
      "loss": 0.087,
      "step": 5119
    },
    {
      "epoch": 9.73846885401807,
      "grad_norm": 0.3195239305496216,
      "learning_rate": 6.7513496348047e-05,
      "loss": 0.0863,
      "step": 5120
    },
    {
      "epoch": 9.74037089871612,
      "grad_norm": 0.39804649353027344,
      "learning_rate": 6.750714512543665e-05,
      "loss": 0.077,
      "step": 5121
    },
    {
      "epoch": 9.74227294341417,
      "grad_norm": 0.20853570103645325,
      "learning_rate": 6.750079390282629e-05,
      "loss": 0.0538,
      "step": 5122
    },
    {
      "epoch": 9.74417498811222,
      "grad_norm": 0.29665881395339966,
      "learning_rate": 6.749444268021594e-05,
      "loss": 0.0811,
      "step": 5123
    },
    {
      "epoch": 9.746077032810271,
      "grad_norm": 0.2606349289417267,
      "learning_rate": 6.74880914576056e-05,
      "loss": 0.0701,
      "step": 5124
    },
    {
      "epoch": 9.747979077508322,
      "grad_norm": 0.297983855009079,
      "learning_rate": 6.748174023499523e-05,
      "loss": 0.0837,
      "step": 5125
    },
    {
      "epoch": 9.749881122206371,
      "grad_norm": 0.4061759412288666,
      "learning_rate": 6.747538901238489e-05,
      "loss": 0.0913,
      "step": 5126
    },
    {
      "epoch": 9.751783166904422,
      "grad_norm": 0.26573270559310913,
      "learning_rate": 6.746903778977454e-05,
      "loss": 0.1043,
      "step": 5127
    },
    {
      "epoch": 9.753685211602473,
      "grad_norm": 0.3666708767414093,
      "learning_rate": 6.746268656716418e-05,
      "loss": 0.0852,
      "step": 5128
    },
    {
      "epoch": 9.755587256300522,
      "grad_norm": 0.29040297865867615,
      "learning_rate": 6.745633534455383e-05,
      "loss": 0.0786,
      "step": 5129
    },
    {
      "epoch": 9.757489300998573,
      "grad_norm": 0.2016884982585907,
      "learning_rate": 6.744998412194348e-05,
      "loss": 0.0695,
      "step": 5130
    },
    {
      "epoch": 9.759391345696624,
      "grad_norm": 0.28894975781440735,
      "learning_rate": 6.744363289933312e-05,
      "loss": 0.0689,
      "step": 5131
    },
    {
      "epoch": 9.761293390394675,
      "grad_norm": 0.3561621606349945,
      "learning_rate": 6.743728167672277e-05,
      "loss": 0.0942,
      "step": 5132
    },
    {
      "epoch": 9.763195435092724,
      "grad_norm": 0.23403863608837128,
      "learning_rate": 6.743093045411242e-05,
      "loss": 0.0946,
      "step": 5133
    },
    {
      "epoch": 9.765097479790775,
      "grad_norm": 0.18953385949134827,
      "learning_rate": 6.742457923150207e-05,
      "loss": 0.0621,
      "step": 5134
    },
    {
      "epoch": 9.766999524488826,
      "grad_norm": 0.2510560154914856,
      "learning_rate": 6.741822800889171e-05,
      "loss": 0.076,
      "step": 5135
    },
    {
      "epoch": 9.768901569186877,
      "grad_norm": 0.23607076704502106,
      "learning_rate": 6.741187678628136e-05,
      "loss": 0.0554,
      "step": 5136
    },
    {
      "epoch": 9.770803613884926,
      "grad_norm": 0.26703163981437683,
      "learning_rate": 6.740552556367102e-05,
      "loss": 0.0917,
      "step": 5137
    },
    {
      "epoch": 9.772705658582977,
      "grad_norm": 0.3677026033401489,
      "learning_rate": 6.739917434106065e-05,
      "loss": 0.0853,
      "step": 5138
    },
    {
      "epoch": 9.774607703281028,
      "grad_norm": 0.22919195890426636,
      "learning_rate": 6.73928231184503e-05,
      "loss": 0.0548,
      "step": 5139
    },
    {
      "epoch": 9.776509747979077,
      "grad_norm": 0.24974371492862701,
      "learning_rate": 6.738647189583996e-05,
      "loss": 0.0657,
      "step": 5140
    },
    {
      "epoch": 9.778411792677128,
      "grad_norm": 0.23522870242595673,
      "learning_rate": 6.73801206732296e-05,
      "loss": 0.0765,
      "step": 5141
    },
    {
      "epoch": 9.780313837375179,
      "grad_norm": 0.3335094451904297,
      "learning_rate": 6.737376945061925e-05,
      "loss": 0.0584,
      "step": 5142
    },
    {
      "epoch": 9.78221588207323,
      "grad_norm": 0.24172602593898773,
      "learning_rate": 6.73674182280089e-05,
      "loss": 0.0613,
      "step": 5143
    },
    {
      "epoch": 9.784117926771279,
      "grad_norm": 0.24692966043949127,
      "learning_rate": 6.736106700539854e-05,
      "loss": 0.0681,
      "step": 5144
    },
    {
      "epoch": 9.78601997146933,
      "grad_norm": 0.1990744024515152,
      "learning_rate": 6.735471578278819e-05,
      "loss": 0.0643,
      "step": 5145
    },
    {
      "epoch": 9.78792201616738,
      "grad_norm": 0.182316854596138,
      "learning_rate": 6.734836456017783e-05,
      "loss": 0.0611,
      "step": 5146
    },
    {
      "epoch": 9.789824060865431,
      "grad_norm": 0.2731354832649231,
      "learning_rate": 6.73420133375675e-05,
      "loss": 0.0725,
      "step": 5147
    },
    {
      "epoch": 9.79172610556348,
      "grad_norm": 0.23140563070774078,
      "learning_rate": 6.733566211495713e-05,
      "loss": 0.075,
      "step": 5148
    },
    {
      "epoch": 9.793628150261531,
      "grad_norm": 0.2696152329444885,
      "learning_rate": 6.732931089234677e-05,
      "loss": 0.0788,
      "step": 5149
    },
    {
      "epoch": 9.795530194959582,
      "grad_norm": 0.22241799533367157,
      "learning_rate": 6.732295966973644e-05,
      "loss": 0.0661,
      "step": 5150
    },
    {
      "epoch": 9.797432239657631,
      "grad_norm": 0.24715057015419006,
      "learning_rate": 6.731660844712607e-05,
      "loss": 0.0768,
      "step": 5151
    },
    {
      "epoch": 9.799334284355682,
      "grad_norm": 0.16379965841770172,
      "learning_rate": 6.731025722451573e-05,
      "loss": 0.0515,
      "step": 5152
    },
    {
      "epoch": 9.801236329053733,
      "grad_norm": 0.25936359167099,
      "learning_rate": 6.730390600190536e-05,
      "loss": 0.0655,
      "step": 5153
    },
    {
      "epoch": 9.803138373751784,
      "grad_norm": 0.29590755701065063,
      "learning_rate": 6.729755477929502e-05,
      "loss": 0.0761,
      "step": 5154
    },
    {
      "epoch": 9.805040418449833,
      "grad_norm": 0.2646804749965668,
      "learning_rate": 6.729120355668467e-05,
      "loss": 0.0717,
      "step": 5155
    },
    {
      "epoch": 9.806942463147884,
      "grad_norm": 0.2943807542324066,
      "learning_rate": 6.72848523340743e-05,
      "loss": 0.0979,
      "step": 5156
    },
    {
      "epoch": 9.808844507845935,
      "grad_norm": 0.25199463963508606,
      "learning_rate": 6.727850111146397e-05,
      "loss": 0.0617,
      "step": 5157
    },
    {
      "epoch": 9.810746552543986,
      "grad_norm": 0.23185868561267853,
      "learning_rate": 6.727214988885361e-05,
      "loss": 0.0615,
      "step": 5158
    },
    {
      "epoch": 9.812648597242035,
      "grad_norm": 0.2331693172454834,
      "learning_rate": 6.726579866624325e-05,
      "loss": 0.0535,
      "step": 5159
    },
    {
      "epoch": 9.814550641940086,
      "grad_norm": 0.23456302285194397,
      "learning_rate": 6.72594474436329e-05,
      "loss": 0.0796,
      "step": 5160
    },
    {
      "epoch": 9.816452686638137,
      "grad_norm": 0.18812789022922516,
      "learning_rate": 6.725309622102255e-05,
      "loss": 0.0626,
      "step": 5161
    },
    {
      "epoch": 9.818354731336186,
      "grad_norm": 0.21228209137916565,
      "learning_rate": 6.724674499841219e-05,
      "loss": 0.0669,
      "step": 5162
    },
    {
      "epoch": 9.820256776034237,
      "grad_norm": 0.24619713425636292,
      "learning_rate": 6.724039377580184e-05,
      "loss": 0.0628,
      "step": 5163
    },
    {
      "epoch": 9.822158820732287,
      "grad_norm": 0.26102298498153687,
      "learning_rate": 6.72340425531915e-05,
      "loss": 0.069,
      "step": 5164
    },
    {
      "epoch": 9.824060865430338,
      "grad_norm": 0.271749883890152,
      "learning_rate": 6.722769133058115e-05,
      "loss": 0.0904,
      "step": 5165
    },
    {
      "epoch": 9.825962910128387,
      "grad_norm": 0.43275517225265503,
      "learning_rate": 6.722134010797078e-05,
      "loss": 0.1015,
      "step": 5166
    },
    {
      "epoch": 9.827864954826438,
      "grad_norm": 0.36582517623901367,
      "learning_rate": 6.721498888536044e-05,
      "loss": 0.0702,
      "step": 5167
    },
    {
      "epoch": 9.82976699952449,
      "grad_norm": 0.24312864243984222,
      "learning_rate": 6.720863766275009e-05,
      "loss": 0.076,
      "step": 5168
    },
    {
      "epoch": 9.83166904422254,
      "grad_norm": 0.26955533027648926,
      "learning_rate": 6.720228644013973e-05,
      "loss": 0.064,
      "step": 5169
    },
    {
      "epoch": 9.83357108892059,
      "grad_norm": 0.24358266592025757,
      "learning_rate": 6.719593521752938e-05,
      "loss": 0.0541,
      "step": 5170
    },
    {
      "epoch": 9.83547313361864,
      "grad_norm": 0.2266244739294052,
      "learning_rate": 6.718958399491903e-05,
      "loss": 0.0716,
      "step": 5171
    },
    {
      "epoch": 9.837375178316691,
      "grad_norm": 0.34034213423728943,
      "learning_rate": 6.718323277230867e-05,
      "loss": 0.0819,
      "step": 5172
    },
    {
      "epoch": 9.83927722301474,
      "grad_norm": 0.2445746511220932,
      "learning_rate": 6.717688154969832e-05,
      "loss": 0.0671,
      "step": 5173
    },
    {
      "epoch": 9.841179267712791,
      "grad_norm": 0.3341140151023865,
      "learning_rate": 6.717053032708797e-05,
      "loss": 0.0818,
      "step": 5174
    },
    {
      "epoch": 9.843081312410842,
      "grad_norm": 0.38900166749954224,
      "learning_rate": 6.716417910447762e-05,
      "loss": 0.0942,
      "step": 5175
    },
    {
      "epoch": 9.844983357108893,
      "grad_norm": 0.3027609884738922,
      "learning_rate": 6.715782788186726e-05,
      "loss": 0.075,
      "step": 5176
    },
    {
      "epoch": 9.846885401806942,
      "grad_norm": 0.29892697930336,
      "learning_rate": 6.71514766592569e-05,
      "loss": 0.0619,
      "step": 5177
    },
    {
      "epoch": 9.848787446504993,
      "grad_norm": 0.2879316508769989,
      "learning_rate": 6.714512543664657e-05,
      "loss": 0.0737,
      "step": 5178
    },
    {
      "epoch": 9.850689491203044,
      "grad_norm": 0.31047359108924866,
      "learning_rate": 6.71387742140362e-05,
      "loss": 0.0809,
      "step": 5179
    },
    {
      "epoch": 9.852591535901094,
      "grad_norm": 0.19660452008247375,
      "learning_rate": 6.713242299142584e-05,
      "loss": 0.0569,
      "step": 5180
    },
    {
      "epoch": 9.854493580599144,
      "grad_norm": 0.2426648586988449,
      "learning_rate": 6.712607176881551e-05,
      "loss": 0.0609,
      "step": 5181
    },
    {
      "epoch": 9.856395625297194,
      "grad_norm": 0.24112775921821594,
      "learning_rate": 6.711972054620515e-05,
      "loss": 0.0585,
      "step": 5182
    },
    {
      "epoch": 9.858297669995245,
      "grad_norm": 0.17509464919567108,
      "learning_rate": 6.71133693235948e-05,
      "loss": 0.0757,
      "step": 5183
    },
    {
      "epoch": 9.860199714693294,
      "grad_norm": 0.23438102006912231,
      "learning_rate": 6.710701810098444e-05,
      "loss": 0.0597,
      "step": 5184
    },
    {
      "epoch": 9.862101759391345,
      "grad_norm": 0.23365238308906555,
      "learning_rate": 6.710066687837409e-05,
      "loss": 0.081,
      "step": 5185
    },
    {
      "epoch": 9.864003804089396,
      "grad_norm": 0.28201591968536377,
      "learning_rate": 6.709431565576374e-05,
      "loss": 0.0711,
      "step": 5186
    },
    {
      "epoch": 9.865905848787447,
      "grad_norm": 0.2844151556491852,
      "learning_rate": 6.708796443315338e-05,
      "loss": 0.0793,
      "step": 5187
    },
    {
      "epoch": 9.867807893485496,
      "grad_norm": 0.2788445055484772,
      "learning_rate": 6.708161321054304e-05,
      "loss": 0.079,
      "step": 5188
    },
    {
      "epoch": 9.869709938183547,
      "grad_norm": 0.17613370716571808,
      "learning_rate": 6.707526198793268e-05,
      "loss": 0.054,
      "step": 5189
    },
    {
      "epoch": 9.871611982881598,
      "grad_norm": 0.40397071838378906,
      "learning_rate": 6.706891076532232e-05,
      "loss": 0.0819,
      "step": 5190
    },
    {
      "epoch": 9.873514027579649,
      "grad_norm": 0.21067620813846588,
      "learning_rate": 6.706255954271197e-05,
      "loss": 0.0609,
      "step": 5191
    },
    {
      "epoch": 9.875416072277698,
      "grad_norm": 0.1924947053194046,
      "learning_rate": 6.705620832010162e-05,
      "loss": 0.0729,
      "step": 5192
    },
    {
      "epoch": 9.877318116975749,
      "grad_norm": 0.317019522190094,
      "learning_rate": 6.704985709749128e-05,
      "loss": 0.0915,
      "step": 5193
    },
    {
      "epoch": 9.8792201616738,
      "grad_norm": 0.27511605620384216,
      "learning_rate": 6.704350587488091e-05,
      "loss": 0.0993,
      "step": 5194
    },
    {
      "epoch": 9.881122206371849,
      "grad_norm": 0.24651335179805756,
      "learning_rate": 6.703715465227057e-05,
      "loss": 0.045,
      "step": 5195
    },
    {
      "epoch": 9.8830242510699,
      "grad_norm": 0.3148553669452667,
      "learning_rate": 6.703080342966022e-05,
      "loss": 0.0787,
      "step": 5196
    },
    {
      "epoch": 9.88492629576795,
      "grad_norm": 0.3132942020893097,
      "learning_rate": 6.702445220704986e-05,
      "loss": 0.0865,
      "step": 5197
    },
    {
      "epoch": 9.886828340466002,
      "grad_norm": 0.23067975044250488,
      "learning_rate": 6.701810098443951e-05,
      "loss": 0.0631,
      "step": 5198
    },
    {
      "epoch": 9.88873038516405,
      "grad_norm": 0.3110620677471161,
      "learning_rate": 6.701174976182916e-05,
      "loss": 0.0849,
      "step": 5199
    },
    {
      "epoch": 9.890632429862102,
      "grad_norm": 0.3161821663379669,
      "learning_rate": 6.70053985392188e-05,
      "loss": 0.1031,
      "step": 5200
    },
    {
      "epoch": 9.892534474560152,
      "grad_norm": 0.36968421936035156,
      "learning_rate": 6.699904731660845e-05,
      "loss": 0.0807,
      "step": 5201
    },
    {
      "epoch": 9.894436519258203,
      "grad_norm": 0.3091064393520355,
      "learning_rate": 6.69926960939981e-05,
      "loss": 0.0816,
      "step": 5202
    },
    {
      "epoch": 9.896338563956252,
      "grad_norm": 0.24623756110668182,
      "learning_rate": 6.698634487138774e-05,
      "loss": 0.0966,
      "step": 5203
    },
    {
      "epoch": 9.898240608654303,
      "grad_norm": 0.21923215687274933,
      "learning_rate": 6.697999364877739e-05,
      "loss": 0.0677,
      "step": 5204
    },
    {
      "epoch": 9.900142653352354,
      "grad_norm": 0.32677069306373596,
      "learning_rate": 6.697364242616704e-05,
      "loss": 0.0863,
      "step": 5205
    },
    {
      "epoch": 9.902044698050403,
      "grad_norm": 0.32407426834106445,
      "learning_rate": 6.69672912035567e-05,
      "loss": 0.0815,
      "step": 5206
    },
    {
      "epoch": 9.903946742748454,
      "grad_norm": 0.20180168747901917,
      "learning_rate": 6.696093998094633e-05,
      "loss": 0.0588,
      "step": 5207
    },
    {
      "epoch": 9.905848787446505,
      "grad_norm": 0.2846551537513733,
      "learning_rate": 6.695458875833599e-05,
      "loss": 0.0707,
      "step": 5208
    },
    {
      "epoch": 9.907750832144556,
      "grad_norm": 0.22882051765918732,
      "learning_rate": 6.694823753572564e-05,
      "loss": 0.0459,
      "step": 5209
    },
    {
      "epoch": 9.909652876842605,
      "grad_norm": 0.2983953356742859,
      "learning_rate": 6.694188631311528e-05,
      "loss": 0.0781,
      "step": 5210
    },
    {
      "epoch": 9.911554921540656,
      "grad_norm": 0.3955616354942322,
      "learning_rate": 6.693553509050493e-05,
      "loss": 0.0782,
      "step": 5211
    },
    {
      "epoch": 9.913456966238707,
      "grad_norm": 0.22890278697013855,
      "learning_rate": 6.692918386789458e-05,
      "loss": 0.0545,
      "step": 5212
    },
    {
      "epoch": 9.915359010936758,
      "grad_norm": 0.2330084592103958,
      "learning_rate": 6.692283264528422e-05,
      "loss": 0.0612,
      "step": 5213
    },
    {
      "epoch": 9.917261055634807,
      "grad_norm": 0.3311925530433655,
      "learning_rate": 6.691648142267387e-05,
      "loss": 0.074,
      "step": 5214
    },
    {
      "epoch": 9.919163100332858,
      "grad_norm": 0.48154211044311523,
      "learning_rate": 6.691013020006352e-05,
      "loss": 0.0739,
      "step": 5215
    },
    {
      "epoch": 9.921065145030909,
      "grad_norm": 0.3412367105484009,
      "learning_rate": 6.690377897745316e-05,
      "loss": 0.0766,
      "step": 5216
    },
    {
      "epoch": 9.922967189728958,
      "grad_norm": 0.18190982937812805,
      "learning_rate": 6.689742775484281e-05,
      "loss": 0.0509,
      "step": 5217
    },
    {
      "epoch": 9.924869234427009,
      "grad_norm": 0.23118863999843597,
      "learning_rate": 6.689107653223245e-05,
      "loss": 0.0655,
      "step": 5218
    },
    {
      "epoch": 9.92677127912506,
      "grad_norm": 0.2357853502035141,
      "learning_rate": 6.688472530962211e-05,
      "loss": 0.0689,
      "step": 5219
    },
    {
      "epoch": 9.92867332382311,
      "grad_norm": 0.3301222026348114,
      "learning_rate": 6.687837408701175e-05,
      "loss": 0.079,
      "step": 5220
    },
    {
      "epoch": 9.93057536852116,
      "grad_norm": 0.3085179030895233,
      "learning_rate": 6.687202286440139e-05,
      "loss": 0.0826,
      "step": 5221
    },
    {
      "epoch": 9.93247741321921,
      "grad_norm": 0.23810113966464996,
      "learning_rate": 6.686567164179106e-05,
      "loss": 0.0773,
      "step": 5222
    },
    {
      "epoch": 9.934379457917261,
      "grad_norm": 0.38858041167259216,
      "learning_rate": 6.68593204191807e-05,
      "loss": 0.0957,
      "step": 5223
    },
    {
      "epoch": 9.936281502615312,
      "grad_norm": 0.15044887363910675,
      "learning_rate": 6.685296919657035e-05,
      "loss": 0.0512,
      "step": 5224
    },
    {
      "epoch": 9.938183547313361,
      "grad_norm": 0.23409341275691986,
      "learning_rate": 6.684661797395999e-05,
      "loss": 0.0822,
      "step": 5225
    },
    {
      "epoch": 9.940085592011412,
      "grad_norm": 0.2580219507217407,
      "learning_rate": 6.684026675134964e-05,
      "loss": 0.0806,
      "step": 5226
    },
    {
      "epoch": 9.941987636709463,
      "grad_norm": 0.21620117127895355,
      "learning_rate": 6.683391552873929e-05,
      "loss": 0.0722,
      "step": 5227
    },
    {
      "epoch": 9.943889681407512,
      "grad_norm": 0.33486407995224,
      "learning_rate": 6.682756430612893e-05,
      "loss": 0.0661,
      "step": 5228
    },
    {
      "epoch": 9.945791726105563,
      "grad_norm": 0.21919208765029907,
      "learning_rate": 6.682121308351858e-05,
      "loss": 0.062,
      "step": 5229
    },
    {
      "epoch": 9.947693770803614,
      "grad_norm": 0.30199721455574036,
      "learning_rate": 6.681486186090823e-05,
      "loss": 0.0926,
      "step": 5230
    },
    {
      "epoch": 9.949595815501665,
      "grad_norm": 0.3589622974395752,
      "learning_rate": 6.680851063829787e-05,
      "loss": 0.0914,
      "step": 5231
    },
    {
      "epoch": 9.951497860199714,
      "grad_norm": 0.32676953077316284,
      "learning_rate": 6.680215941568752e-05,
      "loss": 0.0971,
      "step": 5232
    },
    {
      "epoch": 9.953399904897765,
      "grad_norm": 0.23442213237285614,
      "learning_rate": 6.679580819307717e-05,
      "loss": 0.0597,
      "step": 5233
    },
    {
      "epoch": 9.955301949595816,
      "grad_norm": 0.29047128558158875,
      "learning_rate": 6.678945697046681e-05,
      "loss": 0.0693,
      "step": 5234
    },
    {
      "epoch": 9.957203994293867,
      "grad_norm": 0.17525753378868103,
      "learning_rate": 6.678310574785646e-05,
      "loss": 0.0552,
      "step": 5235
    },
    {
      "epoch": 9.959106038991916,
      "grad_norm": 0.276344358921051,
      "learning_rate": 6.677675452524611e-05,
      "loss": 0.0784,
      "step": 5236
    },
    {
      "epoch": 9.961008083689967,
      "grad_norm": 0.323887437582016,
      "learning_rate": 6.677040330263577e-05,
      "loss": 0.0734,
      "step": 5237
    },
    {
      "epoch": 9.962910128388017,
      "grad_norm": 0.25159624218940735,
      "learning_rate": 6.67640520800254e-05,
      "loss": 0.0683,
      "step": 5238
    },
    {
      "epoch": 9.964812173086067,
      "grad_norm": 0.23952990770339966,
      "learning_rate": 6.675770085741506e-05,
      "loss": 0.0635,
      "step": 5239
    },
    {
      "epoch": 9.966714217784117,
      "grad_norm": 0.21993881464004517,
      "learning_rate": 6.675134963480471e-05,
      "loss": 0.0741,
      "step": 5240
    },
    {
      "epoch": 9.968616262482168,
      "grad_norm": 0.2130298614501953,
      "learning_rate": 6.674499841219435e-05,
      "loss": 0.0696,
      "step": 5241
    },
    {
      "epoch": 9.97051830718022,
      "grad_norm": 0.3035607933998108,
      "learning_rate": 6.6738647189584e-05,
      "loss": 0.0913,
      "step": 5242
    },
    {
      "epoch": 9.972420351878268,
      "grad_norm": 0.34198522567749023,
      "learning_rate": 6.673229596697365e-05,
      "loss": 0.1082,
      "step": 5243
    },
    {
      "epoch": 9.97432239657632,
      "grad_norm": 0.2956897020339966,
      "learning_rate": 6.672594474436329e-05,
      "loss": 0.074,
      "step": 5244
    },
    {
      "epoch": 9.97622444127437,
      "grad_norm": 0.43758097290992737,
      "learning_rate": 6.671959352175294e-05,
      "loss": 0.0786,
      "step": 5245
    },
    {
      "epoch": 9.978126485972421,
      "grad_norm": 0.17064926028251648,
      "learning_rate": 6.671324229914259e-05,
      "loss": 0.0571,
      "step": 5246
    },
    {
      "epoch": 9.98002853067047,
      "grad_norm": 0.3719223737716675,
      "learning_rate": 6.670689107653224e-05,
      "loss": 0.0961,
      "step": 5247
    },
    {
      "epoch": 9.981930575368521,
      "grad_norm": 0.16983850300312042,
      "learning_rate": 6.670053985392188e-05,
      "loss": 0.0459,
      "step": 5248
    },
    {
      "epoch": 9.983832620066572,
      "grad_norm": 0.3689050078392029,
      "learning_rate": 6.669418863131152e-05,
      "loss": 0.099,
      "step": 5249
    },
    {
      "epoch": 9.985734664764623,
      "grad_norm": 0.35418006777763367,
      "learning_rate": 6.668783740870119e-05,
      "loss": 0.0761,
      "step": 5250
    },
    {
      "epoch": 10.00190204469805,
      "grad_norm": 0.39694878458976746,
      "learning_rate": 6.668148618609082e-05,
      "loss": 0.1313,
      "step": 5251
    },
    {
      "epoch": 10.0038040893961,
      "grad_norm": 0.6753419041633606,
      "learning_rate": 6.667513496348046e-05,
      "loss": 0.1505,
      "step": 5252
    },
    {
      "epoch": 10.00570613409415,
      "grad_norm": 0.46378642320632935,
      "learning_rate": 6.666878374087013e-05,
      "loss": 0.1148,
      "step": 5253
    },
    {
      "epoch": 10.007608178792202,
      "grad_norm": 0.342328280210495,
      "learning_rate": 6.666243251825977e-05,
      "loss": 0.1214,
      "step": 5254
    },
    {
      "epoch": 10.009510223490253,
      "grad_norm": 0.41636717319488525,
      "learning_rate": 6.665608129564942e-05,
      "loss": 0.1316,
      "step": 5255
    },
    {
      "epoch": 10.011412268188302,
      "grad_norm": 0.5482467412948608,
      "learning_rate": 6.664973007303906e-05,
      "loss": 0.1513,
      "step": 5256
    },
    {
      "epoch": 10.013314312886353,
      "grad_norm": 0.28404325246810913,
      "learning_rate": 6.664337885042871e-05,
      "loss": 0.1009,
      "step": 5257
    },
    {
      "epoch": 10.015216357584404,
      "grad_norm": 0.14674490690231323,
      "learning_rate": 6.663702762781836e-05,
      "loss": 0.0506,
      "step": 5258
    },
    {
      "epoch": 10.017118402282454,
      "grad_norm": 0.15933150053024292,
      "learning_rate": 6.6630676405208e-05,
      "loss": 0.0569,
      "step": 5259
    },
    {
      "epoch": 10.019020446980504,
      "grad_norm": 0.29720667004585266,
      "learning_rate": 6.662432518259766e-05,
      "loss": 0.0712,
      "step": 5260
    },
    {
      "epoch": 10.020922491678554,
      "grad_norm": 0.22701764106750488,
      "learning_rate": 6.66179739599873e-05,
      "loss": 0.0773,
      "step": 5261
    },
    {
      "epoch": 10.022824536376605,
      "grad_norm": 0.15102310478687286,
      "learning_rate": 6.661162273737694e-05,
      "loss": 0.057,
      "step": 5262
    },
    {
      "epoch": 10.024726581074654,
      "grad_norm": 0.184609517455101,
      "learning_rate": 6.660527151476659e-05,
      "loss": 0.0584,
      "step": 5263
    },
    {
      "epoch": 10.026628625772705,
      "grad_norm": 0.223251610994339,
      "learning_rate": 6.659892029215624e-05,
      "loss": 0.0465,
      "step": 5264
    },
    {
      "epoch": 10.028530670470756,
      "grad_norm": 0.2808699905872345,
      "learning_rate": 6.65925690695459e-05,
      "loss": 0.0836,
      "step": 5265
    },
    {
      "epoch": 10.030432715168807,
      "grad_norm": 0.3460269272327423,
      "learning_rate": 6.658621784693553e-05,
      "loss": 0.0767,
      "step": 5266
    },
    {
      "epoch": 10.032334759866856,
      "grad_norm": 0.24595172703266144,
      "learning_rate": 6.657986662432519e-05,
      "loss": 0.06,
      "step": 5267
    },
    {
      "epoch": 10.034236804564907,
      "grad_norm": 0.252651572227478,
      "learning_rate": 6.657351540171484e-05,
      "loss": 0.0565,
      "step": 5268
    },
    {
      "epoch": 10.036138849262958,
      "grad_norm": 0.23152634501457214,
      "learning_rate": 6.656716417910448e-05,
      "loss": 0.0701,
      "step": 5269
    },
    {
      "epoch": 10.038040893961009,
      "grad_norm": 0.17529435455799103,
      "learning_rate": 6.656081295649413e-05,
      "loss": 0.0527,
      "step": 5270
    },
    {
      "epoch": 10.039942938659058,
      "grad_norm": 0.1904911994934082,
      "learning_rate": 6.655446173388378e-05,
      "loss": 0.0861,
      "step": 5271
    },
    {
      "epoch": 10.041844983357109,
      "grad_norm": 0.35613206028938293,
      "learning_rate": 6.654811051127342e-05,
      "loss": 0.074,
      "step": 5272
    },
    {
      "epoch": 10.04374702805516,
      "grad_norm": 0.2638046145439148,
      "learning_rate": 6.654175928866307e-05,
      "loss": 0.117,
      "step": 5273
    },
    {
      "epoch": 10.045649072753209,
      "grad_norm": 0.21620406210422516,
      "learning_rate": 6.653540806605272e-05,
      "loss": 0.0586,
      "step": 5274
    },
    {
      "epoch": 10.04755111745126,
      "grad_norm": 0.16865353286266327,
      "learning_rate": 6.652905684344236e-05,
      "loss": 0.0665,
      "step": 5275
    },
    {
      "epoch": 10.04945316214931,
      "grad_norm": 0.15396840870380402,
      "learning_rate": 6.652270562083201e-05,
      "loss": 0.0542,
      "step": 5276
    },
    {
      "epoch": 10.051355206847362,
      "grad_norm": 0.3527628779411316,
      "learning_rate": 6.651635439822166e-05,
      "loss": 0.0813,
      "step": 5277
    },
    {
      "epoch": 10.05325725154541,
      "grad_norm": 0.16249553859233856,
      "learning_rate": 6.651000317561132e-05,
      "loss": 0.0546,
      "step": 5278
    },
    {
      "epoch": 10.055159296243461,
      "grad_norm": 0.38107484579086304,
      "learning_rate": 6.650365195300095e-05,
      "loss": 0.0803,
      "step": 5279
    },
    {
      "epoch": 10.057061340941512,
      "grad_norm": 0.14425760507583618,
      "learning_rate": 6.64973007303906e-05,
      "loss": 0.0411,
      "step": 5280
    },
    {
      "epoch": 10.058963385639563,
      "grad_norm": 0.20022647082805634,
      "learning_rate": 6.649094950778026e-05,
      "loss": 0.0641,
      "step": 5281
    },
    {
      "epoch": 10.060865430337612,
      "grad_norm": 0.2576226592063904,
      "learning_rate": 6.64845982851699e-05,
      "loss": 0.0779,
      "step": 5282
    },
    {
      "epoch": 10.062767475035663,
      "grad_norm": 0.23341166973114014,
      "learning_rate": 6.647824706255955e-05,
      "loss": 0.0769,
      "step": 5283
    },
    {
      "epoch": 10.064669519733714,
      "grad_norm": 0.20656423270702362,
      "learning_rate": 6.64718958399492e-05,
      "loss": 0.07,
      "step": 5284
    },
    {
      "epoch": 10.066571564431765,
      "grad_norm": 0.26069366931915283,
      "learning_rate": 6.646554461733884e-05,
      "loss": 0.0513,
      "step": 5285
    },
    {
      "epoch": 10.068473609129814,
      "grad_norm": 0.3721386194229126,
      "learning_rate": 6.645919339472849e-05,
      "loss": 0.0896,
      "step": 5286
    },
    {
      "epoch": 10.070375653827865,
      "grad_norm": 0.36856430768966675,
      "learning_rate": 6.645284217211813e-05,
      "loss": 0.0785,
      "step": 5287
    },
    {
      "epoch": 10.072277698525916,
      "grad_norm": 0.2893162965774536,
      "learning_rate": 6.644649094950778e-05,
      "loss": 0.0697,
      "step": 5288
    },
    {
      "epoch": 10.074179743223965,
      "grad_norm": 0.21396392583847046,
      "learning_rate": 6.644013972689743e-05,
      "loss": 0.0694,
      "step": 5289
    },
    {
      "epoch": 10.076081787922016,
      "grad_norm": 0.49254268407821655,
      "learning_rate": 6.643378850428707e-05,
      "loss": 0.0813,
      "step": 5290
    },
    {
      "epoch": 10.077983832620067,
      "grad_norm": 0.22749567031860352,
      "learning_rate": 6.642743728167674e-05,
      "loss": 0.059,
      "step": 5291
    },
    {
      "epoch": 10.079885877318118,
      "grad_norm": 0.25905197858810425,
      "learning_rate": 6.642108605906637e-05,
      "loss": 0.0504,
      "step": 5292
    },
    {
      "epoch": 10.081787922016167,
      "grad_norm": 0.24032166600227356,
      "learning_rate": 6.641473483645601e-05,
      "loss": 0.0757,
      "step": 5293
    },
    {
      "epoch": 10.083689966714218,
      "grad_norm": 0.34126365184783936,
      "learning_rate": 6.640838361384566e-05,
      "loss": 0.076,
      "step": 5294
    },
    {
      "epoch": 10.085592011412269,
      "grad_norm": 0.12431490421295166,
      "learning_rate": 6.640203239123532e-05,
      "loss": 0.0373,
      "step": 5295
    },
    {
      "epoch": 10.08749405611032,
      "grad_norm": 0.3214389681816101,
      "learning_rate": 6.639568116862497e-05,
      "loss": 0.0746,
      "step": 5296
    },
    {
      "epoch": 10.089396100808369,
      "grad_norm": 0.20561222732067108,
      "learning_rate": 6.63893299460146e-05,
      "loss": 0.0598,
      "step": 5297
    },
    {
      "epoch": 10.09129814550642,
      "grad_norm": 0.19043764472007751,
      "learning_rate": 6.638297872340426e-05,
      "loss": 0.0781,
      "step": 5298
    },
    {
      "epoch": 10.09320019020447,
      "grad_norm": 0.23050783574581146,
      "learning_rate": 6.637662750079391e-05,
      "loss": 0.0877,
      "step": 5299
    },
    {
      "epoch": 10.09510223490252,
      "grad_norm": 0.23546096682548523,
      "learning_rate": 6.637027627818355e-05,
      "loss": 0.063,
      "step": 5300
    },
    {
      "epoch": 10.09700427960057,
      "grad_norm": 0.1960730254650116,
      "learning_rate": 6.63639250555732e-05,
      "loss": 0.0688,
      "step": 5301
    },
    {
      "epoch": 10.098906324298621,
      "grad_norm": 0.2317197173833847,
      "learning_rate": 6.635757383296285e-05,
      "loss": 0.0726,
      "step": 5302
    },
    {
      "epoch": 10.100808368996672,
      "grad_norm": 0.14227913320064545,
      "learning_rate": 6.635122261035249e-05,
      "loss": 0.0646,
      "step": 5303
    },
    {
      "epoch": 10.102710413694721,
      "grad_norm": 0.17636631429195404,
      "learning_rate": 6.634487138774214e-05,
      "loss": 0.0486,
      "step": 5304
    },
    {
      "epoch": 10.104612458392772,
      "grad_norm": 0.2111407071352005,
      "learning_rate": 6.63385201651318e-05,
      "loss": 0.0566,
      "step": 5305
    },
    {
      "epoch": 10.106514503090823,
      "grad_norm": 0.17445050179958344,
      "learning_rate": 6.633216894252143e-05,
      "loss": 0.0696,
      "step": 5306
    },
    {
      "epoch": 10.108416547788874,
      "grad_norm": 0.20955027639865875,
      "learning_rate": 6.632581771991108e-05,
      "loss": 0.0662,
      "step": 5307
    },
    {
      "epoch": 10.110318592486923,
      "grad_norm": 0.2103997766971588,
      "learning_rate": 6.631946649730074e-05,
      "loss": 0.0875,
      "step": 5308
    },
    {
      "epoch": 10.112220637184974,
      "grad_norm": 0.1805890053510666,
      "learning_rate": 6.631311527469039e-05,
      "loss": 0.0602,
      "step": 5309
    },
    {
      "epoch": 10.114122681883025,
      "grad_norm": 0.28872668743133545,
      "learning_rate": 6.630676405208003e-05,
      "loss": 0.065,
      "step": 5310
    },
    {
      "epoch": 10.116024726581074,
      "grad_norm": 0.2273848056793213,
      "learning_rate": 6.630041282946968e-05,
      "loss": 0.0679,
      "step": 5311
    },
    {
      "epoch": 10.117926771279125,
      "grad_norm": 0.2823135554790497,
      "learning_rate": 6.629406160685933e-05,
      "loss": 0.0738,
      "step": 5312
    },
    {
      "epoch": 10.119828815977176,
      "grad_norm": 0.2553102672100067,
      "learning_rate": 6.628771038424897e-05,
      "loss": 0.0748,
      "step": 5313
    },
    {
      "epoch": 10.121730860675227,
      "grad_norm": 0.2882140278816223,
      "learning_rate": 6.628135916163862e-05,
      "loss": 0.0672,
      "step": 5314
    },
    {
      "epoch": 10.123632905373276,
      "grad_norm": 0.2157353013753891,
      "learning_rate": 6.627500793902827e-05,
      "loss": 0.0663,
      "step": 5315
    },
    {
      "epoch": 10.125534950071327,
      "grad_norm": 0.20802216231822968,
      "learning_rate": 6.626865671641791e-05,
      "loss": 0.067,
      "step": 5316
    },
    {
      "epoch": 10.127436994769377,
      "grad_norm": 0.20221203565597534,
      "learning_rate": 6.626230549380756e-05,
      "loss": 0.0742,
      "step": 5317
    },
    {
      "epoch": 10.129339039467428,
      "grad_norm": 0.290554404258728,
      "learning_rate": 6.625595427119721e-05,
      "loss": 0.0563,
      "step": 5318
    },
    {
      "epoch": 10.131241084165477,
      "grad_norm": 0.2569425106048584,
      "learning_rate": 6.624960304858687e-05,
      "loss": 0.0677,
      "step": 5319
    },
    {
      "epoch": 10.133143128863528,
      "grad_norm": 0.20562361180782318,
      "learning_rate": 6.62432518259765e-05,
      "loss": 0.0607,
      "step": 5320
    },
    {
      "epoch": 10.13504517356158,
      "grad_norm": 0.17843393981456757,
      "learning_rate": 6.623690060336614e-05,
      "loss": 0.0944,
      "step": 5321
    },
    {
      "epoch": 10.136947218259628,
      "grad_norm": 0.2664887309074402,
      "learning_rate": 6.623054938075581e-05,
      "loss": 0.0878,
      "step": 5322
    },
    {
      "epoch": 10.13884926295768,
      "grad_norm": 0.2143498659133911,
      "learning_rate": 6.622419815814545e-05,
      "loss": 0.0678,
      "step": 5323
    },
    {
      "epoch": 10.14075130765573,
      "grad_norm": 0.14991843700408936,
      "learning_rate": 6.621784693553508e-05,
      "loss": 0.0695,
      "step": 5324
    },
    {
      "epoch": 10.142653352353781,
      "grad_norm": 0.22047267854213715,
      "learning_rate": 6.621149571292475e-05,
      "loss": 0.0817,
      "step": 5325
    },
    {
      "epoch": 10.14455539705183,
      "grad_norm": 0.20937331020832062,
      "learning_rate": 6.620514449031439e-05,
      "loss": 0.0592,
      "step": 5326
    },
    {
      "epoch": 10.146457441749881,
      "grad_norm": 0.2746511399745941,
      "learning_rate": 6.619879326770404e-05,
      "loss": 0.0583,
      "step": 5327
    },
    {
      "epoch": 10.148359486447932,
      "grad_norm": 0.23711417615413666,
      "learning_rate": 6.619244204509368e-05,
      "loss": 0.0739,
      "step": 5328
    },
    {
      "epoch": 10.150261531145983,
      "grad_norm": 0.4201519787311554,
      "learning_rate": 6.618609082248333e-05,
      "loss": 0.0908,
      "step": 5329
    },
    {
      "epoch": 10.152163575844032,
      "grad_norm": 0.1867586076259613,
      "learning_rate": 6.617973959987298e-05,
      "loss": 0.0488,
      "step": 5330
    },
    {
      "epoch": 10.154065620542083,
      "grad_norm": 0.22101399302482605,
      "learning_rate": 6.617338837726262e-05,
      "loss": 0.0532,
      "step": 5331
    },
    {
      "epoch": 10.155967665240134,
      "grad_norm": 0.18868492543697357,
      "learning_rate": 6.616703715465228e-05,
      "loss": 0.0724,
      "step": 5332
    },
    {
      "epoch": 10.157869709938183,
      "grad_norm": 0.23109176754951477,
      "learning_rate": 6.616068593204192e-05,
      "loss": 0.074,
      "step": 5333
    },
    {
      "epoch": 10.159771754636234,
      "grad_norm": 0.23569205403327942,
      "learning_rate": 6.615433470943156e-05,
      "loss": 0.0833,
      "step": 5334
    },
    {
      "epoch": 10.161673799334284,
      "grad_norm": 0.350978285074234,
      "learning_rate": 6.614798348682121e-05,
      "loss": 0.0692,
      "step": 5335
    },
    {
      "epoch": 10.163575844032335,
      "grad_norm": 0.365357905626297,
      "learning_rate": 6.614163226421087e-05,
      "loss": 0.0927,
      "step": 5336
    },
    {
      "epoch": 10.165477888730384,
      "grad_norm": 0.1944890320301056,
      "learning_rate": 6.613528104160052e-05,
      "loss": 0.0858,
      "step": 5337
    },
    {
      "epoch": 10.167379933428435,
      "grad_norm": 0.20101040601730347,
      "learning_rate": 6.612892981899016e-05,
      "loss": 0.0656,
      "step": 5338
    },
    {
      "epoch": 10.169281978126486,
      "grad_norm": 0.2939762473106384,
      "learning_rate": 6.612257859637981e-05,
      "loss": 0.0918,
      "step": 5339
    },
    {
      "epoch": 10.171184022824537,
      "grad_norm": 0.22608570754528046,
      "learning_rate": 6.611622737376946e-05,
      "loss": 0.0675,
      "step": 5340
    },
    {
      "epoch": 10.173086067522586,
      "grad_norm": 0.49089139699935913,
      "learning_rate": 6.61098761511591e-05,
      "loss": 0.072,
      "step": 5341
    },
    {
      "epoch": 10.174988112220637,
      "grad_norm": 0.28007015585899353,
      "learning_rate": 6.610352492854875e-05,
      "loss": 0.0615,
      "step": 5342
    },
    {
      "epoch": 10.176890156918688,
      "grad_norm": 0.21805545687675476,
      "learning_rate": 6.60971737059384e-05,
      "loss": 0.061,
      "step": 5343
    },
    {
      "epoch": 10.178792201616737,
      "grad_norm": 0.2612074613571167,
      "learning_rate": 6.609082248332804e-05,
      "loss": 0.0716,
      "step": 5344
    },
    {
      "epoch": 10.180694246314788,
      "grad_norm": 0.1996293067932129,
      "learning_rate": 6.608447126071769e-05,
      "loss": 0.0743,
      "step": 5345
    },
    {
      "epoch": 10.182596291012839,
      "grad_norm": 0.18069012463092804,
      "learning_rate": 6.607812003810734e-05,
      "loss": 0.0694,
      "step": 5346
    },
    {
      "epoch": 10.18449833571089,
      "grad_norm": 0.35819369554519653,
      "learning_rate": 6.607176881549698e-05,
      "loss": 0.0903,
      "step": 5347
    },
    {
      "epoch": 10.186400380408939,
      "grad_norm": 0.292188435792923,
      "learning_rate": 6.606541759288663e-05,
      "loss": 0.0868,
      "step": 5348
    },
    {
      "epoch": 10.18830242510699,
      "grad_norm": 0.2514616847038269,
      "learning_rate": 6.605906637027628e-05,
      "loss": 0.0717,
      "step": 5349
    },
    {
      "epoch": 10.19020446980504,
      "grad_norm": 0.2554888427257538,
      "learning_rate": 6.605271514766594e-05,
      "loss": 0.072,
      "step": 5350
    },
    {
      "epoch": 10.192106514503092,
      "grad_norm": 0.24336287379264832,
      "learning_rate": 6.604636392505558e-05,
      "loss": 0.0741,
      "step": 5351
    },
    {
      "epoch": 10.19400855920114,
      "grad_norm": 0.21117272973060608,
      "learning_rate": 6.604001270244521e-05,
      "loss": 0.0476,
      "step": 5352
    },
    {
      "epoch": 10.195910603899192,
      "grad_norm": 0.22045139968395233,
      "learning_rate": 6.603366147983488e-05,
      "loss": 0.0531,
      "step": 5353
    },
    {
      "epoch": 10.197812648597242,
      "grad_norm": 0.19895130395889282,
      "learning_rate": 6.602731025722452e-05,
      "loss": 0.0552,
      "step": 5354
    },
    {
      "epoch": 10.199714693295292,
      "grad_norm": 0.22562842071056366,
      "learning_rate": 6.602095903461417e-05,
      "loss": 0.0652,
      "step": 5355
    },
    {
      "epoch": 10.201616737993342,
      "grad_norm": 0.2797020375728607,
      "learning_rate": 6.601460781200382e-05,
      "loss": 0.0789,
      "step": 5356
    },
    {
      "epoch": 10.203518782691393,
      "grad_norm": 0.1523316651582718,
      "learning_rate": 6.600825658939346e-05,
      "loss": 0.0426,
      "step": 5357
    },
    {
      "epoch": 10.205420827389444,
      "grad_norm": 0.2565470337867737,
      "learning_rate": 6.600190536678311e-05,
      "loss": 0.0659,
      "step": 5358
    },
    {
      "epoch": 10.207322872087493,
      "grad_norm": 0.18802964687347412,
      "learning_rate": 6.599555414417275e-05,
      "loss": 0.0657,
      "step": 5359
    },
    {
      "epoch": 10.209224916785544,
      "grad_norm": 0.20242714881896973,
      "learning_rate": 6.59892029215624e-05,
      "loss": 0.0735,
      "step": 5360
    },
    {
      "epoch": 10.211126961483595,
      "grad_norm": 0.11940667778253555,
      "learning_rate": 6.598285169895205e-05,
      "loss": 0.0642,
      "step": 5361
    },
    {
      "epoch": 10.213029006181646,
      "grad_norm": 0.18162617087364197,
      "learning_rate": 6.597650047634169e-05,
      "loss": 0.0558,
      "step": 5362
    },
    {
      "epoch": 10.214931050879695,
      "grad_norm": 0.24901486933231354,
      "learning_rate": 6.597014925373136e-05,
      "loss": 0.0716,
      "step": 5363
    },
    {
      "epoch": 10.216833095577746,
      "grad_norm": 0.21373523771762848,
      "learning_rate": 6.5963798031121e-05,
      "loss": 0.07,
      "step": 5364
    },
    {
      "epoch": 10.218735140275797,
      "grad_norm": 0.3318414092063904,
      "learning_rate": 6.595744680851063e-05,
      "loss": 0.0876,
      "step": 5365
    },
    {
      "epoch": 10.220637184973846,
      "grad_norm": 0.25907251238822937,
      "learning_rate": 6.595109558590028e-05,
      "loss": 0.0788,
      "step": 5366
    },
    {
      "epoch": 10.222539229671897,
      "grad_norm": 0.16159944236278534,
      "learning_rate": 6.594474436328994e-05,
      "loss": 0.0694,
      "step": 5367
    },
    {
      "epoch": 10.224441274369948,
      "grad_norm": 0.21074874699115753,
      "learning_rate": 6.593839314067959e-05,
      "loss": 0.0784,
      "step": 5368
    },
    {
      "epoch": 10.226343319067999,
      "grad_norm": 0.2328321486711502,
      "learning_rate": 6.593204191806923e-05,
      "loss": 0.0671,
      "step": 5369
    },
    {
      "epoch": 10.228245363766048,
      "grad_norm": 0.39634034037590027,
      "learning_rate": 6.592569069545888e-05,
      "loss": 0.0758,
      "step": 5370
    },
    {
      "epoch": 10.230147408464099,
      "grad_norm": 0.25715863704681396,
      "learning_rate": 6.591933947284853e-05,
      "loss": 0.0527,
      "step": 5371
    },
    {
      "epoch": 10.23204945316215,
      "grad_norm": 0.2564283013343811,
      "learning_rate": 6.591298825023817e-05,
      "loss": 0.0619,
      "step": 5372
    },
    {
      "epoch": 10.2339514978602,
      "grad_norm": 0.24235618114471436,
      "learning_rate": 6.590663702762782e-05,
      "loss": 0.0626,
      "step": 5373
    },
    {
      "epoch": 10.23585354255825,
      "grad_norm": 0.2732139825820923,
      "learning_rate": 6.590028580501747e-05,
      "loss": 0.068,
      "step": 5374
    },
    {
      "epoch": 10.2377555872563,
      "grad_norm": 0.2638366222381592,
      "learning_rate": 6.589393458240711e-05,
      "loss": 0.0578,
      "step": 5375
    },
    {
      "epoch": 10.239657631954351,
      "grad_norm": 0.2025076001882553,
      "learning_rate": 6.588758335979676e-05,
      "loss": 0.0549,
      "step": 5376
    },
    {
      "epoch": 10.241559676652402,
      "grad_norm": 0.24390022456645966,
      "learning_rate": 6.588123213718641e-05,
      "loss": 0.0881,
      "step": 5377
    },
    {
      "epoch": 10.243461721350451,
      "grad_norm": 0.43563568592071533,
      "learning_rate": 6.587488091457605e-05,
      "loss": 0.0741,
      "step": 5378
    },
    {
      "epoch": 10.245363766048502,
      "grad_norm": 0.18210797011852264,
      "learning_rate": 6.58685296919657e-05,
      "loss": 0.0585,
      "step": 5379
    },
    {
      "epoch": 10.247265810746553,
      "grad_norm": 0.17295345664024353,
      "learning_rate": 6.586217846935536e-05,
      "loss": 0.0496,
      "step": 5380
    },
    {
      "epoch": 10.249167855444602,
      "grad_norm": 0.20748892426490784,
      "learning_rate": 6.585582724674501e-05,
      "loss": 0.0505,
      "step": 5381
    },
    {
      "epoch": 10.251069900142653,
      "grad_norm": 0.3153904378414154,
      "learning_rate": 6.584947602413465e-05,
      "loss": 0.0653,
      "step": 5382
    },
    {
      "epoch": 10.252971944840704,
      "grad_norm": 0.16658660769462585,
      "learning_rate": 6.58431248015243e-05,
      "loss": 0.0669,
      "step": 5383
    },
    {
      "epoch": 10.254873989538755,
      "grad_norm": 0.22813019156455994,
      "learning_rate": 6.583677357891395e-05,
      "loss": 0.0519,
      "step": 5384
    },
    {
      "epoch": 10.256776034236804,
      "grad_norm": 0.19745177030563354,
      "learning_rate": 6.583042235630359e-05,
      "loss": 0.0759,
      "step": 5385
    },
    {
      "epoch": 10.258678078934855,
      "grad_norm": 0.2388034611940384,
      "learning_rate": 6.582407113369324e-05,
      "loss": 0.0714,
      "step": 5386
    },
    {
      "epoch": 10.260580123632906,
      "grad_norm": 0.2057228535413742,
      "learning_rate": 6.581771991108289e-05,
      "loss": 0.0641,
      "step": 5387
    },
    {
      "epoch": 10.262482168330957,
      "grad_norm": 0.2861918807029724,
      "learning_rate": 6.581136868847253e-05,
      "loss": 0.082,
      "step": 5388
    },
    {
      "epoch": 10.264384213029006,
      "grad_norm": 0.26108086109161377,
      "learning_rate": 6.580501746586218e-05,
      "loss": 0.0703,
      "step": 5389
    },
    {
      "epoch": 10.266286257727057,
      "grad_norm": 0.331620454788208,
      "learning_rate": 6.579866624325182e-05,
      "loss": 0.0836,
      "step": 5390
    },
    {
      "epoch": 10.268188302425107,
      "grad_norm": 0.24640053510665894,
      "learning_rate": 6.579231502064149e-05,
      "loss": 0.051,
      "step": 5391
    },
    {
      "epoch": 10.270090347123157,
      "grad_norm": 0.27249953150749207,
      "learning_rate": 6.578596379803112e-05,
      "loss": 0.0718,
      "step": 5392
    },
    {
      "epoch": 10.271992391821207,
      "grad_norm": 0.28086280822753906,
      "learning_rate": 6.577961257542076e-05,
      "loss": 0.083,
      "step": 5393
    },
    {
      "epoch": 10.273894436519258,
      "grad_norm": 0.3512130081653595,
      "learning_rate": 6.577326135281043e-05,
      "loss": 0.0763,
      "step": 5394
    },
    {
      "epoch": 10.27579648121731,
      "grad_norm": 0.14599667489528656,
      "learning_rate": 6.576691013020007e-05,
      "loss": 0.0707,
      "step": 5395
    },
    {
      "epoch": 10.277698525915358,
      "grad_norm": 0.19556157290935516,
      "learning_rate": 6.57605589075897e-05,
      "loss": 0.052,
      "step": 5396
    },
    {
      "epoch": 10.27960057061341,
      "grad_norm": 0.24162036180496216,
      "learning_rate": 6.575420768497936e-05,
      "loss": 0.0641,
      "step": 5397
    },
    {
      "epoch": 10.28150261531146,
      "grad_norm": 0.22074343264102936,
      "learning_rate": 6.574785646236901e-05,
      "loss": 0.0683,
      "step": 5398
    },
    {
      "epoch": 10.283404660009511,
      "grad_norm": 0.18643182516098022,
      "learning_rate": 6.574150523975866e-05,
      "loss": 0.0964,
      "step": 5399
    },
    {
      "epoch": 10.28530670470756,
      "grad_norm": 0.22997966408729553,
      "learning_rate": 6.57351540171483e-05,
      "loss": 0.0814,
      "step": 5400
    },
    {
      "epoch": 10.287208749405611,
      "grad_norm": 0.11969192326068878,
      "learning_rate": 6.572880279453795e-05,
      "loss": 0.0678,
      "step": 5401
    },
    {
      "epoch": 10.289110794103662,
      "grad_norm": 0.3286897838115692,
      "learning_rate": 6.57224515719276e-05,
      "loss": 0.0916,
      "step": 5402
    },
    {
      "epoch": 10.291012838801711,
      "grad_norm": 0.2948076128959656,
      "learning_rate": 6.571610034931724e-05,
      "loss": 0.0763,
      "step": 5403
    },
    {
      "epoch": 10.292914883499762,
      "grad_norm": 0.2712877094745636,
      "learning_rate": 6.570974912670689e-05,
      "loss": 0.0816,
      "step": 5404
    },
    {
      "epoch": 10.294816928197813,
      "grad_norm": 0.19706730544567108,
      "learning_rate": 6.570339790409654e-05,
      "loss": 0.0604,
      "step": 5405
    },
    {
      "epoch": 10.296718972895864,
      "grad_norm": 0.27862757444381714,
      "learning_rate": 6.569704668148618e-05,
      "loss": 0.0643,
      "step": 5406
    },
    {
      "epoch": 10.298621017593913,
      "grad_norm": 0.3971151113510132,
      "learning_rate": 6.569069545887583e-05,
      "loss": 0.093,
      "step": 5407
    },
    {
      "epoch": 10.300523062291964,
      "grad_norm": 0.2490602731704712,
      "learning_rate": 6.568434423626549e-05,
      "loss": 0.0675,
      "step": 5408
    },
    {
      "epoch": 10.302425106990015,
      "grad_norm": 0.19942264258861542,
      "learning_rate": 6.567799301365514e-05,
      "loss": 0.0596,
      "step": 5409
    },
    {
      "epoch": 10.304327151688065,
      "grad_norm": 0.17037898302078247,
      "learning_rate": 6.567164179104478e-05,
      "loss": 0.076,
      "step": 5410
    },
    {
      "epoch": 10.306229196386115,
      "grad_norm": 0.20393238961696625,
      "learning_rate": 6.566529056843443e-05,
      "loss": 0.0594,
      "step": 5411
    },
    {
      "epoch": 10.308131241084165,
      "grad_norm": 0.2775140106678009,
      "learning_rate": 6.565893934582408e-05,
      "loss": 0.0589,
      "step": 5412
    },
    {
      "epoch": 10.310033285782216,
      "grad_norm": 0.25280818343162537,
      "learning_rate": 6.565258812321372e-05,
      "loss": 0.0616,
      "step": 5413
    },
    {
      "epoch": 10.311935330480265,
      "grad_norm": 0.22769461572170258,
      "learning_rate": 6.564623690060337e-05,
      "loss": 0.0916,
      "step": 5414
    },
    {
      "epoch": 10.313837375178316,
      "grad_norm": 0.2882596552371979,
      "learning_rate": 6.563988567799302e-05,
      "loss": 0.0752,
      "step": 5415
    },
    {
      "epoch": 10.315739419876367,
      "grad_norm": 0.43357282876968384,
      "learning_rate": 6.563353445538266e-05,
      "loss": 0.0798,
      "step": 5416
    },
    {
      "epoch": 10.317641464574418,
      "grad_norm": 0.26240670680999756,
      "learning_rate": 6.562718323277231e-05,
      "loss": 0.0672,
      "step": 5417
    },
    {
      "epoch": 10.319543509272467,
      "grad_norm": 0.24738280475139618,
      "learning_rate": 6.562083201016196e-05,
      "loss": 0.0774,
      "step": 5418
    },
    {
      "epoch": 10.321445553970518,
      "grad_norm": 0.1921759992837906,
      "learning_rate": 6.56144807875516e-05,
      "loss": 0.0616,
      "step": 5419
    },
    {
      "epoch": 10.323347598668569,
      "grad_norm": 0.1570550799369812,
      "learning_rate": 6.560812956494125e-05,
      "loss": 0.0592,
      "step": 5420
    },
    {
      "epoch": 10.32524964336662,
      "grad_norm": 0.21802720427513123,
      "learning_rate": 6.56017783423309e-05,
      "loss": 0.0929,
      "step": 5421
    },
    {
      "epoch": 10.327151688064669,
      "grad_norm": 0.17816253006458282,
      "learning_rate": 6.559542711972056e-05,
      "loss": 0.072,
      "step": 5422
    },
    {
      "epoch": 10.32905373276272,
      "grad_norm": 0.14564338326454163,
      "learning_rate": 6.55890758971102e-05,
      "loss": 0.0412,
      "step": 5423
    },
    {
      "epoch": 10.33095577746077,
      "grad_norm": 0.1930164396762848,
      "learning_rate": 6.558272467449983e-05,
      "loss": 0.0659,
      "step": 5424
    },
    {
      "epoch": 10.332857822158822,
      "grad_norm": 0.3501245379447937,
      "learning_rate": 6.55763734518895e-05,
      "loss": 0.0617,
      "step": 5425
    },
    {
      "epoch": 10.33475986685687,
      "grad_norm": 0.27747026085853577,
      "learning_rate": 6.557002222927914e-05,
      "loss": 0.0842,
      "step": 5426
    },
    {
      "epoch": 10.336661911554922,
      "grad_norm": 0.24909347295761108,
      "learning_rate": 6.556367100666879e-05,
      "loss": 0.0702,
      "step": 5427
    },
    {
      "epoch": 10.338563956252973,
      "grad_norm": 0.2601378560066223,
      "learning_rate": 6.555731978405844e-05,
      "loss": 0.0732,
      "step": 5428
    },
    {
      "epoch": 10.340466000951022,
      "grad_norm": 0.18222792446613312,
      "learning_rate": 6.555096856144808e-05,
      "loss": 0.0512,
      "step": 5429
    },
    {
      "epoch": 10.342368045649073,
      "grad_norm": 0.1804533451795578,
      "learning_rate": 6.554461733883773e-05,
      "loss": 0.0601,
      "step": 5430
    },
    {
      "epoch": 10.344270090347123,
      "grad_norm": 0.239121213555336,
      "learning_rate": 6.553826611622737e-05,
      "loss": 0.0744,
      "step": 5431
    },
    {
      "epoch": 10.346172135045174,
      "grad_norm": 0.25100961327552795,
      "learning_rate": 6.553191489361702e-05,
      "loss": 0.07,
      "step": 5432
    },
    {
      "epoch": 10.348074179743223,
      "grad_norm": 0.19875814020633698,
      "learning_rate": 6.552556367100667e-05,
      "loss": 0.0781,
      "step": 5433
    },
    {
      "epoch": 10.349976224441274,
      "grad_norm": 0.19966357946395874,
      "learning_rate": 6.551921244839631e-05,
      "loss": 0.0744,
      "step": 5434
    },
    {
      "epoch": 10.351878269139325,
      "grad_norm": 0.2350044846534729,
      "learning_rate": 6.551286122578598e-05,
      "loss": 0.0707,
      "step": 5435
    },
    {
      "epoch": 10.353780313837376,
      "grad_norm": 0.21448549628257751,
      "learning_rate": 6.550651000317562e-05,
      "loss": 0.0725,
      "step": 5436
    },
    {
      "epoch": 10.355682358535425,
      "grad_norm": 0.26761433482170105,
      "learning_rate": 6.550015878056525e-05,
      "loss": 0.0523,
      "step": 5437
    },
    {
      "epoch": 10.357584403233476,
      "grad_norm": 0.30378443002700806,
      "learning_rate": 6.54938075579549e-05,
      "loss": 0.0584,
      "step": 5438
    },
    {
      "epoch": 10.359486447931527,
      "grad_norm": 0.2591283321380615,
      "learning_rate": 6.548745633534456e-05,
      "loss": 0.078,
      "step": 5439
    },
    {
      "epoch": 10.361388492629576,
      "grad_norm": 0.2300068587064743,
      "learning_rate": 6.548110511273421e-05,
      "loss": 0.0706,
      "step": 5440
    },
    {
      "epoch": 10.363290537327627,
      "grad_norm": 0.2989126741886139,
      "learning_rate": 6.547475389012385e-05,
      "loss": 0.0587,
      "step": 5441
    },
    {
      "epoch": 10.365192582025678,
      "grad_norm": 0.2825726568698883,
      "learning_rate": 6.54684026675135e-05,
      "loss": 0.0589,
      "step": 5442
    },
    {
      "epoch": 10.367094626723729,
      "grad_norm": 0.24061256647109985,
      "learning_rate": 6.546205144490315e-05,
      "loss": 0.0759,
      "step": 5443
    },
    {
      "epoch": 10.368996671421778,
      "grad_norm": 0.31294533610343933,
      "learning_rate": 6.545570022229279e-05,
      "loss": 0.0846,
      "step": 5444
    },
    {
      "epoch": 10.370898716119829,
      "grad_norm": 0.13745763897895813,
      "learning_rate": 6.544934899968244e-05,
      "loss": 0.065,
      "step": 5445
    },
    {
      "epoch": 10.37280076081788,
      "grad_norm": 0.33087965846061707,
      "learning_rate": 6.54429977770721e-05,
      "loss": 0.0744,
      "step": 5446
    },
    {
      "epoch": 10.37470280551593,
      "grad_norm": 0.16575999557971954,
      "learning_rate": 6.543664655446173e-05,
      "loss": 0.0637,
      "step": 5447
    },
    {
      "epoch": 10.37660485021398,
      "grad_norm": 0.17727252840995789,
      "learning_rate": 6.543029533185138e-05,
      "loss": 0.0883,
      "step": 5448
    },
    {
      "epoch": 10.37850689491203,
      "grad_norm": 0.29931455850601196,
      "learning_rate": 6.542394410924104e-05,
      "loss": 0.0877,
      "step": 5449
    },
    {
      "epoch": 10.380408939610081,
      "grad_norm": 0.1964782327413559,
      "learning_rate": 6.541759288663067e-05,
      "loss": 0.0585,
      "step": 5450
    },
    {
      "epoch": 10.38231098430813,
      "grad_norm": 0.2581721842288971,
      "learning_rate": 6.541124166402033e-05,
      "loss": 0.0549,
      "step": 5451
    },
    {
      "epoch": 10.384213029006181,
      "grad_norm": 0.2323419451713562,
      "learning_rate": 6.540489044140998e-05,
      "loss": 0.0675,
      "step": 5452
    },
    {
      "epoch": 10.386115073704232,
      "grad_norm": 0.3017885684967041,
      "learning_rate": 6.539853921879963e-05,
      "loss": 0.0737,
      "step": 5453
    },
    {
      "epoch": 10.388017118402283,
      "grad_norm": 0.2838028073310852,
      "learning_rate": 6.539218799618927e-05,
      "loss": 0.0706,
      "step": 5454
    },
    {
      "epoch": 10.389919163100332,
      "grad_norm": 0.20796798169612885,
      "learning_rate": 6.53858367735789e-05,
      "loss": 0.0546,
      "step": 5455
    },
    {
      "epoch": 10.391821207798383,
      "grad_norm": 0.25365984439849854,
      "learning_rate": 6.537948555096857e-05,
      "loss": 0.0914,
      "step": 5456
    },
    {
      "epoch": 10.393723252496434,
      "grad_norm": 0.2627136707305908,
      "learning_rate": 6.537313432835821e-05,
      "loss": 0.0732,
      "step": 5457
    },
    {
      "epoch": 10.395625297194485,
      "grad_norm": 0.16268092393875122,
      "learning_rate": 6.536678310574786e-05,
      "loss": 0.0554,
      "step": 5458
    },
    {
      "epoch": 10.397527341892534,
      "grad_norm": 0.1628711223602295,
      "learning_rate": 6.536043188313751e-05,
      "loss": 0.0719,
      "step": 5459
    },
    {
      "epoch": 10.399429386590585,
      "grad_norm": 0.19809551537036896,
      "learning_rate": 6.535408066052715e-05,
      "loss": 0.0709,
      "step": 5460
    },
    {
      "epoch": 10.401331431288636,
      "grad_norm": 0.14292262494564056,
      "learning_rate": 6.53477294379168e-05,
      "loss": 0.0445,
      "step": 5461
    },
    {
      "epoch": 10.403233475986685,
      "grad_norm": 0.25781774520874023,
      "learning_rate": 6.534137821530644e-05,
      "loss": 0.0749,
      "step": 5462
    },
    {
      "epoch": 10.405135520684736,
      "grad_norm": 0.3060111701488495,
      "learning_rate": 6.533502699269611e-05,
      "loss": 0.0776,
      "step": 5463
    },
    {
      "epoch": 10.407037565382787,
      "grad_norm": 0.2065560519695282,
      "learning_rate": 6.532867577008575e-05,
      "loss": 0.0903,
      "step": 5464
    },
    {
      "epoch": 10.408939610080838,
      "grad_norm": 0.3038816452026367,
      "learning_rate": 6.532232454747538e-05,
      "loss": 0.0803,
      "step": 5465
    },
    {
      "epoch": 10.410841654778887,
      "grad_norm": 0.19374610483646393,
      "learning_rate": 6.531597332486505e-05,
      "loss": 0.0586,
      "step": 5466
    },
    {
      "epoch": 10.412743699476938,
      "grad_norm": 0.3853093981742859,
      "learning_rate": 6.530962210225469e-05,
      "loss": 0.0857,
      "step": 5467
    },
    {
      "epoch": 10.414645744174988,
      "grad_norm": 0.2604793608188629,
      "learning_rate": 6.530327087964433e-05,
      "loss": 0.0683,
      "step": 5468
    },
    {
      "epoch": 10.41654778887304,
      "grad_norm": 0.2732802629470825,
      "learning_rate": 6.529691965703398e-05,
      "loss": 0.079,
      "step": 5469
    },
    {
      "epoch": 10.418449833571088,
      "grad_norm": 0.2541634738445282,
      "learning_rate": 6.529056843442363e-05,
      "loss": 0.0939,
      "step": 5470
    },
    {
      "epoch": 10.42035187826914,
      "grad_norm": 0.23742996156215668,
      "learning_rate": 6.528421721181328e-05,
      "loss": 0.0937,
      "step": 5471
    },
    {
      "epoch": 10.42225392296719,
      "grad_norm": 0.3552636206150055,
      "learning_rate": 6.527786598920292e-05,
      "loss": 0.108,
      "step": 5472
    },
    {
      "epoch": 10.42415596766524,
      "grad_norm": 0.17072491347789764,
      "learning_rate": 6.527151476659257e-05,
      "loss": 0.0529,
      "step": 5473
    },
    {
      "epoch": 10.42605801236329,
      "grad_norm": 0.36401453614234924,
      "learning_rate": 6.526516354398222e-05,
      "loss": 0.0853,
      "step": 5474
    },
    {
      "epoch": 10.427960057061341,
      "grad_norm": 0.27907583117485046,
      "learning_rate": 6.525881232137186e-05,
      "loss": 0.0653,
      "step": 5475
    },
    {
      "epoch": 10.429862101759392,
      "grad_norm": 0.13606156408786774,
      "learning_rate": 6.525246109876151e-05,
      "loss": 0.0469,
      "step": 5476
    },
    {
      "epoch": 10.431764146457441,
      "grad_norm": 0.23634037375450134,
      "learning_rate": 6.524610987615117e-05,
      "loss": 0.0691,
      "step": 5477
    },
    {
      "epoch": 10.433666191155492,
      "grad_norm": 0.35563454031944275,
      "learning_rate": 6.52397586535408e-05,
      "loss": 0.0719,
      "step": 5478
    },
    {
      "epoch": 10.435568235853543,
      "grad_norm": 0.3277428150177002,
      "learning_rate": 6.523340743093046e-05,
      "loss": 0.0521,
      "step": 5479
    },
    {
      "epoch": 10.437470280551594,
      "grad_norm": 0.32706040143966675,
      "learning_rate": 6.522705620832011e-05,
      "loss": 0.0743,
      "step": 5480
    },
    {
      "epoch": 10.439372325249643,
      "grad_norm": 0.44523194432258606,
      "learning_rate": 6.522070498570976e-05,
      "loss": 0.0683,
      "step": 5481
    },
    {
      "epoch": 10.441274369947694,
      "grad_norm": 0.24652712047100067,
      "learning_rate": 6.52143537630994e-05,
      "loss": 0.0589,
      "step": 5482
    },
    {
      "epoch": 10.443176414645745,
      "grad_norm": 0.27444836497306824,
      "learning_rate": 6.520800254048905e-05,
      "loss": 0.0553,
      "step": 5483
    },
    {
      "epoch": 10.445078459343794,
      "grad_norm": 0.17975351214408875,
      "learning_rate": 6.52016513178787e-05,
      "loss": 0.0702,
      "step": 5484
    },
    {
      "epoch": 10.446980504041845,
      "grad_norm": 0.3462238311767578,
      "learning_rate": 6.519530009526834e-05,
      "loss": 0.0834,
      "step": 5485
    },
    {
      "epoch": 10.448882548739896,
      "grad_norm": 0.18925482034683228,
      "learning_rate": 6.518894887265799e-05,
      "loss": 0.0534,
      "step": 5486
    },
    {
      "epoch": 10.450784593437946,
      "grad_norm": 0.2295648753643036,
      "learning_rate": 6.518259765004764e-05,
      "loss": 0.0593,
      "step": 5487
    },
    {
      "epoch": 10.452686638135996,
      "grad_norm": 0.19527499377727509,
      "learning_rate": 6.517624642743728e-05,
      "loss": 0.0415,
      "step": 5488
    },
    {
      "epoch": 10.454588682834046,
      "grad_norm": 0.27798953652381897,
      "learning_rate": 6.516989520482693e-05,
      "loss": 0.0614,
      "step": 5489
    },
    {
      "epoch": 10.456490727532097,
      "grad_norm": 0.3666345477104187,
      "learning_rate": 6.516354398221658e-05,
      "loss": 0.0757,
      "step": 5490
    },
    {
      "epoch": 10.458392772230148,
      "grad_norm": 0.19281333684921265,
      "learning_rate": 6.515719275960622e-05,
      "loss": 0.055,
      "step": 5491
    },
    {
      "epoch": 10.460294816928197,
      "grad_norm": 0.25162625312805176,
      "learning_rate": 6.515084153699587e-05,
      "loss": 0.0727,
      "step": 5492
    },
    {
      "epoch": 10.462196861626248,
      "grad_norm": 0.32369402050971985,
      "learning_rate": 6.514449031438553e-05,
      "loss": 0.0632,
      "step": 5493
    },
    {
      "epoch": 10.464098906324299,
      "grad_norm": 0.3831532299518585,
      "learning_rate": 6.513813909177518e-05,
      "loss": 0.0735,
      "step": 5494
    },
    {
      "epoch": 10.466000951022348,
      "grad_norm": 0.2529776692390442,
      "learning_rate": 6.513178786916482e-05,
      "loss": 0.0838,
      "step": 5495
    },
    {
      "epoch": 10.467902995720399,
      "grad_norm": 0.22363333404064178,
      "learning_rate": 6.512543664655446e-05,
      "loss": 0.071,
      "step": 5496
    },
    {
      "epoch": 10.46980504041845,
      "grad_norm": 0.20970891416072845,
      "learning_rate": 6.511908542394412e-05,
      "loss": 0.0565,
      "step": 5497
    },
    {
      "epoch": 10.4717070851165,
      "grad_norm": 0.14854298532009125,
      "learning_rate": 6.511273420133376e-05,
      "loss": 0.0577,
      "step": 5498
    },
    {
      "epoch": 10.47360912981455,
      "grad_norm": 0.200281023979187,
      "learning_rate": 6.510638297872341e-05,
      "loss": 0.0807,
      "step": 5499
    },
    {
      "epoch": 10.4755111745126,
      "grad_norm": 0.24579498171806335,
      "learning_rate": 6.510003175611305e-05,
      "loss": 0.0698,
      "step": 5500
    },
    {
      "epoch": 10.477413219210652,
      "grad_norm": 0.28687208890914917,
      "learning_rate": 6.50936805335027e-05,
      "loss": 0.0542,
      "step": 5501
    },
    {
      "epoch": 10.479315263908703,
      "grad_norm": 0.2897108793258667,
      "learning_rate": 6.508732931089235e-05,
      "loss": 0.0845,
      "step": 5502
    },
    {
      "epoch": 10.481217308606752,
      "grad_norm": 0.18317654728889465,
      "learning_rate": 6.508097808828199e-05,
      "loss": 0.0556,
      "step": 5503
    },
    {
      "epoch": 10.483119353304803,
      "grad_norm": 0.20223313570022583,
      "learning_rate": 6.507462686567164e-05,
      "loss": 0.076,
      "step": 5504
    },
    {
      "epoch": 10.485021398002853,
      "grad_norm": 0.19548408687114716,
      "learning_rate": 6.50682756430613e-05,
      "loss": 0.0516,
      "step": 5505
    },
    {
      "epoch": 10.486923442700903,
      "grad_norm": 0.24220460653305054,
      "learning_rate": 6.506192442045093e-05,
      "loss": 0.0452,
      "step": 5506
    },
    {
      "epoch": 10.488825487398953,
      "grad_norm": 0.19592654705047607,
      "learning_rate": 6.505557319784058e-05,
      "loss": 0.0606,
      "step": 5507
    },
    {
      "epoch": 10.490727532097004,
      "grad_norm": 0.16772393882274628,
      "learning_rate": 6.504922197523024e-05,
      "loss": 0.0512,
      "step": 5508
    },
    {
      "epoch": 10.492629576795055,
      "grad_norm": 0.17901265621185303,
      "learning_rate": 6.504287075261987e-05,
      "loss": 0.0523,
      "step": 5509
    },
    {
      "epoch": 10.494531621493104,
      "grad_norm": 0.2328658252954483,
      "learning_rate": 6.503651953000953e-05,
      "loss": 0.0514,
      "step": 5510
    },
    {
      "epoch": 10.496433666191155,
      "grad_norm": 0.16231556236743927,
      "learning_rate": 6.503016830739918e-05,
      "loss": 0.0529,
      "step": 5511
    },
    {
      "epoch": 10.498335710889206,
      "grad_norm": 0.2349044382572174,
      "learning_rate": 6.502381708478883e-05,
      "loss": 0.0639,
      "step": 5512
    },
    {
      "epoch": 10.500237755587257,
      "grad_norm": 0.22910793125629425,
      "learning_rate": 6.501746586217847e-05,
      "loss": 0.0819,
      "step": 5513
    },
    {
      "epoch": 10.502139800285306,
      "grad_norm": 0.2594580054283142,
      "learning_rate": 6.501111463956812e-05,
      "loss": 0.0713,
      "step": 5514
    },
    {
      "epoch": 10.504041844983357,
      "grad_norm": 0.21220757067203522,
      "learning_rate": 6.500476341695777e-05,
      "loss": 0.0646,
      "step": 5515
    },
    {
      "epoch": 10.505943889681408,
      "grad_norm": 0.16294394433498383,
      "learning_rate": 6.499841219434741e-05,
      "loss": 0.0406,
      "step": 5516
    },
    {
      "epoch": 10.507845934379457,
      "grad_norm": 0.25099077820777893,
      "learning_rate": 6.499206097173706e-05,
      "loss": 0.0731,
      "step": 5517
    },
    {
      "epoch": 10.509747979077508,
      "grad_norm": 0.2558063268661499,
      "learning_rate": 6.498570974912671e-05,
      "loss": 0.0783,
      "step": 5518
    },
    {
      "epoch": 10.511650023775559,
      "grad_norm": 0.3749690353870392,
      "learning_rate": 6.497935852651635e-05,
      "loss": 0.0638,
      "step": 5519
    },
    {
      "epoch": 10.51355206847361,
      "grad_norm": 0.31419336795806885,
      "learning_rate": 6.4973007303906e-05,
      "loss": 0.085,
      "step": 5520
    },
    {
      "epoch": 10.515454113171659,
      "grad_norm": 0.18491724133491516,
      "learning_rate": 6.496665608129566e-05,
      "loss": 0.0617,
      "step": 5521
    },
    {
      "epoch": 10.51735615786971,
      "grad_norm": 0.1913200169801712,
      "learning_rate": 6.49603048586853e-05,
      "loss": 0.0755,
      "step": 5522
    },
    {
      "epoch": 10.51925820256776,
      "grad_norm": 0.21209709346294403,
      "learning_rate": 6.495395363607495e-05,
      "loss": 0.0809,
      "step": 5523
    },
    {
      "epoch": 10.521160247265811,
      "grad_norm": 0.2636054754257202,
      "learning_rate": 6.49476024134646e-05,
      "loss": 0.0664,
      "step": 5524
    },
    {
      "epoch": 10.52306229196386,
      "grad_norm": 0.27395957708358765,
      "learning_rate": 6.494125119085425e-05,
      "loss": 0.0678,
      "step": 5525
    },
    {
      "epoch": 10.524964336661911,
      "grad_norm": 0.3728583753108978,
      "learning_rate": 6.493489996824389e-05,
      "loss": 0.0793,
      "step": 5526
    },
    {
      "epoch": 10.526866381359962,
      "grad_norm": 0.22992080450057983,
      "learning_rate": 6.492854874563353e-05,
      "loss": 0.09,
      "step": 5527
    },
    {
      "epoch": 10.528768426058011,
      "grad_norm": 0.25271546840667725,
      "learning_rate": 6.492219752302319e-05,
      "loss": 0.0687,
      "step": 5528
    },
    {
      "epoch": 10.530670470756062,
      "grad_norm": 0.2486717402935028,
      "learning_rate": 6.491584630041283e-05,
      "loss": 0.0778,
      "step": 5529
    },
    {
      "epoch": 10.532572515454113,
      "grad_norm": 0.14895717799663544,
      "learning_rate": 6.490949507780248e-05,
      "loss": 0.0693,
      "step": 5530
    },
    {
      "epoch": 10.534474560152164,
      "grad_norm": 0.24672646820545197,
      "learning_rate": 6.490314385519213e-05,
      "loss": 0.0641,
      "step": 5531
    },
    {
      "epoch": 10.536376604850213,
      "grad_norm": 0.21848620474338531,
      "learning_rate": 6.489679263258177e-05,
      "loss": 0.0832,
      "step": 5532
    },
    {
      "epoch": 10.538278649548264,
      "grad_norm": 0.2144383043050766,
      "learning_rate": 6.489044140997142e-05,
      "loss": 0.0599,
      "step": 5533
    },
    {
      "epoch": 10.540180694246315,
      "grad_norm": 0.44709065556526184,
      "learning_rate": 6.488409018736106e-05,
      "loss": 0.0932,
      "step": 5534
    },
    {
      "epoch": 10.542082738944366,
      "grad_norm": 0.23711533844470978,
      "learning_rate": 6.487773896475073e-05,
      "loss": 0.0606,
      "step": 5535
    },
    {
      "epoch": 10.543984783642415,
      "grad_norm": 0.29541435837745667,
      "learning_rate": 6.487138774214037e-05,
      "loss": 0.0881,
      "step": 5536
    },
    {
      "epoch": 10.545886828340466,
      "grad_norm": 0.26311227679252625,
      "learning_rate": 6.486503651953e-05,
      "loss": 0.0698,
      "step": 5537
    },
    {
      "epoch": 10.547788873038517,
      "grad_norm": 0.23279817402362823,
      "learning_rate": 6.485868529691967e-05,
      "loss": 0.0666,
      "step": 5538
    },
    {
      "epoch": 10.549690917736568,
      "grad_norm": 0.1474435180425644,
      "learning_rate": 6.485233407430931e-05,
      "loss": 0.0659,
      "step": 5539
    },
    {
      "epoch": 10.551592962434617,
      "grad_norm": 0.18819576501846313,
      "learning_rate": 6.484598285169895e-05,
      "loss": 0.0546,
      "step": 5540
    },
    {
      "epoch": 10.553495007132668,
      "grad_norm": 0.27992531657218933,
      "learning_rate": 6.48396316290886e-05,
      "loss": 0.0632,
      "step": 5541
    },
    {
      "epoch": 10.555397051830719,
      "grad_norm": 0.22803568840026855,
      "learning_rate": 6.483328040647825e-05,
      "loss": 0.0617,
      "step": 5542
    },
    {
      "epoch": 10.557299096528768,
      "grad_norm": 0.29736748337745667,
      "learning_rate": 6.48269291838679e-05,
      "loss": 0.0773,
      "step": 5543
    },
    {
      "epoch": 10.559201141226819,
      "grad_norm": 0.19592848420143127,
      "learning_rate": 6.482057796125754e-05,
      "loss": 0.086,
      "step": 5544
    },
    {
      "epoch": 10.56110318592487,
      "grad_norm": 0.27602100372314453,
      "learning_rate": 6.481422673864719e-05,
      "loss": 0.0846,
      "step": 5545
    },
    {
      "epoch": 10.56300523062292,
      "grad_norm": 0.2776179015636444,
      "learning_rate": 6.480787551603684e-05,
      "loss": 0.0869,
      "step": 5546
    },
    {
      "epoch": 10.56490727532097,
      "grad_norm": 0.1875569075345993,
      "learning_rate": 6.480152429342648e-05,
      "loss": 0.0682,
      "step": 5547
    },
    {
      "epoch": 10.56680932001902,
      "grad_norm": 0.2264152467250824,
      "learning_rate": 6.479517307081613e-05,
      "loss": 0.0718,
      "step": 5548
    },
    {
      "epoch": 10.568711364717071,
      "grad_norm": 0.19574370980262756,
      "learning_rate": 6.478882184820579e-05,
      "loss": 0.0676,
      "step": 5549
    },
    {
      "epoch": 10.570613409415122,
      "grad_norm": 0.3385010361671448,
      "learning_rate": 6.478247062559542e-05,
      "loss": 0.0978,
      "step": 5550
    },
    {
      "epoch": 10.572515454113171,
      "grad_norm": 0.2784387767314911,
      "learning_rate": 6.477611940298508e-05,
      "loss": 0.0919,
      "step": 5551
    },
    {
      "epoch": 10.574417498811222,
      "grad_norm": 0.23461949825286865,
      "learning_rate": 6.476976818037473e-05,
      "loss": 0.0713,
      "step": 5552
    },
    {
      "epoch": 10.576319543509273,
      "grad_norm": 0.2040696144104004,
      "learning_rate": 6.476341695776438e-05,
      "loss": 0.0631,
      "step": 5553
    },
    {
      "epoch": 10.578221588207322,
      "grad_norm": 0.23008395731449127,
      "learning_rate": 6.475706573515402e-05,
      "loss": 0.0627,
      "step": 5554
    },
    {
      "epoch": 10.580123632905373,
      "grad_norm": 0.20021569728851318,
      "learning_rate": 6.475071451254367e-05,
      "loss": 0.0577,
      "step": 5555
    },
    {
      "epoch": 10.582025677603424,
      "grad_norm": 0.24329201877117157,
      "learning_rate": 6.474436328993332e-05,
      "loss": 0.0774,
      "step": 5556
    },
    {
      "epoch": 10.583927722301475,
      "grad_norm": 0.2651795446872711,
      "learning_rate": 6.473801206732296e-05,
      "loss": 0.0833,
      "step": 5557
    },
    {
      "epoch": 10.585829766999524,
      "grad_norm": 0.2536809742450714,
      "learning_rate": 6.47316608447126e-05,
      "loss": 0.1058,
      "step": 5558
    },
    {
      "epoch": 10.587731811697575,
      "grad_norm": 0.18420420587062836,
      "learning_rate": 6.472530962210226e-05,
      "loss": 0.0821,
      "step": 5559
    },
    {
      "epoch": 10.589633856395626,
      "grad_norm": 0.2313283234834671,
      "learning_rate": 6.47189583994919e-05,
      "loss": 0.0621,
      "step": 5560
    },
    {
      "epoch": 10.591535901093676,
      "grad_norm": 0.28729113936424255,
      "learning_rate": 6.471260717688155e-05,
      "loss": 0.0571,
      "step": 5561
    },
    {
      "epoch": 10.593437945791726,
      "grad_norm": 0.3117208480834961,
      "learning_rate": 6.47062559542712e-05,
      "loss": 0.0824,
      "step": 5562
    },
    {
      "epoch": 10.595339990489776,
      "grad_norm": 0.23070599138736725,
      "learning_rate": 6.469990473166084e-05,
      "loss": 0.0682,
      "step": 5563
    },
    {
      "epoch": 10.597242035187827,
      "grad_norm": 0.18042044341564178,
      "learning_rate": 6.46935535090505e-05,
      "loss": 0.0487,
      "step": 5564
    },
    {
      "epoch": 10.599144079885878,
      "grad_norm": 0.26043298840522766,
      "learning_rate": 6.468720228644013e-05,
      "loss": 0.063,
      "step": 5565
    },
    {
      "epoch": 10.601046124583927,
      "grad_norm": 0.2567049562931061,
      "learning_rate": 6.46808510638298e-05,
      "loss": 0.0643,
      "step": 5566
    },
    {
      "epoch": 10.602948169281978,
      "grad_norm": 0.2641132175922394,
      "learning_rate": 6.467449984121944e-05,
      "loss": 0.0621,
      "step": 5567
    },
    {
      "epoch": 10.60485021398003,
      "grad_norm": 0.24984419345855713,
      "learning_rate": 6.466814861860908e-05,
      "loss": 0.071,
      "step": 5568
    },
    {
      "epoch": 10.606752258678078,
      "grad_norm": 0.29163748025894165,
      "learning_rate": 6.466179739599874e-05,
      "loss": 0.0933,
      "step": 5569
    },
    {
      "epoch": 10.60865430337613,
      "grad_norm": 0.28160735964775085,
      "learning_rate": 6.465544617338838e-05,
      "loss": 0.0745,
      "step": 5570
    },
    {
      "epoch": 10.61055634807418,
      "grad_norm": 0.20157352089881897,
      "learning_rate": 6.464909495077803e-05,
      "loss": 0.0731,
      "step": 5571
    },
    {
      "epoch": 10.612458392772231,
      "grad_norm": 0.3122706115245819,
      "learning_rate": 6.464274372816767e-05,
      "loss": 0.0787,
      "step": 5572
    },
    {
      "epoch": 10.61436043747028,
      "grad_norm": 0.20753252506256104,
      "learning_rate": 6.463639250555732e-05,
      "loss": 0.0654,
      "step": 5573
    },
    {
      "epoch": 10.616262482168331,
      "grad_norm": 0.2078036218881607,
      "learning_rate": 6.463004128294697e-05,
      "loss": 0.0568,
      "step": 5574
    },
    {
      "epoch": 10.618164526866382,
      "grad_norm": 0.2810681462287903,
      "learning_rate": 6.462369006033661e-05,
      "loss": 0.0729,
      "step": 5575
    },
    {
      "epoch": 10.620066571564433,
      "grad_norm": 0.1743522733449936,
      "learning_rate": 6.461733883772626e-05,
      "loss": 0.0479,
      "step": 5576
    },
    {
      "epoch": 10.621968616262482,
      "grad_norm": 0.19856297969818115,
      "learning_rate": 6.461098761511592e-05,
      "loss": 0.0566,
      "step": 5577
    },
    {
      "epoch": 10.623870660960533,
      "grad_norm": 0.25219857692718506,
      "learning_rate": 6.460463639250555e-05,
      "loss": 0.0662,
      "step": 5578
    },
    {
      "epoch": 10.625772705658584,
      "grad_norm": 0.39976876974105835,
      "learning_rate": 6.45982851698952e-05,
      "loss": 0.072,
      "step": 5579
    },
    {
      "epoch": 10.627674750356633,
      "grad_norm": 0.2829294800758362,
      "learning_rate": 6.459193394728486e-05,
      "loss": 0.0574,
      "step": 5580
    },
    {
      "epoch": 10.629576795054684,
      "grad_norm": 0.22536730766296387,
      "learning_rate": 6.45855827246745e-05,
      "loss": 0.0733,
      "step": 5581
    },
    {
      "epoch": 10.631478839752734,
      "grad_norm": 0.1678084135055542,
      "learning_rate": 6.457923150206415e-05,
      "loss": 0.0599,
      "step": 5582
    },
    {
      "epoch": 10.633380884450785,
      "grad_norm": 0.22193792462348938,
      "learning_rate": 6.45728802794538e-05,
      "loss": 0.0646,
      "step": 5583
    },
    {
      "epoch": 10.635282929148834,
      "grad_norm": 0.3831283152103424,
      "learning_rate": 6.456652905684345e-05,
      "loss": 0.0839,
      "step": 5584
    },
    {
      "epoch": 10.637184973846885,
      "grad_norm": 0.17634892463684082,
      "learning_rate": 6.456017783423309e-05,
      "loss": 0.0586,
      "step": 5585
    },
    {
      "epoch": 10.639087018544936,
      "grad_norm": 0.18664374947547913,
      "learning_rate": 6.455382661162274e-05,
      "loss": 0.0496,
      "step": 5586
    },
    {
      "epoch": 10.640989063242987,
      "grad_norm": 0.21163462102413177,
      "learning_rate": 6.45474753890124e-05,
      "loss": 0.0616,
      "step": 5587
    },
    {
      "epoch": 10.642891107941036,
      "grad_norm": 0.1956004649400711,
      "learning_rate": 6.454112416640203e-05,
      "loss": 0.0695,
      "step": 5588
    },
    {
      "epoch": 10.644793152639087,
      "grad_norm": 0.2948111891746521,
      "learning_rate": 6.453477294379168e-05,
      "loss": 0.0684,
      "step": 5589
    },
    {
      "epoch": 10.646695197337138,
      "grad_norm": 0.18460431694984436,
      "learning_rate": 6.452842172118134e-05,
      "loss": 0.0606,
      "step": 5590
    },
    {
      "epoch": 10.648597242035187,
      "grad_norm": 0.2511366903781891,
      "learning_rate": 6.452207049857097e-05,
      "loss": 0.0946,
      "step": 5591
    },
    {
      "epoch": 10.650499286733238,
      "grad_norm": 0.21115030348300934,
      "learning_rate": 6.451571927596063e-05,
      "loss": 0.0552,
      "step": 5592
    },
    {
      "epoch": 10.652401331431289,
      "grad_norm": 0.2667510211467743,
      "learning_rate": 6.450936805335028e-05,
      "loss": 0.0698,
      "step": 5593
    },
    {
      "epoch": 10.65430337612934,
      "grad_norm": 0.35384124517440796,
      "learning_rate": 6.450301683073992e-05,
      "loss": 0.0746,
      "step": 5594
    },
    {
      "epoch": 10.656205420827389,
      "grad_norm": 0.26212817430496216,
      "learning_rate": 6.449666560812957e-05,
      "loss": 0.0589,
      "step": 5595
    },
    {
      "epoch": 10.65810746552544,
      "grad_norm": 0.22370542585849762,
      "learning_rate": 6.449031438551922e-05,
      "loss": 0.0526,
      "step": 5596
    },
    {
      "epoch": 10.66000951022349,
      "grad_norm": 0.362678587436676,
      "learning_rate": 6.448396316290887e-05,
      "loss": 0.1082,
      "step": 5597
    },
    {
      "epoch": 10.661911554921542,
      "grad_norm": 0.3471091389656067,
      "learning_rate": 6.447761194029851e-05,
      "loss": 0.0884,
      "step": 5598
    },
    {
      "epoch": 10.66381359961959,
      "grad_norm": 0.29880112409591675,
      "learning_rate": 6.447126071768815e-05,
      "loss": 0.0734,
      "step": 5599
    },
    {
      "epoch": 10.665715644317642,
      "grad_norm": 0.1709068864583969,
      "learning_rate": 6.446490949507781e-05,
      "loss": 0.0461,
      "step": 5600
    },
    {
      "epoch": 10.667617689015692,
      "grad_norm": 0.23527254164218903,
      "learning_rate": 6.445855827246745e-05,
      "loss": 0.0599,
      "step": 5601
    },
    {
      "epoch": 10.669519733713742,
      "grad_norm": 0.27498099207878113,
      "learning_rate": 6.44522070498571e-05,
      "loss": 0.062,
      "step": 5602
    },
    {
      "epoch": 10.671421778411792,
      "grad_norm": 0.2606363296508789,
      "learning_rate": 6.444585582724675e-05,
      "loss": 0.0702,
      "step": 5603
    },
    {
      "epoch": 10.673323823109843,
      "grad_norm": 0.28660082817077637,
      "learning_rate": 6.44395046046364e-05,
      "loss": 0.0805,
      "step": 5604
    },
    {
      "epoch": 10.675225867807894,
      "grad_norm": 0.22156530618667603,
      "learning_rate": 6.443315338202605e-05,
      "loss": 0.061,
      "step": 5605
    },
    {
      "epoch": 10.677127912505943,
      "grad_norm": 0.20593293011188507,
      "learning_rate": 6.442680215941568e-05,
      "loss": 0.0662,
      "step": 5606
    },
    {
      "epoch": 10.679029957203994,
      "grad_norm": 0.22477205097675323,
      "learning_rate": 6.442045093680535e-05,
      "loss": 0.0728,
      "step": 5607
    },
    {
      "epoch": 10.680932001902045,
      "grad_norm": 0.18077880144119263,
      "learning_rate": 6.441409971419499e-05,
      "loss": 0.06,
      "step": 5608
    },
    {
      "epoch": 10.682834046600096,
      "grad_norm": 0.2662803828716278,
      "learning_rate": 6.440774849158463e-05,
      "loss": 0.0629,
      "step": 5609
    },
    {
      "epoch": 10.684736091298145,
      "grad_norm": 0.26029613614082336,
      "learning_rate": 6.440139726897428e-05,
      "loss": 0.0676,
      "step": 5610
    },
    {
      "epoch": 10.686638135996196,
      "grad_norm": 0.26586633920669556,
      "learning_rate": 6.439504604636393e-05,
      "loss": 0.0532,
      "step": 5611
    },
    {
      "epoch": 10.688540180694247,
      "grad_norm": 0.2061486840248108,
      "learning_rate": 6.438869482375357e-05,
      "loss": 0.0744,
      "step": 5612
    },
    {
      "epoch": 10.690442225392296,
      "grad_norm": 0.2764222025871277,
      "learning_rate": 6.438234360114322e-05,
      "loss": 0.0821,
      "step": 5613
    },
    {
      "epoch": 10.692344270090347,
      "grad_norm": 0.2235095053911209,
      "learning_rate": 6.437599237853287e-05,
      "loss": 0.0598,
      "step": 5614
    },
    {
      "epoch": 10.694246314788398,
      "grad_norm": 0.12553450465202332,
      "learning_rate": 6.436964115592252e-05,
      "loss": 0.0619,
      "step": 5615
    },
    {
      "epoch": 10.696148359486449,
      "grad_norm": 0.23471371829509735,
      "learning_rate": 6.436328993331216e-05,
      "loss": 0.0574,
      "step": 5616
    },
    {
      "epoch": 10.698050404184498,
      "grad_norm": 0.24933548271656036,
      "learning_rate": 6.435693871070181e-05,
      "loss": 0.0638,
      "step": 5617
    },
    {
      "epoch": 10.699952448882549,
      "grad_norm": 0.2832605242729187,
      "learning_rate": 6.435058748809146e-05,
      "loss": 0.0682,
      "step": 5618
    },
    {
      "epoch": 10.7018544935806,
      "grad_norm": 0.3140818178653717,
      "learning_rate": 6.43442362654811e-05,
      "loss": 0.0647,
      "step": 5619
    },
    {
      "epoch": 10.70375653827865,
      "grad_norm": 0.28693559765815735,
      "learning_rate": 6.433788504287075e-05,
      "loss": 0.077,
      "step": 5620
    },
    {
      "epoch": 10.7056585829767,
      "grad_norm": 0.1607866883277893,
      "learning_rate": 6.433153382026041e-05,
      "loss": 0.0549,
      "step": 5621
    },
    {
      "epoch": 10.70756062767475,
      "grad_norm": 0.32727786898612976,
      "learning_rate": 6.432518259765005e-05,
      "loss": 0.0836,
      "step": 5622
    },
    {
      "epoch": 10.709462672372801,
      "grad_norm": 0.26169657707214355,
      "learning_rate": 6.43188313750397e-05,
      "loss": 0.0551,
      "step": 5623
    },
    {
      "epoch": 10.71136471707085,
      "grad_norm": 0.17940069735050201,
      "learning_rate": 6.431248015242935e-05,
      "loss": 0.0548,
      "step": 5624
    },
    {
      "epoch": 10.713266761768901,
      "grad_norm": 0.3701190650463104,
      "learning_rate": 6.4306128929819e-05,
      "loss": 0.0721,
      "step": 5625
    },
    {
      "epoch": 10.715168806466952,
      "grad_norm": 0.15953880548477173,
      "learning_rate": 6.429977770720864e-05,
      "loss": 0.0582,
      "step": 5626
    },
    {
      "epoch": 10.717070851165003,
      "grad_norm": 0.2598779797554016,
      "learning_rate": 6.429342648459829e-05,
      "loss": 0.066,
      "step": 5627
    },
    {
      "epoch": 10.718972895863052,
      "grad_norm": 0.19500410556793213,
      "learning_rate": 6.428707526198794e-05,
      "loss": 0.049,
      "step": 5628
    },
    {
      "epoch": 10.720874940561103,
      "grad_norm": 0.2729705572128296,
      "learning_rate": 6.428072403937758e-05,
      "loss": 0.0742,
      "step": 5629
    },
    {
      "epoch": 10.722776985259154,
      "grad_norm": 0.3123151957988739,
      "learning_rate": 6.427437281676722e-05,
      "loss": 0.0629,
      "step": 5630
    },
    {
      "epoch": 10.724679029957205,
      "grad_norm": 0.23334093391895294,
      "learning_rate": 6.426802159415688e-05,
      "loss": 0.0698,
      "step": 5631
    },
    {
      "epoch": 10.726581074655254,
      "grad_norm": 0.18635810911655426,
      "learning_rate": 6.426167037154652e-05,
      "loss": 0.0712,
      "step": 5632
    },
    {
      "epoch": 10.728483119353305,
      "grad_norm": 0.22240066528320312,
      "learning_rate": 6.425531914893617e-05,
      "loss": 0.0766,
      "step": 5633
    },
    {
      "epoch": 10.730385164051356,
      "grad_norm": 0.1942574381828308,
      "learning_rate": 6.424896792632583e-05,
      "loss": 0.0598,
      "step": 5634
    },
    {
      "epoch": 10.732287208749405,
      "grad_norm": 0.22906853258609772,
      "learning_rate": 6.424261670371546e-05,
      "loss": 0.059,
      "step": 5635
    },
    {
      "epoch": 10.734189253447456,
      "grad_norm": 0.257289856672287,
      "learning_rate": 6.423626548110512e-05,
      "loss": 0.1054,
      "step": 5636
    },
    {
      "epoch": 10.736091298145507,
      "grad_norm": 0.2559765875339508,
      "learning_rate": 6.422991425849475e-05,
      "loss": 0.1258,
      "step": 5637
    },
    {
      "epoch": 10.737993342843557,
      "grad_norm": 0.21526873111724854,
      "learning_rate": 6.422356303588442e-05,
      "loss": 0.0699,
      "step": 5638
    },
    {
      "epoch": 10.739895387541607,
      "grad_norm": 0.23773230612277985,
      "learning_rate": 6.421721181327406e-05,
      "loss": 0.0752,
      "step": 5639
    },
    {
      "epoch": 10.741797432239657,
      "grad_norm": 0.23262368142604828,
      "learning_rate": 6.42108605906637e-05,
      "loss": 0.0627,
      "step": 5640
    },
    {
      "epoch": 10.743699476937708,
      "grad_norm": 0.22525426745414734,
      "learning_rate": 6.420450936805336e-05,
      "loss": 0.0594,
      "step": 5641
    },
    {
      "epoch": 10.74560152163576,
      "grad_norm": 0.2492980659008026,
      "learning_rate": 6.4198158145443e-05,
      "loss": 0.0666,
      "step": 5642
    },
    {
      "epoch": 10.747503566333808,
      "grad_norm": 0.1995389312505722,
      "learning_rate": 6.419180692283265e-05,
      "loss": 0.0621,
      "step": 5643
    },
    {
      "epoch": 10.74940561103186,
      "grad_norm": 0.15982617437839508,
      "learning_rate": 6.418545570022229e-05,
      "loss": 0.068,
      "step": 5644
    },
    {
      "epoch": 10.75130765572991,
      "grad_norm": 0.17556272447109222,
      "learning_rate": 6.417910447761194e-05,
      "loss": 0.0539,
      "step": 5645
    },
    {
      "epoch": 10.75320970042796,
      "grad_norm": 0.2797108292579651,
      "learning_rate": 6.41727532550016e-05,
      "loss": 0.0602,
      "step": 5646
    },
    {
      "epoch": 10.75511174512601,
      "grad_norm": 0.17692141234874725,
      "learning_rate": 6.416640203239123e-05,
      "loss": 0.0541,
      "step": 5647
    },
    {
      "epoch": 10.757013789824061,
      "grad_norm": 0.26082196831703186,
      "learning_rate": 6.416005080978088e-05,
      "loss": 0.0697,
      "step": 5648
    },
    {
      "epoch": 10.758915834522112,
      "grad_norm": 0.2215351164340973,
      "learning_rate": 6.415369958717054e-05,
      "loss": 0.0417,
      "step": 5649
    },
    {
      "epoch": 10.760817879220161,
      "grad_norm": 0.3061333894729614,
      "learning_rate": 6.414734836456017e-05,
      "loss": 0.0884,
      "step": 5650
    },
    {
      "epoch": 10.762719923918212,
      "grad_norm": 0.21420468389987946,
      "learning_rate": 6.414099714194983e-05,
      "loss": 0.0751,
      "step": 5651
    },
    {
      "epoch": 10.764621968616263,
      "grad_norm": 0.23848441243171692,
      "learning_rate": 6.413464591933948e-05,
      "loss": 0.0738,
      "step": 5652
    },
    {
      "epoch": 10.766524013314314,
      "grad_norm": 0.26106590032577515,
      "learning_rate": 6.412829469672912e-05,
      "loss": 0.0667,
      "step": 5653
    },
    {
      "epoch": 10.768426058012363,
      "grad_norm": 0.22626987099647522,
      "learning_rate": 6.412194347411877e-05,
      "loss": 0.0683,
      "step": 5654
    },
    {
      "epoch": 10.770328102710414,
      "grad_norm": 0.19195275008678436,
      "learning_rate": 6.411559225150842e-05,
      "loss": 0.0563,
      "step": 5655
    },
    {
      "epoch": 10.772230147408465,
      "grad_norm": 0.2996932566165924,
      "learning_rate": 6.410924102889807e-05,
      "loss": 0.0789,
      "step": 5656
    },
    {
      "epoch": 10.774132192106514,
      "grad_norm": 0.2504531741142273,
      "learning_rate": 6.410288980628771e-05,
      "loss": 0.0795,
      "step": 5657
    },
    {
      "epoch": 10.776034236804565,
      "grad_norm": 0.2465498298406601,
      "learning_rate": 6.409653858367736e-05,
      "loss": 0.0638,
      "step": 5658
    },
    {
      "epoch": 10.777936281502615,
      "grad_norm": 0.21041840314865112,
      "learning_rate": 6.409018736106701e-05,
      "loss": 0.0669,
      "step": 5659
    },
    {
      "epoch": 10.779838326200666,
      "grad_norm": 0.22026972472667694,
      "learning_rate": 6.408383613845665e-05,
      "loss": 0.0612,
      "step": 5660
    },
    {
      "epoch": 10.781740370898715,
      "grad_norm": 0.22157369554042816,
      "learning_rate": 6.40774849158463e-05,
      "loss": 0.0672,
      "step": 5661
    },
    {
      "epoch": 10.783642415596766,
      "grad_norm": 0.2333994209766388,
      "learning_rate": 6.407113369323596e-05,
      "loss": 0.0834,
      "step": 5662
    },
    {
      "epoch": 10.785544460294817,
      "grad_norm": 0.24471452832221985,
      "learning_rate": 6.40647824706256e-05,
      "loss": 0.0599,
      "step": 5663
    },
    {
      "epoch": 10.787446504992868,
      "grad_norm": 0.1821519434452057,
      "learning_rate": 6.405843124801525e-05,
      "loss": 0.0832,
      "step": 5664
    },
    {
      "epoch": 10.789348549690917,
      "grad_norm": 0.26059895753860474,
      "learning_rate": 6.40520800254049e-05,
      "loss": 0.0892,
      "step": 5665
    },
    {
      "epoch": 10.791250594388968,
      "grad_norm": 0.2928067445755005,
      "learning_rate": 6.404572880279454e-05,
      "loss": 0.0778,
      "step": 5666
    },
    {
      "epoch": 10.793152639087019,
      "grad_norm": 0.28212589025497437,
      "learning_rate": 6.403937758018419e-05,
      "loss": 0.0631,
      "step": 5667
    },
    {
      "epoch": 10.795054683785068,
      "grad_norm": 0.3376213610172272,
      "learning_rate": 6.403302635757383e-05,
      "loss": 0.0674,
      "step": 5668
    },
    {
      "epoch": 10.796956728483119,
      "grad_norm": 0.28411048650741577,
      "learning_rate": 6.402667513496349e-05,
      "loss": 0.0912,
      "step": 5669
    },
    {
      "epoch": 10.79885877318117,
      "grad_norm": 0.20652949810028076,
      "learning_rate": 6.402032391235313e-05,
      "loss": 0.0612,
      "step": 5670
    },
    {
      "epoch": 10.80076081787922,
      "grad_norm": 0.2766312062740326,
      "learning_rate": 6.401397268974277e-05,
      "loss": 0.0801,
      "step": 5671
    },
    {
      "epoch": 10.80266286257727,
      "grad_norm": 0.22918127477169037,
      "learning_rate": 6.400762146713243e-05,
      "loss": 0.0766,
      "step": 5672
    },
    {
      "epoch": 10.80456490727532,
      "grad_norm": 0.13110528886318207,
      "learning_rate": 6.400127024452207e-05,
      "loss": 0.0531,
      "step": 5673
    },
    {
      "epoch": 10.806466951973372,
      "grad_norm": 0.34798091650009155,
      "learning_rate": 6.399491902191172e-05,
      "loss": 0.0846,
      "step": 5674
    },
    {
      "epoch": 10.808368996671422,
      "grad_norm": 0.21080365777015686,
      "learning_rate": 6.398856779930136e-05,
      "loss": 0.0669,
      "step": 5675
    },
    {
      "epoch": 10.810271041369472,
      "grad_norm": 0.18395227193832397,
      "learning_rate": 6.398221657669101e-05,
      "loss": 0.0515,
      "step": 5676
    },
    {
      "epoch": 10.812173086067522,
      "grad_norm": 0.2864764630794525,
      "learning_rate": 6.397586535408067e-05,
      "loss": 0.0655,
      "step": 5677
    },
    {
      "epoch": 10.814075130765573,
      "grad_norm": 0.24587604403495789,
      "learning_rate": 6.39695141314703e-05,
      "loss": 0.0723,
      "step": 5678
    },
    {
      "epoch": 10.815977175463622,
      "grad_norm": 0.26868876814842224,
      "learning_rate": 6.396316290885997e-05,
      "loss": 0.0668,
      "step": 5679
    },
    {
      "epoch": 10.817879220161673,
      "grad_norm": 0.24933819472789764,
      "learning_rate": 6.395681168624961e-05,
      "loss": 0.0811,
      "step": 5680
    },
    {
      "epoch": 10.819781264859724,
      "grad_norm": 0.26463747024536133,
      "learning_rate": 6.395046046363925e-05,
      "loss": 0.0582,
      "step": 5681
    },
    {
      "epoch": 10.821683309557775,
      "grad_norm": 0.19800810515880585,
      "learning_rate": 6.39441092410289e-05,
      "loss": 0.0624,
      "step": 5682
    },
    {
      "epoch": 10.823585354255824,
      "grad_norm": 0.230701744556427,
      "learning_rate": 6.393775801841855e-05,
      "loss": 0.0566,
      "step": 5683
    },
    {
      "epoch": 10.825487398953875,
      "grad_norm": 0.2717706263065338,
      "learning_rate": 6.393140679580819e-05,
      "loss": 0.0842,
      "step": 5684
    },
    {
      "epoch": 10.827389443651926,
      "grad_norm": 0.20510073006153107,
      "learning_rate": 6.392505557319784e-05,
      "loss": 0.0735,
      "step": 5685
    },
    {
      "epoch": 10.829291488349977,
      "grad_norm": 0.17784926295280457,
      "learning_rate": 6.391870435058749e-05,
      "loss": 0.0746,
      "step": 5686
    },
    {
      "epoch": 10.831193533048026,
      "grad_norm": 0.30288398265838623,
      "learning_rate": 6.391235312797714e-05,
      "loss": 0.0667,
      "step": 5687
    },
    {
      "epoch": 10.833095577746077,
      "grad_norm": 0.10946258902549744,
      "learning_rate": 6.390600190536678e-05,
      "loss": 0.0515,
      "step": 5688
    },
    {
      "epoch": 10.834997622444128,
      "grad_norm": 0.24755284190177917,
      "learning_rate": 6.389965068275643e-05,
      "loss": 0.0555,
      "step": 5689
    },
    {
      "epoch": 10.836899667142179,
      "grad_norm": 0.231001615524292,
      "learning_rate": 6.389329946014609e-05,
      "loss": 0.0629,
      "step": 5690
    },
    {
      "epoch": 10.838801711840228,
      "grad_norm": 0.2223047912120819,
      "learning_rate": 6.388694823753572e-05,
      "loss": 0.0638,
      "step": 5691
    },
    {
      "epoch": 10.840703756538279,
      "grad_norm": 0.40585601329803467,
      "learning_rate": 6.388059701492538e-05,
      "loss": 0.0897,
      "step": 5692
    },
    {
      "epoch": 10.84260580123633,
      "grad_norm": 0.3459436893463135,
      "learning_rate": 6.387424579231503e-05,
      "loss": 0.08,
      "step": 5693
    },
    {
      "epoch": 10.844507845934379,
      "grad_norm": 0.19375737011432648,
      "learning_rate": 6.386789456970467e-05,
      "loss": 0.0645,
      "step": 5694
    },
    {
      "epoch": 10.84640989063243,
      "grad_norm": 0.280331552028656,
      "learning_rate": 6.386154334709432e-05,
      "loss": 0.092,
      "step": 5695
    },
    {
      "epoch": 10.84831193533048,
      "grad_norm": 0.24692851305007935,
      "learning_rate": 6.385519212448397e-05,
      "loss": 0.0631,
      "step": 5696
    },
    {
      "epoch": 10.850213980028531,
      "grad_norm": 0.23048484325408936,
      "learning_rate": 6.384884090187362e-05,
      "loss": 0.0675,
      "step": 5697
    },
    {
      "epoch": 10.85211602472658,
      "grad_norm": 0.17280757427215576,
      "learning_rate": 6.384248967926326e-05,
      "loss": 0.0642,
      "step": 5698
    },
    {
      "epoch": 10.854018069424631,
      "grad_norm": 0.18811771273612976,
      "learning_rate": 6.383613845665291e-05,
      "loss": 0.0718,
      "step": 5699
    },
    {
      "epoch": 10.855920114122682,
      "grad_norm": 0.21706104278564453,
      "learning_rate": 6.382978723404256e-05,
      "loss": 0.0615,
      "step": 5700
    },
    {
      "epoch": 10.857822158820733,
      "grad_norm": 0.34146109223365784,
      "learning_rate": 6.38234360114322e-05,
      "loss": 0.0715,
      "step": 5701
    },
    {
      "epoch": 10.859724203518782,
      "grad_norm": 0.1904754787683487,
      "learning_rate": 6.381708478882184e-05,
      "loss": 0.068,
      "step": 5702
    },
    {
      "epoch": 10.861626248216833,
      "grad_norm": 0.26520657539367676,
      "learning_rate": 6.38107335662115e-05,
      "loss": 0.0644,
      "step": 5703
    },
    {
      "epoch": 10.863528292914884,
      "grad_norm": 0.18787312507629395,
      "learning_rate": 6.380438234360114e-05,
      "loss": 0.048,
      "step": 5704
    },
    {
      "epoch": 10.865430337612933,
      "grad_norm": 0.22588405013084412,
      "learning_rate": 6.37980311209908e-05,
      "loss": 0.0581,
      "step": 5705
    },
    {
      "epoch": 10.867332382310984,
      "grad_norm": 0.2537688910961151,
      "learning_rate": 6.379167989838045e-05,
      "loss": 0.0779,
      "step": 5706
    },
    {
      "epoch": 10.869234427009035,
      "grad_norm": 0.25890594720840454,
      "learning_rate": 6.378532867577009e-05,
      "loss": 0.0649,
      "step": 5707
    },
    {
      "epoch": 10.871136471707086,
      "grad_norm": 0.14498251676559448,
      "learning_rate": 6.377897745315974e-05,
      "loss": 0.0734,
      "step": 5708
    },
    {
      "epoch": 10.873038516405135,
      "grad_norm": 0.2548612654209137,
      "learning_rate": 6.377262623054938e-05,
      "loss": 0.0813,
      "step": 5709
    },
    {
      "epoch": 10.874940561103186,
      "grad_norm": 0.19444790482521057,
      "learning_rate": 6.376627500793904e-05,
      "loss": 0.067,
      "step": 5710
    },
    {
      "epoch": 10.876842605801237,
      "grad_norm": 0.22690126299858093,
      "learning_rate": 6.375992378532868e-05,
      "loss": 0.0743,
      "step": 5711
    },
    {
      "epoch": 10.878744650499288,
      "grad_norm": 0.2915293872356415,
      "learning_rate": 6.375357256271832e-05,
      "loss": 0.0657,
      "step": 5712
    },
    {
      "epoch": 10.880646695197337,
      "grad_norm": 0.3559345006942749,
      "learning_rate": 6.374722134010798e-05,
      "loss": 0.0864,
      "step": 5713
    },
    {
      "epoch": 10.882548739895388,
      "grad_norm": 0.16378241777420044,
      "learning_rate": 6.374087011749762e-05,
      "loss": 0.0379,
      "step": 5714
    },
    {
      "epoch": 10.884450784593438,
      "grad_norm": 0.43481773138046265,
      "learning_rate": 6.373451889488727e-05,
      "loss": 0.1046,
      "step": 5715
    },
    {
      "epoch": 10.88635282929149,
      "grad_norm": 0.31856125593185425,
      "learning_rate": 6.372816767227691e-05,
      "loss": 0.0795,
      "step": 5716
    },
    {
      "epoch": 10.888254873989538,
      "grad_norm": 0.2719641327857971,
      "learning_rate": 6.372181644966656e-05,
      "loss": 0.0866,
      "step": 5717
    },
    {
      "epoch": 10.89015691868759,
      "grad_norm": 0.25337544083595276,
      "learning_rate": 6.371546522705622e-05,
      "loss": 0.063,
      "step": 5718
    },
    {
      "epoch": 10.89205896338564,
      "grad_norm": 0.18715976178646088,
      "learning_rate": 6.370911400444585e-05,
      "loss": 0.0703,
      "step": 5719
    },
    {
      "epoch": 10.89396100808369,
      "grad_norm": 0.17753618955612183,
      "learning_rate": 6.37027627818355e-05,
      "loss": 0.0657,
      "step": 5720
    },
    {
      "epoch": 10.89586305278174,
      "grad_norm": 0.17351071536540985,
      "learning_rate": 6.369641155922516e-05,
      "loss": 0.0625,
      "step": 5721
    },
    {
      "epoch": 10.897765097479791,
      "grad_norm": 0.3891054391860962,
      "learning_rate": 6.36900603366148e-05,
      "loss": 0.0823,
      "step": 5722
    },
    {
      "epoch": 10.899667142177842,
      "grad_norm": 0.3254895806312561,
      "learning_rate": 6.368370911400445e-05,
      "loss": 0.0696,
      "step": 5723
    },
    {
      "epoch": 10.901569186875891,
      "grad_norm": 0.3056901693344116,
      "learning_rate": 6.36773578913941e-05,
      "loss": 0.0859,
      "step": 5724
    },
    {
      "epoch": 10.903471231573942,
      "grad_norm": 0.4118444621562958,
      "learning_rate": 6.367100666878374e-05,
      "loss": 0.0926,
      "step": 5725
    },
    {
      "epoch": 10.905373276271993,
      "grad_norm": 0.23973733186721802,
      "learning_rate": 6.366465544617339e-05,
      "loss": 0.0759,
      "step": 5726
    },
    {
      "epoch": 10.907275320970044,
      "grad_norm": 0.2237926423549652,
      "learning_rate": 6.365830422356304e-05,
      "loss": 0.061,
      "step": 5727
    },
    {
      "epoch": 10.909177365668093,
      "grad_norm": 0.28075718879699707,
      "learning_rate": 6.365195300095269e-05,
      "loss": 0.083,
      "step": 5728
    },
    {
      "epoch": 10.911079410366144,
      "grad_norm": 0.24081255495548248,
      "learning_rate": 6.364560177834233e-05,
      "loss": 0.0543,
      "step": 5729
    },
    {
      "epoch": 10.912981455064195,
      "grad_norm": 0.1605367660522461,
      "learning_rate": 6.363925055573198e-05,
      "loss": 0.0598,
      "step": 5730
    },
    {
      "epoch": 10.914883499762244,
      "grad_norm": 0.2056272178888321,
      "learning_rate": 6.363289933312164e-05,
      "loss": 0.0569,
      "step": 5731
    },
    {
      "epoch": 10.916785544460295,
      "grad_norm": 0.1978829950094223,
      "learning_rate": 6.362654811051127e-05,
      "loss": 0.06,
      "step": 5732
    },
    {
      "epoch": 10.918687589158345,
      "grad_norm": 0.27313923835754395,
      "learning_rate": 6.362019688790093e-05,
      "loss": 0.0939,
      "step": 5733
    },
    {
      "epoch": 10.920589633856396,
      "grad_norm": 0.4229736030101776,
      "learning_rate": 6.361384566529058e-05,
      "loss": 0.0779,
      "step": 5734
    },
    {
      "epoch": 10.922491678554445,
      "grad_norm": 0.23305979371070862,
      "learning_rate": 6.360749444268022e-05,
      "loss": 0.0556,
      "step": 5735
    },
    {
      "epoch": 10.924393723252496,
      "grad_norm": 0.2642943561077118,
      "learning_rate": 6.360114322006987e-05,
      "loss": 0.0831,
      "step": 5736
    },
    {
      "epoch": 10.926295767950547,
      "grad_norm": 0.2663883864879608,
      "learning_rate": 6.359479199745952e-05,
      "loss": 0.072,
      "step": 5737
    },
    {
      "epoch": 10.928197812648598,
      "grad_norm": 0.25676098465919495,
      "learning_rate": 6.358844077484916e-05,
      "loss": 0.0689,
      "step": 5738
    },
    {
      "epoch": 10.930099857346647,
      "grad_norm": 0.28164270520210266,
      "learning_rate": 6.358208955223881e-05,
      "loss": 0.0707,
      "step": 5739
    },
    {
      "epoch": 10.932001902044698,
      "grad_norm": 0.19156311452388763,
      "learning_rate": 6.357573832962845e-05,
      "loss": 0.0643,
      "step": 5740
    },
    {
      "epoch": 10.933903946742749,
      "grad_norm": 0.37269774079322815,
      "learning_rate": 6.356938710701811e-05,
      "loss": 0.0854,
      "step": 5741
    },
    {
      "epoch": 10.935805991440798,
      "grad_norm": 0.19018201529979706,
      "learning_rate": 6.356303588440775e-05,
      "loss": 0.0746,
      "step": 5742
    },
    {
      "epoch": 10.937708036138849,
      "grad_norm": 0.272897869348526,
      "learning_rate": 6.355668466179739e-05,
      "loss": 0.0653,
      "step": 5743
    },
    {
      "epoch": 10.9396100808369,
      "grad_norm": 0.24475888907909393,
      "learning_rate": 6.355033343918705e-05,
      "loss": 0.062,
      "step": 5744
    },
    {
      "epoch": 10.94151212553495,
      "grad_norm": 0.21947970986366272,
      "learning_rate": 6.354398221657669e-05,
      "loss": 0.0675,
      "step": 5745
    },
    {
      "epoch": 10.943414170233,
      "grad_norm": 0.41693615913391113,
      "learning_rate": 6.353763099396634e-05,
      "loss": 0.0798,
      "step": 5746
    },
    {
      "epoch": 10.94531621493105,
      "grad_norm": 0.2103748619556427,
      "learning_rate": 6.353127977135598e-05,
      "loss": 0.0561,
      "step": 5747
    },
    {
      "epoch": 10.947218259629102,
      "grad_norm": 0.40578147768974304,
      "learning_rate": 6.352492854874564e-05,
      "loss": 0.0809,
      "step": 5748
    },
    {
      "epoch": 10.949120304327153,
      "grad_norm": 0.22502633929252625,
      "learning_rate": 6.351857732613529e-05,
      "loss": 0.0503,
      "step": 5749
    },
    {
      "epoch": 10.951022349025202,
      "grad_norm": 0.2478155642747879,
      "learning_rate": 6.351222610352493e-05,
      "loss": 0.0542,
      "step": 5750
    },
    {
      "epoch": 10.952924393723253,
      "grad_norm": 0.22941410541534424,
      "learning_rate": 6.350587488091459e-05,
      "loss": 0.0712,
      "step": 5751
    },
    {
      "epoch": 10.954826438421303,
      "grad_norm": 0.2138836532831192,
      "learning_rate": 6.349952365830423e-05,
      "loss": 0.0617,
      "step": 5752
    },
    {
      "epoch": 10.956728483119353,
      "grad_norm": 0.1656169891357422,
      "learning_rate": 6.349317243569387e-05,
      "loss": 0.0573,
      "step": 5753
    },
    {
      "epoch": 10.958630527817403,
      "grad_norm": 0.20173387229442596,
      "learning_rate": 6.348682121308352e-05,
      "loss": 0.0667,
      "step": 5754
    },
    {
      "epoch": 10.960532572515454,
      "grad_norm": 0.23104602098464966,
      "learning_rate": 6.348046999047317e-05,
      "loss": 0.0728,
      "step": 5755
    },
    {
      "epoch": 10.962434617213505,
      "grad_norm": 0.2905726134777069,
      "learning_rate": 6.347411876786281e-05,
      "loss": 0.0786,
      "step": 5756
    },
    {
      "epoch": 10.964336661911554,
      "grad_norm": 0.24987512826919556,
      "learning_rate": 6.346776754525246e-05,
      "loss": 0.1001,
      "step": 5757
    },
    {
      "epoch": 10.966238706609605,
      "grad_norm": 0.2912735939025879,
      "learning_rate": 6.346141632264211e-05,
      "loss": 0.1025,
      "step": 5758
    },
    {
      "epoch": 10.968140751307656,
      "grad_norm": 0.40985098481178284,
      "learning_rate": 6.345506510003176e-05,
      "loss": 0.0943,
      "step": 5759
    },
    {
      "epoch": 10.970042796005707,
      "grad_norm": 0.26428380608558655,
      "learning_rate": 6.34487138774214e-05,
      "loss": 0.0693,
      "step": 5760
    },
    {
      "epoch": 10.971944840703756,
      "grad_norm": 0.29792144894599915,
      "learning_rate": 6.344236265481105e-05,
      "loss": 0.1097,
      "step": 5761
    },
    {
      "epoch": 10.973846885401807,
      "grad_norm": 0.21788249909877777,
      "learning_rate": 6.34360114322007e-05,
      "loss": 0.0709,
      "step": 5762
    },
    {
      "epoch": 10.975748930099858,
      "grad_norm": 0.16237157583236694,
      "learning_rate": 6.342966020959034e-05,
      "loss": 0.0689,
      "step": 5763
    },
    {
      "epoch": 10.977650974797907,
      "grad_norm": 0.16543160378932953,
      "learning_rate": 6.342330898698e-05,
      "loss": 0.0638,
      "step": 5764
    },
    {
      "epoch": 10.979553019495958,
      "grad_norm": 0.24611502885818481,
      "learning_rate": 6.341695776436965e-05,
      "loss": 0.0938,
      "step": 5765
    },
    {
      "epoch": 10.981455064194009,
      "grad_norm": 0.17132847011089325,
      "learning_rate": 6.341060654175929e-05,
      "loss": 0.0471,
      "step": 5766
    },
    {
      "epoch": 10.98335710889206,
      "grad_norm": 0.3223244845867157,
      "learning_rate": 6.340425531914894e-05,
      "loss": 0.2433,
      "step": 5767
    },
    {
      "epoch": 10.985259153590109,
      "grad_norm": 0.37807127833366394,
      "learning_rate": 6.339790409653859e-05,
      "loss": 0.0779,
      "step": 5768
    },
    {
      "epoch": 10.98716119828816,
      "grad_norm": 0.30722638964653015,
      "learning_rate": 6.339155287392824e-05,
      "loss": 0.0833,
      "step": 5769
    },
    {
      "epoch": 10.98906324298621,
      "grad_norm": 0.16426853835582733,
      "learning_rate": 6.338520165131788e-05,
      "loss": 0.0615,
      "step": 5770
    },
    {
      "epoch": 10.990965287684261,
      "grad_norm": 0.17217318713665009,
      "learning_rate": 6.337885042870753e-05,
      "loss": 0.0548,
      "step": 5771
    },
    {
      "epoch": 10.99286733238231,
      "grad_norm": 0.24564428627490997,
      "learning_rate": 6.337249920609718e-05,
      "loss": 0.0566,
      "step": 5772
    },
    {
      "epoch": 10.994769377080361,
      "grad_norm": 0.26936957240104675,
      "learning_rate": 6.336614798348682e-05,
      "loss": 0.0648,
      "step": 5773
    },
    {
      "epoch": 10.996671421778412,
      "grad_norm": 0.21403786540031433,
      "learning_rate": 6.335979676087646e-05,
      "loss": 0.049,
      "step": 5774
    },
    {
      "epoch": 10.998573466476461,
      "grad_norm": 0.2741341292858124,
      "learning_rate": 6.335344553826613e-05,
      "loss": 0.0736,
      "step": 5775
    },
    {
      "epoch": 11.000475511174512,
      "grad_norm": 0.2551502287387848,
      "learning_rate": 6.334709431565576e-05,
      "loss": 0.0683,
      "step": 5776
    },
    {
      "epoch": 11.002377555872563,
      "grad_norm": 0.23173901438713074,
      "learning_rate": 6.334074309304542e-05,
      "loss": 0.072,
      "step": 5777
    },
    {
      "epoch": 11.004279600570614,
      "grad_norm": 0.1508857011795044,
      "learning_rate": 6.333439187043505e-05,
      "loss": 0.0782,
      "step": 5778
    },
    {
      "epoch": 11.006181645268663,
      "grad_norm": 0.17948079109191895,
      "learning_rate": 6.33280406478247e-05,
      "loss": 0.0658,
      "step": 5779
    },
    {
      "epoch": 11.008083689966714,
      "grad_norm": 0.1505243331193924,
      "learning_rate": 6.332168942521436e-05,
      "loss": 0.0351,
      "step": 5780
    },
    {
      "epoch": 11.009985734664765,
      "grad_norm": 0.19849568605422974,
      "learning_rate": 6.3315338202604e-05,
      "loss": 0.0602,
      "step": 5781
    },
    {
      "epoch": 11.011887779362816,
      "grad_norm": 0.2975490093231201,
      "learning_rate": 6.330898697999366e-05,
      "loss": 0.0693,
      "step": 5782
    },
    {
      "epoch": 11.013789824060865,
      "grad_norm": 0.2738147974014282,
      "learning_rate": 6.33026357573833e-05,
      "loss": 0.0974,
      "step": 5783
    },
    {
      "epoch": 11.015691868758916,
      "grad_norm": 0.11051417142152786,
      "learning_rate": 6.329628453477294e-05,
      "loss": 0.0597,
      "step": 5784
    },
    {
      "epoch": 11.017593913456967,
      "grad_norm": 0.28424665331840515,
      "learning_rate": 6.328993331216259e-05,
      "loss": 0.0704,
      "step": 5785
    },
    {
      "epoch": 11.019495958155016,
      "grad_norm": 0.1699250191450119,
      "learning_rate": 6.328358208955224e-05,
      "loss": 0.0625,
      "step": 5786
    },
    {
      "epoch": 11.021398002853067,
      "grad_norm": 0.23104004561901093,
      "learning_rate": 6.32772308669419e-05,
      "loss": 0.0612,
      "step": 5787
    },
    {
      "epoch": 11.023300047551118,
      "grad_norm": 0.22150813043117523,
      "learning_rate": 6.327087964433153e-05,
      "loss": 0.0616,
      "step": 5788
    },
    {
      "epoch": 11.025202092249168,
      "grad_norm": 0.24865034222602844,
      "learning_rate": 6.326452842172118e-05,
      "loss": 0.0589,
      "step": 5789
    },
    {
      "epoch": 11.027104136947218,
      "grad_norm": 0.14965787529945374,
      "learning_rate": 6.325817719911084e-05,
      "loss": 0.0652,
      "step": 5790
    },
    {
      "epoch": 11.029006181645268,
      "grad_norm": 0.22783611714839935,
      "learning_rate": 6.325182597650047e-05,
      "loss": 0.0545,
      "step": 5791
    },
    {
      "epoch": 11.03090822634332,
      "grad_norm": 0.34645095467567444,
      "learning_rate": 6.324547475389013e-05,
      "loss": 0.0643,
      "step": 5792
    },
    {
      "epoch": 11.03281027104137,
      "grad_norm": 0.18645666539669037,
      "learning_rate": 6.323912353127978e-05,
      "loss": 0.0603,
      "step": 5793
    },
    {
      "epoch": 11.03471231573942,
      "grad_norm": 0.13025419414043427,
      "learning_rate": 6.323277230866942e-05,
      "loss": 0.0434,
      "step": 5794
    },
    {
      "epoch": 11.03661436043747,
      "grad_norm": 0.24453479051589966,
      "learning_rate": 6.322642108605907e-05,
      "loss": 0.0606,
      "step": 5795
    },
    {
      "epoch": 11.038516405135521,
      "grad_norm": 0.29672566056251526,
      "learning_rate": 6.322006986344872e-05,
      "loss": 0.0697,
      "step": 5796
    },
    {
      "epoch": 11.04041844983357,
      "grad_norm": 0.2354295700788498,
      "learning_rate": 6.321371864083836e-05,
      "loss": 0.061,
      "step": 5797
    },
    {
      "epoch": 11.042320494531621,
      "grad_norm": 0.2746586799621582,
      "learning_rate": 6.320736741822801e-05,
      "loss": 0.0486,
      "step": 5798
    },
    {
      "epoch": 11.044222539229672,
      "grad_norm": 0.5878947973251343,
      "learning_rate": 6.320101619561766e-05,
      "loss": 0.1112,
      "step": 5799
    },
    {
      "epoch": 11.046124583927723,
      "grad_norm": 0.19655099511146545,
      "learning_rate": 6.319466497300731e-05,
      "loss": 0.0582,
      "step": 5800
    },
    {
      "epoch": 11.048026628625772,
      "grad_norm": 0.17464180290699005,
      "learning_rate": 6.318831375039695e-05,
      "loss": 0.0595,
      "step": 5801
    },
    {
      "epoch": 11.049928673323823,
      "grad_norm": 0.31295058131217957,
      "learning_rate": 6.31819625277866e-05,
      "loss": 0.0695,
      "step": 5802
    },
    {
      "epoch": 11.051830718021874,
      "grad_norm": 0.18527528643608093,
      "learning_rate": 6.317561130517626e-05,
      "loss": 0.0813,
      "step": 5803
    },
    {
      "epoch": 11.053732762719925,
      "grad_norm": 0.24729035794734955,
      "learning_rate": 6.31692600825659e-05,
      "loss": 0.053,
      "step": 5804
    },
    {
      "epoch": 11.055634807417974,
      "grad_norm": 0.24672453105449677,
      "learning_rate": 6.316290885995555e-05,
      "loss": 0.0678,
      "step": 5805
    },
    {
      "epoch": 11.057536852116025,
      "grad_norm": 0.3050698935985565,
      "learning_rate": 6.31565576373452e-05,
      "loss": 0.0641,
      "step": 5806
    },
    {
      "epoch": 11.059438896814076,
      "grad_norm": 0.20497417449951172,
      "learning_rate": 6.315020641473484e-05,
      "loss": 0.0874,
      "step": 5807
    },
    {
      "epoch": 11.061340941512125,
      "grad_norm": 0.2367674559354782,
      "learning_rate": 6.314385519212449e-05,
      "loss": 0.0716,
      "step": 5808
    },
    {
      "epoch": 11.063242986210176,
      "grad_norm": 0.315449595451355,
      "learning_rate": 6.313750396951414e-05,
      "loss": 0.0681,
      "step": 5809
    },
    {
      "epoch": 11.065145030908226,
      "grad_norm": 0.19548985362052917,
      "learning_rate": 6.313115274690378e-05,
      "loss": 0.0749,
      "step": 5810
    },
    {
      "epoch": 11.067047075606277,
      "grad_norm": 0.2502439320087433,
      "learning_rate": 6.312480152429343e-05,
      "loss": 0.0923,
      "step": 5811
    },
    {
      "epoch": 11.068949120304326,
      "grad_norm": 0.27582550048828125,
      "learning_rate": 6.311845030168307e-05,
      "loss": 0.0694,
      "step": 5812
    },
    {
      "epoch": 11.070851165002377,
      "grad_norm": 0.19255176186561584,
      "learning_rate": 6.311209907907273e-05,
      "loss": 0.0664,
      "step": 5813
    },
    {
      "epoch": 11.072753209700428,
      "grad_norm": 0.316190630197525,
      "learning_rate": 6.310574785646237e-05,
      "loss": 0.1079,
      "step": 5814
    },
    {
      "epoch": 11.074655254398479,
      "grad_norm": 0.20605596899986267,
      "learning_rate": 6.309939663385201e-05,
      "loss": 0.0579,
      "step": 5815
    },
    {
      "epoch": 11.076557299096528,
      "grad_norm": 0.23222453892230988,
      "learning_rate": 6.309304541124168e-05,
      "loss": 0.051,
      "step": 5816
    },
    {
      "epoch": 11.078459343794579,
      "grad_norm": 0.24020570516586304,
      "learning_rate": 6.308669418863131e-05,
      "loss": 0.0639,
      "step": 5817
    },
    {
      "epoch": 11.08036138849263,
      "grad_norm": 0.25853022933006287,
      "learning_rate": 6.308034296602097e-05,
      "loss": 0.0623,
      "step": 5818
    },
    {
      "epoch": 11.08226343319068,
      "grad_norm": 0.20045647025108337,
      "learning_rate": 6.30739917434106e-05,
      "loss": 0.0501,
      "step": 5819
    },
    {
      "epoch": 11.08416547788873,
      "grad_norm": 0.22093622386455536,
      "learning_rate": 6.306764052080026e-05,
      "loss": 0.0672,
      "step": 5820
    },
    {
      "epoch": 11.08606752258678,
      "grad_norm": 0.21147139370441437,
      "learning_rate": 6.306128929818991e-05,
      "loss": 0.0595,
      "step": 5821
    },
    {
      "epoch": 11.087969567284832,
      "grad_norm": 0.19494277238845825,
      "learning_rate": 6.305493807557955e-05,
      "loss": 0.0687,
      "step": 5822
    },
    {
      "epoch": 11.08987161198288,
      "grad_norm": 0.2994452118873596,
      "learning_rate": 6.304858685296921e-05,
      "loss": 0.0588,
      "step": 5823
    },
    {
      "epoch": 11.091773656680932,
      "grad_norm": 0.3022996485233307,
      "learning_rate": 6.304223563035885e-05,
      "loss": 0.0742,
      "step": 5824
    },
    {
      "epoch": 11.093675701378983,
      "grad_norm": 0.1606903374195099,
      "learning_rate": 6.303588440774849e-05,
      "loss": 0.0511,
      "step": 5825
    },
    {
      "epoch": 11.095577746077034,
      "grad_norm": 0.2217331975698471,
      "learning_rate": 6.302953318513814e-05,
      "loss": 0.0645,
      "step": 5826
    },
    {
      "epoch": 11.097479790775083,
      "grad_norm": 0.18725329637527466,
      "learning_rate": 6.302318196252779e-05,
      "loss": 0.066,
      "step": 5827
    },
    {
      "epoch": 11.099381835473134,
      "grad_norm": 0.17725598812103271,
      "learning_rate": 6.301683073991743e-05,
      "loss": 0.0502,
      "step": 5828
    },
    {
      "epoch": 11.101283880171184,
      "grad_norm": 0.1918100267648697,
      "learning_rate": 6.301047951730708e-05,
      "loss": 0.0616,
      "step": 5829
    },
    {
      "epoch": 11.103185924869235,
      "grad_norm": 0.2146424800157547,
      "learning_rate": 6.300412829469673e-05,
      "loss": 0.0629,
      "step": 5830
    },
    {
      "epoch": 11.105087969567284,
      "grad_norm": 0.20890381932258606,
      "learning_rate": 6.299777707208639e-05,
      "loss": 0.0567,
      "step": 5831
    },
    {
      "epoch": 11.106990014265335,
      "grad_norm": 0.19044464826583862,
      "learning_rate": 6.299142584947602e-05,
      "loss": 0.0497,
      "step": 5832
    },
    {
      "epoch": 11.108892058963386,
      "grad_norm": 0.2516787350177765,
      "learning_rate": 6.298507462686568e-05,
      "loss": 0.0568,
      "step": 5833
    },
    {
      "epoch": 11.110794103661435,
      "grad_norm": 0.23537124693393707,
      "learning_rate": 6.297872340425533e-05,
      "loss": 0.0664,
      "step": 5834
    },
    {
      "epoch": 11.112696148359486,
      "grad_norm": 0.27227261662483215,
      "learning_rate": 6.297237218164497e-05,
      "loss": 0.0628,
      "step": 5835
    },
    {
      "epoch": 11.114598193057537,
      "grad_norm": 0.24858912825584412,
      "learning_rate": 6.296602095903462e-05,
      "loss": 0.0837,
      "step": 5836
    },
    {
      "epoch": 11.116500237755588,
      "grad_norm": 0.23548686504364014,
      "learning_rate": 6.295966973642427e-05,
      "loss": 0.0812,
      "step": 5837
    },
    {
      "epoch": 11.118402282453637,
      "grad_norm": 0.2094622701406479,
      "learning_rate": 6.295331851381391e-05,
      "loss": 0.0523,
      "step": 5838
    },
    {
      "epoch": 11.120304327151688,
      "grad_norm": 0.2200409173965454,
      "learning_rate": 6.294696729120356e-05,
      "loss": 0.0663,
      "step": 5839
    },
    {
      "epoch": 11.122206371849739,
      "grad_norm": 0.2630188465118408,
      "learning_rate": 6.294061606859321e-05,
      "loss": 0.0799,
      "step": 5840
    },
    {
      "epoch": 11.12410841654779,
      "grad_norm": 0.17814132571220398,
      "learning_rate": 6.293426484598286e-05,
      "loss": 0.0582,
      "step": 5841
    },
    {
      "epoch": 11.126010461245839,
      "grad_norm": 0.18654607236385345,
      "learning_rate": 6.29279136233725e-05,
      "loss": 0.0812,
      "step": 5842
    },
    {
      "epoch": 11.12791250594389,
      "grad_norm": 0.2401699423789978,
      "learning_rate": 6.292156240076214e-05,
      "loss": 0.0606,
      "step": 5843
    },
    {
      "epoch": 11.12981455064194,
      "grad_norm": 0.15061615407466888,
      "learning_rate": 6.29152111781518e-05,
      "loss": 0.0574,
      "step": 5844
    },
    {
      "epoch": 11.13171659533999,
      "grad_norm": 0.096590057015419,
      "learning_rate": 6.290885995554144e-05,
      "loss": 0.0534,
      "step": 5845
    },
    {
      "epoch": 11.13361864003804,
      "grad_norm": 0.13583779335021973,
      "learning_rate": 6.290250873293108e-05,
      "loss": 0.0677,
      "step": 5846
    },
    {
      "epoch": 11.135520684736091,
      "grad_norm": 0.2538928687572479,
      "learning_rate": 6.289615751032075e-05,
      "loss": 0.0754,
      "step": 5847
    },
    {
      "epoch": 11.137422729434142,
      "grad_norm": 0.19595909118652344,
      "learning_rate": 6.288980628771039e-05,
      "loss": 0.0668,
      "step": 5848
    },
    {
      "epoch": 11.139324774132191,
      "grad_norm": 0.2578146159648895,
      "learning_rate": 6.288345506510004e-05,
      "loss": 0.0577,
      "step": 5849
    },
    {
      "epoch": 11.141226818830242,
      "grad_norm": 0.13636615872383118,
      "learning_rate": 6.287710384248968e-05,
      "loss": 0.0453,
      "step": 5850
    },
    {
      "epoch": 11.143128863528293,
      "grad_norm": 0.26093125343322754,
      "learning_rate": 6.287075261987933e-05,
      "loss": 0.0587,
      "step": 5851
    },
    {
      "epoch": 11.145030908226344,
      "grad_norm": 0.356971800327301,
      "learning_rate": 6.286440139726898e-05,
      "loss": 0.0575,
      "step": 5852
    },
    {
      "epoch": 11.146932952924393,
      "grad_norm": 0.25038501620292664,
      "learning_rate": 6.285805017465862e-05,
      "loss": 0.067,
      "step": 5853
    },
    {
      "epoch": 11.148834997622444,
      "grad_norm": 0.19863051176071167,
      "learning_rate": 6.285169895204828e-05,
      "loss": 0.0696,
      "step": 5854
    },
    {
      "epoch": 11.150737042320495,
      "grad_norm": 0.183120995759964,
      "learning_rate": 6.284534772943792e-05,
      "loss": 0.0579,
      "step": 5855
    },
    {
      "epoch": 11.152639087018544,
      "grad_norm": 0.16368988156318665,
      "learning_rate": 6.283899650682756e-05,
      "loss": 0.0694,
      "step": 5856
    },
    {
      "epoch": 11.154541131716595,
      "grad_norm": 0.4161554276943207,
      "learning_rate": 6.283264528421721e-05,
      "loss": 0.0811,
      "step": 5857
    },
    {
      "epoch": 11.156443176414646,
      "grad_norm": 0.2229108363389969,
      "learning_rate": 6.282629406160686e-05,
      "loss": 0.0694,
      "step": 5858
    },
    {
      "epoch": 11.158345221112697,
      "grad_norm": 0.39521241188049316,
      "learning_rate": 6.281994283899652e-05,
      "loss": 0.0755,
      "step": 5859
    },
    {
      "epoch": 11.160247265810746,
      "grad_norm": 0.17866304516792297,
      "learning_rate": 6.281359161638615e-05,
      "loss": 0.0602,
      "step": 5860
    },
    {
      "epoch": 11.162149310508797,
      "grad_norm": 0.21596479415893555,
      "learning_rate": 6.28072403937758e-05,
      "loss": 0.0337,
      "step": 5861
    },
    {
      "epoch": 11.164051355206848,
      "grad_norm": 0.21442416310310364,
      "learning_rate": 6.280088917116546e-05,
      "loss": 0.0639,
      "step": 5862
    },
    {
      "epoch": 11.165953399904899,
      "grad_norm": 0.24515525996685028,
      "learning_rate": 6.27945379485551e-05,
      "loss": 0.0851,
      "step": 5863
    },
    {
      "epoch": 11.167855444602948,
      "grad_norm": 0.3923701047897339,
      "learning_rate": 6.278818672594475e-05,
      "loss": 0.0787,
      "step": 5864
    },
    {
      "epoch": 11.169757489300999,
      "grad_norm": 0.13907523453235626,
      "learning_rate": 6.27818355033344e-05,
      "loss": 0.0994,
      "step": 5865
    },
    {
      "epoch": 11.17165953399905,
      "grad_norm": 0.22324611246585846,
      "learning_rate": 6.277548428072404e-05,
      "loss": 0.0684,
      "step": 5866
    },
    {
      "epoch": 11.173561578697099,
      "grad_norm": 0.20631495118141174,
      "learning_rate": 6.276913305811369e-05,
      "loss": 0.0605,
      "step": 5867
    },
    {
      "epoch": 11.17546362339515,
      "grad_norm": 0.15408509969711304,
      "learning_rate": 6.276278183550334e-05,
      "loss": 0.057,
      "step": 5868
    },
    {
      "epoch": 11.1773656680932,
      "grad_norm": 0.2680682837963104,
      "learning_rate": 6.275643061289298e-05,
      "loss": 0.0606,
      "step": 5869
    },
    {
      "epoch": 11.179267712791251,
      "grad_norm": 0.22061650454998016,
      "learning_rate": 6.275007939028263e-05,
      "loss": 0.0615,
      "step": 5870
    },
    {
      "epoch": 11.1811697574893,
      "grad_norm": 0.23415584862232208,
      "learning_rate": 6.274372816767228e-05,
      "loss": 0.067,
      "step": 5871
    },
    {
      "epoch": 11.183071802187351,
      "grad_norm": 0.25447916984558105,
      "learning_rate": 6.273737694506193e-05,
      "loss": 0.0595,
      "step": 5872
    },
    {
      "epoch": 11.184973846885402,
      "grad_norm": 0.21616342663764954,
      "learning_rate": 6.273102572245157e-05,
      "loss": 0.0573,
      "step": 5873
    },
    {
      "epoch": 11.186875891583453,
      "grad_norm": 0.2780592739582062,
      "learning_rate": 6.272467449984122e-05,
      "loss": 0.1027,
      "step": 5874
    },
    {
      "epoch": 11.188777936281502,
      "grad_norm": 0.1951284557580948,
      "learning_rate": 6.271832327723088e-05,
      "loss": 0.0686,
      "step": 5875
    },
    {
      "epoch": 11.190679980979553,
      "grad_norm": 0.25119706988334656,
      "learning_rate": 6.271197205462052e-05,
      "loss": 0.0727,
      "step": 5876
    },
    {
      "epoch": 11.192582025677604,
      "grad_norm": 0.17263376712799072,
      "learning_rate": 6.270562083201017e-05,
      "loss": 0.0717,
      "step": 5877
    },
    {
      "epoch": 11.194484070375653,
      "grad_norm": 0.24591860175132751,
      "learning_rate": 6.269926960939982e-05,
      "loss": 0.0794,
      "step": 5878
    },
    {
      "epoch": 11.196386115073704,
      "grad_norm": 0.21584731340408325,
      "learning_rate": 6.269291838678946e-05,
      "loss": 0.0759,
      "step": 5879
    },
    {
      "epoch": 11.198288159771755,
      "grad_norm": 0.20340055227279663,
      "learning_rate": 6.268656716417911e-05,
      "loss": 0.0579,
      "step": 5880
    },
    {
      "epoch": 11.200190204469806,
      "grad_norm": 0.23323297500610352,
      "learning_rate": 6.268021594156876e-05,
      "loss": 0.0683,
      "step": 5881
    },
    {
      "epoch": 11.202092249167855,
      "grad_norm": 0.26051902770996094,
      "learning_rate": 6.26738647189584e-05,
      "loss": 0.0615,
      "step": 5882
    },
    {
      "epoch": 11.203994293865906,
      "grad_norm": 0.1682775467634201,
      "learning_rate": 6.266751349634805e-05,
      "loss": 0.0721,
      "step": 5883
    },
    {
      "epoch": 11.205896338563957,
      "grad_norm": 0.13243019580841064,
      "learning_rate": 6.266116227373769e-05,
      "loss": 0.0512,
      "step": 5884
    },
    {
      "epoch": 11.207798383262007,
      "grad_norm": 0.18601296842098236,
      "learning_rate": 6.265481105112735e-05,
      "loss": 0.053,
      "step": 5885
    },
    {
      "epoch": 11.209700427960057,
      "grad_norm": 0.14149025082588196,
      "learning_rate": 6.264845982851699e-05,
      "loss": 0.0629,
      "step": 5886
    },
    {
      "epoch": 11.211602472658107,
      "grad_norm": 0.12751762568950653,
      "learning_rate": 6.264210860590663e-05,
      "loss": 0.0582,
      "step": 5887
    },
    {
      "epoch": 11.213504517356158,
      "grad_norm": 0.20913732051849365,
      "learning_rate": 6.263575738329628e-05,
      "loss": 0.0564,
      "step": 5888
    },
    {
      "epoch": 11.21540656205421,
      "grad_norm": 0.2129901647567749,
      "learning_rate": 6.262940616068593e-05,
      "loss": 0.0887,
      "step": 5889
    },
    {
      "epoch": 11.217308606752258,
      "grad_norm": 0.2636778652667999,
      "learning_rate": 6.262305493807559e-05,
      "loss": 0.0714,
      "step": 5890
    },
    {
      "epoch": 11.21921065145031,
      "grad_norm": 0.20317071676254272,
      "learning_rate": 6.261670371546522e-05,
      "loss": 0.0539,
      "step": 5891
    },
    {
      "epoch": 11.22111269614836,
      "grad_norm": 0.13354192674160004,
      "learning_rate": 6.261035249285488e-05,
      "loss": 0.0637,
      "step": 5892
    },
    {
      "epoch": 11.22301474084641,
      "grad_norm": 0.17662078142166138,
      "learning_rate": 6.260400127024453e-05,
      "loss": 0.0594,
      "step": 5893
    },
    {
      "epoch": 11.22491678554446,
      "grad_norm": 0.22467391192913055,
      "learning_rate": 6.259765004763417e-05,
      "loss": 0.0635,
      "step": 5894
    },
    {
      "epoch": 11.226818830242511,
      "grad_norm": 0.23158535361289978,
      "learning_rate": 6.259129882502382e-05,
      "loss": 0.0479,
      "step": 5895
    },
    {
      "epoch": 11.228720874940562,
      "grad_norm": 0.2320852428674698,
      "learning_rate": 6.258494760241347e-05,
      "loss": 0.0787,
      "step": 5896
    },
    {
      "epoch": 11.230622919638611,
      "grad_norm": 0.26288601756095886,
      "learning_rate": 6.257859637980311e-05,
      "loss": 0.0807,
      "step": 5897
    },
    {
      "epoch": 11.232524964336662,
      "grad_norm": 0.15741541981697083,
      "learning_rate": 6.257224515719276e-05,
      "loss": 0.0616,
      "step": 5898
    },
    {
      "epoch": 11.234427009034713,
      "grad_norm": 0.17461591958999634,
      "learning_rate": 6.256589393458241e-05,
      "loss": 0.0524,
      "step": 5899
    },
    {
      "epoch": 11.236329053732764,
      "grad_norm": 0.2302362471818924,
      "learning_rate": 6.255954271197205e-05,
      "loss": 0.0652,
      "step": 5900
    },
    {
      "epoch": 11.238231098430813,
      "grad_norm": 0.18687783181667328,
      "learning_rate": 6.25531914893617e-05,
      "loss": 0.0526,
      "step": 5901
    },
    {
      "epoch": 11.240133143128864,
      "grad_norm": 0.22332634031772614,
      "learning_rate": 6.254684026675135e-05,
      "loss": 0.0668,
      "step": 5902
    },
    {
      "epoch": 11.242035187826914,
      "grad_norm": 0.2076207846403122,
      "learning_rate": 6.2540489044141e-05,
      "loss": 0.0583,
      "step": 5903
    },
    {
      "epoch": 11.243937232524964,
      "grad_norm": 0.2083384096622467,
      "learning_rate": 6.253413782153064e-05,
      "loss": 0.0623,
      "step": 5904
    },
    {
      "epoch": 11.245839277223014,
      "grad_norm": 0.28720328211784363,
      "learning_rate": 6.25277865989203e-05,
      "loss": 0.0853,
      "step": 5905
    },
    {
      "epoch": 11.247741321921065,
      "grad_norm": 0.24125772714614868,
      "learning_rate": 6.252143537630995e-05,
      "loss": 0.0643,
      "step": 5906
    },
    {
      "epoch": 11.249643366619116,
      "grad_norm": 0.13032442331314087,
      "learning_rate": 6.251508415369959e-05,
      "loss": 0.0647,
      "step": 5907
    },
    {
      "epoch": 11.251545411317165,
      "grad_norm": 0.25721585750579834,
      "learning_rate": 6.250873293108924e-05,
      "loss": 0.0678,
      "step": 5908
    },
    {
      "epoch": 11.253447456015216,
      "grad_norm": 0.28258436918258667,
      "learning_rate": 6.250238170847889e-05,
      "loss": 0.0573,
      "step": 5909
    },
    {
      "epoch": 11.255349500713267,
      "grad_norm": 0.20877690613269806,
      "learning_rate": 6.249603048586853e-05,
      "loss": 0.0682,
      "step": 5910
    },
    {
      "epoch": 11.257251545411318,
      "grad_norm": 0.2083483189344406,
      "learning_rate": 6.248967926325818e-05,
      "loss": 0.068,
      "step": 5911
    },
    {
      "epoch": 11.259153590109367,
      "grad_norm": 0.1268092691898346,
      "learning_rate": 6.248332804064783e-05,
      "loss": 0.0624,
      "step": 5912
    },
    {
      "epoch": 11.261055634807418,
      "grad_norm": 0.18287968635559082,
      "learning_rate": 6.247697681803748e-05,
      "loss": 0.0581,
      "step": 5913
    },
    {
      "epoch": 11.262957679505469,
      "grad_norm": 0.27067553997039795,
      "learning_rate": 6.247062559542712e-05,
      "loss": 0.0549,
      "step": 5914
    },
    {
      "epoch": 11.264859724203518,
      "grad_norm": 0.23166772723197937,
      "learning_rate": 6.246427437281676e-05,
      "loss": 0.0617,
      "step": 5915
    },
    {
      "epoch": 11.266761768901569,
      "grad_norm": 0.2581332325935364,
      "learning_rate": 6.245792315020643e-05,
      "loss": 0.0632,
      "step": 5916
    },
    {
      "epoch": 11.26866381359962,
      "grad_norm": 0.14740905165672302,
      "learning_rate": 6.245157192759606e-05,
      "loss": 0.0505,
      "step": 5917
    },
    {
      "epoch": 11.27056585829767,
      "grad_norm": 0.2767736613750458,
      "learning_rate": 6.24452207049857e-05,
      "loss": 0.0766,
      "step": 5918
    },
    {
      "epoch": 11.27246790299572,
      "grad_norm": 0.2171592116355896,
      "learning_rate": 6.243886948237537e-05,
      "loss": 0.0722,
      "step": 5919
    },
    {
      "epoch": 11.27436994769377,
      "grad_norm": 0.2533954381942749,
      "learning_rate": 6.2432518259765e-05,
      "loss": 0.0748,
      "step": 5920
    },
    {
      "epoch": 11.276271992391822,
      "grad_norm": 0.30213838815689087,
      "learning_rate": 6.242616703715466e-05,
      "loss": 0.0741,
      "step": 5921
    },
    {
      "epoch": 11.278174037089872,
      "grad_norm": 0.2629832327365875,
      "learning_rate": 6.24198158145443e-05,
      "loss": 0.0613,
      "step": 5922
    },
    {
      "epoch": 11.280076081787922,
      "grad_norm": 0.16284258663654327,
      "learning_rate": 6.241346459193395e-05,
      "loss": 0.0562,
      "step": 5923
    },
    {
      "epoch": 11.281978126485972,
      "grad_norm": 0.23348303139209747,
      "learning_rate": 6.24071133693236e-05,
      "loss": 0.0698,
      "step": 5924
    },
    {
      "epoch": 11.283880171184023,
      "grad_norm": 0.2706508934497833,
      "learning_rate": 6.240076214671324e-05,
      "loss": 0.0806,
      "step": 5925
    },
    {
      "epoch": 11.285782215882072,
      "grad_norm": 0.14358559250831604,
      "learning_rate": 6.23944109241029e-05,
      "loss": 0.0485,
      "step": 5926
    },
    {
      "epoch": 11.287684260580123,
      "grad_norm": 0.19689619541168213,
      "learning_rate": 6.238805970149254e-05,
      "loss": 0.0785,
      "step": 5927
    },
    {
      "epoch": 11.289586305278174,
      "grad_norm": 0.2577039301395416,
      "learning_rate": 6.238170847888218e-05,
      "loss": 0.1031,
      "step": 5928
    },
    {
      "epoch": 11.291488349976225,
      "grad_norm": 0.18427234888076782,
      "learning_rate": 6.237535725627183e-05,
      "loss": 0.0651,
      "step": 5929
    },
    {
      "epoch": 11.293390394674274,
      "grad_norm": 0.2167913019657135,
      "learning_rate": 6.236900603366148e-05,
      "loss": 0.0556,
      "step": 5930
    },
    {
      "epoch": 11.295292439372325,
      "grad_norm": 0.1679394543170929,
      "learning_rate": 6.236265481105114e-05,
      "loss": 0.0697,
      "step": 5931
    },
    {
      "epoch": 11.297194484070376,
      "grad_norm": 0.15036652982234955,
      "learning_rate": 6.235630358844077e-05,
      "loss": 0.0572,
      "step": 5932
    },
    {
      "epoch": 11.299096528768427,
      "grad_norm": 0.285196989774704,
      "learning_rate": 6.234995236583043e-05,
      "loss": 0.0848,
      "step": 5933
    },
    {
      "epoch": 11.300998573466476,
      "grad_norm": 0.17883095145225525,
      "learning_rate": 6.234360114322008e-05,
      "loss": 0.0508,
      "step": 5934
    },
    {
      "epoch": 11.302900618164527,
      "grad_norm": 0.21533527970314026,
      "learning_rate": 6.233724992060972e-05,
      "loss": 0.0493,
      "step": 5935
    },
    {
      "epoch": 11.304802662862578,
      "grad_norm": 0.19913125038146973,
      "learning_rate": 6.233089869799937e-05,
      "loss": 0.0754,
      "step": 5936
    },
    {
      "epoch": 11.306704707560627,
      "grad_norm": 0.20028865337371826,
      "learning_rate": 6.232454747538902e-05,
      "loss": 0.0516,
      "step": 5937
    },
    {
      "epoch": 11.308606752258678,
      "grad_norm": 0.22111880779266357,
      "learning_rate": 6.231819625277866e-05,
      "loss": 0.0631,
      "step": 5938
    },
    {
      "epoch": 11.310508796956729,
      "grad_norm": 0.24731788039207458,
      "learning_rate": 6.231184503016831e-05,
      "loss": 0.0715,
      "step": 5939
    },
    {
      "epoch": 11.31241084165478,
      "grad_norm": 0.13716444373130798,
      "learning_rate": 6.230549380755796e-05,
      "loss": 0.0511,
      "step": 5940
    },
    {
      "epoch": 11.314312886352829,
      "grad_norm": 0.19190862774848938,
      "learning_rate": 6.22991425849476e-05,
      "loss": 0.0606,
      "step": 5941
    },
    {
      "epoch": 11.31621493105088,
      "grad_norm": 0.2019043266773224,
      "learning_rate": 6.229279136233725e-05,
      "loss": 0.0483,
      "step": 5942
    },
    {
      "epoch": 11.31811697574893,
      "grad_norm": 0.17971554398536682,
      "learning_rate": 6.22864401397269e-05,
      "loss": 0.0536,
      "step": 5943
    },
    {
      "epoch": 11.320019020446981,
      "grad_norm": 0.19725970923900604,
      "learning_rate": 6.228008891711656e-05,
      "loss": 0.0824,
      "step": 5944
    },
    {
      "epoch": 11.32192106514503,
      "grad_norm": 0.3006778955459595,
      "learning_rate": 6.22737376945062e-05,
      "loss": 0.0767,
      "step": 5945
    },
    {
      "epoch": 11.323823109843081,
      "grad_norm": 0.1861002892255783,
      "learning_rate": 6.226738647189583e-05,
      "loss": 0.0626,
      "step": 5946
    },
    {
      "epoch": 11.325725154541132,
      "grad_norm": 0.22243209183216095,
      "learning_rate": 6.22610352492855e-05,
      "loss": 0.0664,
      "step": 5947
    },
    {
      "epoch": 11.327627199239181,
      "grad_norm": 0.20367178320884705,
      "learning_rate": 6.225468402667514e-05,
      "loss": 0.0541,
      "step": 5948
    },
    {
      "epoch": 11.329529243937232,
      "grad_norm": 0.23790258169174194,
      "learning_rate": 6.224833280406479e-05,
      "loss": 0.0684,
      "step": 5949
    },
    {
      "epoch": 11.331431288635283,
      "grad_norm": 0.2381439507007599,
      "learning_rate": 6.224198158145444e-05,
      "loss": 0.0833,
      "step": 5950
    },
    {
      "epoch": 11.333333333333334,
      "grad_norm": 0.2214234173297882,
      "learning_rate": 6.223563035884408e-05,
      "loss": 0.0623,
      "step": 5951
    },
    {
      "epoch": 11.335235378031383,
      "grad_norm": 0.24603192508220673,
      "learning_rate": 6.222927913623373e-05,
      "loss": 0.0916,
      "step": 5952
    },
    {
      "epoch": 11.337137422729434,
      "grad_norm": 0.23510603606700897,
      "learning_rate": 6.222292791362337e-05,
      "loss": 0.0894,
      "step": 5953
    },
    {
      "epoch": 11.339039467427485,
      "grad_norm": 0.2977652847766876,
      "learning_rate": 6.221657669101302e-05,
      "loss": 0.0824,
      "step": 5954
    },
    {
      "epoch": 11.340941512125536,
      "grad_norm": 0.15970540046691895,
      "learning_rate": 6.221022546840267e-05,
      "loss": 0.0599,
      "step": 5955
    },
    {
      "epoch": 11.342843556823585,
      "grad_norm": 0.12921825051307678,
      "learning_rate": 6.220387424579231e-05,
      "loss": 0.0651,
      "step": 5956
    },
    {
      "epoch": 11.344745601521636,
      "grad_norm": 0.16822974383831024,
      "learning_rate": 6.219752302318198e-05,
      "loss": 0.0669,
      "step": 5957
    },
    {
      "epoch": 11.346647646219687,
      "grad_norm": 0.1776653677225113,
      "learning_rate": 6.219117180057161e-05,
      "loss": 0.0584,
      "step": 5958
    },
    {
      "epoch": 11.348549690917736,
      "grad_norm": 0.19017675518989563,
      "learning_rate": 6.218482057796125e-05,
      "loss": 0.0642,
      "step": 5959
    },
    {
      "epoch": 11.350451735615787,
      "grad_norm": 0.2337440699338913,
      "learning_rate": 6.21784693553509e-05,
      "loss": 0.0648,
      "step": 5960
    },
    {
      "epoch": 11.352353780313837,
      "grad_norm": 0.2026948630809784,
      "learning_rate": 6.217211813274056e-05,
      "loss": 0.0603,
      "step": 5961
    },
    {
      "epoch": 11.354255825011888,
      "grad_norm": 0.3038780391216278,
      "learning_rate": 6.216576691013021e-05,
      "loss": 0.2317,
      "step": 5962
    },
    {
      "epoch": 11.356157869709937,
      "grad_norm": 0.2820383906364441,
      "learning_rate": 6.215941568751985e-05,
      "loss": 0.1064,
      "step": 5963
    },
    {
      "epoch": 11.358059914407988,
      "grad_norm": 0.361411452293396,
      "learning_rate": 6.21530644649095e-05,
      "loss": 0.1318,
      "step": 5964
    },
    {
      "epoch": 11.35996195910604,
      "grad_norm": 0.17481759190559387,
      "learning_rate": 6.214671324229915e-05,
      "loss": 0.0419,
      "step": 5965
    },
    {
      "epoch": 11.36186400380409,
      "grad_norm": 0.3438234329223633,
      "learning_rate": 6.214036201968879e-05,
      "loss": 0.0683,
      "step": 5966
    },
    {
      "epoch": 11.36376604850214,
      "grad_norm": 0.2617831826210022,
      "learning_rate": 6.213401079707844e-05,
      "loss": 0.0683,
      "step": 5967
    },
    {
      "epoch": 11.36566809320019,
      "grad_norm": 0.21333356201648712,
      "learning_rate": 6.212765957446809e-05,
      "loss": 0.0591,
      "step": 5968
    },
    {
      "epoch": 11.367570137898241,
      "grad_norm": 0.16227330267429352,
      "learning_rate": 6.212130835185773e-05,
      "loss": 0.061,
      "step": 5969
    },
    {
      "epoch": 11.36947218259629,
      "grad_norm": 0.13485564291477203,
      "learning_rate": 6.211495712924738e-05,
      "loss": 0.0865,
      "step": 5970
    },
    {
      "epoch": 11.371374227294341,
      "grad_norm": 0.16355431079864502,
      "learning_rate": 6.210860590663703e-05,
      "loss": 0.1023,
      "step": 5971
    },
    {
      "epoch": 11.373276271992392,
      "grad_norm": 0.20975372195243835,
      "learning_rate": 6.210225468402667e-05,
      "loss": 0.061,
      "step": 5972
    },
    {
      "epoch": 11.375178316690443,
      "grad_norm": 0.22960126399993896,
      "learning_rate": 6.209590346141632e-05,
      "loss": 0.0706,
      "step": 5973
    },
    {
      "epoch": 11.377080361388492,
      "grad_norm": 0.19620640575885773,
      "learning_rate": 6.208955223880598e-05,
      "loss": 0.0595,
      "step": 5974
    },
    {
      "epoch": 11.378982406086543,
      "grad_norm": 0.17666909098625183,
      "learning_rate": 6.208320101619563e-05,
      "loss": 0.0581,
      "step": 5975
    },
    {
      "epoch": 11.380884450784594,
      "grad_norm": 0.18484234809875488,
      "learning_rate": 6.207684979358527e-05,
      "loss": 0.0614,
      "step": 5976
    },
    {
      "epoch": 11.382786495482645,
      "grad_norm": 0.18495036661624908,
      "learning_rate": 6.207049857097492e-05,
      "loss": 0.061,
      "step": 5977
    },
    {
      "epoch": 11.384688540180694,
      "grad_norm": 0.19153162837028503,
      "learning_rate": 6.206414734836457e-05,
      "loss": 0.0624,
      "step": 5978
    },
    {
      "epoch": 11.386590584878745,
      "grad_norm": 0.20312979817390442,
      "learning_rate": 6.205779612575421e-05,
      "loss": 0.0728,
      "step": 5979
    },
    {
      "epoch": 11.388492629576795,
      "grad_norm": 0.19322583079338074,
      "learning_rate": 6.205144490314386e-05,
      "loss": 0.0545,
      "step": 5980
    },
    {
      "epoch": 11.390394674274846,
      "grad_norm": 0.16400569677352905,
      "learning_rate": 6.204509368053351e-05,
      "loss": 0.0508,
      "step": 5981
    },
    {
      "epoch": 11.392296718972895,
      "grad_norm": 0.19000458717346191,
      "learning_rate": 6.203874245792315e-05,
      "loss": 0.0747,
      "step": 5982
    },
    {
      "epoch": 11.394198763670946,
      "grad_norm": 0.2547617554664612,
      "learning_rate": 6.20323912353128e-05,
      "loss": 0.0712,
      "step": 5983
    },
    {
      "epoch": 11.396100808368997,
      "grad_norm": 0.23105841875076294,
      "learning_rate": 6.202604001270245e-05,
      "loss": 0.0629,
      "step": 5984
    },
    {
      "epoch": 11.398002853067046,
      "grad_norm": 0.2565494477748871,
      "learning_rate": 6.20196887900921e-05,
      "loss": 0.0691,
      "step": 5985
    },
    {
      "epoch": 11.399904897765097,
      "grad_norm": 0.35790374875068665,
      "learning_rate": 6.201333756748174e-05,
      "loss": 0.1012,
      "step": 5986
    },
    {
      "epoch": 11.401806942463148,
      "grad_norm": 0.21136760711669922,
      "learning_rate": 6.200698634487138e-05,
      "loss": 0.0689,
      "step": 5987
    },
    {
      "epoch": 11.403708987161199,
      "grad_norm": 0.3775070011615753,
      "learning_rate": 6.200063512226105e-05,
      "loss": 0.0807,
      "step": 5988
    },
    {
      "epoch": 11.405611031859248,
      "grad_norm": 0.1303168386220932,
      "learning_rate": 6.199428389965069e-05,
      "loss": 0.0422,
      "step": 5989
    },
    {
      "epoch": 11.407513076557299,
      "grad_norm": 0.233234703540802,
      "learning_rate": 6.198793267704032e-05,
      "loss": 0.0829,
      "step": 5990
    },
    {
      "epoch": 11.40941512125535,
      "grad_norm": 0.18227219581604004,
      "learning_rate": 6.198158145442999e-05,
      "loss": 0.0711,
      "step": 5991
    },
    {
      "epoch": 11.4113171659534,
      "grad_norm": 0.18231059610843658,
      "learning_rate": 6.197523023181963e-05,
      "loss": 0.0441,
      "step": 5992
    },
    {
      "epoch": 11.41321921065145,
      "grad_norm": 0.2130158692598343,
      "learning_rate": 6.196887900920928e-05,
      "loss": 0.083,
      "step": 5993
    },
    {
      "epoch": 11.4151212553495,
      "grad_norm": 0.18165136873722076,
      "learning_rate": 6.196252778659892e-05,
      "loss": 0.0659,
      "step": 5994
    },
    {
      "epoch": 11.417023300047552,
      "grad_norm": 0.1659289002418518,
      "learning_rate": 6.195617656398857e-05,
      "loss": 0.0444,
      "step": 5995
    },
    {
      "epoch": 11.4189253447456,
      "grad_norm": 0.2109125852584839,
      "learning_rate": 6.194982534137822e-05,
      "loss": 0.0707,
      "step": 5996
    },
    {
      "epoch": 11.420827389443652,
      "grad_norm": 0.24702899158000946,
      "learning_rate": 6.194347411876786e-05,
      "loss": 0.0764,
      "step": 5997
    },
    {
      "epoch": 11.422729434141702,
      "grad_norm": 0.26356109976768494,
      "learning_rate": 6.193712289615751e-05,
      "loss": 0.0847,
      "step": 5998
    },
    {
      "epoch": 11.424631478839753,
      "grad_norm": 0.2942047119140625,
      "learning_rate": 6.193077167354716e-05,
      "loss": 0.069,
      "step": 5999
    },
    {
      "epoch": 11.426533523537802,
      "grad_norm": 0.19051453471183777,
      "learning_rate": 6.19244204509368e-05,
      "loss": 0.0714,
      "step": 6000
    },
    {
      "epoch": 11.428435568235853,
      "grad_norm": 0.24365760385990143,
      "learning_rate": 6.191806922832645e-05,
      "loss": 0.0772,
      "step": 6001
    },
    {
      "epoch": 11.430337612933904,
      "grad_norm": 0.19398783147335052,
      "learning_rate": 6.19117180057161e-05,
      "loss": 0.0523,
      "step": 6002
    },
    {
      "epoch": 11.432239657631955,
      "grad_norm": 0.23657700419425964,
      "learning_rate": 6.190536678310576e-05,
      "loss": 0.0653,
      "step": 6003
    },
    {
      "epoch": 11.434141702330004,
      "grad_norm": 0.21983860433101654,
      "learning_rate": 6.18990155604954e-05,
      "loss": 0.0706,
      "step": 6004
    },
    {
      "epoch": 11.436043747028055,
      "grad_norm": 0.17263579368591309,
      "learning_rate": 6.189266433788505e-05,
      "loss": 0.0641,
      "step": 6005
    },
    {
      "epoch": 11.437945791726106,
      "grad_norm": 0.45329514145851135,
      "learning_rate": 6.18863131152747e-05,
      "loss": 0.0845,
      "step": 6006
    },
    {
      "epoch": 11.439847836424155,
      "grad_norm": 0.2627212703227997,
      "learning_rate": 6.187996189266434e-05,
      "loss": 0.0679,
      "step": 6007
    },
    {
      "epoch": 11.441749881122206,
      "grad_norm": 0.2166578322649002,
      "learning_rate": 6.187361067005399e-05,
      "loss": 0.0726,
      "step": 6008
    },
    {
      "epoch": 11.443651925820257,
      "grad_norm": 0.23665811121463776,
      "learning_rate": 6.186725944744364e-05,
      "loss": 0.055,
      "step": 6009
    },
    {
      "epoch": 11.445553970518308,
      "grad_norm": 0.19395120441913605,
      "learning_rate": 6.186090822483328e-05,
      "loss": 0.0775,
      "step": 6010
    },
    {
      "epoch": 11.447456015216357,
      "grad_norm": 0.2175108790397644,
      "learning_rate": 6.185455700222293e-05,
      "loss": 0.061,
      "step": 6011
    },
    {
      "epoch": 11.449358059914408,
      "grad_norm": 0.25349605083465576,
      "learning_rate": 6.184820577961258e-05,
      "loss": 0.0624,
      "step": 6012
    },
    {
      "epoch": 11.451260104612459,
      "grad_norm": 0.22870026528835297,
      "learning_rate": 6.184185455700222e-05,
      "loss": 0.0701,
      "step": 6013
    },
    {
      "epoch": 11.45316214931051,
      "grad_norm": 0.2290843427181244,
      "learning_rate": 6.183550333439187e-05,
      "loss": 0.0703,
      "step": 6014
    },
    {
      "epoch": 11.455064194008559,
      "grad_norm": 0.2366465926170349,
      "learning_rate": 6.182915211178152e-05,
      "loss": 0.0896,
      "step": 6015
    },
    {
      "epoch": 11.45696623870661,
      "grad_norm": 0.18010316789150238,
      "learning_rate": 6.182280088917118e-05,
      "loss": 0.0661,
      "step": 6016
    },
    {
      "epoch": 11.45886828340466,
      "grad_norm": 0.1060408502817154,
      "learning_rate": 6.181644966656081e-05,
      "loss": 0.0629,
      "step": 6017
    },
    {
      "epoch": 11.46077032810271,
      "grad_norm": 0.24757403135299683,
      "learning_rate": 6.181009844395045e-05,
      "loss": 0.0557,
      "step": 6018
    },
    {
      "epoch": 11.46267237280076,
      "grad_norm": 0.21169930696487427,
      "learning_rate": 6.180374722134012e-05,
      "loss": 0.0692,
      "step": 6019
    },
    {
      "epoch": 11.464574417498811,
      "grad_norm": 0.19375531375408173,
      "learning_rate": 6.179739599872976e-05,
      "loss": 0.0625,
      "step": 6020
    },
    {
      "epoch": 11.466476462196862,
      "grad_norm": 0.2409812957048416,
      "learning_rate": 6.179104477611941e-05,
      "loss": 0.0438,
      "step": 6021
    },
    {
      "epoch": 11.468378506894911,
      "grad_norm": 0.18499095737934113,
      "learning_rate": 6.178469355350906e-05,
      "loss": 0.0741,
      "step": 6022
    },
    {
      "epoch": 11.470280551592962,
      "grad_norm": 0.26087823510169983,
      "learning_rate": 6.17783423308987e-05,
      "loss": 0.1246,
      "step": 6023
    },
    {
      "epoch": 11.472182596291013,
      "grad_norm": 0.21440978348255157,
      "learning_rate": 6.177199110828835e-05,
      "loss": 0.058,
      "step": 6024
    },
    {
      "epoch": 11.474084640989064,
      "grad_norm": 0.17176677286624908,
      "learning_rate": 6.176563988567799e-05,
      "loss": 0.0616,
      "step": 6025
    },
    {
      "epoch": 11.475986685687113,
      "grad_norm": 0.3351137042045593,
      "learning_rate": 6.175928866306764e-05,
      "loss": 0.0703,
      "step": 6026
    },
    {
      "epoch": 11.477888730385164,
      "grad_norm": 0.22019176185131073,
      "learning_rate": 6.175293744045729e-05,
      "loss": 0.0614,
      "step": 6027
    },
    {
      "epoch": 11.479790775083215,
      "grad_norm": 0.2089587152004242,
      "learning_rate": 6.174658621784693e-05,
      "loss": 0.0662,
      "step": 6028
    },
    {
      "epoch": 11.481692819781266,
      "grad_norm": 0.21846778690814972,
      "learning_rate": 6.17402349952366e-05,
      "loss": 0.0718,
      "step": 6029
    },
    {
      "epoch": 11.483594864479315,
      "grad_norm": 0.235796257853508,
      "learning_rate": 6.173388377262623e-05,
      "loss": 0.0806,
      "step": 6030
    },
    {
      "epoch": 11.485496909177366,
      "grad_norm": 0.17812518775463104,
      "learning_rate": 6.172753255001587e-05,
      "loss": 0.0638,
      "step": 6031
    },
    {
      "epoch": 11.487398953875417,
      "grad_norm": 0.3472003638744354,
      "learning_rate": 6.172118132740552e-05,
      "loss": 0.0726,
      "step": 6032
    },
    {
      "epoch": 11.489300998573466,
      "grad_norm": 0.22589357197284698,
      "learning_rate": 6.171483010479518e-05,
      "loss": 0.0789,
      "step": 6033
    },
    {
      "epoch": 11.491203043271517,
      "grad_norm": 0.15773439407348633,
      "learning_rate": 6.170847888218483e-05,
      "loss": 0.0628,
      "step": 6034
    },
    {
      "epoch": 11.493105087969568,
      "grad_norm": 0.20131023228168488,
      "learning_rate": 6.170212765957447e-05,
      "loss": 0.0671,
      "step": 6035
    },
    {
      "epoch": 11.495007132667618,
      "grad_norm": 0.27972444891929626,
      "learning_rate": 6.169577643696412e-05,
      "loss": 0.0747,
      "step": 6036
    },
    {
      "epoch": 11.496909177365668,
      "grad_norm": 0.14555224776268005,
      "learning_rate": 6.168942521435377e-05,
      "loss": 0.0625,
      "step": 6037
    },
    {
      "epoch": 11.498811222063718,
      "grad_norm": 0.1408931314945221,
      "learning_rate": 6.168307399174341e-05,
      "loss": 0.0775,
      "step": 6038
    },
    {
      "epoch": 11.50071326676177,
      "grad_norm": 0.2279130518436432,
      "learning_rate": 6.167672276913306e-05,
      "loss": 0.0592,
      "step": 6039
    },
    {
      "epoch": 11.50261531145982,
      "grad_norm": 0.33479738235473633,
      "learning_rate": 6.167037154652271e-05,
      "loss": 0.0841,
      "step": 6040
    },
    {
      "epoch": 11.50451735615787,
      "grad_norm": 0.34034156799316406,
      "learning_rate": 6.166402032391235e-05,
      "loss": 0.0792,
      "step": 6041
    },
    {
      "epoch": 11.50641940085592,
      "grad_norm": 0.3499704599380493,
      "learning_rate": 6.1657669101302e-05,
      "loss": 0.0841,
      "step": 6042
    },
    {
      "epoch": 11.508321445553971,
      "grad_norm": 0.2372085303068161,
      "learning_rate": 6.165131787869165e-05,
      "loss": 0.0597,
      "step": 6043
    },
    {
      "epoch": 11.51022349025202,
      "grad_norm": 0.2807712256908417,
      "learning_rate": 6.164496665608129e-05,
      "loss": 0.0565,
      "step": 6044
    },
    {
      "epoch": 11.512125534950071,
      "grad_norm": 0.2463337928056717,
      "learning_rate": 6.163861543347094e-05,
      "loss": 0.069,
      "step": 6045
    },
    {
      "epoch": 11.514027579648122,
      "grad_norm": 0.19078849256038666,
      "learning_rate": 6.16322642108606e-05,
      "loss": 0.0503,
      "step": 6046
    },
    {
      "epoch": 11.515929624346173,
      "grad_norm": 0.2622719705104828,
      "learning_rate": 6.162591298825025e-05,
      "loss": 0.0877,
      "step": 6047
    },
    {
      "epoch": 11.517831669044222,
      "grad_norm": 0.16791850328445435,
      "learning_rate": 6.161956176563989e-05,
      "loss": 0.0764,
      "step": 6048
    },
    {
      "epoch": 11.519733713742273,
      "grad_norm": 0.15910501778125763,
      "learning_rate": 6.161321054302952e-05,
      "loss": 0.0728,
      "step": 6049
    },
    {
      "epoch": 11.521635758440324,
      "grad_norm": 0.13227106630802155,
      "learning_rate": 6.160685932041919e-05,
      "loss": 0.0574,
      "step": 6050
    },
    {
      "epoch": 11.523537803138375,
      "grad_norm": 0.20476002991199493,
      "learning_rate": 6.160050809780883e-05,
      "loss": 0.0523,
      "step": 6051
    },
    {
      "epoch": 11.525439847836424,
      "grad_norm": 0.20992450416088104,
      "learning_rate": 6.159415687519848e-05,
      "loss": 0.0625,
      "step": 6052
    },
    {
      "epoch": 11.527341892534475,
      "grad_norm": 0.19773100316524506,
      "learning_rate": 6.158780565258813e-05,
      "loss": 0.0705,
      "step": 6053
    },
    {
      "epoch": 11.529243937232525,
      "grad_norm": 0.20119865238666534,
      "learning_rate": 6.158145442997777e-05,
      "loss": 0.0512,
      "step": 6054
    },
    {
      "epoch": 11.531145981930575,
      "grad_norm": 0.1957743763923645,
      "learning_rate": 6.157510320736742e-05,
      "loss": 0.0553,
      "step": 6055
    },
    {
      "epoch": 11.533048026628625,
      "grad_norm": 0.18962952494621277,
      "learning_rate": 6.156875198475706e-05,
      "loss": 0.0679,
      "step": 6056
    },
    {
      "epoch": 11.534950071326676,
      "grad_norm": 0.29589298367500305,
      "learning_rate": 6.156240076214673e-05,
      "loss": 0.0769,
      "step": 6057
    },
    {
      "epoch": 11.536852116024727,
      "grad_norm": 0.22001637518405914,
      "learning_rate": 6.155604953953636e-05,
      "loss": 0.0712,
      "step": 6058
    },
    {
      "epoch": 11.538754160722776,
      "grad_norm": 0.16552984714508057,
      "learning_rate": 6.1549698316926e-05,
      "loss": 0.0724,
      "step": 6059
    },
    {
      "epoch": 11.540656205420827,
      "grad_norm": 0.27614063024520874,
      "learning_rate": 6.154334709431567e-05,
      "loss": 0.0616,
      "step": 6060
    },
    {
      "epoch": 11.542558250118878,
      "grad_norm": 0.3362590968608856,
      "learning_rate": 6.15369958717053e-05,
      "loss": 0.076,
      "step": 6061
    },
    {
      "epoch": 11.544460294816929,
      "grad_norm": 0.31461089849472046,
      "learning_rate": 6.153064464909494e-05,
      "loss": 0.0668,
      "step": 6062
    },
    {
      "epoch": 11.546362339514978,
      "grad_norm": 0.2893616557121277,
      "learning_rate": 6.15242934264846e-05,
      "loss": 0.0526,
      "step": 6063
    },
    {
      "epoch": 11.548264384213029,
      "grad_norm": 0.19768260419368744,
      "learning_rate": 6.151794220387425e-05,
      "loss": 0.0504,
      "step": 6064
    },
    {
      "epoch": 11.55016642891108,
      "grad_norm": 0.21805433928966522,
      "learning_rate": 6.15115909812639e-05,
      "loss": 0.0522,
      "step": 6065
    },
    {
      "epoch": 11.552068473609129,
      "grad_norm": 0.2028207778930664,
      "learning_rate": 6.150523975865354e-05,
      "loss": 0.0526,
      "step": 6066
    },
    {
      "epoch": 11.55397051830718,
      "grad_norm": 0.11325007677078247,
      "learning_rate": 6.149888853604319e-05,
      "loss": 0.0516,
      "step": 6067
    },
    {
      "epoch": 11.55587256300523,
      "grad_norm": 0.24469703435897827,
      "learning_rate": 6.149253731343284e-05,
      "loss": 0.059,
      "step": 6068
    },
    {
      "epoch": 11.557774607703282,
      "grad_norm": 0.21558788418769836,
      "learning_rate": 6.148618609082248e-05,
      "loss": 0.0621,
      "step": 6069
    },
    {
      "epoch": 11.55967665240133,
      "grad_norm": 0.17346946895122528,
      "learning_rate": 6.147983486821213e-05,
      "loss": 0.06,
      "step": 6070
    },
    {
      "epoch": 11.561578697099382,
      "grad_norm": 0.24504558742046356,
      "learning_rate": 6.147348364560178e-05,
      "loss": 0.0723,
      "step": 6071
    },
    {
      "epoch": 11.563480741797433,
      "grad_norm": 0.1529906839132309,
      "learning_rate": 6.146713242299142e-05,
      "loss": 0.0782,
      "step": 6072
    },
    {
      "epoch": 11.565382786495483,
      "grad_norm": 0.2217208594083786,
      "learning_rate": 6.146078120038107e-05,
      "loss": 0.0485,
      "step": 6073
    },
    {
      "epoch": 11.567284831193533,
      "grad_norm": 0.23646852374076843,
      "learning_rate": 6.145442997777073e-05,
      "loss": 0.0632,
      "step": 6074
    },
    {
      "epoch": 11.569186875891583,
      "grad_norm": 0.24585440754890442,
      "learning_rate": 6.144807875516038e-05,
      "loss": 0.0679,
      "step": 6075
    },
    {
      "epoch": 11.571088920589634,
      "grad_norm": 0.22161781787872314,
      "learning_rate": 6.144172753255002e-05,
      "loss": 0.0743,
      "step": 6076
    },
    {
      "epoch": 11.572990965287683,
      "grad_norm": 0.20924946665763855,
      "learning_rate": 6.143537630993967e-05,
      "loss": 0.0625,
      "step": 6077
    },
    {
      "epoch": 11.574893009985734,
      "grad_norm": 0.1991322785615921,
      "learning_rate": 6.142902508732932e-05,
      "loss": 0.0647,
      "step": 6078
    },
    {
      "epoch": 11.576795054683785,
      "grad_norm": 0.14237317442893982,
      "learning_rate": 6.142267386471896e-05,
      "loss": 0.059,
      "step": 6079
    },
    {
      "epoch": 11.578697099381836,
      "grad_norm": 0.17688287794589996,
      "learning_rate": 6.141632264210861e-05,
      "loss": 0.071,
      "step": 6080
    },
    {
      "epoch": 11.580599144079885,
      "grad_norm": 0.3290722966194153,
      "learning_rate": 6.140997141949826e-05,
      "loss": 0.1133,
      "step": 6081
    },
    {
      "epoch": 11.582501188777936,
      "grad_norm": 0.2535814642906189,
      "learning_rate": 6.14036201968879e-05,
      "loss": 0.0844,
      "step": 6082
    },
    {
      "epoch": 11.584403233475987,
      "grad_norm": 0.17662294209003448,
      "learning_rate": 6.139726897427755e-05,
      "loss": 0.088,
      "step": 6083
    },
    {
      "epoch": 11.586305278174038,
      "grad_norm": 0.42101120948791504,
      "learning_rate": 6.13909177516672e-05,
      "loss": 0.0983,
      "step": 6084
    },
    {
      "epoch": 11.588207322872087,
      "grad_norm": 0.15525983273983002,
      "learning_rate": 6.138456652905684e-05,
      "loss": 0.0685,
      "step": 6085
    },
    {
      "epoch": 11.590109367570138,
      "grad_norm": 0.19920846819877625,
      "learning_rate": 6.13782153064465e-05,
      "loss": 0.0664,
      "step": 6086
    },
    {
      "epoch": 11.592011412268189,
      "grad_norm": 0.17457710206508636,
      "learning_rate": 6.137186408383615e-05,
      "loss": 0.0697,
      "step": 6087
    },
    {
      "epoch": 11.593913456966238,
      "grad_norm": 0.12437298893928528,
      "learning_rate": 6.13655128612258e-05,
      "loss": 0.0557,
      "step": 6088
    },
    {
      "epoch": 11.595815501664289,
      "grad_norm": 0.25111961364746094,
      "learning_rate": 6.135916163861544e-05,
      "loss": 0.0669,
      "step": 6089
    },
    {
      "epoch": 11.59771754636234,
      "grad_norm": 0.15608292818069458,
      "learning_rate": 6.135281041600507e-05,
      "loss": 0.0836,
      "step": 6090
    },
    {
      "epoch": 11.59961959106039,
      "grad_norm": 0.24482889473438263,
      "learning_rate": 6.134645919339474e-05,
      "loss": 0.0685,
      "step": 6091
    },
    {
      "epoch": 11.60152163575844,
      "grad_norm": 0.27814391255378723,
      "learning_rate": 6.134010797078438e-05,
      "loss": 0.083,
      "step": 6092
    },
    {
      "epoch": 11.60342368045649,
      "grad_norm": 0.25812262296676636,
      "learning_rate": 6.133375674817403e-05,
      "loss": 0.0642,
      "step": 6093
    },
    {
      "epoch": 11.605325725154541,
      "grad_norm": 0.14833666384220123,
      "learning_rate": 6.132740552556368e-05,
      "loss": 0.0636,
      "step": 6094
    },
    {
      "epoch": 11.607227769852592,
      "grad_norm": 0.21950408816337585,
      "learning_rate": 6.132105430295332e-05,
      "loss": 0.052,
      "step": 6095
    },
    {
      "epoch": 11.609129814550641,
      "grad_norm": 0.2043406218290329,
      "learning_rate": 6.131470308034297e-05,
      "loss": 0.0664,
      "step": 6096
    },
    {
      "epoch": 11.611031859248692,
      "grad_norm": 0.12655547261238098,
      "learning_rate": 6.130835185773261e-05,
      "loss": 0.044,
      "step": 6097
    },
    {
      "epoch": 11.612933903946743,
      "grad_norm": 0.30291444063186646,
      "learning_rate": 6.130200063512226e-05,
      "loss": 0.0844,
      "step": 6098
    },
    {
      "epoch": 11.614835948644792,
      "grad_norm": 0.18023452162742615,
      "learning_rate": 6.129564941251191e-05,
      "loss": 0.065,
      "step": 6099
    },
    {
      "epoch": 11.616737993342843,
      "grad_norm": 0.1753673255443573,
      "learning_rate": 6.128929818990155e-05,
      "loss": 0.0577,
      "step": 6100
    },
    {
      "epoch": 11.618640038040894,
      "grad_norm": 0.2921854853630066,
      "learning_rate": 6.12829469672912e-05,
      "loss": 0.0798,
      "step": 6101
    },
    {
      "epoch": 11.620542082738945,
      "grad_norm": 0.2377692610025406,
      "learning_rate": 6.127659574468086e-05,
      "loss": 0.0592,
      "step": 6102
    },
    {
      "epoch": 11.622444127436994,
      "grad_norm": 0.1325494349002838,
      "learning_rate": 6.12702445220705e-05,
      "loss": 0.0464,
      "step": 6103
    },
    {
      "epoch": 11.624346172135045,
      "grad_norm": 0.2446260154247284,
      "learning_rate": 6.126389329946015e-05,
      "loss": 0.0591,
      "step": 6104
    },
    {
      "epoch": 11.626248216833096,
      "grad_norm": 0.24112503230571747,
      "learning_rate": 6.12575420768498e-05,
      "loss": 0.0636,
      "step": 6105
    },
    {
      "epoch": 11.628150261531147,
      "grad_norm": 0.21624989807605743,
      "learning_rate": 6.125119085423945e-05,
      "loss": 0.0573,
      "step": 6106
    },
    {
      "epoch": 11.630052306229196,
      "grad_norm": 0.29112082719802856,
      "learning_rate": 6.124483963162909e-05,
      "loss": 0.055,
      "step": 6107
    },
    {
      "epoch": 11.631954350927247,
      "grad_norm": 0.256797730922699,
      "learning_rate": 6.123848840901874e-05,
      "loss": 0.0586,
      "step": 6108
    },
    {
      "epoch": 11.633856395625298,
      "grad_norm": 0.2393050193786621,
      "learning_rate": 6.123213718640839e-05,
      "loss": 0.0776,
      "step": 6109
    },
    {
      "epoch": 11.635758440323347,
      "grad_norm": 0.1504524052143097,
      "learning_rate": 6.122578596379803e-05,
      "loss": 0.0749,
      "step": 6110
    },
    {
      "epoch": 11.637660485021398,
      "grad_norm": 0.3078169822692871,
      "learning_rate": 6.121943474118768e-05,
      "loss": 0.0655,
      "step": 6111
    },
    {
      "epoch": 11.639562529719448,
      "grad_norm": 0.33488377928733826,
      "learning_rate": 6.121308351857733e-05,
      "loss": 0.0708,
      "step": 6112
    },
    {
      "epoch": 11.6414645744175,
      "grad_norm": 0.23643803596496582,
      "learning_rate": 6.120673229596697e-05,
      "loss": 0.0711,
      "step": 6113
    },
    {
      "epoch": 11.643366619115548,
      "grad_norm": 0.22984230518341064,
      "learning_rate": 6.120038107335662e-05,
      "loss": 0.0676,
      "step": 6114
    },
    {
      "epoch": 11.6452686638136,
      "grad_norm": 0.25781843066215515,
      "learning_rate": 6.119402985074628e-05,
      "loss": 0.0803,
      "step": 6115
    },
    {
      "epoch": 11.64717070851165,
      "grad_norm": 0.2645505964756012,
      "learning_rate": 6.118767862813591e-05,
      "loss": 0.0491,
      "step": 6116
    },
    {
      "epoch": 11.649072753209701,
      "grad_norm": 0.18775779008865356,
      "learning_rate": 6.118132740552557e-05,
      "loss": 0.0639,
      "step": 6117
    },
    {
      "epoch": 11.65097479790775,
      "grad_norm": 0.21034850180149078,
      "learning_rate": 6.117497618291522e-05,
      "loss": 0.0883,
      "step": 6118
    },
    {
      "epoch": 11.652876842605801,
      "grad_norm": 0.2566367983818054,
      "learning_rate": 6.116862496030487e-05,
      "loss": 0.0591,
      "step": 6119
    },
    {
      "epoch": 11.654778887303852,
      "grad_norm": 0.22888152301311493,
      "learning_rate": 6.116227373769451e-05,
      "loss": 0.0591,
      "step": 6120
    },
    {
      "epoch": 11.656680932001901,
      "grad_norm": 0.3078417181968689,
      "learning_rate": 6.115592251508415e-05,
      "loss": 0.0904,
      "step": 6121
    },
    {
      "epoch": 11.658582976699952,
      "grad_norm": 0.3688507676124573,
      "learning_rate": 6.114957129247381e-05,
      "loss": 0.0792,
      "step": 6122
    },
    {
      "epoch": 11.660485021398003,
      "grad_norm": 0.15711909532546997,
      "learning_rate": 6.114322006986345e-05,
      "loss": 0.0576,
      "step": 6123
    },
    {
      "epoch": 11.662387066096054,
      "grad_norm": 0.22273437678813934,
      "learning_rate": 6.11368688472531e-05,
      "loss": 0.0745,
      "step": 6124
    },
    {
      "epoch": 11.664289110794103,
      "grad_norm": 0.3121403753757477,
      "learning_rate": 6.113051762464275e-05,
      "loss": 0.0654,
      "step": 6125
    },
    {
      "epoch": 11.666191155492154,
      "grad_norm": 0.3767518997192383,
      "learning_rate": 6.112416640203239e-05,
      "loss": 0.0829,
      "step": 6126
    },
    {
      "epoch": 11.668093200190205,
      "grad_norm": 0.2748688757419586,
      "learning_rate": 6.111781517942204e-05,
      "loss": 0.0773,
      "step": 6127
    },
    {
      "epoch": 11.669995244888256,
      "grad_norm": 0.24371102452278137,
      "learning_rate": 6.111146395681168e-05,
      "loss": 0.0689,
      "step": 6128
    },
    {
      "epoch": 11.671897289586305,
      "grad_norm": 0.2633253335952759,
      "learning_rate": 6.110511273420135e-05,
      "loss": 0.0924,
      "step": 6129
    },
    {
      "epoch": 11.673799334284356,
      "grad_norm": 0.1922878921031952,
      "learning_rate": 6.109876151159099e-05,
      "loss": 0.0653,
      "step": 6130
    },
    {
      "epoch": 11.675701378982406,
      "grad_norm": 0.305594801902771,
      "learning_rate": 6.109241028898062e-05,
      "loss": 0.0625,
      "step": 6131
    },
    {
      "epoch": 11.677603423680456,
      "grad_norm": 0.1557915359735489,
      "learning_rate": 6.108605906637029e-05,
      "loss": 0.0555,
      "step": 6132
    },
    {
      "epoch": 11.679505468378506,
      "grad_norm": 0.2857607901096344,
      "learning_rate": 6.107970784375993e-05,
      "loss": 0.0755,
      "step": 6133
    },
    {
      "epoch": 11.681407513076557,
      "grad_norm": 0.2326517254114151,
      "learning_rate": 6.107335662114957e-05,
      "loss": 0.0717,
      "step": 6134
    },
    {
      "epoch": 11.683309557774608,
      "grad_norm": 0.3538796305656433,
      "learning_rate": 6.106700539853922e-05,
      "loss": 0.0812,
      "step": 6135
    },
    {
      "epoch": 11.685211602472657,
      "grad_norm": 0.17662444710731506,
      "learning_rate": 6.106065417592887e-05,
      "loss": 0.0792,
      "step": 6136
    },
    {
      "epoch": 11.687113647170708,
      "grad_norm": 0.3378899097442627,
      "learning_rate": 6.105430295331852e-05,
      "loss": 0.0608,
      "step": 6137
    },
    {
      "epoch": 11.689015691868759,
      "grad_norm": 0.14349856972694397,
      "learning_rate": 6.104795173070816e-05,
      "loss": 0.0428,
      "step": 6138
    },
    {
      "epoch": 11.69091773656681,
      "grad_norm": 0.35674959421157837,
      "learning_rate": 6.104160050809781e-05,
      "loss": 0.0936,
      "step": 6139
    },
    {
      "epoch": 11.692819781264859,
      "grad_norm": 0.2726203203201294,
      "learning_rate": 6.103524928548746e-05,
      "loss": 0.0628,
      "step": 6140
    },
    {
      "epoch": 11.69472182596291,
      "grad_norm": 0.24450047314167023,
      "learning_rate": 6.102889806287711e-05,
      "loss": 0.0539,
      "step": 6141
    },
    {
      "epoch": 11.69662387066096,
      "grad_norm": 0.223331019282341,
      "learning_rate": 6.102254684026676e-05,
      "loss": 0.058,
      "step": 6142
    },
    {
      "epoch": 11.698525915359012,
      "grad_norm": 0.22839614748954773,
      "learning_rate": 6.1016195617656405e-05,
      "loss": 0.058,
      "step": 6143
    },
    {
      "epoch": 11.70042796005706,
      "grad_norm": 0.21672922372817993,
      "learning_rate": 6.100984439504604e-05,
      "loss": 0.0782,
      "step": 6144
    },
    {
      "epoch": 11.702330004755112,
      "grad_norm": 0.18630997836589813,
      "learning_rate": 6.10034931724357e-05,
      "loss": 0.0767,
      "step": 6145
    },
    {
      "epoch": 11.704232049453163,
      "grad_norm": 0.20281566679477692,
      "learning_rate": 6.099714194982534e-05,
      "loss": 0.0924,
      "step": 6146
    },
    {
      "epoch": 11.706134094151212,
      "grad_norm": 0.2289772927761078,
      "learning_rate": 6.0990790727215e-05,
      "loss": 0.0603,
      "step": 6147
    },
    {
      "epoch": 11.708036138849263,
      "grad_norm": 0.3022541403770447,
      "learning_rate": 6.098443950460464e-05,
      "loss": 0.094,
      "step": 6148
    },
    {
      "epoch": 11.709938183547314,
      "grad_norm": 0.23966433107852936,
      "learning_rate": 6.097808828199428e-05,
      "loss": 0.0481,
      "step": 6149
    },
    {
      "epoch": 11.711840228245364,
      "grad_norm": 0.2377452850341797,
      "learning_rate": 6.097173705938394e-05,
      "loss": 0.0634,
      "step": 6150
    },
    {
      "epoch": 11.713742272943414,
      "grad_norm": 0.24796810746192932,
      "learning_rate": 6.096538583677358e-05,
      "loss": 0.078,
      "step": 6151
    },
    {
      "epoch": 11.715644317641464,
      "grad_norm": 0.18740686774253845,
      "learning_rate": 6.0959034614163224e-05,
      "loss": 0.0541,
      "step": 6152
    },
    {
      "epoch": 11.717546362339515,
      "grad_norm": 0.18637046217918396,
      "learning_rate": 6.0952683391552876e-05,
      "loss": 0.0589,
      "step": 6153
    },
    {
      "epoch": 11.719448407037566,
      "grad_norm": 0.19956031441688538,
      "learning_rate": 6.094633216894252e-05,
      "loss": 0.0686,
      "step": 6154
    },
    {
      "epoch": 11.721350451735615,
      "grad_norm": 0.25902166962623596,
      "learning_rate": 6.093998094633217e-05,
      "loss": 0.0509,
      "step": 6155
    },
    {
      "epoch": 11.723252496433666,
      "grad_norm": 0.18729211390018463,
      "learning_rate": 6.093362972372182e-05,
      "loss": 0.0846,
      "step": 6156
    },
    {
      "epoch": 11.725154541131717,
      "grad_norm": 0.22573569416999817,
      "learning_rate": 6.092727850111146e-05,
      "loss": 0.071,
      "step": 6157
    },
    {
      "epoch": 11.727056585829766,
      "grad_norm": 0.3560294806957245,
      "learning_rate": 6.0920927278501115e-05,
      "loss": 0.0857,
      "step": 6158
    },
    {
      "epoch": 11.728958630527817,
      "grad_norm": 0.15637195110321045,
      "learning_rate": 6.091457605589076e-05,
      "loss": 0.0748,
      "step": 6159
    },
    {
      "epoch": 11.730860675225868,
      "grad_norm": 0.24328051507472992,
      "learning_rate": 6.090822483328041e-05,
      "loss": 0.0702,
      "step": 6160
    },
    {
      "epoch": 11.732762719923919,
      "grad_norm": 0.17262868583202362,
      "learning_rate": 6.090187361067006e-05,
      "loss": 0.0601,
      "step": 6161
    },
    {
      "epoch": 11.734664764621968,
      "grad_norm": 0.21686933934688568,
      "learning_rate": 6.08955223880597e-05,
      "loss": 0.0809,
      "step": 6162
    },
    {
      "epoch": 11.736566809320019,
      "grad_norm": 0.2194422036409378,
      "learning_rate": 6.0889171165449354e-05,
      "loss": 0.0534,
      "step": 6163
    },
    {
      "epoch": 11.73846885401807,
      "grad_norm": 0.2666342854499817,
      "learning_rate": 6.0882819942839e-05,
      "loss": 0.0611,
      "step": 6164
    },
    {
      "epoch": 11.74037089871612,
      "grad_norm": 0.20401927828788757,
      "learning_rate": 6.087646872022865e-05,
      "loss": 0.0613,
      "step": 6165
    },
    {
      "epoch": 11.74227294341417,
      "grad_norm": 0.156899556517601,
      "learning_rate": 6.0870117497618296e-05,
      "loss": 0.0698,
      "step": 6166
    },
    {
      "epoch": 11.74417498811222,
      "grad_norm": 0.39890050888061523,
      "learning_rate": 6.086376627500794e-05,
      "loss": 0.0778,
      "step": 6167
    },
    {
      "epoch": 11.746077032810271,
      "grad_norm": 0.2203887552022934,
      "learning_rate": 6.085741505239759e-05,
      "loss": 0.071,
      "step": 6168
    },
    {
      "epoch": 11.747979077508322,
      "grad_norm": 0.22864332795143127,
      "learning_rate": 6.085106382978724e-05,
      "loss": 0.0784,
      "step": 6169
    },
    {
      "epoch": 11.749881122206371,
      "grad_norm": 0.20026814937591553,
      "learning_rate": 6.0844712607176876e-05,
      "loss": 0.0475,
      "step": 6170
    },
    {
      "epoch": 11.751783166904422,
      "grad_norm": 0.1660257875919342,
      "learning_rate": 6.0838361384566534e-05,
      "loss": 0.0513,
      "step": 6171
    },
    {
      "epoch": 11.753685211602473,
      "grad_norm": 0.17982488870620728,
      "learning_rate": 6.083201016195618e-05,
      "loss": 0.0601,
      "step": 6172
    },
    {
      "epoch": 11.755587256300522,
      "grad_norm": 0.20312677323818207,
      "learning_rate": 6.082565893934583e-05,
      "loss": 0.0605,
      "step": 6173
    },
    {
      "epoch": 11.757489300998573,
      "grad_norm": 0.2112540304660797,
      "learning_rate": 6.0819307716735476e-05,
      "loss": 0.0544,
      "step": 6174
    },
    {
      "epoch": 11.759391345696624,
      "grad_norm": 0.16244660317897797,
      "learning_rate": 6.0812956494125115e-05,
      "loss": 0.0588,
      "step": 6175
    },
    {
      "epoch": 11.761293390394675,
      "grad_norm": 0.17628826200962067,
      "learning_rate": 6.080660527151477e-05,
      "loss": 0.0646,
      "step": 6176
    },
    {
      "epoch": 11.763195435092724,
      "grad_norm": 0.11459421366453171,
      "learning_rate": 6.080025404890441e-05,
      "loss": 0.0431,
      "step": 6177
    },
    {
      "epoch": 11.765097479790775,
      "grad_norm": 0.2594488263130188,
      "learning_rate": 6.079390282629407e-05,
      "loss": 0.07,
      "step": 6178
    },
    {
      "epoch": 11.766999524488826,
      "grad_norm": 0.16236400604248047,
      "learning_rate": 6.0787551603683715e-05,
      "loss": 0.0516,
      "step": 6179
    },
    {
      "epoch": 11.768901569186877,
      "grad_norm": 0.22279828786849976,
      "learning_rate": 6.0781200381073354e-05,
      "loss": 0.0665,
      "step": 6180
    },
    {
      "epoch": 11.770803613884926,
      "grad_norm": 0.3085854947566986,
      "learning_rate": 6.077484915846301e-05,
      "loss": 0.0654,
      "step": 6181
    },
    {
      "epoch": 11.772705658582977,
      "grad_norm": 0.23716288805007935,
      "learning_rate": 6.076849793585265e-05,
      "loss": 0.0664,
      "step": 6182
    },
    {
      "epoch": 11.774607703281028,
      "grad_norm": 0.3094969689846039,
      "learning_rate": 6.076214671324231e-05,
      "loss": 0.0782,
      "step": 6183
    },
    {
      "epoch": 11.776509747979077,
      "grad_norm": 0.23545250296592712,
      "learning_rate": 6.075579549063195e-05,
      "loss": 0.0704,
      "step": 6184
    },
    {
      "epoch": 11.778411792677128,
      "grad_norm": 0.2429591715335846,
      "learning_rate": 6.074944426802159e-05,
      "loss": 0.0588,
      "step": 6185
    },
    {
      "epoch": 11.780313837375179,
      "grad_norm": 0.13099047541618347,
      "learning_rate": 6.074309304541125e-05,
      "loss": 0.0624,
      "step": 6186
    },
    {
      "epoch": 11.78221588207323,
      "grad_norm": 0.18458104133605957,
      "learning_rate": 6.073674182280089e-05,
      "loss": 0.05,
      "step": 6187
    },
    {
      "epoch": 11.784117926771279,
      "grad_norm": 0.16507314145565033,
      "learning_rate": 6.0730390600190534e-05,
      "loss": 0.0724,
      "step": 6188
    },
    {
      "epoch": 11.78601997146933,
      "grad_norm": 0.29678866267204285,
      "learning_rate": 6.0724039377580186e-05,
      "loss": 0.0681,
      "step": 6189
    },
    {
      "epoch": 11.78792201616738,
      "grad_norm": 0.2282082587480545,
      "learning_rate": 6.071768815496983e-05,
      "loss": 0.0717,
      "step": 6190
    },
    {
      "epoch": 11.789824060865431,
      "grad_norm": 0.25922587513923645,
      "learning_rate": 6.071133693235948e-05,
      "loss": 0.091,
      "step": 6191
    },
    {
      "epoch": 11.79172610556348,
      "grad_norm": 0.14782756567001343,
      "learning_rate": 6.070498570974913e-05,
      "loss": 0.0469,
      "step": 6192
    },
    {
      "epoch": 11.793628150261531,
      "grad_norm": 0.1892451047897339,
      "learning_rate": 6.069863448713877e-05,
      "loss": 0.0502,
      "step": 6193
    },
    {
      "epoch": 11.795530194959582,
      "grad_norm": 0.1491728127002716,
      "learning_rate": 6.0692283264528425e-05,
      "loss": 0.0513,
      "step": 6194
    },
    {
      "epoch": 11.797432239657631,
      "grad_norm": 0.20024257898330688,
      "learning_rate": 6.068593204191807e-05,
      "loss": 0.0671,
      "step": 6195
    },
    {
      "epoch": 11.799334284355682,
      "grad_norm": 0.2266501933336258,
      "learning_rate": 6.067958081930772e-05,
      "loss": 0.0704,
      "step": 6196
    },
    {
      "epoch": 11.801236329053733,
      "grad_norm": 0.15230099856853485,
      "learning_rate": 6.067322959669737e-05,
      "loss": 0.0547,
      "step": 6197
    },
    {
      "epoch": 11.803138373751784,
      "grad_norm": 0.33129194378852844,
      "learning_rate": 6.066687837408701e-05,
      "loss": 0.0669,
      "step": 6198
    },
    {
      "epoch": 11.805040418449833,
      "grad_norm": 0.1871880143880844,
      "learning_rate": 6.0660527151476664e-05,
      "loss": 0.0597,
      "step": 6199
    },
    {
      "epoch": 11.806942463147884,
      "grad_norm": 0.2125813364982605,
      "learning_rate": 6.065417592886631e-05,
      "loss": 0.0713,
      "step": 6200
    },
    {
      "epoch": 11.808844507845935,
      "grad_norm": 0.1894565224647522,
      "learning_rate": 6.064782470625596e-05,
      "loss": 0.053,
      "step": 6201
    },
    {
      "epoch": 11.810746552543986,
      "grad_norm": 0.18981118500232697,
      "learning_rate": 6.0641473483645606e-05,
      "loss": 0.066,
      "step": 6202
    },
    {
      "epoch": 11.812648597242035,
      "grad_norm": 0.1811751425266266,
      "learning_rate": 6.063512226103525e-05,
      "loss": 0.0551,
      "step": 6203
    },
    {
      "epoch": 11.814550641940086,
      "grad_norm": 0.23820361495018005,
      "learning_rate": 6.06287710384249e-05,
      "loss": 0.072,
      "step": 6204
    },
    {
      "epoch": 11.816452686638137,
      "grad_norm": 0.14921443164348602,
      "learning_rate": 6.062241981581455e-05,
      "loss": 0.0919,
      "step": 6205
    },
    {
      "epoch": 11.818354731336186,
      "grad_norm": 0.23790062963962555,
      "learning_rate": 6.0616068593204186e-05,
      "loss": 0.0573,
      "step": 6206
    },
    {
      "epoch": 11.820256776034237,
      "grad_norm": 0.18912047147750854,
      "learning_rate": 6.0609717370593845e-05,
      "loss": 0.0564,
      "step": 6207
    },
    {
      "epoch": 11.822158820732287,
      "grad_norm": 0.2074270099401474,
      "learning_rate": 6.060336614798349e-05,
      "loss": 0.0583,
      "step": 6208
    },
    {
      "epoch": 11.824060865430338,
      "grad_norm": 0.15312808752059937,
      "learning_rate": 6.059701492537314e-05,
      "loss": 0.0617,
      "step": 6209
    },
    {
      "epoch": 11.825962910128387,
      "grad_norm": 0.2151595503091812,
      "learning_rate": 6.059066370276279e-05,
      "loss": 0.0597,
      "step": 6210
    },
    {
      "epoch": 11.827864954826438,
      "grad_norm": 0.2581579089164734,
      "learning_rate": 6.0584312480152425e-05,
      "loss": 0.0678,
      "step": 6211
    },
    {
      "epoch": 11.82976699952449,
      "grad_norm": 0.16070884466171265,
      "learning_rate": 6.0577961257542084e-05,
      "loss": 0.0555,
      "step": 6212
    },
    {
      "epoch": 11.83166904422254,
      "grad_norm": 0.1928481012582779,
      "learning_rate": 6.057161003493172e-05,
      "loss": 0.0547,
      "step": 6213
    },
    {
      "epoch": 11.83357108892059,
      "grad_norm": 0.15289796888828278,
      "learning_rate": 6.056525881232138e-05,
      "loss": 0.0473,
      "step": 6214
    },
    {
      "epoch": 11.83547313361864,
      "grad_norm": 0.2093266397714615,
      "learning_rate": 6.0558907589711026e-05,
      "loss": 0.0844,
      "step": 6215
    },
    {
      "epoch": 11.837375178316691,
      "grad_norm": 0.259589821100235,
      "learning_rate": 6.0552556367100664e-05,
      "loss": 0.0646,
      "step": 6216
    },
    {
      "epoch": 11.83927722301474,
      "grad_norm": 0.20682299137115479,
      "learning_rate": 6.054620514449032e-05,
      "loss": 0.0715,
      "step": 6217
    },
    {
      "epoch": 11.841179267712791,
      "grad_norm": 0.16990284621715546,
      "learning_rate": 6.053985392187996e-05,
      "loss": 0.0527,
      "step": 6218
    },
    {
      "epoch": 11.843081312410842,
      "grad_norm": 0.1800818145275116,
      "learning_rate": 6.053350269926962e-05,
      "loss": 0.0584,
      "step": 6219
    },
    {
      "epoch": 11.844983357108893,
      "grad_norm": 0.31309738755226135,
      "learning_rate": 6.052715147665926e-05,
      "loss": 0.0688,
      "step": 6220
    },
    {
      "epoch": 11.846885401806942,
      "grad_norm": 0.23532579839229584,
      "learning_rate": 6.05208002540489e-05,
      "loss": 0.0611,
      "step": 6221
    },
    {
      "epoch": 11.848787446504993,
      "grad_norm": 0.20161910355091095,
      "learning_rate": 6.051444903143856e-05,
      "loss": 0.0703,
      "step": 6222
    },
    {
      "epoch": 11.850689491203044,
      "grad_norm": 0.21405024826526642,
      "learning_rate": 6.05080978088282e-05,
      "loss": 0.0669,
      "step": 6223
    },
    {
      "epoch": 11.852591535901094,
      "grad_norm": 0.33080706000328064,
      "learning_rate": 6.0501746586217845e-05,
      "loss": 0.0689,
      "step": 6224
    },
    {
      "epoch": 11.854493580599144,
      "grad_norm": 0.29687973856925964,
      "learning_rate": 6.04953953636075e-05,
      "loss": 0.0697,
      "step": 6225
    },
    {
      "epoch": 11.856395625297194,
      "grad_norm": 0.3404380679130554,
      "learning_rate": 6.048904414099714e-05,
      "loss": 0.0576,
      "step": 6226
    },
    {
      "epoch": 11.858297669995245,
      "grad_norm": 0.26443350315093994,
      "learning_rate": 6.0482692918386794e-05,
      "loss": 0.0718,
      "step": 6227
    },
    {
      "epoch": 11.860199714693294,
      "grad_norm": 0.22917382419109344,
      "learning_rate": 6.047634169577644e-05,
      "loss": 0.0833,
      "step": 6228
    },
    {
      "epoch": 11.862101759391345,
      "grad_norm": 0.3504857122898102,
      "learning_rate": 6.0469990473166084e-05,
      "loss": 0.0833,
      "step": 6229
    },
    {
      "epoch": 11.864003804089396,
      "grad_norm": 0.2867469787597656,
      "learning_rate": 6.0463639250555736e-05,
      "loss": 0.0707,
      "step": 6230
    },
    {
      "epoch": 11.865905848787447,
      "grad_norm": 0.22201254963874817,
      "learning_rate": 6.045728802794538e-05,
      "loss": 0.0761,
      "step": 6231
    },
    {
      "epoch": 11.867807893485496,
      "grad_norm": 0.20512932538986206,
      "learning_rate": 6.045093680533503e-05,
      "loss": 0.0685,
      "step": 6232
    },
    {
      "epoch": 11.869709938183547,
      "grad_norm": 0.23396115005016327,
      "learning_rate": 6.044458558272468e-05,
      "loss": 0.0619,
      "step": 6233
    },
    {
      "epoch": 11.871611982881598,
      "grad_norm": 0.19524352252483368,
      "learning_rate": 6.043823436011432e-05,
      "loss": 0.0484,
      "step": 6234
    },
    {
      "epoch": 11.873514027579649,
      "grad_norm": 0.3588017523288727,
      "learning_rate": 6.0431883137503974e-05,
      "loss": 0.0832,
      "step": 6235
    },
    {
      "epoch": 11.875416072277698,
      "grad_norm": 0.24636587500572205,
      "learning_rate": 6.042553191489362e-05,
      "loss": 0.0983,
      "step": 6236
    },
    {
      "epoch": 11.877318116975749,
      "grad_norm": 0.17110472917556763,
      "learning_rate": 6.041918069228327e-05,
      "loss": 0.0779,
      "step": 6237
    },
    {
      "epoch": 11.8792201616738,
      "grad_norm": 0.2017090618610382,
      "learning_rate": 6.0412829469672916e-05,
      "loss": 0.0645,
      "step": 6238
    },
    {
      "epoch": 11.881122206371849,
      "grad_norm": 0.26978799700737,
      "learning_rate": 6.040647824706256e-05,
      "loss": 0.0542,
      "step": 6239
    },
    {
      "epoch": 11.8830242510699,
      "grad_norm": 0.16964603960514069,
      "learning_rate": 6.040012702445221e-05,
      "loss": 0.064,
      "step": 6240
    },
    {
      "epoch": 11.88492629576795,
      "grad_norm": 0.22872652113437653,
      "learning_rate": 6.039377580184186e-05,
      "loss": 0.0655,
      "step": 6241
    },
    {
      "epoch": 11.886828340466002,
      "grad_norm": 0.2875925600528717,
      "learning_rate": 6.03874245792315e-05,
      "loss": 0.0754,
      "step": 6242
    },
    {
      "epoch": 11.88873038516405,
      "grad_norm": 0.2056693732738495,
      "learning_rate": 6.0381073356621155e-05,
      "loss": 0.0781,
      "step": 6243
    },
    {
      "epoch": 11.890632429862102,
      "grad_norm": 0.18020100891590118,
      "learning_rate": 6.03747221340108e-05,
      "loss": 0.047,
      "step": 6244
    },
    {
      "epoch": 11.892534474560152,
      "grad_norm": 0.20994068682193756,
      "learning_rate": 6.036837091140045e-05,
      "loss": 0.0528,
      "step": 6245
    },
    {
      "epoch": 11.894436519258203,
      "grad_norm": 0.2020370066165924,
      "learning_rate": 6.03620196887901e-05,
      "loss": 0.0607,
      "step": 6246
    },
    {
      "epoch": 11.896338563956252,
      "grad_norm": 0.22844986617565155,
      "learning_rate": 6.0355668466179736e-05,
      "loss": 0.0939,
      "step": 6247
    },
    {
      "epoch": 11.898240608654303,
      "grad_norm": 0.2161705046892166,
      "learning_rate": 6.0349317243569394e-05,
      "loss": 0.0408,
      "step": 6248
    },
    {
      "epoch": 11.900142653352354,
      "grad_norm": 0.202499121427536,
      "learning_rate": 6.034296602095903e-05,
      "loss": 0.063,
      "step": 6249
    },
    {
      "epoch": 11.902044698050403,
      "grad_norm": 0.2268448770046234,
      "learning_rate": 6.033661479834869e-05,
      "loss": 0.0916,
      "step": 6250
    },
    {
      "epoch": 11.903946742748454,
      "grad_norm": 0.20890720188617706,
      "learning_rate": 6.0330263575738336e-05,
      "loss": 0.0673,
      "step": 6251
    },
    {
      "epoch": 11.905848787446505,
      "grad_norm": 0.16075345873832703,
      "learning_rate": 6.0323912353127974e-05,
      "loss": 0.0671,
      "step": 6252
    },
    {
      "epoch": 11.907750832144556,
      "grad_norm": 0.19686825573444366,
      "learning_rate": 6.031756113051763e-05,
      "loss": 0.0554,
      "step": 6253
    },
    {
      "epoch": 11.909652876842605,
      "grad_norm": 0.15151557326316833,
      "learning_rate": 6.031120990790727e-05,
      "loss": 0.0482,
      "step": 6254
    },
    {
      "epoch": 11.911554921540656,
      "grad_norm": 0.3714834451675415,
      "learning_rate": 6.030485868529693e-05,
      "loss": 0.0697,
      "step": 6255
    },
    {
      "epoch": 11.913456966238707,
      "grad_norm": 0.18377116322517395,
      "learning_rate": 6.029850746268657e-05,
      "loss": 0.066,
      "step": 6256
    },
    {
      "epoch": 11.915359010936758,
      "grad_norm": 0.21345947682857513,
      "learning_rate": 6.029215624007621e-05,
      "loss": 0.0698,
      "step": 6257
    },
    {
      "epoch": 11.917261055634807,
      "grad_norm": 0.215050607919693,
      "learning_rate": 6.0285805017465865e-05,
      "loss": 0.0755,
      "step": 6258
    },
    {
      "epoch": 11.919163100332858,
      "grad_norm": 0.1479867845773697,
      "learning_rate": 6.027945379485551e-05,
      "loss": 0.0519,
      "step": 6259
    },
    {
      "epoch": 11.921065145030909,
      "grad_norm": 0.2630639374256134,
      "learning_rate": 6.0273102572245155e-05,
      "loss": 0.0523,
      "step": 6260
    },
    {
      "epoch": 11.922967189728958,
      "grad_norm": 0.2974599003791809,
      "learning_rate": 6.026675134963481e-05,
      "loss": 0.0531,
      "step": 6261
    },
    {
      "epoch": 11.924869234427009,
      "grad_norm": 0.09849124401807785,
      "learning_rate": 6.026040012702445e-05,
      "loss": 0.047,
      "step": 6262
    },
    {
      "epoch": 11.92677127912506,
      "grad_norm": 0.18132846057415009,
      "learning_rate": 6.0254048904414104e-05,
      "loss": 0.063,
      "step": 6263
    },
    {
      "epoch": 11.92867332382311,
      "grad_norm": 0.2807028591632843,
      "learning_rate": 6.024769768180375e-05,
      "loss": 0.0774,
      "step": 6264
    },
    {
      "epoch": 11.93057536852116,
      "grad_norm": 0.2209707349538803,
      "learning_rate": 6.0241346459193394e-05,
      "loss": 0.0665,
      "step": 6265
    },
    {
      "epoch": 11.93247741321921,
      "grad_norm": 0.21792303025722504,
      "learning_rate": 6.0234995236583046e-05,
      "loss": 0.0612,
      "step": 6266
    },
    {
      "epoch": 11.934379457917261,
      "grad_norm": 0.216604083776474,
      "learning_rate": 6.022864401397269e-05,
      "loss": 0.0543,
      "step": 6267
    },
    {
      "epoch": 11.936281502615312,
      "grad_norm": 0.40272215008735657,
      "learning_rate": 6.022229279136234e-05,
      "loss": 0.0773,
      "step": 6268
    },
    {
      "epoch": 11.938183547313361,
      "grad_norm": 0.2663206160068512,
      "learning_rate": 6.021594156875199e-05,
      "loss": 0.0746,
      "step": 6269
    },
    {
      "epoch": 11.940085592011412,
      "grad_norm": 0.2070496529340744,
      "learning_rate": 6.020959034614163e-05,
      "loss": 0.0441,
      "step": 6270
    },
    {
      "epoch": 11.941987636709463,
      "grad_norm": 0.11446495354175568,
      "learning_rate": 6.0203239123531285e-05,
      "loss": 0.028,
      "step": 6271
    },
    {
      "epoch": 11.943889681407512,
      "grad_norm": 0.2797466516494751,
      "learning_rate": 6.019688790092093e-05,
      "loss": 0.0703,
      "step": 6272
    },
    {
      "epoch": 11.945791726105563,
      "grad_norm": 0.1635390669107437,
      "learning_rate": 6.019053667831058e-05,
      "loss": 0.0625,
      "step": 6273
    },
    {
      "epoch": 11.947693770803614,
      "grad_norm": 0.24397385120391846,
      "learning_rate": 6.018418545570023e-05,
      "loss": 0.07,
      "step": 6274
    },
    {
      "epoch": 11.949595815501665,
      "grad_norm": 0.19531305134296417,
      "learning_rate": 6.017783423308987e-05,
      "loss": 0.0726,
      "step": 6275
    },
    {
      "epoch": 11.951497860199714,
      "grad_norm": 0.23823881149291992,
      "learning_rate": 6.0171483010479524e-05,
      "loss": 0.066,
      "step": 6276
    },
    {
      "epoch": 11.953399904897765,
      "grad_norm": 0.28544214367866516,
      "learning_rate": 6.016513178786917e-05,
      "loss": 0.0846,
      "step": 6277
    },
    {
      "epoch": 11.955301949595816,
      "grad_norm": 0.3207830488681793,
      "learning_rate": 6.015878056525881e-05,
      "loss": 0.0746,
      "step": 6278
    },
    {
      "epoch": 11.957203994293867,
      "grad_norm": 0.22940942645072937,
      "learning_rate": 6.0152429342648466e-05,
      "loss": 0.0648,
      "step": 6279
    },
    {
      "epoch": 11.959106038991916,
      "grad_norm": 0.24860794842243195,
      "learning_rate": 6.0146078120038104e-05,
      "loss": 0.0766,
      "step": 6280
    },
    {
      "epoch": 11.961008083689967,
      "grad_norm": 0.1024889349937439,
      "learning_rate": 6.013972689742776e-05,
      "loss": 0.0379,
      "step": 6281
    },
    {
      "epoch": 11.962910128388017,
      "grad_norm": 0.22909225523471832,
      "learning_rate": 6.013337567481741e-05,
      "loss": 0.0703,
      "step": 6282
    },
    {
      "epoch": 11.964812173086067,
      "grad_norm": 0.18012385070323944,
      "learning_rate": 6.0127024452207046e-05,
      "loss": 0.0596,
      "step": 6283
    },
    {
      "epoch": 11.966714217784117,
      "grad_norm": 0.31610363721847534,
      "learning_rate": 6.0120673229596705e-05,
      "loss": 0.0758,
      "step": 6284
    },
    {
      "epoch": 11.968616262482168,
      "grad_norm": 0.3586494028568268,
      "learning_rate": 6.011432200698634e-05,
      "loss": 0.0977,
      "step": 6285
    },
    {
      "epoch": 11.97051830718022,
      "grad_norm": 0.24757806956768036,
      "learning_rate": 6.0107970784376e-05,
      "loss": 0.0829,
      "step": 6286
    },
    {
      "epoch": 11.972420351878268,
      "grad_norm": 0.2095833718776703,
      "learning_rate": 6.010161956176564e-05,
      "loss": 0.0691,
      "step": 6287
    },
    {
      "epoch": 11.97432239657632,
      "grad_norm": 0.337371826171875,
      "learning_rate": 6.0095268339155285e-05,
      "loss": 0.0835,
      "step": 6288
    },
    {
      "epoch": 11.97622444127437,
      "grad_norm": 0.1434415727853775,
      "learning_rate": 6.0088917116544944e-05,
      "loss": 0.0393,
      "step": 6289
    },
    {
      "epoch": 11.978126485972421,
      "grad_norm": 0.1812358945608139,
      "learning_rate": 6.008256589393458e-05,
      "loss": 0.076,
      "step": 6290
    },
    {
      "epoch": 11.98002853067047,
      "grad_norm": 0.21688790619373322,
      "learning_rate": 6.007621467132424e-05,
      "loss": 0.0581,
      "step": 6291
    },
    {
      "epoch": 11.981930575368521,
      "grad_norm": 0.227838397026062,
      "learning_rate": 6.006986344871388e-05,
      "loss": 0.0694,
      "step": 6292
    },
    {
      "epoch": 11.983832620066572,
      "grad_norm": 0.29435136914253235,
      "learning_rate": 6.0063512226103524e-05,
      "loss": 0.0856,
      "step": 6293
    },
    {
      "epoch": 11.985734664764623,
      "grad_norm": 0.36537912487983704,
      "learning_rate": 6.0057161003493176e-05,
      "loss": 0.0635,
      "step": 6294
    },
    {
      "epoch": 11.987636709462672,
      "grad_norm": 0.26229700446128845,
      "learning_rate": 6.005080978088282e-05,
      "loss": 0.0579,
      "step": 6295
    },
    {
      "epoch": 11.989538754160723,
      "grad_norm": 0.17395451664924622,
      "learning_rate": 6.0044458558272466e-05,
      "loss": 0.0638,
      "step": 6296
    },
    {
      "epoch": 11.991440798858774,
      "grad_norm": 0.16267430782318115,
      "learning_rate": 6.003810733566212e-05,
      "loss": 0.0652,
      "step": 6297
    },
    {
      "epoch": 11.993342843556823,
      "grad_norm": 0.28298768401145935,
      "learning_rate": 6.003175611305176e-05,
      "loss": 0.0639,
      "step": 6298
    },
    {
      "epoch": 11.995244888254874,
      "grad_norm": 0.23265522718429565,
      "learning_rate": 6.0025404890441414e-05,
      "loss": 0.0675,
      "step": 6299
    },
    {
      "epoch": 11.997146932952925,
      "grad_norm": 0.1655040979385376,
      "learning_rate": 6.001905366783106e-05,
      "loss": 0.0492,
      "step": 6300
    },
    {
      "epoch": 11.999048977650975,
      "grad_norm": 0.27211880683898926,
      "learning_rate": 6.0012702445220705e-05,
      "loss": 0.0648,
      "step": 6301
    },
    {
      "epoch": 12.000951022349025,
      "grad_norm": 0.29517486691474915,
      "learning_rate": 6.0006351222610356e-05,
      "loss": 0.0727,
      "step": 6302
    },
    {
      "epoch": 12.002853067047075,
      "grad_norm": 0.19214580953121185,
      "learning_rate": 6e-05,
      "loss": 0.0641,
      "step": 6303
    },
    {
      "epoch": 12.004755111745126,
      "grad_norm": 0.1292438805103302,
      "learning_rate": 5.999364877738965e-05,
      "loss": 0.0493,
      "step": 6304
    },
    {
      "epoch": 12.006657156443177,
      "grad_norm": 0.19623462855815887,
      "learning_rate": 5.99872975547793e-05,
      "loss": 0.078,
      "step": 6305
    },
    {
      "epoch": 12.008559201141226,
      "grad_norm": 0.3401356041431427,
      "learning_rate": 5.9980946332168943e-05,
      "loss": 0.0795,
      "step": 6306
    },
    {
      "epoch": 12.010461245839277,
      "grad_norm": 0.20049375295639038,
      "learning_rate": 5.9974595109558595e-05,
      "loss": 0.0629,
      "step": 6307
    },
    {
      "epoch": 12.012363290537328,
      "grad_norm": 0.13026028871536255,
      "learning_rate": 5.996824388694824e-05,
      "loss": 0.0499,
      "step": 6308
    },
    {
      "epoch": 12.014265335235377,
      "grad_norm": 0.245549276471138,
      "learning_rate": 5.996189266433789e-05,
      "loss": 0.0634,
      "step": 6309
    },
    {
      "epoch": 12.016167379933428,
      "grad_norm": 0.12406376004219055,
      "learning_rate": 5.995554144172754e-05,
      "loss": 0.0408,
      "step": 6310
    },
    {
      "epoch": 12.018069424631479,
      "grad_norm": 0.16947674751281738,
      "learning_rate": 5.994919021911718e-05,
      "loss": 0.065,
      "step": 6311
    },
    {
      "epoch": 12.01997146932953,
      "grad_norm": 0.14021891355514526,
      "learning_rate": 5.9942838996506834e-05,
      "loss": 0.0771,
      "step": 6312
    },
    {
      "epoch": 12.021873514027579,
      "grad_norm": 0.12413173168897629,
      "learning_rate": 5.993648777389648e-05,
      "loss": 0.054,
      "step": 6313
    },
    {
      "epoch": 12.02377555872563,
      "grad_norm": 0.11317545920610428,
      "learning_rate": 5.993013655128612e-05,
      "loss": 0.0738,
      "step": 6314
    },
    {
      "epoch": 12.02567760342368,
      "grad_norm": 0.1977335810661316,
      "learning_rate": 5.9923785328675776e-05,
      "loss": 0.0588,
      "step": 6315
    },
    {
      "epoch": 12.027579648121732,
      "grad_norm": 0.23832173645496368,
      "learning_rate": 5.9917434106065414e-05,
      "loss": 0.0626,
      "step": 6316
    },
    {
      "epoch": 12.02948169281978,
      "grad_norm": 0.23073810338974,
      "learning_rate": 5.991108288345507e-05,
      "loss": 0.0534,
      "step": 6317
    },
    {
      "epoch": 12.031383737517832,
      "grad_norm": 0.1403868943452835,
      "learning_rate": 5.990473166084472e-05,
      "loss": 0.0786,
      "step": 6318
    },
    {
      "epoch": 12.033285782215883,
      "grad_norm": 0.25834405422210693,
      "learning_rate": 5.9898380438234356e-05,
      "loss": 0.0566,
      "step": 6319
    },
    {
      "epoch": 12.035187826913932,
      "grad_norm": 0.18973225355148315,
      "learning_rate": 5.9892029215624015e-05,
      "loss": 0.0484,
      "step": 6320
    },
    {
      "epoch": 12.037089871611983,
      "grad_norm": 0.18137171864509583,
      "learning_rate": 5.988567799301365e-05,
      "loss": 0.0717,
      "step": 6321
    },
    {
      "epoch": 12.038991916310033,
      "grad_norm": 0.23322740197181702,
      "learning_rate": 5.987932677040331e-05,
      "loss": 0.2109,
      "step": 6322
    },
    {
      "epoch": 12.040893961008084,
      "grad_norm": 0.20067892968654633,
      "learning_rate": 5.987297554779295e-05,
      "loss": 0.0475,
      "step": 6323
    },
    {
      "epoch": 12.042796005706133,
      "grad_norm": 0.13799051940441132,
      "learning_rate": 5.9866624325182595e-05,
      "loss": 0.0529,
      "step": 6324
    },
    {
      "epoch": 12.044698050404184,
      "grad_norm": 0.21583084762096405,
      "learning_rate": 5.9860273102572254e-05,
      "loss": 0.0732,
      "step": 6325
    },
    {
      "epoch": 12.046600095102235,
      "grad_norm": 0.21806849539279938,
      "learning_rate": 5.985392187996189e-05,
      "loss": 0.0911,
      "step": 6326
    },
    {
      "epoch": 12.048502139800286,
      "grad_norm": 0.33009856939315796,
      "learning_rate": 5.984757065735155e-05,
      "loss": 0.0738,
      "step": 6327
    },
    {
      "epoch": 12.050404184498335,
      "grad_norm": 0.14276517927646637,
      "learning_rate": 5.984121943474119e-05,
      "loss": 0.0441,
      "step": 6328
    },
    {
      "epoch": 12.052306229196386,
      "grad_norm": 0.14218777418136597,
      "learning_rate": 5.9834868212130834e-05,
      "loss": 0.0672,
      "step": 6329
    },
    {
      "epoch": 12.054208273894437,
      "grad_norm": 0.09867583215236664,
      "learning_rate": 5.9828516989520486e-05,
      "loss": 0.0498,
      "step": 6330
    },
    {
      "epoch": 12.056110318592486,
      "grad_norm": 0.11911500245332718,
      "learning_rate": 5.982216576691013e-05,
      "loss": 0.0503,
      "step": 6331
    },
    {
      "epoch": 12.058012363290537,
      "grad_norm": 0.21429555118083954,
      "learning_rate": 5.9815814544299776e-05,
      "loss": 0.0746,
      "step": 6332
    },
    {
      "epoch": 12.059914407988588,
      "grad_norm": 0.21537581086158752,
      "learning_rate": 5.980946332168943e-05,
      "loss": 0.0896,
      "step": 6333
    },
    {
      "epoch": 12.061816452686639,
      "grad_norm": 0.11404319852590561,
      "learning_rate": 5.980311209907907e-05,
      "loss": 0.0468,
      "step": 6334
    },
    {
      "epoch": 12.063718497384688,
      "grad_norm": 0.1308046579360962,
      "learning_rate": 5.9796760876468725e-05,
      "loss": 0.0697,
      "step": 6335
    },
    {
      "epoch": 12.065620542082739,
      "grad_norm": 0.15974490344524384,
      "learning_rate": 5.979040965385837e-05,
      "loss": 0.0586,
      "step": 6336
    },
    {
      "epoch": 12.06752258678079,
      "grad_norm": 0.19112728536128998,
      "learning_rate": 5.9784058431248015e-05,
      "loss": 0.0471,
      "step": 6337
    },
    {
      "epoch": 12.06942463147884,
      "grad_norm": 0.1520841419696808,
      "learning_rate": 5.977770720863767e-05,
      "loss": 0.0383,
      "step": 6338
    },
    {
      "epoch": 12.07132667617689,
      "grad_norm": 0.2817116379737854,
      "learning_rate": 5.977135598602731e-05,
      "loss": 0.0721,
      "step": 6339
    },
    {
      "epoch": 12.07322872087494,
      "grad_norm": 0.13829343020915985,
      "learning_rate": 5.9765004763416964e-05,
      "loss": 0.06,
      "step": 6340
    },
    {
      "epoch": 12.075130765572991,
      "grad_norm": 0.1855231076478958,
      "learning_rate": 5.975865354080661e-05,
      "loss": 0.0683,
      "step": 6341
    },
    {
      "epoch": 12.077032810271042,
      "grad_norm": 0.19953955709934235,
      "learning_rate": 5.9752302318196254e-05,
      "loss": 0.0674,
      "step": 6342
    },
    {
      "epoch": 12.078934854969091,
      "grad_norm": 0.1148846447467804,
      "learning_rate": 5.9745951095585906e-05,
      "loss": 0.0501,
      "step": 6343
    },
    {
      "epoch": 12.080836899667142,
      "grad_norm": 0.1192585900425911,
      "learning_rate": 5.973959987297555e-05,
      "loss": 0.0706,
      "step": 6344
    },
    {
      "epoch": 12.082738944365193,
      "grad_norm": 0.20792797207832336,
      "learning_rate": 5.97332486503652e-05,
      "loss": 0.0679,
      "step": 6345
    },
    {
      "epoch": 12.084640989063242,
      "grad_norm": 0.30921128392219543,
      "learning_rate": 5.972689742775485e-05,
      "loss": 0.0922,
      "step": 6346
    },
    {
      "epoch": 12.086543033761293,
      "grad_norm": 0.18978790938854218,
      "learning_rate": 5.972054620514449e-05,
      "loss": 0.0555,
      "step": 6347
    },
    {
      "epoch": 12.088445078459344,
      "grad_norm": 0.156063511967659,
      "learning_rate": 5.9714194982534145e-05,
      "loss": 0.069,
      "step": 6348
    },
    {
      "epoch": 12.090347123157395,
      "grad_norm": 0.10468289256095886,
      "learning_rate": 5.970784375992379e-05,
      "loss": 0.0391,
      "step": 6349
    },
    {
      "epoch": 12.092249167855444,
      "grad_norm": 0.26243624091148376,
      "learning_rate": 5.970149253731343e-05,
      "loss": 0.0671,
      "step": 6350
    },
    {
      "epoch": 12.094151212553495,
      "grad_norm": 0.149452343583107,
      "learning_rate": 5.969514131470309e-05,
      "loss": 0.0625,
      "step": 6351
    },
    {
      "epoch": 12.096053257251546,
      "grad_norm": 0.15380904078483582,
      "learning_rate": 5.9688790092092725e-05,
      "loss": 0.0701,
      "step": 6352
    },
    {
      "epoch": 12.097955301949597,
      "grad_norm": 0.25332167744636536,
      "learning_rate": 5.9682438869482384e-05,
      "loss": 0.0597,
      "step": 6353
    },
    {
      "epoch": 12.099857346647646,
      "grad_norm": 0.20952828228473663,
      "learning_rate": 5.967608764687203e-05,
      "loss": 0.0715,
      "step": 6354
    },
    {
      "epoch": 12.101759391345697,
      "grad_norm": 0.19237349927425385,
      "learning_rate": 5.966973642426167e-05,
      "loss": 0.0607,
      "step": 6355
    },
    {
      "epoch": 12.103661436043748,
      "grad_norm": 0.1403336375951767,
      "learning_rate": 5.9663385201651326e-05,
      "loss": 0.0667,
      "step": 6356
    },
    {
      "epoch": 12.105563480741797,
      "grad_norm": 0.2567223310470581,
      "learning_rate": 5.9657033979040964e-05,
      "loss": 0.0587,
      "step": 6357
    },
    {
      "epoch": 12.107465525439848,
      "grad_norm": 0.15880762040615082,
      "learning_rate": 5.965068275643062e-05,
      "loss": 0.0588,
      "step": 6358
    },
    {
      "epoch": 12.109367570137898,
      "grad_norm": 0.2392554134130478,
      "learning_rate": 5.964433153382026e-05,
      "loss": 0.0647,
      "step": 6359
    },
    {
      "epoch": 12.11126961483595,
      "grad_norm": 0.22182698547840118,
      "learning_rate": 5.9637980311209906e-05,
      "loss": 0.0631,
      "step": 6360
    },
    {
      "epoch": 12.113171659533998,
      "grad_norm": 0.1435650885105133,
      "learning_rate": 5.963162908859956e-05,
      "loss": 0.0505,
      "step": 6361
    },
    {
      "epoch": 12.11507370423205,
      "grad_norm": 0.17061443626880646,
      "learning_rate": 5.96252778659892e-05,
      "loss": 0.0682,
      "step": 6362
    },
    {
      "epoch": 12.1169757489301,
      "grad_norm": 0.14622712135314941,
      "learning_rate": 5.961892664337886e-05,
      "loss": 0.0528,
      "step": 6363
    },
    {
      "epoch": 12.118877793628151,
      "grad_norm": 0.16312119364738464,
      "learning_rate": 5.96125754207685e-05,
      "loss": 0.058,
      "step": 6364
    },
    {
      "epoch": 12.1207798383262,
      "grad_norm": 0.13628754019737244,
      "learning_rate": 5.9606224198158145e-05,
      "loss": 0.0492,
      "step": 6365
    },
    {
      "epoch": 12.122681883024251,
      "grad_norm": 0.1798691302537918,
      "learning_rate": 5.9599872975547796e-05,
      "loss": 0.0829,
      "step": 6366
    },
    {
      "epoch": 12.124583927722302,
      "grad_norm": 0.3143334686756134,
      "learning_rate": 5.959352175293744e-05,
      "loss": 0.0772,
      "step": 6367
    },
    {
      "epoch": 12.126485972420351,
      "grad_norm": 0.1629781275987625,
      "learning_rate": 5.958717053032709e-05,
      "loss": 0.0538,
      "step": 6368
    },
    {
      "epoch": 12.128388017118402,
      "grad_norm": 0.15789663791656494,
      "learning_rate": 5.958081930771674e-05,
      "loss": 0.0534,
      "step": 6369
    },
    {
      "epoch": 12.130290061816453,
      "grad_norm": 0.30617648363113403,
      "learning_rate": 5.9574468085106384e-05,
      "loss": 0.107,
      "step": 6370
    },
    {
      "epoch": 12.132192106514504,
      "grad_norm": 0.24667520821094513,
      "learning_rate": 5.9568116862496035e-05,
      "loss": 0.0722,
      "step": 6371
    },
    {
      "epoch": 12.134094151212553,
      "grad_norm": 0.1449286937713623,
      "learning_rate": 5.956176563988568e-05,
      "loss": 0.088,
      "step": 6372
    },
    {
      "epoch": 12.135996195910604,
      "grad_norm": 0.13355699181556702,
      "learning_rate": 5.9555414417275325e-05,
      "loss": 0.066,
      "step": 6373
    },
    {
      "epoch": 12.137898240608655,
      "grad_norm": 0.20543327927589417,
      "learning_rate": 5.954906319466498e-05,
      "loss": 0.0674,
      "step": 6374
    },
    {
      "epoch": 12.139800285306706,
      "grad_norm": 0.18319639563560486,
      "learning_rate": 5.954271197205462e-05,
      "loss": 0.0745,
      "step": 6375
    },
    {
      "epoch": 12.141702330004755,
      "grad_norm": 0.18459048867225647,
      "learning_rate": 5.9536360749444274e-05,
      "loss": 0.064,
      "step": 6376
    },
    {
      "epoch": 12.143604374702806,
      "grad_norm": 0.16675764322280884,
      "learning_rate": 5.953000952683392e-05,
      "loss": 0.052,
      "step": 6377
    },
    {
      "epoch": 12.145506419400856,
      "grad_norm": 0.12792012095451355,
      "learning_rate": 5.9523658304223564e-05,
      "loss": 0.0523,
      "step": 6378
    },
    {
      "epoch": 12.147408464098906,
      "grad_norm": 0.1535326987504959,
      "learning_rate": 5.9517307081613216e-05,
      "loss": 0.0552,
      "step": 6379
    },
    {
      "epoch": 12.149310508796956,
      "grad_norm": 0.15143269300460815,
      "learning_rate": 5.951095585900286e-05,
      "loss": 0.06,
      "step": 6380
    },
    {
      "epoch": 12.151212553495007,
      "grad_norm": 0.20260484516620636,
      "learning_rate": 5.950460463639251e-05,
      "loss": 0.0664,
      "step": 6381
    },
    {
      "epoch": 12.153114598193058,
      "grad_norm": 0.19385959208011627,
      "learning_rate": 5.949825341378216e-05,
      "loss": 0.0573,
      "step": 6382
    },
    {
      "epoch": 12.155016642891107,
      "grad_norm": 0.23560649156570435,
      "learning_rate": 5.94919021911718e-05,
      "loss": 0.0446,
      "step": 6383
    },
    {
      "epoch": 12.156918687589158,
      "grad_norm": 0.20179638266563416,
      "learning_rate": 5.9485550968561455e-05,
      "loss": 0.0603,
      "step": 6384
    },
    {
      "epoch": 12.158820732287209,
      "grad_norm": 0.21200603246688843,
      "learning_rate": 5.94791997459511e-05,
      "loss": 0.0679,
      "step": 6385
    },
    {
      "epoch": 12.16072277698526,
      "grad_norm": 0.1812458634376526,
      "learning_rate": 5.947284852334074e-05,
      "loss": 0.0648,
      "step": 6386
    },
    {
      "epoch": 12.162624821683309,
      "grad_norm": 0.23138156533241272,
      "learning_rate": 5.94664973007304e-05,
      "loss": 0.076,
      "step": 6387
    },
    {
      "epoch": 12.16452686638136,
      "grad_norm": 0.12320508062839508,
      "learning_rate": 5.9460146078120035e-05,
      "loss": 0.0624,
      "step": 6388
    },
    {
      "epoch": 12.16642891107941,
      "grad_norm": 0.24851442873477936,
      "learning_rate": 5.9453794855509694e-05,
      "loss": 0.0432,
      "step": 6389
    },
    {
      "epoch": 12.16833095577746,
      "grad_norm": 0.14283230900764465,
      "learning_rate": 5.944744363289933e-05,
      "loss": 0.0561,
      "step": 6390
    },
    {
      "epoch": 12.17023300047551,
      "grad_norm": 0.3072549104690552,
      "learning_rate": 5.944109241028898e-05,
      "loss": 0.1046,
      "step": 6391
    },
    {
      "epoch": 12.172135045173562,
      "grad_norm": 0.1524524837732315,
      "learning_rate": 5.9434741187678636e-05,
      "loss": 0.0736,
      "step": 6392
    },
    {
      "epoch": 12.174037089871613,
      "grad_norm": 0.24835297465324402,
      "learning_rate": 5.9428389965068274e-05,
      "loss": 0.0675,
      "step": 6393
    },
    {
      "epoch": 12.175939134569662,
      "grad_norm": 0.19676776230335236,
      "learning_rate": 5.942203874245793e-05,
      "loss": 0.062,
      "step": 6394
    },
    {
      "epoch": 12.177841179267713,
      "grad_norm": 0.3479282855987549,
      "learning_rate": 5.941568751984757e-05,
      "loss": 0.085,
      "step": 6395
    },
    {
      "epoch": 12.179743223965763,
      "grad_norm": 0.3322353661060333,
      "learning_rate": 5.9409336297237216e-05,
      "loss": 0.0578,
      "step": 6396
    },
    {
      "epoch": 12.181645268663814,
      "grad_norm": 0.164231076836586,
      "learning_rate": 5.940298507462687e-05,
      "loss": 0.049,
      "step": 6397
    },
    {
      "epoch": 12.183547313361863,
      "grad_norm": 0.21767644584178925,
      "learning_rate": 5.939663385201651e-05,
      "loss": 0.0578,
      "step": 6398
    },
    {
      "epoch": 12.185449358059914,
      "grad_norm": 0.16830454766750336,
      "learning_rate": 5.939028262940617e-05,
      "loss": 0.0664,
      "step": 6399
    },
    {
      "epoch": 12.187351402757965,
      "grad_norm": 0.19520054757595062,
      "learning_rate": 5.938393140679581e-05,
      "loss": 0.0614,
      "step": 6400
    },
    {
      "epoch": 12.189253447456014,
      "grad_norm": 0.1620379537343979,
      "learning_rate": 5.9377580184185455e-05,
      "loss": 0.0724,
      "step": 6401
    },
    {
      "epoch": 12.191155492154065,
      "grad_norm": 0.1629243791103363,
      "learning_rate": 5.937122896157511e-05,
      "loss": 0.0479,
      "step": 6402
    },
    {
      "epoch": 12.193057536852116,
      "grad_norm": 0.13102157413959503,
      "learning_rate": 5.936487773896475e-05,
      "loss": 0.0568,
      "step": 6403
    },
    {
      "epoch": 12.194959581550167,
      "grad_norm": 0.30659785866737366,
      "learning_rate": 5.93585265163544e-05,
      "loss": 0.066,
      "step": 6404
    },
    {
      "epoch": 12.196861626248216,
      "grad_norm": 0.3091185986995697,
      "learning_rate": 5.935217529374405e-05,
      "loss": 0.0935,
      "step": 6405
    },
    {
      "epoch": 12.198763670946267,
      "grad_norm": 0.34564849734306335,
      "learning_rate": 5.9345824071133694e-05,
      "loss": 0.0775,
      "step": 6406
    },
    {
      "epoch": 12.200665715644318,
      "grad_norm": 0.21233195066452026,
      "learning_rate": 5.9339472848523346e-05,
      "loss": 0.0528,
      "step": 6407
    },
    {
      "epoch": 12.202567760342369,
      "grad_norm": 0.12436842918395996,
      "learning_rate": 5.933312162591299e-05,
      "loss": 0.0616,
      "step": 6408
    },
    {
      "epoch": 12.204469805040418,
      "grad_norm": 0.1318405121564865,
      "learning_rate": 5.9326770403302636e-05,
      "loss": 0.0406,
      "step": 6409
    },
    {
      "epoch": 12.206371849738469,
      "grad_norm": 0.16262254118919373,
      "learning_rate": 5.932041918069229e-05,
      "loss": 0.0573,
      "step": 6410
    },
    {
      "epoch": 12.20827389443652,
      "grad_norm": 0.26165643334388733,
      "learning_rate": 5.931406795808193e-05,
      "loss": 0.0639,
      "step": 6411
    },
    {
      "epoch": 12.210175939134569,
      "grad_norm": 0.1864023208618164,
      "learning_rate": 5.9307716735471585e-05,
      "loss": 0.0692,
      "step": 6412
    },
    {
      "epoch": 12.21207798383262,
      "grad_norm": 0.20596599578857422,
      "learning_rate": 5.930136551286123e-05,
      "loss": 0.0654,
      "step": 6413
    },
    {
      "epoch": 12.21398002853067,
      "grad_norm": 0.1777569204568863,
      "learning_rate": 5.9295014290250875e-05,
      "loss": 0.061,
      "step": 6414
    },
    {
      "epoch": 12.215882073228721,
      "grad_norm": 0.24012942612171173,
      "learning_rate": 5.928866306764053e-05,
      "loss": 0.0711,
      "step": 6415
    },
    {
      "epoch": 12.21778411792677,
      "grad_norm": 0.19146688282489777,
      "learning_rate": 5.928231184503017e-05,
      "loss": 0.0627,
      "step": 6416
    },
    {
      "epoch": 12.219686162624821,
      "grad_norm": 0.17938733100891113,
      "learning_rate": 5.9275960622419824e-05,
      "loss": 0.0672,
      "step": 6417
    },
    {
      "epoch": 12.221588207322872,
      "grad_norm": 0.1901213824748993,
      "learning_rate": 5.926960939980947e-05,
      "loss": 0.0539,
      "step": 6418
    },
    {
      "epoch": 12.223490252020923,
      "grad_norm": 0.22269144654273987,
      "learning_rate": 5.926325817719911e-05,
      "loss": 0.0457,
      "step": 6419
    },
    {
      "epoch": 12.225392296718972,
      "grad_norm": 0.2517926096916199,
      "learning_rate": 5.9256906954588766e-05,
      "loss": 0.0483,
      "step": 6420
    },
    {
      "epoch": 12.227294341417023,
      "grad_norm": 0.2087641805410385,
      "learning_rate": 5.925055573197841e-05,
      "loss": 0.0624,
      "step": 6421
    },
    {
      "epoch": 12.229196386115074,
      "grad_norm": 0.3165057599544525,
      "learning_rate": 5.924420450936805e-05,
      "loss": 0.0733,
      "step": 6422
    },
    {
      "epoch": 12.231098430813125,
      "grad_norm": 0.1193968653678894,
      "learning_rate": 5.923785328675771e-05,
      "loss": 0.0496,
      "step": 6423
    },
    {
      "epoch": 12.233000475511174,
      "grad_norm": 0.21130363643169403,
      "learning_rate": 5.9231502064147346e-05,
      "loss": 0.0672,
      "step": 6424
    },
    {
      "epoch": 12.234902520209225,
      "grad_norm": 0.1780078411102295,
      "learning_rate": 5.9225150841537004e-05,
      "loss": 0.0556,
      "step": 6425
    },
    {
      "epoch": 12.236804564907276,
      "grad_norm": 0.1848141849040985,
      "learning_rate": 5.921879961892664e-05,
      "loss": 0.0528,
      "step": 6426
    },
    {
      "epoch": 12.238706609605325,
      "grad_norm": 0.2436516433954239,
      "learning_rate": 5.921244839631629e-05,
      "loss": 0.0728,
      "step": 6427
    },
    {
      "epoch": 12.240608654303376,
      "grad_norm": 0.14385049045085907,
      "learning_rate": 5.9206097173705946e-05,
      "loss": 0.0684,
      "step": 6428
    },
    {
      "epoch": 12.242510699001427,
      "grad_norm": 0.10670439153909683,
      "learning_rate": 5.9199745951095585e-05,
      "loss": 0.0511,
      "step": 6429
    },
    {
      "epoch": 12.244412743699478,
      "grad_norm": 0.13781152665615082,
      "learning_rate": 5.919339472848524e-05,
      "loss": 0.0619,
      "step": 6430
    },
    {
      "epoch": 12.246314788397527,
      "grad_norm": 0.18462249636650085,
      "learning_rate": 5.918704350587488e-05,
      "loss": 0.0663,
      "step": 6431
    },
    {
      "epoch": 12.248216833095578,
      "grad_norm": 0.17309071123600006,
      "learning_rate": 5.918069228326453e-05,
      "loss": 0.0647,
      "step": 6432
    },
    {
      "epoch": 12.250118877793629,
      "grad_norm": 0.11195661127567291,
      "learning_rate": 5.917434106065418e-05,
      "loss": 0.0505,
      "step": 6433
    },
    {
      "epoch": 12.25202092249168,
      "grad_norm": 0.21312302350997925,
      "learning_rate": 5.9167989838043824e-05,
      "loss": 0.0622,
      "step": 6434
    },
    {
      "epoch": 12.253922967189729,
      "grad_norm": 0.14799503982067108,
      "learning_rate": 5.916163861543348e-05,
      "loss": 0.0512,
      "step": 6435
    },
    {
      "epoch": 12.25582501188778,
      "grad_norm": 0.24153608083724976,
      "learning_rate": 5.915528739282312e-05,
      "loss": 0.0662,
      "step": 6436
    },
    {
      "epoch": 12.25772705658583,
      "grad_norm": 0.1948433816432953,
      "learning_rate": 5.9148936170212766e-05,
      "loss": 0.0518,
      "step": 6437
    },
    {
      "epoch": 12.25962910128388,
      "grad_norm": 0.11079234629869461,
      "learning_rate": 5.914258494760242e-05,
      "loss": 0.0561,
      "step": 6438
    },
    {
      "epoch": 12.26153114598193,
      "grad_norm": 0.23933576047420502,
      "learning_rate": 5.913623372499206e-05,
      "loss": 0.0755,
      "step": 6439
    },
    {
      "epoch": 12.263433190679981,
      "grad_norm": 0.2561612129211426,
      "learning_rate": 5.912988250238171e-05,
      "loss": 0.0776,
      "step": 6440
    },
    {
      "epoch": 12.265335235378032,
      "grad_norm": 0.19443702697753906,
      "learning_rate": 5.912353127977136e-05,
      "loss": 0.0593,
      "step": 6441
    },
    {
      "epoch": 12.267237280076081,
      "grad_norm": 0.2021740823984146,
      "learning_rate": 5.9117180057161004e-05,
      "loss": 0.0532,
      "step": 6442
    },
    {
      "epoch": 12.269139324774132,
      "grad_norm": 0.3714454472064972,
      "learning_rate": 5.9110828834550656e-05,
      "loss": 0.095,
      "step": 6443
    },
    {
      "epoch": 12.271041369472183,
      "grad_norm": 0.1737510859966278,
      "learning_rate": 5.91044776119403e-05,
      "loss": 0.061,
      "step": 6444
    },
    {
      "epoch": 12.272943414170234,
      "grad_norm": 0.19303299486637115,
      "learning_rate": 5.9098126389329946e-05,
      "loss": 0.056,
      "step": 6445
    },
    {
      "epoch": 12.274845458868283,
      "grad_norm": 0.23309148848056793,
      "learning_rate": 5.90917751667196e-05,
      "loss": 0.0765,
      "step": 6446
    },
    {
      "epoch": 12.276747503566334,
      "grad_norm": 0.1525864601135254,
      "learning_rate": 5.908542394410924e-05,
      "loss": 0.053,
      "step": 6447
    },
    {
      "epoch": 12.278649548264385,
      "grad_norm": 0.18437476456165314,
      "learning_rate": 5.9079072721498895e-05,
      "loss": 0.0576,
      "step": 6448
    },
    {
      "epoch": 12.280551592962434,
      "grad_norm": 0.200849249958992,
      "learning_rate": 5.907272149888854e-05,
      "loss": 0.0605,
      "step": 6449
    },
    {
      "epoch": 12.282453637660485,
      "grad_norm": 0.16408275067806244,
      "learning_rate": 5.9066370276278185e-05,
      "loss": 0.0857,
      "step": 6450
    },
    {
      "epoch": 12.284355682358536,
      "grad_norm": 0.19141173362731934,
      "learning_rate": 5.906001905366784e-05,
      "loss": 0.0515,
      "step": 6451
    },
    {
      "epoch": 12.286257727056586,
      "grad_norm": 0.1545351892709732,
      "learning_rate": 5.905366783105748e-05,
      "loss": 0.0498,
      "step": 6452
    },
    {
      "epoch": 12.288159771754636,
      "grad_norm": 0.1542082577943802,
      "learning_rate": 5.9047316608447134e-05,
      "loss": 0.0617,
      "step": 6453
    },
    {
      "epoch": 12.290061816452686,
      "grad_norm": 0.2996247112751007,
      "learning_rate": 5.904096538583678e-05,
      "loss": 0.0759,
      "step": 6454
    },
    {
      "epoch": 12.291963861150737,
      "grad_norm": 0.09020984172821045,
      "learning_rate": 5.903461416322642e-05,
      "loss": 0.046,
      "step": 6455
    },
    {
      "epoch": 12.293865905848788,
      "grad_norm": 0.1882646679878235,
      "learning_rate": 5.9028262940616076e-05,
      "loss": 0.046,
      "step": 6456
    },
    {
      "epoch": 12.295767950546837,
      "grad_norm": 0.17779584228992462,
      "learning_rate": 5.902191171800572e-05,
      "loss": 0.0586,
      "step": 6457
    },
    {
      "epoch": 12.297669995244888,
      "grad_norm": 0.21012863516807556,
      "learning_rate": 5.901556049539536e-05,
      "loss": 0.0503,
      "step": 6458
    },
    {
      "epoch": 12.29957203994294,
      "grad_norm": 0.2423000931739807,
      "learning_rate": 5.900920927278502e-05,
      "loss": 0.0584,
      "step": 6459
    },
    {
      "epoch": 12.301474084640988,
      "grad_norm": 0.2120276242494583,
      "learning_rate": 5.9002858050174656e-05,
      "loss": 0.067,
      "step": 6460
    },
    {
      "epoch": 12.30337612933904,
      "grad_norm": 0.20125135779380798,
      "learning_rate": 5.8996506827564315e-05,
      "loss": 0.0686,
      "step": 6461
    },
    {
      "epoch": 12.30527817403709,
      "grad_norm": 0.13728126883506775,
      "learning_rate": 5.899015560495395e-05,
      "loss": 0.0547,
      "step": 6462
    },
    {
      "epoch": 12.307180218735141,
      "grad_norm": 0.1871999353170395,
      "learning_rate": 5.89838043823436e-05,
      "loss": 0.0634,
      "step": 6463
    },
    {
      "epoch": 12.30908226343319,
      "grad_norm": 0.38858938217163086,
      "learning_rate": 5.897745315973326e-05,
      "loss": 0.1177,
      "step": 6464
    },
    {
      "epoch": 12.310984308131241,
      "grad_norm": 0.1638013869524002,
      "learning_rate": 5.8971101937122895e-05,
      "loss": 0.1012,
      "step": 6465
    },
    {
      "epoch": 12.312886352829292,
      "grad_norm": 0.2869760990142822,
      "learning_rate": 5.8964750714512554e-05,
      "loss": 0.1207,
      "step": 6466
    },
    {
      "epoch": 12.314788397527343,
      "grad_norm": 0.15024712681770325,
      "learning_rate": 5.895839949190219e-05,
      "loss": 0.059,
      "step": 6467
    },
    {
      "epoch": 12.316690442225392,
      "grad_norm": 0.21937395632266998,
      "learning_rate": 5.895204826929184e-05,
      "loss": 0.0839,
      "step": 6468
    },
    {
      "epoch": 12.318592486923443,
      "grad_norm": 0.0890878215432167,
      "learning_rate": 5.894569704668149e-05,
      "loss": 0.0606,
      "step": 6469
    },
    {
      "epoch": 12.320494531621494,
      "grad_norm": 0.1961527168750763,
      "learning_rate": 5.8939345824071134e-05,
      "loss": 0.0759,
      "step": 6470
    },
    {
      "epoch": 12.322396576319543,
      "grad_norm": 0.14583544433116913,
      "learning_rate": 5.8932994601460786e-05,
      "loss": 0.0622,
      "step": 6471
    },
    {
      "epoch": 12.324298621017594,
      "grad_norm": 0.2313392609357834,
      "learning_rate": 5.892664337885043e-05,
      "loss": 0.0693,
      "step": 6472
    },
    {
      "epoch": 12.326200665715644,
      "grad_norm": 0.16019153594970703,
      "learning_rate": 5.8920292156240076e-05,
      "loss": 0.0538,
      "step": 6473
    },
    {
      "epoch": 12.328102710413695,
      "grad_norm": 0.22677358984947205,
      "learning_rate": 5.891394093362973e-05,
      "loss": 0.1244,
      "step": 6474
    },
    {
      "epoch": 12.330004755111744,
      "grad_norm": 0.20695747435092926,
      "learning_rate": 5.890758971101937e-05,
      "loss": 0.066,
      "step": 6475
    },
    {
      "epoch": 12.331906799809795,
      "grad_norm": 0.1801384836435318,
      "learning_rate": 5.890123848840902e-05,
      "loss": 0.0492,
      "step": 6476
    },
    {
      "epoch": 12.333808844507846,
      "grad_norm": 0.23628447949886322,
      "learning_rate": 5.889488726579867e-05,
      "loss": 0.0686,
      "step": 6477
    },
    {
      "epoch": 12.335710889205897,
      "grad_norm": 0.1047058179974556,
      "learning_rate": 5.8888536043188315e-05,
      "loss": 0.0622,
      "step": 6478
    },
    {
      "epoch": 12.337612933903946,
      "grad_norm": 0.19160684943199158,
      "learning_rate": 5.888218482057797e-05,
      "loss": 0.0791,
      "step": 6479
    },
    {
      "epoch": 12.339514978601997,
      "grad_norm": 0.19324207305908203,
      "learning_rate": 5.887583359796761e-05,
      "loss": 0.0835,
      "step": 6480
    },
    {
      "epoch": 12.341417023300048,
      "grad_norm": 0.18310590088367462,
      "learning_rate": 5.886948237535726e-05,
      "loss": 0.0604,
      "step": 6481
    },
    {
      "epoch": 12.343319067998099,
      "grad_norm": 0.2560455799102783,
      "learning_rate": 5.886313115274691e-05,
      "loss": 0.0815,
      "step": 6482
    },
    {
      "epoch": 12.345221112696148,
      "grad_norm": 0.37622329592704773,
      "learning_rate": 5.8856779930136554e-05,
      "loss": 0.0725,
      "step": 6483
    },
    {
      "epoch": 12.347123157394199,
      "grad_norm": 0.14701174199581146,
      "learning_rate": 5.8850428707526206e-05,
      "loss": 0.0511,
      "step": 6484
    },
    {
      "epoch": 12.34902520209225,
      "grad_norm": 0.22057802975177765,
      "learning_rate": 5.884407748491585e-05,
      "loss": 0.0583,
      "step": 6485
    },
    {
      "epoch": 12.350927246790299,
      "grad_norm": 0.16882608830928802,
      "learning_rate": 5.8837726262305496e-05,
      "loss": 0.0721,
      "step": 6486
    },
    {
      "epoch": 12.35282929148835,
      "grad_norm": 0.21905134618282318,
      "learning_rate": 5.883137503969515e-05,
      "loss": 0.0457,
      "step": 6487
    },
    {
      "epoch": 12.3547313361864,
      "grad_norm": 0.28024494647979736,
      "learning_rate": 5.882502381708479e-05,
      "loss": 0.0851,
      "step": 6488
    },
    {
      "epoch": 12.356633380884452,
      "grad_norm": 0.12102214246988297,
      "learning_rate": 5.8818672594474444e-05,
      "loss": 0.0646,
      "step": 6489
    },
    {
      "epoch": 12.3585354255825,
      "grad_norm": 0.21057242155075073,
      "learning_rate": 5.881232137186409e-05,
      "loss": 0.0577,
      "step": 6490
    },
    {
      "epoch": 12.360437470280552,
      "grad_norm": 0.20247074961662292,
      "learning_rate": 5.880597014925373e-05,
      "loss": 0.0555,
      "step": 6491
    },
    {
      "epoch": 12.362339514978602,
      "grad_norm": 0.36881357431411743,
      "learning_rate": 5.8799618926643386e-05,
      "loss": 0.0832,
      "step": 6492
    },
    {
      "epoch": 12.364241559676653,
      "grad_norm": 0.19223707914352417,
      "learning_rate": 5.879326770403303e-05,
      "loss": 0.0571,
      "step": 6493
    },
    {
      "epoch": 12.366143604374702,
      "grad_norm": 0.14942830801010132,
      "learning_rate": 5.878691648142267e-05,
      "loss": 0.0565,
      "step": 6494
    },
    {
      "epoch": 12.368045649072753,
      "grad_norm": 0.14054648578166962,
      "learning_rate": 5.878056525881233e-05,
      "loss": 0.0504,
      "step": 6495
    },
    {
      "epoch": 12.369947693770804,
      "grad_norm": 0.22053012251853943,
      "learning_rate": 5.877421403620197e-05,
      "loss": 0.0552,
      "step": 6496
    },
    {
      "epoch": 12.371849738468853,
      "grad_norm": 0.23704233765602112,
      "learning_rate": 5.8767862813591625e-05,
      "loss": 0.0667,
      "step": 6497
    },
    {
      "epoch": 12.373751783166904,
      "grad_norm": 0.17568421363830566,
      "learning_rate": 5.8761511590981264e-05,
      "loss": 0.0699,
      "step": 6498
    },
    {
      "epoch": 12.375653827864955,
      "grad_norm": 0.27614712715148926,
      "learning_rate": 5.875516036837091e-05,
      "loss": 0.0682,
      "step": 6499
    },
    {
      "epoch": 12.377555872563006,
      "grad_norm": 0.2671426832675934,
      "learning_rate": 5.874880914576056e-05,
      "loss": 0.0776,
      "step": 6500
    },
    {
      "epoch": 12.379457917261055,
      "grad_norm": 0.19971761107444763,
      "learning_rate": 5.8742457923150206e-05,
      "loss": 0.0601,
      "step": 6501
    },
    {
      "epoch": 12.381359961959106,
      "grad_norm": 0.15626168251037598,
      "learning_rate": 5.8736106700539864e-05,
      "loss": 0.0683,
      "step": 6502
    },
    {
      "epoch": 12.383262006657157,
      "grad_norm": 0.1768084019422531,
      "learning_rate": 5.87297554779295e-05,
      "loss": 0.061,
      "step": 6503
    },
    {
      "epoch": 12.385164051355208,
      "grad_norm": 0.17245520651340485,
      "learning_rate": 5.872340425531915e-05,
      "loss": 0.0494,
      "step": 6504
    },
    {
      "epoch": 12.387066096053257,
      "grad_norm": 0.14765174686908722,
      "learning_rate": 5.87170530327088e-05,
      "loss": 0.1017,
      "step": 6505
    },
    {
      "epoch": 12.388968140751308,
      "grad_norm": 0.21613261103630066,
      "learning_rate": 5.8710701810098444e-05,
      "loss": 0.0665,
      "step": 6506
    },
    {
      "epoch": 12.390870185449359,
      "grad_norm": 0.10022646188735962,
      "learning_rate": 5.8704350587488096e-05,
      "loss": 0.0818,
      "step": 6507
    },
    {
      "epoch": 12.392772230147408,
      "grad_norm": 0.1353398710489273,
      "learning_rate": 5.869799936487774e-05,
      "loss": 0.0521,
      "step": 6508
    },
    {
      "epoch": 12.394674274845459,
      "grad_norm": 0.291027307510376,
      "learning_rate": 5.8691648142267386e-05,
      "loss": 0.0542,
      "step": 6509
    },
    {
      "epoch": 12.39657631954351,
      "grad_norm": 0.2245619297027588,
      "learning_rate": 5.868529691965704e-05,
      "loss": 0.0711,
      "step": 6510
    },
    {
      "epoch": 12.39847836424156,
      "grad_norm": 0.11868350952863693,
      "learning_rate": 5.867894569704668e-05,
      "loss": 0.0782,
      "step": 6511
    },
    {
      "epoch": 12.40038040893961,
      "grad_norm": 0.24280130863189697,
      "learning_rate": 5.867259447443633e-05,
      "loss": 0.0843,
      "step": 6512
    },
    {
      "epoch": 12.40228245363766,
      "grad_norm": 0.16018423438072205,
      "learning_rate": 5.866624325182598e-05,
      "loss": 0.0559,
      "step": 6513
    },
    {
      "epoch": 12.404184498335711,
      "grad_norm": 0.25252264738082886,
      "learning_rate": 5.8659892029215625e-05,
      "loss": 0.0612,
      "step": 6514
    },
    {
      "epoch": 12.406086543033762,
      "grad_norm": 0.19043385982513428,
      "learning_rate": 5.865354080660528e-05,
      "loss": 0.0699,
      "step": 6515
    },
    {
      "epoch": 12.407988587731811,
      "grad_norm": 0.16969317197799683,
      "learning_rate": 5.864718958399492e-05,
      "loss": 0.0805,
      "step": 6516
    },
    {
      "epoch": 12.409890632429862,
      "grad_norm": 0.20662227272987366,
      "learning_rate": 5.864083836138457e-05,
      "loss": 0.0657,
      "step": 6517
    },
    {
      "epoch": 12.411792677127913,
      "grad_norm": 0.1850903332233429,
      "learning_rate": 5.863448713877422e-05,
      "loss": 0.0698,
      "step": 6518
    },
    {
      "epoch": 12.413694721825962,
      "grad_norm": 0.2603982388973236,
      "learning_rate": 5.8628135916163864e-05,
      "loss": 0.0693,
      "step": 6519
    },
    {
      "epoch": 12.415596766524013,
      "grad_norm": 0.25423145294189453,
      "learning_rate": 5.8621784693553516e-05,
      "loss": 0.0576,
      "step": 6520
    },
    {
      "epoch": 12.417498811222064,
      "grad_norm": 0.24416786432266235,
      "learning_rate": 5.861543347094316e-05,
      "loss": 0.0623,
      "step": 6521
    },
    {
      "epoch": 12.419400855920115,
      "grad_norm": 0.11997820436954498,
      "learning_rate": 5.8609082248332806e-05,
      "loss": 0.0576,
      "step": 6522
    },
    {
      "epoch": 12.421302900618164,
      "grad_norm": 0.20214514434337616,
      "learning_rate": 5.860273102572246e-05,
      "loss": 0.0598,
      "step": 6523
    },
    {
      "epoch": 12.423204945316215,
      "grad_norm": 0.21587751805782318,
      "learning_rate": 5.85963798031121e-05,
      "loss": 0.0655,
      "step": 6524
    },
    {
      "epoch": 12.425106990014266,
      "grad_norm": 0.11338116228580475,
      "learning_rate": 5.8590028580501755e-05,
      "loss": 0.0439,
      "step": 6525
    },
    {
      "epoch": 12.427009034712317,
      "grad_norm": 0.26029443740844727,
      "learning_rate": 5.85836773578914e-05,
      "loss": 0.0662,
      "step": 6526
    },
    {
      "epoch": 12.428911079410366,
      "grad_norm": 0.19062893092632294,
      "learning_rate": 5.857732613528104e-05,
      "loss": 0.084,
      "step": 6527
    },
    {
      "epoch": 12.430813124108417,
      "grad_norm": 0.16516171395778656,
      "learning_rate": 5.85709749126707e-05,
      "loss": 0.0629,
      "step": 6528
    },
    {
      "epoch": 12.432715168806467,
      "grad_norm": 0.20229387283325195,
      "learning_rate": 5.8564623690060335e-05,
      "loss": 0.0525,
      "step": 6529
    },
    {
      "epoch": 12.434617213504517,
      "grad_norm": 0.16246093809604645,
      "learning_rate": 5.855827246744998e-05,
      "loss": 0.0539,
      "step": 6530
    },
    {
      "epoch": 12.436519258202567,
      "grad_norm": 0.20271427929401398,
      "learning_rate": 5.855192124483964e-05,
      "loss": 0.0703,
      "step": 6531
    },
    {
      "epoch": 12.438421302900618,
      "grad_norm": 0.38267767429351807,
      "learning_rate": 5.854557002222928e-05,
      "loss": 0.0817,
      "step": 6532
    },
    {
      "epoch": 12.44032334759867,
      "grad_norm": 0.17112848162651062,
      "learning_rate": 5.8539218799618936e-05,
      "loss": 0.065,
      "step": 6533
    },
    {
      "epoch": 12.442225392296718,
      "grad_norm": 0.29853564500808716,
      "learning_rate": 5.8532867577008574e-05,
      "loss": 0.0604,
      "step": 6534
    },
    {
      "epoch": 12.44412743699477,
      "grad_norm": 0.1856275498867035,
      "learning_rate": 5.852651635439822e-05,
      "loss": 0.0523,
      "step": 6535
    },
    {
      "epoch": 12.44602948169282,
      "grad_norm": 0.2045893520116806,
      "learning_rate": 5.852016513178787e-05,
      "loss": 0.0572,
      "step": 6536
    },
    {
      "epoch": 12.447931526390871,
      "grad_norm": 0.1976349800825119,
      "learning_rate": 5.8513813909177516e-05,
      "loss": 0.0583,
      "step": 6537
    },
    {
      "epoch": 12.44983357108892,
      "grad_norm": 0.2721671164035797,
      "learning_rate": 5.8507462686567175e-05,
      "loss": 0.059,
      "step": 6538
    },
    {
      "epoch": 12.451735615786971,
      "grad_norm": 0.22028490900993347,
      "learning_rate": 5.850111146395681e-05,
      "loss": 0.0497,
      "step": 6539
    },
    {
      "epoch": 12.453637660485022,
      "grad_norm": 0.18933860957622528,
      "learning_rate": 5.849476024134646e-05,
      "loss": 0.0444,
      "step": 6540
    },
    {
      "epoch": 12.455539705183071,
      "grad_norm": 0.13993880152702332,
      "learning_rate": 5.848840901873611e-05,
      "loss": 0.046,
      "step": 6541
    },
    {
      "epoch": 12.457441749881122,
      "grad_norm": 0.1706465780735016,
      "learning_rate": 5.8482057796125755e-05,
      "loss": 0.0635,
      "step": 6542
    },
    {
      "epoch": 12.459343794579173,
      "grad_norm": 0.10777102410793304,
      "learning_rate": 5.847570657351541e-05,
      "loss": 0.0568,
      "step": 6543
    },
    {
      "epoch": 12.461245839277224,
      "grad_norm": 0.2629159390926361,
      "learning_rate": 5.846935535090505e-05,
      "loss": 0.0549,
      "step": 6544
    },
    {
      "epoch": 12.463147883975273,
      "grad_norm": 0.2961212992668152,
      "learning_rate": 5.84630041282947e-05,
      "loss": 0.0595,
      "step": 6545
    },
    {
      "epoch": 12.465049928673324,
      "grad_norm": 0.22317998111248016,
      "learning_rate": 5.845665290568435e-05,
      "loss": 0.0729,
      "step": 6546
    },
    {
      "epoch": 12.466951973371375,
      "grad_norm": 0.24215160310268402,
      "learning_rate": 5.8450301683073994e-05,
      "loss": 0.0706,
      "step": 6547
    },
    {
      "epoch": 12.468854018069425,
      "grad_norm": 0.22650720179080963,
      "learning_rate": 5.844395046046364e-05,
      "loss": 0.0717,
      "step": 6548
    },
    {
      "epoch": 12.470756062767475,
      "grad_norm": 0.1432311236858368,
      "learning_rate": 5.843759923785329e-05,
      "loss": 0.0609,
      "step": 6549
    },
    {
      "epoch": 12.472658107465525,
      "grad_norm": 0.2215825915336609,
      "learning_rate": 5.8431248015242936e-05,
      "loss": 0.0709,
      "step": 6550
    },
    {
      "epoch": 12.474560152163576,
      "grad_norm": 0.31182411313056946,
      "learning_rate": 5.842489679263259e-05,
      "loss": 0.0952,
      "step": 6551
    },
    {
      "epoch": 12.476462196861625,
      "grad_norm": 0.1482132524251938,
      "learning_rate": 5.841854557002223e-05,
      "loss": 0.0996,
      "step": 6552
    },
    {
      "epoch": 12.478364241559676,
      "grad_norm": 0.17739084362983704,
      "learning_rate": 5.841219434741188e-05,
      "loss": 0.07,
      "step": 6553
    },
    {
      "epoch": 12.480266286257727,
      "grad_norm": 0.2121700793504715,
      "learning_rate": 5.840584312480153e-05,
      "loss": 0.0768,
      "step": 6554
    },
    {
      "epoch": 12.482168330955778,
      "grad_norm": 0.1372910887002945,
      "learning_rate": 5.8399491902191175e-05,
      "loss": 0.053,
      "step": 6555
    },
    {
      "epoch": 12.484070375653827,
      "grad_norm": 0.1774810403585434,
      "learning_rate": 5.8393140679580826e-05,
      "loss": 0.0661,
      "step": 6556
    },
    {
      "epoch": 12.485972420351878,
      "grad_norm": 0.25764548778533936,
      "learning_rate": 5.838678945697047e-05,
      "loss": 0.0706,
      "step": 6557
    },
    {
      "epoch": 12.487874465049929,
      "grad_norm": 0.16082552075386047,
      "learning_rate": 5.838043823436011e-05,
      "loss": 0.0528,
      "step": 6558
    },
    {
      "epoch": 12.48977650974798,
      "grad_norm": 0.27846723794937134,
      "learning_rate": 5.837408701174977e-05,
      "loss": 0.0806,
      "step": 6559
    },
    {
      "epoch": 12.491678554446029,
      "grad_norm": 0.1597774624824524,
      "learning_rate": 5.8367735789139413e-05,
      "loss": 0.0466,
      "step": 6560
    },
    {
      "epoch": 12.49358059914408,
      "grad_norm": 0.18083947896957397,
      "learning_rate": 5.8361384566529065e-05,
      "loss": 0.0633,
      "step": 6561
    },
    {
      "epoch": 12.49548264384213,
      "grad_norm": 0.14837045967578888,
      "learning_rate": 5.835503334391871e-05,
      "loss": 0.0658,
      "step": 6562
    },
    {
      "epoch": 12.49738468854018,
      "grad_norm": 0.20318514108657837,
      "learning_rate": 5.834868212130835e-05,
      "loss": 0.0727,
      "step": 6563
    },
    {
      "epoch": 12.49928673323823,
      "grad_norm": 0.23750798404216766,
      "learning_rate": 5.834233089869801e-05,
      "loss": 0.066,
      "step": 6564
    },
    {
      "epoch": 12.501188777936282,
      "grad_norm": 0.1972920298576355,
      "learning_rate": 5.8335979676087646e-05,
      "loss": 0.063,
      "step": 6565
    },
    {
      "epoch": 12.503090822634332,
      "grad_norm": 0.1612776815891266,
      "learning_rate": 5.832962845347729e-05,
      "loss": 0.0388,
      "step": 6566
    },
    {
      "epoch": 12.504992867332382,
      "grad_norm": 0.22913222014904022,
      "learning_rate": 5.832327723086695e-05,
      "loss": 0.0666,
      "step": 6567
    },
    {
      "epoch": 12.506894912030432,
      "grad_norm": 0.1786653995513916,
      "learning_rate": 5.831692600825659e-05,
      "loss": 0.0828,
      "step": 6568
    },
    {
      "epoch": 12.508796956728483,
      "grad_norm": 0.16014467179775238,
      "learning_rate": 5.8310574785646246e-05,
      "loss": 0.0473,
      "step": 6569
    },
    {
      "epoch": 12.510699001426534,
      "grad_norm": 0.1647956818342209,
      "learning_rate": 5.8304223563035884e-05,
      "loss": 0.0628,
      "step": 6570
    },
    {
      "epoch": 12.512601046124583,
      "grad_norm": 0.3335977792739868,
      "learning_rate": 5.829787234042553e-05,
      "loss": 0.0541,
      "step": 6571
    },
    {
      "epoch": 12.514503090822634,
      "grad_norm": 0.19307781755924225,
      "learning_rate": 5.829152111781518e-05,
      "loss": 0.054,
      "step": 6572
    },
    {
      "epoch": 12.516405135520685,
      "grad_norm": 0.20400182902812958,
      "learning_rate": 5.8285169895204826e-05,
      "loss": 0.0475,
      "step": 6573
    },
    {
      "epoch": 12.518307180218734,
      "grad_norm": 0.2718433737754822,
      "learning_rate": 5.8278818672594485e-05,
      "loss": 0.071,
      "step": 6574
    },
    {
      "epoch": 12.520209224916785,
      "grad_norm": 0.26126694679260254,
      "learning_rate": 5.827246744998412e-05,
      "loss": 0.1035,
      "step": 6575
    },
    {
      "epoch": 12.522111269614836,
      "grad_norm": 0.2177870124578476,
      "learning_rate": 5.826611622737377e-05,
      "loss": 0.0665,
      "step": 6576
    },
    {
      "epoch": 12.524013314312887,
      "grad_norm": 0.2201864868402481,
      "learning_rate": 5.825976500476342e-05,
      "loss": 0.0589,
      "step": 6577
    },
    {
      "epoch": 12.525915359010936,
      "grad_norm": 0.264030396938324,
      "learning_rate": 5.8253413782153065e-05,
      "loss": 0.0553,
      "step": 6578
    },
    {
      "epoch": 12.527817403708987,
      "grad_norm": 0.24069593846797943,
      "learning_rate": 5.824706255954272e-05,
      "loss": 0.0669,
      "step": 6579
    },
    {
      "epoch": 12.529719448407038,
      "grad_norm": 0.16773609817028046,
      "learning_rate": 5.824071133693236e-05,
      "loss": 0.0736,
      "step": 6580
    },
    {
      "epoch": 12.531621493105089,
      "grad_norm": 0.2015453428030014,
      "learning_rate": 5.823436011432201e-05,
      "loss": 0.0606,
      "step": 6581
    },
    {
      "epoch": 12.533523537803138,
      "grad_norm": 0.17908424139022827,
      "learning_rate": 5.822800889171166e-05,
      "loss": 0.0692,
      "step": 6582
    },
    {
      "epoch": 12.535425582501189,
      "grad_norm": 0.11076918244361877,
      "learning_rate": 5.8221657669101304e-05,
      "loss": 0.0622,
      "step": 6583
    },
    {
      "epoch": 12.53732762719924,
      "grad_norm": 0.24794691801071167,
      "learning_rate": 5.821530644649095e-05,
      "loss": 0.0702,
      "step": 6584
    },
    {
      "epoch": 12.539229671897289,
      "grad_norm": 0.15396280586719513,
      "learning_rate": 5.82089552238806e-05,
      "loss": 0.0575,
      "step": 6585
    },
    {
      "epoch": 12.54113171659534,
      "grad_norm": 0.2552478611469269,
      "learning_rate": 5.8202604001270246e-05,
      "loss": 0.057,
      "step": 6586
    },
    {
      "epoch": 12.54303376129339,
      "grad_norm": 0.14009839296340942,
      "learning_rate": 5.81962527786599e-05,
      "loss": 0.0487,
      "step": 6587
    },
    {
      "epoch": 12.544935805991441,
      "grad_norm": 0.2083524912595749,
      "learning_rate": 5.818990155604954e-05,
      "loss": 0.0589,
      "step": 6588
    },
    {
      "epoch": 12.54683785068949,
      "grad_norm": 0.24637573957443237,
      "learning_rate": 5.818355033343919e-05,
      "loss": 0.0678,
      "step": 6589
    },
    {
      "epoch": 12.548739895387541,
      "grad_norm": 0.23935122787952423,
      "learning_rate": 5.817719911082884e-05,
      "loss": 0.0725,
      "step": 6590
    },
    {
      "epoch": 12.550641940085592,
      "grad_norm": 0.1685015857219696,
      "learning_rate": 5.8170847888218485e-05,
      "loss": 0.0625,
      "step": 6591
    },
    {
      "epoch": 12.552543984783643,
      "grad_norm": 0.26168474555015564,
      "learning_rate": 5.816449666560814e-05,
      "loss": 0.0683,
      "step": 6592
    },
    {
      "epoch": 12.554446029481692,
      "grad_norm": 0.2873637080192566,
      "learning_rate": 5.815814544299778e-05,
      "loss": 0.0743,
      "step": 6593
    },
    {
      "epoch": 12.556348074179743,
      "grad_norm": 0.1569848358631134,
      "learning_rate": 5.815179422038742e-05,
      "loss": 0.0919,
      "step": 6594
    },
    {
      "epoch": 12.558250118877794,
      "grad_norm": 0.18452291190624237,
      "learning_rate": 5.814544299777708e-05,
      "loss": 0.0589,
      "step": 6595
    },
    {
      "epoch": 12.560152163575845,
      "grad_norm": 0.1471710354089737,
      "learning_rate": 5.8139091775166724e-05,
      "loss": 0.0641,
      "step": 6596
    },
    {
      "epoch": 12.562054208273894,
      "grad_norm": 0.1817297339439392,
      "learning_rate": 5.8132740552556376e-05,
      "loss": 0.1083,
      "step": 6597
    },
    {
      "epoch": 12.563956252971945,
      "grad_norm": 0.21600444614887238,
      "learning_rate": 5.812638932994602e-05,
      "loss": 0.0774,
      "step": 6598
    },
    {
      "epoch": 12.565858297669996,
      "grad_norm": 0.1959545612335205,
      "learning_rate": 5.812003810733566e-05,
      "loss": 0.0751,
      "step": 6599
    },
    {
      "epoch": 12.567760342368045,
      "grad_norm": 0.23535875976085663,
      "learning_rate": 5.811368688472532e-05,
      "loss": 0.0791,
      "step": 6600
    },
    {
      "epoch": 12.569662387066096,
      "grad_norm": 0.2223137617111206,
      "learning_rate": 5.8107335662114956e-05,
      "loss": 0.0687,
      "step": 6601
    },
    {
      "epoch": 12.571564431764147,
      "grad_norm": 0.258556067943573,
      "learning_rate": 5.81009844395046e-05,
      "loss": 0.1151,
      "step": 6602
    },
    {
      "epoch": 12.573466476462198,
      "grad_norm": 0.15880589187145233,
      "learning_rate": 5.809463321689426e-05,
      "loss": 0.0607,
      "step": 6603
    },
    {
      "epoch": 12.575368521160247,
      "grad_norm": 0.2969416379928589,
      "learning_rate": 5.80882819942839e-05,
      "loss": 0.0591,
      "step": 6604
    },
    {
      "epoch": 12.577270565858298,
      "grad_norm": 0.2408699095249176,
      "learning_rate": 5.808193077167356e-05,
      "loss": 0.0465,
      "step": 6605
    },
    {
      "epoch": 12.579172610556348,
      "grad_norm": 0.14702194929122925,
      "learning_rate": 5.8075579549063195e-05,
      "loss": 0.0376,
      "step": 6606
    },
    {
      "epoch": 12.5810746552544,
      "grad_norm": 0.2562587559223175,
      "learning_rate": 5.806922832645284e-05,
      "loss": 0.0612,
      "step": 6607
    },
    {
      "epoch": 12.582976699952448,
      "grad_norm": 0.17348787188529968,
      "learning_rate": 5.806287710384249e-05,
      "loss": 0.0527,
      "step": 6608
    },
    {
      "epoch": 12.5848787446505,
      "grad_norm": 0.3318137526512146,
      "learning_rate": 5.805652588123214e-05,
      "loss": 0.0837,
      "step": 6609
    },
    {
      "epoch": 12.58678078934855,
      "grad_norm": 0.16464082896709442,
      "learning_rate": 5.805017465862179e-05,
      "loss": 0.0611,
      "step": 6610
    },
    {
      "epoch": 12.5886828340466,
      "grad_norm": 0.18480993807315826,
      "learning_rate": 5.8043823436011434e-05,
      "loss": 0.0562,
      "step": 6611
    },
    {
      "epoch": 12.59058487874465,
      "grad_norm": 0.2223837673664093,
      "learning_rate": 5.803747221340108e-05,
      "loss": 0.0498,
      "step": 6612
    },
    {
      "epoch": 12.592486923442701,
      "grad_norm": 0.242923304438591,
      "learning_rate": 5.803112099079073e-05,
      "loss": 0.0685,
      "step": 6613
    },
    {
      "epoch": 12.594388968140752,
      "grad_norm": 0.2212572693824768,
      "learning_rate": 5.8024769768180376e-05,
      "loss": 0.0659,
      "step": 6614
    },
    {
      "epoch": 12.596291012838801,
      "grad_norm": 0.13670332729816437,
      "learning_rate": 5.801841854557003e-05,
      "loss": 0.0418,
      "step": 6615
    },
    {
      "epoch": 12.598193057536852,
      "grad_norm": 0.13037995994091034,
      "learning_rate": 5.801206732295967e-05,
      "loss": 0.0492,
      "step": 6616
    },
    {
      "epoch": 12.600095102234903,
      "grad_norm": 0.21356116235256195,
      "learning_rate": 5.800571610034932e-05,
      "loss": 0.0579,
      "step": 6617
    },
    {
      "epoch": 12.601997146932954,
      "grad_norm": 0.20228992402553558,
      "learning_rate": 5.799936487773897e-05,
      "loss": 0.0575,
      "step": 6618
    },
    {
      "epoch": 12.603899191631003,
      "grad_norm": 0.24074339866638184,
      "learning_rate": 5.7993013655128615e-05,
      "loss": 0.0601,
      "step": 6619
    },
    {
      "epoch": 12.605801236329054,
      "grad_norm": 0.18884117901325226,
      "learning_rate": 5.798666243251826e-05,
      "loss": 0.0941,
      "step": 6620
    },
    {
      "epoch": 12.607703281027105,
      "grad_norm": 0.29533296823501587,
      "learning_rate": 5.798031120990791e-05,
      "loss": 0.0597,
      "step": 6621
    },
    {
      "epoch": 12.609605325725155,
      "grad_norm": 0.3215848207473755,
      "learning_rate": 5.797395998729756e-05,
      "loss": 0.076,
      "step": 6622
    },
    {
      "epoch": 12.611507370423205,
      "grad_norm": 0.20919732749462128,
      "learning_rate": 5.796760876468721e-05,
      "loss": 0.0765,
      "step": 6623
    },
    {
      "epoch": 12.613409415121255,
      "grad_norm": 0.1770048439502716,
      "learning_rate": 5.7961257542076854e-05,
      "loss": 0.0747,
      "step": 6624
    },
    {
      "epoch": 12.615311459819306,
      "grad_norm": 0.17611941695213318,
      "learning_rate": 5.79549063194665e-05,
      "loss": 0.0725,
      "step": 6625
    },
    {
      "epoch": 12.617213504517355,
      "grad_norm": 0.21046479046344757,
      "learning_rate": 5.794855509685615e-05,
      "loss": 0.0916,
      "step": 6626
    },
    {
      "epoch": 12.619115549215406,
      "grad_norm": 0.11053504049777985,
      "learning_rate": 5.7942203874245796e-05,
      "loss": 0.0519,
      "step": 6627
    },
    {
      "epoch": 12.621017593913457,
      "grad_norm": 0.0993650034070015,
      "learning_rate": 5.793585265163545e-05,
      "loss": 0.0418,
      "step": 6628
    },
    {
      "epoch": 12.622919638611508,
      "grad_norm": 0.19049400091171265,
      "learning_rate": 5.792950142902509e-05,
      "loss": 0.0824,
      "step": 6629
    },
    {
      "epoch": 12.624821683309557,
      "grad_norm": 0.21655261516571045,
      "learning_rate": 5.792315020641473e-05,
      "loss": 0.0589,
      "step": 6630
    },
    {
      "epoch": 12.626723728007608,
      "grad_norm": 0.12360772490501404,
      "learning_rate": 5.791679898380439e-05,
      "loss": 0.0578,
      "step": 6631
    },
    {
      "epoch": 12.628625772705659,
      "grad_norm": 0.16151952743530273,
      "learning_rate": 5.7910447761194034e-05,
      "loss": 0.0813,
      "step": 6632
    },
    {
      "epoch": 12.63052781740371,
      "grad_norm": 0.25301963090896606,
      "learning_rate": 5.7904096538583686e-05,
      "loss": 0.0563,
      "step": 6633
    },
    {
      "epoch": 12.632429862101759,
      "grad_norm": 0.2590464651584625,
      "learning_rate": 5.789774531597333e-05,
      "loss": 0.0685,
      "step": 6634
    },
    {
      "epoch": 12.63433190679981,
      "grad_norm": 0.18846406042575836,
      "learning_rate": 5.789139409336297e-05,
      "loss": 0.0678,
      "step": 6635
    },
    {
      "epoch": 12.63623395149786,
      "grad_norm": 0.21730905771255493,
      "learning_rate": 5.788504287075263e-05,
      "loss": 0.072,
      "step": 6636
    },
    {
      "epoch": 12.63813599619591,
      "grad_norm": 0.23395763337612152,
      "learning_rate": 5.7878691648142266e-05,
      "loss": 0.0682,
      "step": 6637
    },
    {
      "epoch": 12.64003804089396,
      "grad_norm": 0.23463581502437592,
      "learning_rate": 5.787234042553191e-05,
      "loss": 0.0764,
      "step": 6638
    },
    {
      "epoch": 12.641940085592012,
      "grad_norm": 0.1801515519618988,
      "learning_rate": 5.7865989202921563e-05,
      "loss": 0.0494,
      "step": 6639
    },
    {
      "epoch": 12.643842130290063,
      "grad_norm": 0.30223360657691956,
      "learning_rate": 5.785963798031121e-05,
      "loss": 0.0686,
      "step": 6640
    },
    {
      "epoch": 12.645744174988112,
      "grad_norm": 0.17319241166114807,
      "learning_rate": 5.785328675770087e-05,
      "loss": 0.0631,
      "step": 6641
    },
    {
      "epoch": 12.647646219686163,
      "grad_norm": 0.17781338095664978,
      "learning_rate": 5.7846935535090505e-05,
      "loss": 0.061,
      "step": 6642
    },
    {
      "epoch": 12.649548264384213,
      "grad_norm": 0.34193697571754456,
      "learning_rate": 5.784058431248015e-05,
      "loss": 0.0789,
      "step": 6643
    },
    {
      "epoch": 12.651450309082264,
      "grad_norm": 0.2082579880952835,
      "learning_rate": 5.78342330898698e-05,
      "loss": 0.0496,
      "step": 6644
    },
    {
      "epoch": 12.653352353780313,
      "grad_norm": 0.36952418088912964,
      "learning_rate": 5.782788186725945e-05,
      "loss": 0.0874,
      "step": 6645
    },
    {
      "epoch": 12.655254398478364,
      "grad_norm": 0.23420372605323792,
      "learning_rate": 5.78215306446491e-05,
      "loss": 0.0723,
      "step": 6646
    },
    {
      "epoch": 12.657156443176415,
      "grad_norm": 0.19578108191490173,
      "learning_rate": 5.7815179422038744e-05,
      "loss": 0.0692,
      "step": 6647
    },
    {
      "epoch": 12.659058487874464,
      "grad_norm": 0.24436496198177338,
      "learning_rate": 5.780882819942839e-05,
      "loss": 0.0659,
      "step": 6648
    },
    {
      "epoch": 12.660960532572515,
      "grad_norm": 0.2588810622692108,
      "learning_rate": 5.780247697681804e-05,
      "loss": 0.059,
      "step": 6649
    },
    {
      "epoch": 12.662862577270566,
      "grad_norm": 0.19169163703918457,
      "learning_rate": 5.7796125754207686e-05,
      "loss": 0.0457,
      "step": 6650
    },
    {
      "epoch": 12.664764621968617,
      "grad_norm": 0.31427711248397827,
      "learning_rate": 5.778977453159734e-05,
      "loss": 0.0721,
      "step": 6651
    },
    {
      "epoch": 12.666666666666666,
      "grad_norm": 0.28431782126426697,
      "learning_rate": 5.778342330898698e-05,
      "loss": 0.0636,
      "step": 6652
    },
    {
      "epoch": 12.668568711364717,
      "grad_norm": 0.19275827705860138,
      "learning_rate": 5.777707208637663e-05,
      "loss": 0.0598,
      "step": 6653
    },
    {
      "epoch": 12.670470756062768,
      "grad_norm": 0.28678253293037415,
      "learning_rate": 5.777072086376628e-05,
      "loss": 0.0799,
      "step": 6654
    },
    {
      "epoch": 12.672372800760819,
      "grad_norm": 0.279844731092453,
      "learning_rate": 5.7764369641155925e-05,
      "loss": 0.0634,
      "step": 6655
    },
    {
      "epoch": 12.674274845458868,
      "grad_norm": 0.16320635378360748,
      "learning_rate": 5.775801841854557e-05,
      "loss": 0.0355,
      "step": 6656
    },
    {
      "epoch": 12.676176890156919,
      "grad_norm": 0.30712923407554626,
      "learning_rate": 5.775166719593522e-05,
      "loss": 0.0888,
      "step": 6657
    },
    {
      "epoch": 12.67807893485497,
      "grad_norm": 0.1668136864900589,
      "learning_rate": 5.774531597332487e-05,
      "loss": 0.0527,
      "step": 6658
    },
    {
      "epoch": 12.679980979553019,
      "grad_norm": 0.17858512699604034,
      "learning_rate": 5.773896475071452e-05,
      "loss": 0.0638,
      "step": 6659
    },
    {
      "epoch": 12.68188302425107,
      "grad_norm": 0.19142888486385345,
      "learning_rate": 5.7732613528104164e-05,
      "loss": 0.0601,
      "step": 6660
    },
    {
      "epoch": 12.68378506894912,
      "grad_norm": 0.2255232036113739,
      "learning_rate": 5.77262623054938e-05,
      "loss": 0.0577,
      "step": 6661
    },
    {
      "epoch": 12.685687113647171,
      "grad_norm": 0.2920807898044586,
      "learning_rate": 5.771991108288346e-05,
      "loss": 0.078,
      "step": 6662
    },
    {
      "epoch": 12.68758915834522,
      "grad_norm": 0.1767326146364212,
      "learning_rate": 5.7713559860273106e-05,
      "loss": 0.0508,
      "step": 6663
    },
    {
      "epoch": 12.689491203043271,
      "grad_norm": 0.14078831672668457,
      "learning_rate": 5.770720863766276e-05,
      "loss": 0.0664,
      "step": 6664
    },
    {
      "epoch": 12.691393247741322,
      "grad_norm": 0.2606064975261688,
      "learning_rate": 5.77008574150524e-05,
      "loss": 0.0693,
      "step": 6665
    },
    {
      "epoch": 12.693295292439373,
      "grad_norm": 0.22922249138355255,
      "learning_rate": 5.769450619244204e-05,
      "loss": 0.0546,
      "step": 6666
    },
    {
      "epoch": 12.695197337137422,
      "grad_norm": 0.14279182255268097,
      "learning_rate": 5.76881549698317e-05,
      "loss": 0.0531,
      "step": 6667
    },
    {
      "epoch": 12.697099381835473,
      "grad_norm": 0.1705865114927292,
      "learning_rate": 5.768180374722134e-05,
      "loss": 0.0628,
      "step": 6668
    },
    {
      "epoch": 12.699001426533524,
      "grad_norm": 0.2879997193813324,
      "learning_rate": 5.7675452524611e-05,
      "loss": 0.065,
      "step": 6669
    },
    {
      "epoch": 12.700903471231573,
      "grad_norm": 0.2573727071285248,
      "learning_rate": 5.766910130200064e-05,
      "loss": 0.0716,
      "step": 6670
    },
    {
      "epoch": 12.702805515929624,
      "grad_norm": 0.17233164608478546,
      "learning_rate": 5.766275007939028e-05,
      "loss": 0.0493,
      "step": 6671
    },
    {
      "epoch": 12.704707560627675,
      "grad_norm": 0.16349801421165466,
      "learning_rate": 5.765639885677994e-05,
      "loss": 0.0523,
      "step": 6672
    },
    {
      "epoch": 12.706609605325726,
      "grad_norm": 0.14094343781471252,
      "learning_rate": 5.765004763416958e-05,
      "loss": 0.0527,
      "step": 6673
    },
    {
      "epoch": 12.708511650023775,
      "grad_norm": 0.16286692023277283,
      "learning_rate": 5.764369641155922e-05,
      "loss": 0.0644,
      "step": 6674
    },
    {
      "epoch": 12.710413694721826,
      "grad_norm": 0.3345188796520233,
      "learning_rate": 5.7637345188948874e-05,
      "loss": 0.0684,
      "step": 6675
    },
    {
      "epoch": 12.712315739419877,
      "grad_norm": 0.2787039875984192,
      "learning_rate": 5.763099396633852e-05,
      "loss": 0.0881,
      "step": 6676
    },
    {
      "epoch": 12.714217784117928,
      "grad_norm": 0.46529752016067505,
      "learning_rate": 5.762464274372818e-05,
      "loss": 0.1022,
      "step": 6677
    },
    {
      "epoch": 12.716119828815977,
      "grad_norm": 0.33916109800338745,
      "learning_rate": 5.7618291521117816e-05,
      "loss": 0.0715,
      "step": 6678
    },
    {
      "epoch": 12.718021873514028,
      "grad_norm": 0.22834204137325287,
      "learning_rate": 5.761194029850746e-05,
      "loss": 0.0581,
      "step": 6679
    },
    {
      "epoch": 12.719923918212078,
      "grad_norm": 0.26732945442199707,
      "learning_rate": 5.760558907589711e-05,
      "loss": 0.0597,
      "step": 6680
    },
    {
      "epoch": 12.721825962910128,
      "grad_norm": 0.18561658263206482,
      "learning_rate": 5.759923785328676e-05,
      "loss": 0.0624,
      "step": 6681
    },
    {
      "epoch": 12.723728007608178,
      "grad_norm": 0.2095700353384018,
      "learning_rate": 5.759288663067641e-05,
      "loss": 0.0563,
      "step": 6682
    },
    {
      "epoch": 12.72563005230623,
      "grad_norm": 0.20030710101127625,
      "learning_rate": 5.7586535408066055e-05,
      "loss": 0.0558,
      "step": 6683
    },
    {
      "epoch": 12.72753209700428,
      "grad_norm": 0.25263649225234985,
      "learning_rate": 5.75801841854557e-05,
      "loss": 0.0831,
      "step": 6684
    },
    {
      "epoch": 12.72943414170233,
      "grad_norm": 0.2224830538034439,
      "learning_rate": 5.757383296284535e-05,
      "loss": 0.0692,
      "step": 6685
    },
    {
      "epoch": 12.73133618640038,
      "grad_norm": 0.25267601013183594,
      "learning_rate": 5.7567481740235e-05,
      "loss": 0.052,
      "step": 6686
    },
    {
      "epoch": 12.733238231098431,
      "grad_norm": 0.23251911997795105,
      "learning_rate": 5.756113051762465e-05,
      "loss": 0.0375,
      "step": 6687
    },
    {
      "epoch": 12.735140275796482,
      "grad_norm": 0.1670525074005127,
      "learning_rate": 5.7554779295014294e-05,
      "loss": 0.0453,
      "step": 6688
    },
    {
      "epoch": 12.737042320494531,
      "grad_norm": 0.25991198420524597,
      "learning_rate": 5.754842807240394e-05,
      "loss": 0.0634,
      "step": 6689
    },
    {
      "epoch": 12.738944365192582,
      "grad_norm": 0.25706109404563904,
      "learning_rate": 5.754207684979359e-05,
      "loss": 0.0636,
      "step": 6690
    },
    {
      "epoch": 12.740846409890633,
      "grad_norm": 0.18260982632637024,
      "learning_rate": 5.7535725627183236e-05,
      "loss": 0.0675,
      "step": 6691
    },
    {
      "epoch": 12.742748454588682,
      "grad_norm": 0.18274103105068207,
      "learning_rate": 5.752937440457288e-05,
      "loss": 0.0686,
      "step": 6692
    },
    {
      "epoch": 12.744650499286733,
      "grad_norm": 0.13347792625427246,
      "learning_rate": 5.752302318196253e-05,
      "loss": 0.0592,
      "step": 6693
    },
    {
      "epoch": 12.746552543984784,
      "grad_norm": 0.2516426742076874,
      "learning_rate": 5.751667195935218e-05,
      "loss": 0.0571,
      "step": 6694
    },
    {
      "epoch": 12.748454588682835,
      "grad_norm": 0.1681799739599228,
      "learning_rate": 5.751032073674183e-05,
      "loss": 0.0779,
      "step": 6695
    },
    {
      "epoch": 12.750356633380884,
      "grad_norm": 0.14614692330360413,
      "learning_rate": 5.7503969514131474e-05,
      "loss": 0.0715,
      "step": 6696
    },
    {
      "epoch": 12.752258678078935,
      "grad_norm": 0.17188875377178192,
      "learning_rate": 5.749761829152111e-05,
      "loss": 0.0705,
      "step": 6697
    },
    {
      "epoch": 12.754160722776986,
      "grad_norm": 0.16121084988117218,
      "learning_rate": 5.749126706891077e-05,
      "loss": 0.0606,
      "step": 6698
    },
    {
      "epoch": 12.756062767475036,
      "grad_norm": 0.1163937896490097,
      "learning_rate": 5.7484915846300416e-05,
      "loss": 0.0499,
      "step": 6699
    },
    {
      "epoch": 12.757964812173086,
      "grad_norm": 0.2049596607685089,
      "learning_rate": 5.747856462369007e-05,
      "loss": 0.0616,
      "step": 6700
    },
    {
      "epoch": 12.759866856871136,
      "grad_norm": 0.13135495781898499,
      "learning_rate": 5.747221340107971e-05,
      "loss": 0.0565,
      "step": 6701
    },
    {
      "epoch": 12.761768901569187,
      "grad_norm": 0.1636398732662201,
      "learning_rate": 5.746586217846935e-05,
      "loss": 0.0517,
      "step": 6702
    },
    {
      "epoch": 12.763670946267236,
      "grad_norm": 0.15648093819618225,
      "learning_rate": 5.745951095585901e-05,
      "loss": 0.0503,
      "step": 6703
    },
    {
      "epoch": 12.765572990965287,
      "grad_norm": 0.15833771228790283,
      "learning_rate": 5.745315973324865e-05,
      "loss": 0.0375,
      "step": 6704
    },
    {
      "epoch": 12.767475035663338,
      "grad_norm": 0.2346632480621338,
      "learning_rate": 5.744680851063831e-05,
      "loss": 0.074,
      "step": 6705
    },
    {
      "epoch": 12.769377080361389,
      "grad_norm": 0.11698388308286667,
      "learning_rate": 5.744045728802795e-05,
      "loss": 0.0629,
      "step": 6706
    },
    {
      "epoch": 12.771279125059438,
      "grad_norm": 0.21215496957302094,
      "learning_rate": 5.743410606541759e-05,
      "loss": 0.0593,
      "step": 6707
    },
    {
      "epoch": 12.773181169757489,
      "grad_norm": 0.18357087671756744,
      "learning_rate": 5.742775484280725e-05,
      "loss": 0.0698,
      "step": 6708
    },
    {
      "epoch": 12.77508321445554,
      "grad_norm": 0.283631294965744,
      "learning_rate": 5.742140362019689e-05,
      "loss": 0.0672,
      "step": 6709
    },
    {
      "epoch": 12.77698525915359,
      "grad_norm": 0.15704897046089172,
      "learning_rate": 5.741505239758653e-05,
      "loss": 0.0568,
      "step": 6710
    },
    {
      "epoch": 12.77888730385164,
      "grad_norm": 0.23621056973934174,
      "learning_rate": 5.7408701174976184e-05,
      "loss": 0.0824,
      "step": 6711
    },
    {
      "epoch": 12.78078934854969,
      "grad_norm": 0.25481176376342773,
      "learning_rate": 5.740234995236583e-05,
      "loss": 0.0598,
      "step": 6712
    },
    {
      "epoch": 12.782691393247742,
      "grad_norm": 0.17756560444831848,
      "learning_rate": 5.739599872975549e-05,
      "loss": 0.0736,
      "step": 6713
    },
    {
      "epoch": 12.78459343794579,
      "grad_norm": 0.23665587604045868,
      "learning_rate": 5.7389647507145126e-05,
      "loss": 0.0666,
      "step": 6714
    },
    {
      "epoch": 12.786495482643842,
      "grad_norm": 0.25118178129196167,
      "learning_rate": 5.738329628453477e-05,
      "loss": 0.0633,
      "step": 6715
    },
    {
      "epoch": 12.788397527341893,
      "grad_norm": 0.16724514961242676,
      "learning_rate": 5.737694506192442e-05,
      "loss": 0.0587,
      "step": 6716
    },
    {
      "epoch": 12.790299572039943,
      "grad_norm": 0.24320180714130402,
      "learning_rate": 5.737059383931407e-05,
      "loss": 0.0912,
      "step": 6717
    },
    {
      "epoch": 12.792201616737993,
      "grad_norm": 0.27849307656288147,
      "learning_rate": 5.736424261670372e-05,
      "loss": 0.0874,
      "step": 6718
    },
    {
      "epoch": 12.794103661436043,
      "grad_norm": 0.12784165143966675,
      "learning_rate": 5.7357891394093365e-05,
      "loss": 0.0605,
      "step": 6719
    },
    {
      "epoch": 12.796005706134094,
      "grad_norm": 0.1868617981672287,
      "learning_rate": 5.735154017148301e-05,
      "loss": 0.0909,
      "step": 6720
    },
    {
      "epoch": 12.797907750832145,
      "grad_norm": 0.1971626728773117,
      "learning_rate": 5.734518894887266e-05,
      "loss": 0.0585,
      "step": 6721
    },
    {
      "epoch": 12.799809795530194,
      "grad_norm": 0.37201204895973206,
      "learning_rate": 5.733883772626231e-05,
      "loss": 0.0801,
      "step": 6722
    },
    {
      "epoch": 12.801711840228245,
      "grad_norm": 0.34986621141433716,
      "learning_rate": 5.733248650365196e-05,
      "loss": 0.0757,
      "step": 6723
    },
    {
      "epoch": 12.803613884926296,
      "grad_norm": 0.18505094945430756,
      "learning_rate": 5.7326135281041604e-05,
      "loss": 0.0591,
      "step": 6724
    },
    {
      "epoch": 12.805515929624345,
      "grad_norm": 0.19197897613048553,
      "learning_rate": 5.731978405843125e-05,
      "loss": 0.0509,
      "step": 6725
    },
    {
      "epoch": 12.807417974322396,
      "grad_norm": 0.17455685138702393,
      "learning_rate": 5.73134328358209e-05,
      "loss": 0.0522,
      "step": 6726
    },
    {
      "epoch": 12.809320019020447,
      "grad_norm": 0.1809726059436798,
      "learning_rate": 5.7307081613210546e-05,
      "loss": 0.0662,
      "step": 6727
    },
    {
      "epoch": 12.811222063718498,
      "grad_norm": 0.2538421154022217,
      "learning_rate": 5.730073039060019e-05,
      "loss": 0.0511,
      "step": 6728
    },
    {
      "epoch": 12.813124108416547,
      "grad_norm": 0.35618454217910767,
      "learning_rate": 5.729437916798984e-05,
      "loss": 0.0665,
      "step": 6729
    },
    {
      "epoch": 12.815026153114598,
      "grad_norm": 0.21073471009731293,
      "learning_rate": 5.728802794537949e-05,
      "loss": 0.0613,
      "step": 6730
    },
    {
      "epoch": 12.816928197812649,
      "grad_norm": 0.17091955244541168,
      "learning_rate": 5.728167672276914e-05,
      "loss": 0.0684,
      "step": 6731
    },
    {
      "epoch": 12.8188302425107,
      "grad_norm": 0.30342546105384827,
      "learning_rate": 5.7275325500158785e-05,
      "loss": 0.1016,
      "step": 6732
    },
    {
      "epoch": 12.820732287208749,
      "grad_norm": 0.16343329846858978,
      "learning_rate": 5.726897427754842e-05,
      "loss": 0.0558,
      "step": 6733
    },
    {
      "epoch": 12.8226343319068,
      "grad_norm": 0.20818068087100983,
      "learning_rate": 5.726262305493808e-05,
      "loss": 0.0525,
      "step": 6734
    },
    {
      "epoch": 12.82453637660485,
      "grad_norm": 0.1833098977804184,
      "learning_rate": 5.725627183232773e-05,
      "loss": 0.0445,
      "step": 6735
    },
    {
      "epoch": 12.8264384213029,
      "grad_norm": 0.23190821707248688,
      "learning_rate": 5.724992060971738e-05,
      "loss": 0.0761,
      "step": 6736
    },
    {
      "epoch": 12.82834046600095,
      "grad_norm": 0.25338640809059143,
      "learning_rate": 5.7243569387107024e-05,
      "loss": 0.0685,
      "step": 6737
    },
    {
      "epoch": 12.830242510699001,
      "grad_norm": 0.15732528269290924,
      "learning_rate": 5.723721816449666e-05,
      "loss": 0.0626,
      "step": 6738
    },
    {
      "epoch": 12.832144555397052,
      "grad_norm": 0.20255662500858307,
      "learning_rate": 5.723086694188632e-05,
      "loss": 0.0554,
      "step": 6739
    },
    {
      "epoch": 12.834046600095101,
      "grad_norm": 0.22862756252288818,
      "learning_rate": 5.722451571927596e-05,
      "loss": 0.056,
      "step": 6740
    },
    {
      "epoch": 12.835948644793152,
      "grad_norm": 0.164798304438591,
      "learning_rate": 5.721816449666562e-05,
      "loss": 0.048,
      "step": 6741
    },
    {
      "epoch": 12.837850689491203,
      "grad_norm": 0.23823440074920654,
      "learning_rate": 5.721181327405526e-05,
      "loss": 0.0574,
      "step": 6742
    },
    {
      "epoch": 12.839752734189254,
      "grad_norm": 0.18618272244930267,
      "learning_rate": 5.72054620514449e-05,
      "loss": 0.054,
      "step": 6743
    },
    {
      "epoch": 12.841654778887303,
      "grad_norm": 0.2012937366962433,
      "learning_rate": 5.719911082883456e-05,
      "loss": 0.0629,
      "step": 6744
    },
    {
      "epoch": 12.843556823585354,
      "grad_norm": 0.26844924688339233,
      "learning_rate": 5.71927596062242e-05,
      "loss": 0.0771,
      "step": 6745
    },
    {
      "epoch": 12.845458868283405,
      "grad_norm": 0.14869700372219086,
      "learning_rate": 5.718640838361384e-05,
      "loss": 0.0462,
      "step": 6746
    },
    {
      "epoch": 12.847360912981456,
      "grad_norm": 0.24835607409477234,
      "learning_rate": 5.7180057161003495e-05,
      "loss": 0.0575,
      "step": 6747
    },
    {
      "epoch": 12.849262957679505,
      "grad_norm": 0.1934400200843811,
      "learning_rate": 5.717370593839314e-05,
      "loss": 0.0588,
      "step": 6748
    },
    {
      "epoch": 12.851165002377556,
      "grad_norm": 0.10977878421545029,
      "learning_rate": 5.716735471578279e-05,
      "loss": 0.0605,
      "step": 6749
    },
    {
      "epoch": 12.853067047075607,
      "grad_norm": 0.24564503133296967,
      "learning_rate": 5.716100349317244e-05,
      "loss": 0.0702,
      "step": 6750
    },
    {
      "epoch": 12.854969091773656,
      "grad_norm": 0.2185717672109604,
      "learning_rate": 5.715465227056208e-05,
      "loss": 0.0549,
      "step": 6751
    },
    {
      "epoch": 12.856871136471707,
      "grad_norm": 0.19493113458156586,
      "learning_rate": 5.7148301047951734e-05,
      "loss": 0.0528,
      "step": 6752
    },
    {
      "epoch": 12.858773181169758,
      "grad_norm": 0.2067340910434723,
      "learning_rate": 5.714194982534138e-05,
      "loss": 0.0603,
      "step": 6753
    },
    {
      "epoch": 12.860675225867809,
      "grad_norm": 0.21597956120967865,
      "learning_rate": 5.713559860273103e-05,
      "loss": 0.0724,
      "step": 6754
    },
    {
      "epoch": 12.862577270565858,
      "grad_norm": 0.1443902999162674,
      "learning_rate": 5.7129247380120676e-05,
      "loss": 0.0477,
      "step": 6755
    },
    {
      "epoch": 12.864479315263909,
      "grad_norm": 0.2689155042171478,
      "learning_rate": 5.712289615751032e-05,
      "loss": 0.0632,
      "step": 6756
    },
    {
      "epoch": 12.86638135996196,
      "grad_norm": 0.2500086724758148,
      "learning_rate": 5.711654493489997e-05,
      "loss": 0.0685,
      "step": 6757
    },
    {
      "epoch": 12.86828340466001,
      "grad_norm": 0.2317550778388977,
      "learning_rate": 5.711019371228962e-05,
      "loss": 0.0762,
      "step": 6758
    },
    {
      "epoch": 12.87018544935806,
      "grad_norm": 0.1884494423866272,
      "learning_rate": 5.710384248967927e-05,
      "loss": 0.0441,
      "step": 6759
    },
    {
      "epoch": 12.87208749405611,
      "grad_norm": 0.22511914372444153,
      "learning_rate": 5.7097491267068914e-05,
      "loss": 0.0659,
      "step": 6760
    },
    {
      "epoch": 12.873989538754161,
      "grad_norm": 0.15401434898376465,
      "learning_rate": 5.709114004445856e-05,
      "loss": 0.0706,
      "step": 6761
    },
    {
      "epoch": 12.87589158345221,
      "grad_norm": 0.32594338059425354,
      "learning_rate": 5.708478882184821e-05,
      "loss": 0.0572,
      "step": 6762
    },
    {
      "epoch": 12.877793628150261,
      "grad_norm": 0.2316208928823471,
      "learning_rate": 5.7078437599237856e-05,
      "loss": 0.0639,
      "step": 6763
    },
    {
      "epoch": 12.879695672848312,
      "grad_norm": 0.17722217738628387,
      "learning_rate": 5.70720863766275e-05,
      "loss": 0.0673,
      "step": 6764
    },
    {
      "epoch": 12.881597717546363,
      "grad_norm": 0.1882850080728531,
      "learning_rate": 5.706573515401715e-05,
      "loss": 0.0526,
      "step": 6765
    },
    {
      "epoch": 12.883499762244412,
      "grad_norm": 0.2745272219181061,
      "learning_rate": 5.70593839314068e-05,
      "loss": 0.0669,
      "step": 6766
    },
    {
      "epoch": 12.885401806942463,
      "grad_norm": 0.11121740192174911,
      "learning_rate": 5.705303270879645e-05,
      "loss": 0.0524,
      "step": 6767
    },
    {
      "epoch": 12.887303851640514,
      "grad_norm": 0.25695016980171204,
      "learning_rate": 5.7046681486186095e-05,
      "loss": 0.0619,
      "step": 6768
    },
    {
      "epoch": 12.889205896338565,
      "grad_norm": 0.2739970088005066,
      "learning_rate": 5.7040330263575734e-05,
      "loss": 0.0784,
      "step": 6769
    },
    {
      "epoch": 12.891107941036614,
      "grad_norm": 0.35055750608444214,
      "learning_rate": 5.703397904096539e-05,
      "loss": 0.0932,
      "step": 6770
    },
    {
      "epoch": 12.893009985734665,
      "grad_norm": 0.16527678072452545,
      "learning_rate": 5.702762781835503e-05,
      "loss": 0.044,
      "step": 6771
    },
    {
      "epoch": 12.894912030432716,
      "grad_norm": 0.19253818690776825,
      "learning_rate": 5.702127659574469e-05,
      "loss": 0.0612,
      "step": 6772
    },
    {
      "epoch": 12.896814075130766,
      "grad_norm": 0.24265630543231964,
      "learning_rate": 5.7014925373134334e-05,
      "loss": 0.0736,
      "step": 6773
    },
    {
      "epoch": 12.898716119828816,
      "grad_norm": 0.31677743792533875,
      "learning_rate": 5.700857415052397e-05,
      "loss": 0.0684,
      "step": 6774
    },
    {
      "epoch": 12.900618164526866,
      "grad_norm": 0.18557855486869812,
      "learning_rate": 5.700222292791363e-05,
      "loss": 0.0661,
      "step": 6775
    },
    {
      "epoch": 12.902520209224917,
      "grad_norm": 0.13152940571308136,
      "learning_rate": 5.699587170530327e-05,
      "loss": 0.0682,
      "step": 6776
    },
    {
      "epoch": 12.904422253922966,
      "grad_norm": 0.16339555382728577,
      "learning_rate": 5.698952048269293e-05,
      "loss": 0.0729,
      "step": 6777
    },
    {
      "epoch": 12.906324298621017,
      "grad_norm": 0.14895136654376984,
      "learning_rate": 5.6983169260082566e-05,
      "loss": 0.0647,
      "step": 6778
    },
    {
      "epoch": 12.908226343319068,
      "grad_norm": 0.18370980024337769,
      "learning_rate": 5.697681803747221e-05,
      "loss": 0.0676,
      "step": 6779
    },
    {
      "epoch": 12.91012838801712,
      "grad_norm": 0.2819598317146301,
      "learning_rate": 5.697046681486187e-05,
      "loss": 0.0698,
      "step": 6780
    },
    {
      "epoch": 12.912030432715168,
      "grad_norm": 0.2503798305988312,
      "learning_rate": 5.696411559225151e-05,
      "loss": 0.0609,
      "step": 6781
    },
    {
      "epoch": 12.91393247741322,
      "grad_norm": 0.21035026013851166,
      "learning_rate": 5.695776436964115e-05,
      "loss": 0.0779,
      "step": 6782
    },
    {
      "epoch": 12.91583452211127,
      "grad_norm": 0.20218412578105927,
      "learning_rate": 5.6951413147030805e-05,
      "loss": 0.0544,
      "step": 6783
    },
    {
      "epoch": 12.917736566809321,
      "grad_norm": 0.1813918650150299,
      "learning_rate": 5.694506192442045e-05,
      "loss": 0.0736,
      "step": 6784
    },
    {
      "epoch": 12.91963861150737,
      "grad_norm": 0.24606718122959137,
      "learning_rate": 5.69387107018101e-05,
      "loss": 0.0602,
      "step": 6785
    },
    {
      "epoch": 12.921540656205421,
      "grad_norm": 0.26272520422935486,
      "learning_rate": 5.693235947919975e-05,
      "loss": 0.119,
      "step": 6786
    },
    {
      "epoch": 12.923442700903472,
      "grad_norm": 0.236623153090477,
      "learning_rate": 5.692600825658939e-05,
      "loss": 0.0748,
      "step": 6787
    },
    {
      "epoch": 12.925344745601521,
      "grad_norm": 0.20204751193523407,
      "learning_rate": 5.6919657033979044e-05,
      "loss": 0.0672,
      "step": 6788
    },
    {
      "epoch": 12.927246790299572,
      "grad_norm": 0.20411472022533417,
      "learning_rate": 5.691330581136869e-05,
      "loss": 0.0616,
      "step": 6789
    },
    {
      "epoch": 12.929148834997623,
      "grad_norm": 0.2538439929485321,
      "learning_rate": 5.690695458875834e-05,
      "loss": 0.0664,
      "step": 6790
    },
    {
      "epoch": 12.931050879695674,
      "grad_norm": 0.14548756182193756,
      "learning_rate": 5.6900603366147986e-05,
      "loss": 0.0713,
      "step": 6791
    },
    {
      "epoch": 12.932952924393723,
      "grad_norm": 0.2531229555606842,
      "learning_rate": 5.689425214353763e-05,
      "loss": 0.0797,
      "step": 6792
    },
    {
      "epoch": 12.934854969091774,
      "grad_norm": 0.14427652955055237,
      "learning_rate": 5.688790092092728e-05,
      "loss": 0.0643,
      "step": 6793
    },
    {
      "epoch": 12.936757013789824,
      "grad_norm": 0.14956879615783691,
      "learning_rate": 5.688154969831693e-05,
      "loss": 0.078,
      "step": 6794
    },
    {
      "epoch": 12.938659058487875,
      "grad_norm": 0.2142200469970703,
      "learning_rate": 5.687519847570658e-05,
      "loss": 0.0764,
      "step": 6795
    },
    {
      "epoch": 12.940561103185924,
      "grad_norm": 0.3186730146408081,
      "learning_rate": 5.6868847253096225e-05,
      "loss": 0.0628,
      "step": 6796
    },
    {
      "epoch": 12.942463147883975,
      "grad_norm": 0.20074574649333954,
      "learning_rate": 5.686249603048587e-05,
      "loss": 0.0576,
      "step": 6797
    },
    {
      "epoch": 12.944365192582026,
      "grad_norm": 0.18254826962947845,
      "learning_rate": 5.685614480787552e-05,
      "loss": 0.0641,
      "step": 6798
    },
    {
      "epoch": 12.946267237280075,
      "grad_norm": 0.16693158447742462,
      "learning_rate": 5.684979358526517e-05,
      "loss": 0.0701,
      "step": 6799
    },
    {
      "epoch": 12.948169281978126,
      "grad_norm": 0.16787877678871155,
      "learning_rate": 5.6843442362654805e-05,
      "loss": 0.0484,
      "step": 6800
    },
    {
      "epoch": 12.950071326676177,
      "grad_norm": 0.24660000205039978,
      "learning_rate": 5.6837091140044464e-05,
      "loss": 0.0587,
      "step": 6801
    },
    {
      "epoch": 12.951973371374228,
      "grad_norm": 0.18229947984218597,
      "learning_rate": 5.683073991743411e-05,
      "loss": 0.073,
      "step": 6802
    },
    {
      "epoch": 12.953875416072277,
      "grad_norm": 0.16272900998592377,
      "learning_rate": 5.682438869482376e-05,
      "loss": 0.0604,
      "step": 6803
    },
    {
      "epoch": 12.955777460770328,
      "grad_norm": 0.34907060861587524,
      "learning_rate": 5.6818037472213406e-05,
      "loss": 0.0667,
      "step": 6804
    },
    {
      "epoch": 12.957679505468379,
      "grad_norm": 0.18044497072696686,
      "learning_rate": 5.6811686249603044e-05,
      "loss": 0.0738,
      "step": 6805
    },
    {
      "epoch": 12.95958155016643,
      "grad_norm": 0.22940124571323395,
      "learning_rate": 5.68053350269927e-05,
      "loss": 0.0549,
      "step": 6806
    },
    {
      "epoch": 12.961483594864479,
      "grad_norm": 0.2308824360370636,
      "learning_rate": 5.679898380438234e-05,
      "loss": 0.0707,
      "step": 6807
    },
    {
      "epoch": 12.96338563956253,
      "grad_norm": 0.16795487701892853,
      "learning_rate": 5.6792632581772e-05,
      "loss": 0.0697,
      "step": 6808
    },
    {
      "epoch": 12.96528768426058,
      "grad_norm": 0.12602196633815765,
      "learning_rate": 5.6786281359161645e-05,
      "loss": 0.0684,
      "step": 6809
    },
    {
      "epoch": 12.96718972895863,
      "grad_norm": 0.3651646673679352,
      "learning_rate": 5.677993013655128e-05,
      "loss": 0.0826,
      "step": 6810
    },
    {
      "epoch": 12.96909177365668,
      "grad_norm": 0.1852756291627884,
      "learning_rate": 5.677357891394094e-05,
      "loss": 0.0611,
      "step": 6811
    },
    {
      "epoch": 12.970993818354732,
      "grad_norm": 0.18153737485408783,
      "learning_rate": 5.676722769133058e-05,
      "loss": 0.0456,
      "step": 6812
    },
    {
      "epoch": 12.972895863052782,
      "grad_norm": 0.2669750154018402,
      "learning_rate": 5.676087646872024e-05,
      "loss": 0.0751,
      "step": 6813
    },
    {
      "epoch": 12.974797907750832,
      "grad_norm": 0.1596618890762329,
      "learning_rate": 5.675452524610988e-05,
      "loss": 0.0502,
      "step": 6814
    },
    {
      "epoch": 12.976699952448882,
      "grad_norm": 0.28879284858703613,
      "learning_rate": 5.674817402349952e-05,
      "loss": 0.0732,
      "step": 6815
    },
    {
      "epoch": 12.978601997146933,
      "grad_norm": 0.14341185986995697,
      "learning_rate": 5.674182280088918e-05,
      "loss": 0.0654,
      "step": 6816
    },
    {
      "epoch": 12.980504041844984,
      "grad_norm": 0.27027440071105957,
      "learning_rate": 5.673547157827882e-05,
      "loss": 0.0882,
      "step": 6817
    },
    {
      "epoch": 12.982406086543033,
      "grad_norm": 0.18623539805412292,
      "learning_rate": 5.6729120355668464e-05,
      "loss": 0.0525,
      "step": 6818
    },
    {
      "epoch": 12.984308131241084,
      "grad_norm": 0.3178105354309082,
      "learning_rate": 5.6722769133058116e-05,
      "loss": 0.0671,
      "step": 6819
    },
    {
      "epoch": 12.986210175939135,
      "grad_norm": 0.17404824495315552,
      "learning_rate": 5.671641791044776e-05,
      "loss": 0.0524,
      "step": 6820
    },
    {
      "epoch": 12.988112220637184,
      "grad_norm": 0.26026493310928345,
      "learning_rate": 5.671006668783741e-05,
      "loss": 0.0612,
      "step": 6821
    },
    {
      "epoch": 12.990014265335235,
      "grad_norm": 0.22481663525104523,
      "learning_rate": 5.670371546522706e-05,
      "loss": 0.0582,
      "step": 6822
    },
    {
      "epoch": 12.991916310033286,
      "grad_norm": 0.2761218547821045,
      "learning_rate": 5.66973642426167e-05,
      "loss": 0.0805,
      "step": 6823
    },
    {
      "epoch": 12.993818354731337,
      "grad_norm": 0.28394970297813416,
      "learning_rate": 5.6691013020006354e-05,
      "loss": 0.0675,
      "step": 6824
    },
    {
      "epoch": 12.995720399429386,
      "grad_norm": 0.2090967446565628,
      "learning_rate": 5.6684661797396e-05,
      "loss": 0.0471,
      "step": 6825
    },
    {
      "epoch": 12.997622444127437,
      "grad_norm": 0.13171760737895966,
      "learning_rate": 5.667831057478565e-05,
      "loss": 0.0535,
      "step": 6826
    },
    {
      "epoch": 12.999524488825488,
      "grad_norm": 0.2904289662837982,
      "learning_rate": 5.6671959352175296e-05,
      "loss": 0.0671,
      "step": 6827
    },
    {
      "epoch": 13.001426533523539,
      "grad_norm": 0.19046546518802643,
      "learning_rate": 5.666560812956494e-05,
      "loss": 0.0539,
      "step": 6828
    },
    {
      "epoch": 13.003328578221588,
      "grad_norm": 0.22456832230091095,
      "learning_rate": 5.665925690695459e-05,
      "loss": 0.0717,
      "step": 6829
    },
    {
      "epoch": 13.005230622919639,
      "grad_norm": 0.1796262115240097,
      "learning_rate": 5.665290568434424e-05,
      "loss": 0.0637,
      "step": 6830
    },
    {
      "epoch": 13.00713266761769,
      "grad_norm": 0.15445412695407867,
      "learning_rate": 5.664655446173389e-05,
      "loss": 0.0533,
      "step": 6831
    },
    {
      "epoch": 13.009034712315739,
      "grad_norm": 0.1539810299873352,
      "learning_rate": 5.6640203239123535e-05,
      "loss": 0.0606,
      "step": 6832
    },
    {
      "epoch": 13.01093675701379,
      "grad_norm": 0.12736746668815613,
      "learning_rate": 5.663385201651318e-05,
      "loss": 0.059,
      "step": 6833
    },
    {
      "epoch": 13.01283880171184,
      "grad_norm": 0.14891792833805084,
      "learning_rate": 5.662750079390283e-05,
      "loss": 0.0619,
      "step": 6834
    },
    {
      "epoch": 13.014740846409891,
      "grad_norm": 0.2501888871192932,
      "learning_rate": 5.662114957129248e-05,
      "loss": 0.0684,
      "step": 6835
    },
    {
      "epoch": 13.01664289110794,
      "grad_norm": 0.1664845049381256,
      "learning_rate": 5.6614798348682116e-05,
      "loss": 0.038,
      "step": 6836
    },
    {
      "epoch": 13.018544935805991,
      "grad_norm": 0.23480184376239777,
      "learning_rate": 5.6608447126071774e-05,
      "loss": 0.0641,
      "step": 6837
    },
    {
      "epoch": 13.020446980504042,
      "grad_norm": 0.43155115842819214,
      "learning_rate": 5.660209590346142e-05,
      "loss": 0.0728,
      "step": 6838
    },
    {
      "epoch": 13.022349025202093,
      "grad_norm": 0.13202181458473206,
      "learning_rate": 5.659574468085107e-05,
      "loss": 0.0662,
      "step": 6839
    },
    {
      "epoch": 13.024251069900142,
      "grad_norm": 0.22883914411067963,
      "learning_rate": 5.6589393458240716e-05,
      "loss": 0.0761,
      "step": 6840
    },
    {
      "epoch": 13.026153114598193,
      "grad_norm": 0.2755950391292572,
      "learning_rate": 5.6583042235630354e-05,
      "loss": 0.0561,
      "step": 6841
    },
    {
      "epoch": 13.028055159296244,
      "grad_norm": 0.264454185962677,
      "learning_rate": 5.657669101302001e-05,
      "loss": 0.0748,
      "step": 6842
    },
    {
      "epoch": 13.029957203994293,
      "grad_norm": 0.14492832124233246,
      "learning_rate": 5.657033979040965e-05,
      "loss": 0.0572,
      "step": 6843
    },
    {
      "epoch": 13.031859248692344,
      "grad_norm": 0.20079435408115387,
      "learning_rate": 5.656398856779931e-05,
      "loss": 0.0594,
      "step": 6844
    },
    {
      "epoch": 13.033761293390395,
      "grad_norm": 0.26874688267707825,
      "learning_rate": 5.6557637345188955e-05,
      "loss": 0.0679,
      "step": 6845
    },
    {
      "epoch": 13.035663338088446,
      "grad_norm": 0.12486004829406738,
      "learning_rate": 5.655128612257859e-05,
      "loss": 0.0768,
      "step": 6846
    },
    {
      "epoch": 13.037565382786495,
      "grad_norm": 0.1992534101009369,
      "learning_rate": 5.654493489996825e-05,
      "loss": 0.0716,
      "step": 6847
    },
    {
      "epoch": 13.039467427484546,
      "grad_norm": 0.20536714792251587,
      "learning_rate": 5.653858367735789e-05,
      "loss": 0.0653,
      "step": 6848
    },
    {
      "epoch": 13.041369472182597,
      "grad_norm": 0.18057265877723694,
      "learning_rate": 5.653223245474755e-05,
      "loss": 0.0879,
      "step": 6849
    },
    {
      "epoch": 13.043271516880647,
      "grad_norm": 0.2120429128408432,
      "learning_rate": 5.652588123213719e-05,
      "loss": 0.0556,
      "step": 6850
    },
    {
      "epoch": 13.045173561578697,
      "grad_norm": 0.08657226711511612,
      "learning_rate": 5.651953000952683e-05,
      "loss": 0.0394,
      "step": 6851
    },
    {
      "epoch": 13.047075606276747,
      "grad_norm": 0.2886129915714264,
      "learning_rate": 5.651317878691649e-05,
      "loss": 0.0573,
      "step": 6852
    },
    {
      "epoch": 13.048977650974798,
      "grad_norm": 0.14705555140972137,
      "learning_rate": 5.650682756430613e-05,
      "loss": 0.0598,
      "step": 6853
    },
    {
      "epoch": 13.050879695672847,
      "grad_norm": 0.13532306253910065,
      "learning_rate": 5.6500476341695774e-05,
      "loss": 0.046,
      "step": 6854
    },
    {
      "epoch": 13.052781740370898,
      "grad_norm": 0.1289561241865158,
      "learning_rate": 5.6494125119085426e-05,
      "loss": 0.055,
      "step": 6855
    },
    {
      "epoch": 13.05468378506895,
      "grad_norm": 0.22951653599739075,
      "learning_rate": 5.648777389647507e-05,
      "loss": 0.0565,
      "step": 6856
    },
    {
      "epoch": 13.056585829767,
      "grad_norm": 0.2005440890789032,
      "learning_rate": 5.648142267386472e-05,
      "loss": 0.0723,
      "step": 6857
    },
    {
      "epoch": 13.05848787446505,
      "grad_norm": 0.18460001051425934,
      "learning_rate": 5.647507145125437e-05,
      "loss": 0.0673,
      "step": 6858
    },
    {
      "epoch": 13.0603899191631,
      "grad_norm": 0.16628994047641754,
      "learning_rate": 5.646872022864401e-05,
      "loss": 0.0655,
      "step": 6859
    },
    {
      "epoch": 13.062291963861151,
      "grad_norm": 0.11353509873151779,
      "learning_rate": 5.6462369006033665e-05,
      "loss": 0.0724,
      "step": 6860
    },
    {
      "epoch": 13.064194008559202,
      "grad_norm": 0.16154246032238007,
      "learning_rate": 5.645601778342331e-05,
      "loss": 0.0549,
      "step": 6861
    },
    {
      "epoch": 13.066096053257251,
      "grad_norm": 0.20058204233646393,
      "learning_rate": 5.644966656081296e-05,
      "loss": 0.0529,
      "step": 6862
    },
    {
      "epoch": 13.067998097955302,
      "grad_norm": 0.09383098036050797,
      "learning_rate": 5.644331533820261e-05,
      "loss": 0.0483,
      "step": 6863
    },
    {
      "epoch": 13.069900142653353,
      "grad_norm": 0.23607304692268372,
      "learning_rate": 5.643696411559225e-05,
      "loss": 0.0793,
      "step": 6864
    },
    {
      "epoch": 13.071802187351402,
      "grad_norm": 0.1318606436252594,
      "learning_rate": 5.6430612892981904e-05,
      "loss": 0.0509,
      "step": 6865
    },
    {
      "epoch": 13.073704232049453,
      "grad_norm": 0.25798821449279785,
      "learning_rate": 5.642426167037155e-05,
      "loss": 0.0718,
      "step": 6866
    },
    {
      "epoch": 13.075606276747504,
      "grad_norm": 0.21423642337322235,
      "learning_rate": 5.64179104477612e-05,
      "loss": 0.0569,
      "step": 6867
    },
    {
      "epoch": 13.077508321445555,
      "grad_norm": 0.16581445932388306,
      "learning_rate": 5.6411559225150846e-05,
      "loss": 0.0542,
      "step": 6868
    },
    {
      "epoch": 13.079410366143604,
      "grad_norm": 0.15126816928386688,
      "learning_rate": 5.640520800254049e-05,
      "loss": 0.0478,
      "step": 6869
    },
    {
      "epoch": 13.081312410841655,
      "grad_norm": 0.1429680585861206,
      "learning_rate": 5.639885677993014e-05,
      "loss": 0.062,
      "step": 6870
    },
    {
      "epoch": 13.083214455539705,
      "grad_norm": 0.049759067595005035,
      "learning_rate": 5.639250555731979e-05,
      "loss": 0.0326,
      "step": 6871
    },
    {
      "epoch": 13.085116500237756,
      "grad_norm": 0.19529815018177032,
      "learning_rate": 5.6386154334709426e-05,
      "loss": 0.0561,
      "step": 6872
    },
    {
      "epoch": 13.087018544935805,
      "grad_norm": 0.17486746609210968,
      "learning_rate": 5.6379803112099085e-05,
      "loss": 0.0596,
      "step": 6873
    },
    {
      "epoch": 13.088920589633856,
      "grad_norm": 0.12108680605888367,
      "learning_rate": 5.637345188948873e-05,
      "loss": 0.0475,
      "step": 6874
    },
    {
      "epoch": 13.090822634331907,
      "grad_norm": 0.16674591600894928,
      "learning_rate": 5.636710066687838e-05,
      "loss": 0.0877,
      "step": 6875
    },
    {
      "epoch": 13.092724679029958,
      "grad_norm": 0.14456391334533691,
      "learning_rate": 5.636074944426803e-05,
      "loss": 0.0416,
      "step": 6876
    },
    {
      "epoch": 13.094626723728007,
      "grad_norm": 0.22981326282024384,
      "learning_rate": 5.6354398221657665e-05,
      "loss": 0.0885,
      "step": 6877
    },
    {
      "epoch": 13.096528768426058,
      "grad_norm": 0.14590546488761902,
      "learning_rate": 5.6348046999047324e-05,
      "loss": 0.0377,
      "step": 6878
    },
    {
      "epoch": 13.098430813124109,
      "grad_norm": 0.18224582076072693,
      "learning_rate": 5.634169577643696e-05,
      "loss": 0.059,
      "step": 6879
    },
    {
      "epoch": 13.100332857822158,
      "grad_norm": 0.26473188400268555,
      "learning_rate": 5.633534455382662e-05,
      "loss": 0.0562,
      "step": 6880
    },
    {
      "epoch": 13.102234902520209,
      "grad_norm": 0.2478216588497162,
      "learning_rate": 5.632899333121626e-05,
      "loss": 0.0563,
      "step": 6881
    },
    {
      "epoch": 13.10413694721826,
      "grad_norm": 0.29242587089538574,
      "learning_rate": 5.6322642108605904e-05,
      "loss": 0.0608,
      "step": 6882
    },
    {
      "epoch": 13.10603899191631,
      "grad_norm": 0.16479073464870453,
      "learning_rate": 5.631629088599556e-05,
      "loss": 0.0678,
      "step": 6883
    },
    {
      "epoch": 13.10794103661436,
      "grad_norm": 0.23486103117465973,
      "learning_rate": 5.63099396633852e-05,
      "loss": 0.0652,
      "step": 6884
    },
    {
      "epoch": 13.10984308131241,
      "grad_norm": 0.154318168759346,
      "learning_rate": 5.630358844077486e-05,
      "loss": 0.1446,
      "step": 6885
    },
    {
      "epoch": 13.111745126010462,
      "grad_norm": 0.14569495618343353,
      "learning_rate": 5.62972372181645e-05,
      "loss": 0.053,
      "step": 6886
    },
    {
      "epoch": 13.113647170708512,
      "grad_norm": 0.17716017365455627,
      "learning_rate": 5.629088599555414e-05,
      "loss": 0.0676,
      "step": 6887
    },
    {
      "epoch": 13.115549215406562,
      "grad_norm": 0.20627303421497345,
      "learning_rate": 5.6284534772943795e-05,
      "loss": 0.0589,
      "step": 6888
    },
    {
      "epoch": 13.117451260104612,
      "grad_norm": 0.1511334627866745,
      "learning_rate": 5.627818355033344e-05,
      "loss": 0.058,
      "step": 6889
    },
    {
      "epoch": 13.119353304802663,
      "grad_norm": 0.17161938548088074,
      "learning_rate": 5.6271832327723085e-05,
      "loss": 0.0459,
      "step": 6890
    },
    {
      "epoch": 13.121255349500712,
      "grad_norm": 0.193283349275589,
      "learning_rate": 5.6265481105112736e-05,
      "loss": 0.0838,
      "step": 6891
    },
    {
      "epoch": 13.123157394198763,
      "grad_norm": 0.12341416627168655,
      "learning_rate": 5.625912988250238e-05,
      "loss": 0.0658,
      "step": 6892
    },
    {
      "epoch": 13.125059438896814,
      "grad_norm": 0.1398099958896637,
      "learning_rate": 5.6252778659892033e-05,
      "loss": 0.0475,
      "step": 6893
    },
    {
      "epoch": 13.126961483594865,
      "grad_norm": 0.1959352046251297,
      "learning_rate": 5.624642743728168e-05,
      "loss": 0.053,
      "step": 6894
    },
    {
      "epoch": 13.128863528292914,
      "grad_norm": 0.13947531580924988,
      "learning_rate": 5.6240076214671324e-05,
      "loss": 0.0479,
      "step": 6895
    },
    {
      "epoch": 13.130765572990965,
      "grad_norm": 0.15292446315288544,
      "learning_rate": 5.6233724992060975e-05,
      "loss": 0.052,
      "step": 6896
    },
    {
      "epoch": 13.132667617689016,
      "grad_norm": 0.20733055472373962,
      "learning_rate": 5.622737376945062e-05,
      "loss": 0.0528,
      "step": 6897
    },
    {
      "epoch": 13.134569662387067,
      "grad_norm": 0.18862584233283997,
      "learning_rate": 5.622102254684027e-05,
      "loss": 0.0677,
      "step": 6898
    },
    {
      "epoch": 13.136471707085116,
      "grad_norm": 0.15024633705615997,
      "learning_rate": 5.621467132422992e-05,
      "loss": 0.0485,
      "step": 6899
    },
    {
      "epoch": 13.138373751783167,
      "grad_norm": 0.2352614849805832,
      "learning_rate": 5.620832010161956e-05,
      "loss": 0.0782,
      "step": 6900
    },
    {
      "epoch": 13.140275796481218,
      "grad_norm": 0.23415392637252808,
      "learning_rate": 5.6201968879009214e-05,
      "loss": 0.064,
      "step": 6901
    },
    {
      "epoch": 13.142177841179267,
      "grad_norm": 0.11979921162128448,
      "learning_rate": 5.619561765639886e-05,
      "loss": 0.0543,
      "step": 6902
    },
    {
      "epoch": 13.144079885877318,
      "grad_norm": 0.14752525091171265,
      "learning_rate": 5.618926643378851e-05,
      "loss": 0.0593,
      "step": 6903
    },
    {
      "epoch": 13.145981930575369,
      "grad_norm": 0.19310492277145386,
      "learning_rate": 5.6182915211178156e-05,
      "loss": 0.0586,
      "step": 6904
    },
    {
      "epoch": 13.14788397527342,
      "grad_norm": 0.12309253960847855,
      "learning_rate": 5.61765639885678e-05,
      "loss": 0.0548,
      "step": 6905
    },
    {
      "epoch": 13.149786019971469,
      "grad_norm": 0.09493910521268845,
      "learning_rate": 5.617021276595745e-05,
      "loss": 0.0878,
      "step": 6906
    },
    {
      "epoch": 13.15168806466952,
      "grad_norm": 0.161015585064888,
      "learning_rate": 5.61638615433471e-05,
      "loss": 0.0655,
      "step": 6907
    },
    {
      "epoch": 13.15359010936757,
      "grad_norm": 0.31709450483322144,
      "learning_rate": 5.6157510320736736e-05,
      "loss": 0.0841,
      "step": 6908
    },
    {
      "epoch": 13.155492154065621,
      "grad_norm": 0.20902904868125916,
      "learning_rate": 5.6151159098126395e-05,
      "loss": 0.0471,
      "step": 6909
    },
    {
      "epoch": 13.15739419876367,
      "grad_norm": 0.13743600249290466,
      "learning_rate": 5.6144807875516033e-05,
      "loss": 0.058,
      "step": 6910
    },
    {
      "epoch": 13.159296243461721,
      "grad_norm": 0.1610523909330368,
      "learning_rate": 5.613845665290569e-05,
      "loss": 0.0612,
      "step": 6911
    },
    {
      "epoch": 13.161198288159772,
      "grad_norm": 0.13575279712677002,
      "learning_rate": 5.613210543029534e-05,
      "loss": 0.0683,
      "step": 6912
    },
    {
      "epoch": 13.163100332857821,
      "grad_norm": 0.15719729661941528,
      "learning_rate": 5.6125754207684975e-05,
      "loss": 0.0565,
      "step": 6913
    },
    {
      "epoch": 13.165002377555872,
      "grad_norm": 0.17677704989910126,
      "learning_rate": 5.6119402985074634e-05,
      "loss": 0.0441,
      "step": 6914
    },
    {
      "epoch": 13.166904422253923,
      "grad_norm": 0.11826075613498688,
      "learning_rate": 5.611305176246427e-05,
      "loss": 0.0513,
      "step": 6915
    },
    {
      "epoch": 13.168806466951974,
      "grad_norm": 0.13202832639217377,
      "learning_rate": 5.610670053985393e-05,
      "loss": 0.0492,
      "step": 6916
    },
    {
      "epoch": 13.170708511650023,
      "grad_norm": 0.264354944229126,
      "learning_rate": 5.610034931724357e-05,
      "loss": 0.1007,
      "step": 6917
    },
    {
      "epoch": 13.172610556348074,
      "grad_norm": 0.19659414887428284,
      "learning_rate": 5.6093998094633214e-05,
      "loss": 0.0618,
      "step": 6918
    },
    {
      "epoch": 13.174512601046125,
      "grad_norm": 0.2893676459789276,
      "learning_rate": 5.608764687202287e-05,
      "loss": 0.0606,
      "step": 6919
    },
    {
      "epoch": 13.176414645744176,
      "grad_norm": 0.12958897650241852,
      "learning_rate": 5.608129564941251e-05,
      "loss": 0.0488,
      "step": 6920
    },
    {
      "epoch": 13.178316690442225,
      "grad_norm": 0.18396779894828796,
      "learning_rate": 5.607494442680217e-05,
      "loss": 0.0595,
      "step": 6921
    },
    {
      "epoch": 13.180218735140276,
      "grad_norm": 0.14515328407287598,
      "learning_rate": 5.606859320419181e-05,
      "loss": 0.0506,
      "step": 6922
    },
    {
      "epoch": 13.182120779838327,
      "grad_norm": 0.21712662279605865,
      "learning_rate": 5.606224198158145e-05,
      "loss": 0.0807,
      "step": 6923
    },
    {
      "epoch": 13.184022824536376,
      "grad_norm": 0.10604386031627655,
      "learning_rate": 5.6055890758971105e-05,
      "loss": 0.0654,
      "step": 6924
    },
    {
      "epoch": 13.185924869234427,
      "grad_norm": 0.25209611654281616,
      "learning_rate": 5.604953953636075e-05,
      "loss": 0.0643,
      "step": 6925
    },
    {
      "epoch": 13.187826913932478,
      "grad_norm": 0.22648891806602478,
      "learning_rate": 5.6043188313750395e-05,
      "loss": 0.0522,
      "step": 6926
    },
    {
      "epoch": 13.189728958630528,
      "grad_norm": 0.202748104929924,
      "learning_rate": 5.603683709114005e-05,
      "loss": 0.0352,
      "step": 6927
    },
    {
      "epoch": 13.191631003328578,
      "grad_norm": 0.23080334067344666,
      "learning_rate": 5.603048586852969e-05,
      "loss": 0.0616,
      "step": 6928
    },
    {
      "epoch": 13.193533048026628,
      "grad_norm": 0.09643086045980453,
      "learning_rate": 5.6024134645919344e-05,
      "loss": 0.0688,
      "step": 6929
    },
    {
      "epoch": 13.19543509272468,
      "grad_norm": 0.3101188540458679,
      "learning_rate": 5.601778342330899e-05,
      "loss": 0.0709,
      "step": 6930
    },
    {
      "epoch": 13.19733713742273,
      "grad_norm": 0.21519017219543457,
      "learning_rate": 5.6011432200698634e-05,
      "loss": 0.0642,
      "step": 6931
    },
    {
      "epoch": 13.19923918212078,
      "grad_norm": 0.34213313460350037,
      "learning_rate": 5.6005080978088286e-05,
      "loss": 0.0737,
      "step": 6932
    },
    {
      "epoch": 13.20114122681883,
      "grad_norm": 0.24912647902965546,
      "learning_rate": 5.599872975547793e-05,
      "loss": 0.0525,
      "step": 6933
    },
    {
      "epoch": 13.203043271516881,
      "grad_norm": 0.29063260555267334,
      "learning_rate": 5.599237853286758e-05,
      "loss": 0.0818,
      "step": 6934
    },
    {
      "epoch": 13.204945316214932,
      "grad_norm": 0.38843879103660583,
      "learning_rate": 5.598602731025723e-05,
      "loss": 0.0916,
      "step": 6935
    },
    {
      "epoch": 13.206847360912981,
      "grad_norm": 0.13748934864997864,
      "learning_rate": 5.597967608764687e-05,
      "loss": 0.0387,
      "step": 6936
    },
    {
      "epoch": 13.208749405611032,
      "grad_norm": 0.23910416662693024,
      "learning_rate": 5.5973324865036525e-05,
      "loss": 0.0579,
      "step": 6937
    },
    {
      "epoch": 13.210651450309083,
      "grad_norm": 0.12952692806720734,
      "learning_rate": 5.596697364242617e-05,
      "loss": 0.0485,
      "step": 6938
    },
    {
      "epoch": 13.212553495007132,
      "grad_norm": 0.14758580923080444,
      "learning_rate": 5.596062241981582e-05,
      "loss": 0.063,
      "step": 6939
    },
    {
      "epoch": 13.214455539705183,
      "grad_norm": 0.19575528800487518,
      "learning_rate": 5.595427119720547e-05,
      "loss": 0.0773,
      "step": 6940
    },
    {
      "epoch": 13.216357584403234,
      "grad_norm": 0.15974241495132446,
      "learning_rate": 5.594791997459511e-05,
      "loss": 0.0873,
      "step": 6941
    },
    {
      "epoch": 13.218259629101285,
      "grad_norm": 0.145432248711586,
      "learning_rate": 5.5941568751984764e-05,
      "loss": 0.0524,
      "step": 6942
    },
    {
      "epoch": 13.220161673799334,
      "grad_norm": 0.21533656120300293,
      "learning_rate": 5.593521752937441e-05,
      "loss": 0.0599,
      "step": 6943
    },
    {
      "epoch": 13.222063718497385,
      "grad_norm": 0.16924172639846802,
      "learning_rate": 5.592886630676405e-05,
      "loss": 0.0677,
      "step": 6944
    },
    {
      "epoch": 13.223965763195435,
      "grad_norm": 0.18145938217639923,
      "learning_rate": 5.5922515084153706e-05,
      "loss": 0.0612,
      "step": 6945
    },
    {
      "epoch": 13.225867807893486,
      "grad_norm": 0.10504286736249924,
      "learning_rate": 5.5916163861543344e-05,
      "loss": 0.0781,
      "step": 6946
    },
    {
      "epoch": 13.227769852591535,
      "grad_norm": 0.18427816033363342,
      "learning_rate": 5.5909812638933e-05,
      "loss": 0.0584,
      "step": 6947
    },
    {
      "epoch": 13.229671897289586,
      "grad_norm": 0.1417822390794754,
      "learning_rate": 5.590346141632265e-05,
      "loss": 0.0591,
      "step": 6948
    },
    {
      "epoch": 13.231573941987637,
      "grad_norm": 0.19255954027175903,
      "learning_rate": 5.5897110193712286e-05,
      "loss": 0.0752,
      "step": 6949
    },
    {
      "epoch": 13.233475986685686,
      "grad_norm": 0.1782093495130539,
      "learning_rate": 5.5890758971101944e-05,
      "loss": 0.0452,
      "step": 6950
    },
    {
      "epoch": 13.235378031383737,
      "grad_norm": 0.2388116419315338,
      "learning_rate": 5.588440774849158e-05,
      "loss": 0.0738,
      "step": 6951
    },
    {
      "epoch": 13.237280076081788,
      "grad_norm": 0.2225741297006607,
      "learning_rate": 5.587805652588124e-05,
      "loss": 0.0606,
      "step": 6952
    },
    {
      "epoch": 13.239182120779839,
      "grad_norm": 0.12010703235864639,
      "learning_rate": 5.587170530327088e-05,
      "loss": 0.0701,
      "step": 6953
    },
    {
      "epoch": 13.241084165477888,
      "grad_norm": 0.10702623426914215,
      "learning_rate": 5.5865354080660525e-05,
      "loss": 0.0665,
      "step": 6954
    },
    {
      "epoch": 13.242986210175939,
      "grad_norm": 0.20028775930404663,
      "learning_rate": 5.585900285805018e-05,
      "loss": 0.0596,
      "step": 6955
    },
    {
      "epoch": 13.24488825487399,
      "grad_norm": 0.12375194579362869,
      "learning_rate": 5.585265163543982e-05,
      "loss": 0.0528,
      "step": 6956
    },
    {
      "epoch": 13.24679029957204,
      "grad_norm": 0.15475822985172272,
      "learning_rate": 5.584630041282948e-05,
      "loss": 0.0643,
      "step": 6957
    },
    {
      "epoch": 13.24869234427009,
      "grad_norm": 0.14023445546627045,
      "learning_rate": 5.583994919021912e-05,
      "loss": 0.0493,
      "step": 6958
    },
    {
      "epoch": 13.25059438896814,
      "grad_norm": 0.34267377853393555,
      "learning_rate": 5.5833597967608764e-05,
      "loss": 0.0888,
      "step": 6959
    },
    {
      "epoch": 13.252496433666192,
      "grad_norm": 0.22459900379180908,
      "learning_rate": 5.5827246744998415e-05,
      "loss": 0.072,
      "step": 6960
    },
    {
      "epoch": 13.25439847836424,
      "grad_norm": 0.16262765228748322,
      "learning_rate": 5.582089552238806e-05,
      "loss": 0.041,
      "step": 6961
    },
    {
      "epoch": 13.256300523062292,
      "grad_norm": 0.16136904060840607,
      "learning_rate": 5.5814544299777706e-05,
      "loss": 0.0512,
      "step": 6962
    },
    {
      "epoch": 13.258202567760343,
      "grad_norm": 0.24696044623851776,
      "learning_rate": 5.580819307716736e-05,
      "loss": 0.0577,
      "step": 6963
    },
    {
      "epoch": 13.260104612458393,
      "grad_norm": 0.27188289165496826,
      "learning_rate": 5.5801841854557e-05,
      "loss": 0.0564,
      "step": 6964
    },
    {
      "epoch": 13.262006657156443,
      "grad_norm": 0.2794724106788635,
      "learning_rate": 5.5795490631946654e-05,
      "loss": 0.0614,
      "step": 6965
    },
    {
      "epoch": 13.263908701854493,
      "grad_norm": 0.181308776140213,
      "learning_rate": 5.57891394093363e-05,
      "loss": 0.0531,
      "step": 6966
    },
    {
      "epoch": 13.265810746552544,
      "grad_norm": 0.2799053490161896,
      "learning_rate": 5.5782788186725944e-05,
      "loss": 0.0701,
      "step": 6967
    },
    {
      "epoch": 13.267712791250595,
      "grad_norm": 0.34972909092903137,
      "learning_rate": 5.5776436964115596e-05,
      "loss": 0.1403,
      "step": 6968
    },
    {
      "epoch": 13.269614835948644,
      "grad_norm": 0.31298545002937317,
      "learning_rate": 5.577008574150524e-05,
      "loss": 0.1189,
      "step": 6969
    },
    {
      "epoch": 13.271516880646695,
      "grad_norm": 0.15706603229045868,
      "learning_rate": 5.576373451889489e-05,
      "loss": 0.0698,
      "step": 6970
    },
    {
      "epoch": 13.273418925344746,
      "grad_norm": 0.20245127379894257,
      "learning_rate": 5.575738329628454e-05,
      "loss": 0.0658,
      "step": 6971
    },
    {
      "epoch": 13.275320970042795,
      "grad_norm": 0.2303195297718048,
      "learning_rate": 5.575103207367418e-05,
      "loss": 0.0678,
      "step": 6972
    },
    {
      "epoch": 13.277223014740846,
      "grad_norm": 0.16143448650836945,
      "learning_rate": 5.5744680851063835e-05,
      "loss": 0.053,
      "step": 6973
    },
    {
      "epoch": 13.279125059438897,
      "grad_norm": 0.14536364376544952,
      "learning_rate": 5.573832962845348e-05,
      "loss": 0.0762,
      "step": 6974
    },
    {
      "epoch": 13.281027104136948,
      "grad_norm": 0.11136563122272491,
      "learning_rate": 5.573197840584313e-05,
      "loss": 0.0643,
      "step": 6975
    },
    {
      "epoch": 13.282929148834997,
      "grad_norm": 0.11188220232725143,
      "learning_rate": 5.572562718323278e-05,
      "loss": 0.0507,
      "step": 6976
    },
    {
      "epoch": 13.284831193533048,
      "grad_norm": 0.1301429122686386,
      "learning_rate": 5.571927596062242e-05,
      "loss": 0.0502,
      "step": 6977
    },
    {
      "epoch": 13.286733238231099,
      "grad_norm": 0.21348558366298676,
      "learning_rate": 5.5712924738012074e-05,
      "loss": 0.0819,
      "step": 6978
    },
    {
      "epoch": 13.28863528292915,
      "grad_norm": 0.19746661186218262,
      "learning_rate": 5.570657351540172e-05,
      "loss": 0.0736,
      "step": 6979
    },
    {
      "epoch": 13.290537327627199,
      "grad_norm": 0.21061228215694427,
      "learning_rate": 5.570022229279136e-05,
      "loss": 0.0724,
      "step": 6980
    },
    {
      "epoch": 13.29243937232525,
      "grad_norm": 0.19609305262565613,
      "learning_rate": 5.5693871070181016e-05,
      "loss": 0.0768,
      "step": 6981
    },
    {
      "epoch": 13.2943414170233,
      "grad_norm": 0.3237515985965729,
      "learning_rate": 5.5687519847570654e-05,
      "loss": 0.0907,
      "step": 6982
    },
    {
      "epoch": 13.29624346172135,
      "grad_norm": 0.2335703819990158,
      "learning_rate": 5.568116862496031e-05,
      "loss": 0.0758,
      "step": 6983
    },
    {
      "epoch": 13.2981455064194,
      "grad_norm": 0.19985871016979218,
      "learning_rate": 5.567481740234996e-05,
      "loss": 0.0549,
      "step": 6984
    },
    {
      "epoch": 13.300047551117451,
      "grad_norm": 0.1393788903951645,
      "learning_rate": 5.5668466179739596e-05,
      "loss": 0.0535,
      "step": 6985
    },
    {
      "epoch": 13.301949595815502,
      "grad_norm": 0.21756048500537872,
      "learning_rate": 5.5662114957129255e-05,
      "loss": 0.0675,
      "step": 6986
    },
    {
      "epoch": 13.303851640513551,
      "grad_norm": 0.3030356466770172,
      "learning_rate": 5.565576373451889e-05,
      "loss": 0.0777,
      "step": 6987
    },
    {
      "epoch": 13.305753685211602,
      "grad_norm": 0.22126442193984985,
      "learning_rate": 5.564941251190855e-05,
      "loss": 0.059,
      "step": 6988
    },
    {
      "epoch": 13.307655729909653,
      "grad_norm": 0.18576373159885406,
      "learning_rate": 5.564306128929819e-05,
      "loss": 0.0573,
      "step": 6989
    },
    {
      "epoch": 13.309557774607704,
      "grad_norm": 0.11999959498643875,
      "learning_rate": 5.5636710066687835e-05,
      "loss": 0.0502,
      "step": 6990
    },
    {
      "epoch": 13.311459819305753,
      "grad_norm": 0.1743020862340927,
      "learning_rate": 5.563035884407749e-05,
      "loss": 0.0431,
      "step": 6991
    },
    {
      "epoch": 13.313361864003804,
      "grad_norm": 0.1762196123600006,
      "learning_rate": 5.562400762146713e-05,
      "loss": 0.0785,
      "step": 6992
    },
    {
      "epoch": 13.315263908701855,
      "grad_norm": 0.17211808264255524,
      "learning_rate": 5.561765639885679e-05,
      "loss": 0.062,
      "step": 6993
    },
    {
      "epoch": 13.317165953399904,
      "grad_norm": 0.18189425766468048,
      "learning_rate": 5.561130517624643e-05,
      "loss": 0.0509,
      "step": 6994
    },
    {
      "epoch": 13.319067998097955,
      "grad_norm": 0.25210216641426086,
      "learning_rate": 5.5604953953636074e-05,
      "loss": 0.0502,
      "step": 6995
    },
    {
      "epoch": 13.320970042796006,
      "grad_norm": 0.16832081973552704,
      "learning_rate": 5.5598602731025726e-05,
      "loss": 0.0775,
      "step": 6996
    },
    {
      "epoch": 13.322872087494057,
      "grad_norm": 0.2224736213684082,
      "learning_rate": 5.559225150841537e-05,
      "loss": 0.058,
      "step": 6997
    },
    {
      "epoch": 13.324774132192106,
      "grad_norm": 0.17822198569774628,
      "learning_rate": 5.5585900285805016e-05,
      "loss": 0.0648,
      "step": 6998
    },
    {
      "epoch": 13.326676176890157,
      "grad_norm": 0.12076257914304733,
      "learning_rate": 5.557954906319467e-05,
      "loss": 0.0414,
      "step": 6999
    },
    {
      "epoch": 13.328578221588208,
      "grad_norm": 0.2498164027929306,
      "learning_rate": 5.557319784058431e-05,
      "loss": 0.0601,
      "step": 7000
    },
    {
      "epoch": 13.330480266286258,
      "grad_norm": 0.128321573138237,
      "learning_rate": 5.5566846617973965e-05,
      "loss": 0.0526,
      "step": 7001
    },
    {
      "epoch": 13.332382310984308,
      "grad_norm": 0.4024381935596466,
      "learning_rate": 5.556049539536361e-05,
      "loss": 0.0775,
      "step": 7002
    },
    {
      "epoch": 13.334284355682358,
      "grad_norm": 0.15158173441886902,
      "learning_rate": 5.5554144172753255e-05,
      "loss": 0.046,
      "step": 7003
    },
    {
      "epoch": 13.33618640038041,
      "grad_norm": 0.20958949625492096,
      "learning_rate": 5.554779295014291e-05,
      "loss": 0.0629,
      "step": 7004
    },
    {
      "epoch": 13.338088445078458,
      "grad_norm": 0.20752930641174316,
      "learning_rate": 5.554144172753255e-05,
      "loss": 0.0659,
      "step": 7005
    },
    {
      "epoch": 13.33999048977651,
      "grad_norm": 0.18724681437015533,
      "learning_rate": 5.5535090504922204e-05,
      "loss": 0.051,
      "step": 7006
    },
    {
      "epoch": 13.34189253447456,
      "grad_norm": 0.1965479701757431,
      "learning_rate": 5.552873928231185e-05,
      "loss": 0.0591,
      "step": 7007
    },
    {
      "epoch": 13.343794579172611,
      "grad_norm": 0.12148141860961914,
      "learning_rate": 5.5522388059701494e-05,
      "loss": 0.0607,
      "step": 7008
    },
    {
      "epoch": 13.34569662387066,
      "grad_norm": 0.19634923338890076,
      "learning_rate": 5.5516036837091146e-05,
      "loss": 0.0603,
      "step": 7009
    },
    {
      "epoch": 13.347598668568711,
      "grad_norm": 0.16275207698345184,
      "learning_rate": 5.550968561448079e-05,
      "loss": 0.0504,
      "step": 7010
    },
    {
      "epoch": 13.349500713266762,
      "grad_norm": 0.1558467447757721,
      "learning_rate": 5.550333439187044e-05,
      "loss": 0.0836,
      "step": 7011
    },
    {
      "epoch": 13.351402757964813,
      "grad_norm": 0.12177915871143341,
      "learning_rate": 5.549698316926009e-05,
      "loss": 0.0455,
      "step": 7012
    },
    {
      "epoch": 13.353304802662862,
      "grad_norm": 0.341022789478302,
      "learning_rate": 5.549063194664973e-05,
      "loss": 0.092,
      "step": 7013
    },
    {
      "epoch": 13.355206847360913,
      "grad_norm": 0.21205639839172363,
      "learning_rate": 5.5484280724039384e-05,
      "loss": 0.0562,
      "step": 7014
    },
    {
      "epoch": 13.357108892058964,
      "grad_norm": 0.2380506545305252,
      "learning_rate": 5.547792950142903e-05,
      "loss": 0.0593,
      "step": 7015
    },
    {
      "epoch": 13.359010936757013,
      "grad_norm": 0.24583934247493744,
      "learning_rate": 5.547157827881867e-05,
      "loss": 0.0501,
      "step": 7016
    },
    {
      "epoch": 13.360912981455064,
      "grad_norm": 0.11172381788492203,
      "learning_rate": 5.5465227056208326e-05,
      "loss": 0.0679,
      "step": 7017
    },
    {
      "epoch": 13.362815026153115,
      "grad_norm": 0.1802452802658081,
      "learning_rate": 5.5458875833597965e-05,
      "loss": 0.0647,
      "step": 7018
    },
    {
      "epoch": 13.364717070851166,
      "grad_norm": 0.22848699986934662,
      "learning_rate": 5.545252461098762e-05,
      "loss": 0.0521,
      "step": 7019
    },
    {
      "epoch": 13.366619115549215,
      "grad_norm": 0.16188445687294006,
      "learning_rate": 5.544617338837726e-05,
      "loss": 0.0473,
      "step": 7020
    },
    {
      "epoch": 13.368521160247266,
      "grad_norm": 0.2097487598657608,
      "learning_rate": 5.543982216576691e-05,
      "loss": 0.0594,
      "step": 7021
    },
    {
      "epoch": 13.370423204945316,
      "grad_norm": 0.24954724311828613,
      "learning_rate": 5.5433470943156565e-05,
      "loss": 0.0532,
      "step": 7022
    },
    {
      "epoch": 13.372325249643367,
      "grad_norm": 0.1943424791097641,
      "learning_rate": 5.5427119720546204e-05,
      "loss": 0.0583,
      "step": 7023
    },
    {
      "epoch": 13.374227294341416,
      "grad_norm": 0.21266767382621765,
      "learning_rate": 5.542076849793586e-05,
      "loss": 0.0446,
      "step": 7024
    },
    {
      "epoch": 13.376129339039467,
      "grad_norm": 0.24260656535625458,
      "learning_rate": 5.54144172753255e-05,
      "loss": 0.0572,
      "step": 7025
    },
    {
      "epoch": 13.378031383737518,
      "grad_norm": 0.16277165710926056,
      "learning_rate": 5.5408066052715146e-05,
      "loss": 0.0468,
      "step": 7026
    },
    {
      "epoch": 13.379933428435567,
      "grad_norm": 0.33338937163352966,
      "learning_rate": 5.54017148301048e-05,
      "loss": 0.0645,
      "step": 7027
    },
    {
      "epoch": 13.381835473133618,
      "grad_norm": 0.19156119227409363,
      "learning_rate": 5.539536360749444e-05,
      "loss": 0.0589,
      "step": 7028
    },
    {
      "epoch": 13.383737517831669,
      "grad_norm": 0.18192563951015472,
      "learning_rate": 5.53890123848841e-05,
      "loss": 0.0793,
      "step": 7029
    },
    {
      "epoch": 13.38563956252972,
      "grad_norm": 0.10455767810344696,
      "learning_rate": 5.538266116227374e-05,
      "loss": 0.0547,
      "step": 7030
    },
    {
      "epoch": 13.387541607227769,
      "grad_norm": 0.2381904125213623,
      "learning_rate": 5.5376309939663384e-05,
      "loss": 0.0556,
      "step": 7031
    },
    {
      "epoch": 13.38944365192582,
      "grad_norm": 0.1895589977502823,
      "learning_rate": 5.5369958717053036e-05,
      "loss": 0.0442,
      "step": 7032
    },
    {
      "epoch": 13.39134569662387,
      "grad_norm": 0.11963589489459991,
      "learning_rate": 5.536360749444268e-05,
      "loss": 0.0476,
      "step": 7033
    },
    {
      "epoch": 13.393247741321922,
      "grad_norm": 0.13365913927555084,
      "learning_rate": 5.5357256271832326e-05,
      "loss": 0.0399,
      "step": 7034
    },
    {
      "epoch": 13.39514978601997,
      "grad_norm": 0.2600623369216919,
      "learning_rate": 5.535090504922198e-05,
      "loss": 0.0673,
      "step": 7035
    },
    {
      "epoch": 13.397051830718022,
      "grad_norm": 0.20314230024814606,
      "learning_rate": 5.534455382661162e-05,
      "loss": 0.0698,
      "step": 7036
    },
    {
      "epoch": 13.398953875416073,
      "grad_norm": 0.29874828457832336,
      "learning_rate": 5.5338202604001275e-05,
      "loss": 0.075,
      "step": 7037
    },
    {
      "epoch": 13.400855920114124,
      "grad_norm": 0.18833285570144653,
      "learning_rate": 5.533185138139092e-05,
      "loss": 0.0596,
      "step": 7038
    },
    {
      "epoch": 13.402757964812173,
      "grad_norm": 0.1932767629623413,
      "learning_rate": 5.5325500158780565e-05,
      "loss": 0.0531,
      "step": 7039
    },
    {
      "epoch": 13.404660009510224,
      "grad_norm": 0.2641963064670563,
      "learning_rate": 5.531914893617022e-05,
      "loss": 0.0631,
      "step": 7040
    },
    {
      "epoch": 13.406562054208274,
      "grad_norm": 0.1732291430234909,
      "learning_rate": 5.531279771355986e-05,
      "loss": 0.0409,
      "step": 7041
    },
    {
      "epoch": 13.408464098906324,
      "grad_norm": 0.13920660316944122,
      "learning_rate": 5.5306446490949514e-05,
      "loss": 0.0634,
      "step": 7042
    },
    {
      "epoch": 13.410366143604374,
      "grad_norm": 0.20135769248008728,
      "learning_rate": 5.530009526833916e-05,
      "loss": 0.0524,
      "step": 7043
    },
    {
      "epoch": 13.412268188302425,
      "grad_norm": 0.2123275250196457,
      "learning_rate": 5.5293744045728804e-05,
      "loss": 0.0617,
      "step": 7044
    },
    {
      "epoch": 13.414170233000476,
      "grad_norm": 0.18639571964740753,
      "learning_rate": 5.5287392823118456e-05,
      "loss": 0.0567,
      "step": 7045
    },
    {
      "epoch": 13.416072277698525,
      "grad_norm": 0.1802852600812912,
      "learning_rate": 5.52810416005081e-05,
      "loss": 0.0557,
      "step": 7046
    },
    {
      "epoch": 13.417974322396576,
      "grad_norm": 0.2503332793712616,
      "learning_rate": 5.527469037789775e-05,
      "loss": 0.0819,
      "step": 7047
    },
    {
      "epoch": 13.419876367094627,
      "grad_norm": 0.1591678261756897,
      "learning_rate": 5.52683391552874e-05,
      "loss": 0.0587,
      "step": 7048
    },
    {
      "epoch": 13.421778411792678,
      "grad_norm": 0.10254660993814468,
      "learning_rate": 5.5261987932677036e-05,
      "loss": 0.0609,
      "step": 7049
    },
    {
      "epoch": 13.423680456490727,
      "grad_norm": 0.24069607257843018,
      "learning_rate": 5.5255636710066695e-05,
      "loss": 0.0776,
      "step": 7050
    },
    {
      "epoch": 13.425582501188778,
      "grad_norm": 0.13990217447280884,
      "learning_rate": 5.524928548745634e-05,
      "loss": 0.0618,
      "step": 7051
    },
    {
      "epoch": 13.427484545886829,
      "grad_norm": 0.15880687534809113,
      "learning_rate": 5.524293426484598e-05,
      "loss": 0.041,
      "step": 7052
    },
    {
      "epoch": 13.429386590584878,
      "grad_norm": 0.1640283614397049,
      "learning_rate": 5.523658304223564e-05,
      "loss": 0.0569,
      "step": 7053
    },
    {
      "epoch": 13.431288635282929,
      "grad_norm": 0.11456301063299179,
      "learning_rate": 5.5230231819625275e-05,
      "loss": 0.0604,
      "step": 7054
    },
    {
      "epoch": 13.43319067998098,
      "grad_norm": 0.18163856863975525,
      "learning_rate": 5.5223880597014934e-05,
      "loss": 0.0525,
      "step": 7055
    },
    {
      "epoch": 13.43509272467903,
      "grad_norm": 0.172135591506958,
      "learning_rate": 5.521752937440457e-05,
      "loss": 0.0732,
      "step": 7056
    },
    {
      "epoch": 13.43699476937708,
      "grad_norm": 0.14534039795398712,
      "learning_rate": 5.521117815179422e-05,
      "loss": 0.0527,
      "step": 7057
    },
    {
      "epoch": 13.43889681407513,
      "grad_norm": 0.10526961088180542,
      "learning_rate": 5.5204826929183876e-05,
      "loss": 0.0452,
      "step": 7058
    },
    {
      "epoch": 13.440798858773181,
      "grad_norm": 0.2756957411766052,
      "learning_rate": 5.5198475706573514e-05,
      "loss": 0.1274,
      "step": 7059
    },
    {
      "epoch": 13.442700903471232,
      "grad_norm": 0.21903552114963531,
      "learning_rate": 5.519212448396317e-05,
      "loss": 0.0579,
      "step": 7060
    },
    {
      "epoch": 13.444602948169281,
      "grad_norm": 0.16684231162071228,
      "learning_rate": 5.518577326135281e-05,
      "loss": 0.0581,
      "step": 7061
    },
    {
      "epoch": 13.446504992867332,
      "grad_norm": 0.14972932636737823,
      "learning_rate": 5.5179422038742456e-05,
      "loss": 0.0484,
      "step": 7062
    },
    {
      "epoch": 13.448407037565383,
      "grad_norm": 0.19577839970588684,
      "learning_rate": 5.517307081613211e-05,
      "loss": 0.0649,
      "step": 7063
    },
    {
      "epoch": 13.450309082263432,
      "grad_norm": 0.21453331410884857,
      "learning_rate": 5.516671959352175e-05,
      "loss": 0.0846,
      "step": 7064
    },
    {
      "epoch": 13.452211126961483,
      "grad_norm": 0.13317766785621643,
      "learning_rate": 5.516036837091141e-05,
      "loss": 0.0544,
      "step": 7065
    },
    {
      "epoch": 13.454113171659534,
      "grad_norm": 0.15833154320716858,
      "learning_rate": 5.515401714830105e-05,
      "loss": 0.0639,
      "step": 7066
    },
    {
      "epoch": 13.456015216357585,
      "grad_norm": 0.15371407568454742,
      "learning_rate": 5.5147665925690695e-05,
      "loss": 0.0665,
      "step": 7067
    },
    {
      "epoch": 13.457917261055634,
      "grad_norm": 0.15756915509700775,
      "learning_rate": 5.514131470308035e-05,
      "loss": 0.0502,
      "step": 7068
    },
    {
      "epoch": 13.459819305753685,
      "grad_norm": 0.15613050758838654,
      "learning_rate": 5.513496348046999e-05,
      "loss": 0.0805,
      "step": 7069
    },
    {
      "epoch": 13.461721350451736,
      "grad_norm": 0.2498931884765625,
      "learning_rate": 5.512861225785964e-05,
      "loss": 0.0624,
      "step": 7070
    },
    {
      "epoch": 13.463623395149787,
      "grad_norm": 0.1730266809463501,
      "learning_rate": 5.512226103524929e-05,
      "loss": 0.0419,
      "step": 7071
    },
    {
      "epoch": 13.465525439847836,
      "grad_norm": 0.15835736691951752,
      "learning_rate": 5.5115909812638934e-05,
      "loss": 0.0533,
      "step": 7072
    },
    {
      "epoch": 13.467427484545887,
      "grad_norm": 0.22773846983909607,
      "learning_rate": 5.5109558590028586e-05,
      "loss": 0.0575,
      "step": 7073
    },
    {
      "epoch": 13.469329529243938,
      "grad_norm": 0.2337859719991684,
      "learning_rate": 5.510320736741823e-05,
      "loss": 0.0611,
      "step": 7074
    },
    {
      "epoch": 13.471231573941987,
      "grad_norm": 0.31625816226005554,
      "learning_rate": 5.5096856144807876e-05,
      "loss": 0.0967,
      "step": 7075
    },
    {
      "epoch": 13.473133618640038,
      "grad_norm": 0.18602289259433746,
      "learning_rate": 5.509050492219753e-05,
      "loss": 0.055,
      "step": 7076
    },
    {
      "epoch": 13.475035663338089,
      "grad_norm": 0.16424252092838287,
      "learning_rate": 5.508415369958717e-05,
      "loss": 0.0407,
      "step": 7077
    },
    {
      "epoch": 13.47693770803614,
      "grad_norm": 0.1765555590391159,
      "learning_rate": 5.5077802476976824e-05,
      "loss": 0.0674,
      "step": 7078
    },
    {
      "epoch": 13.478839752734189,
      "grad_norm": 0.26882824301719666,
      "learning_rate": 5.507145125436647e-05,
      "loss": 0.0779,
      "step": 7079
    },
    {
      "epoch": 13.48074179743224,
      "grad_norm": 0.320591002702713,
      "learning_rate": 5.5065100031756115e-05,
      "loss": 0.1318,
      "step": 7080
    },
    {
      "epoch": 13.48264384213029,
      "grad_norm": 0.1425531655550003,
      "learning_rate": 5.5058748809145766e-05,
      "loss": 0.0624,
      "step": 7081
    },
    {
      "epoch": 13.484545886828341,
      "grad_norm": 0.2594924569129944,
      "learning_rate": 5.505239758653541e-05,
      "loss": 0.0806,
      "step": 7082
    },
    {
      "epoch": 13.48644793152639,
      "grad_norm": 0.18772482872009277,
      "learning_rate": 5.504604636392506e-05,
      "loss": 0.0777,
      "step": 7083
    },
    {
      "epoch": 13.488349976224441,
      "grad_norm": 0.29713594913482666,
      "learning_rate": 5.503969514131471e-05,
      "loss": 0.102,
      "step": 7084
    },
    {
      "epoch": 13.490252020922492,
      "grad_norm": 0.21502874791622162,
      "learning_rate": 5.503334391870435e-05,
      "loss": 0.0693,
      "step": 7085
    },
    {
      "epoch": 13.492154065620543,
      "grad_norm": 0.162057027220726,
      "learning_rate": 5.5026992696094005e-05,
      "loss": 0.0544,
      "step": 7086
    },
    {
      "epoch": 13.494056110318592,
      "grad_norm": 0.22181911766529083,
      "learning_rate": 5.502064147348365e-05,
      "loss": 0.0667,
      "step": 7087
    },
    {
      "epoch": 13.495958155016643,
      "grad_norm": 0.15365485846996307,
      "learning_rate": 5.501429025087329e-05,
      "loss": 0.0605,
      "step": 7088
    },
    {
      "epoch": 13.497860199714694,
      "grad_norm": 0.10755611211061478,
      "learning_rate": 5.500793902826295e-05,
      "loss": 0.0389,
      "step": 7089
    },
    {
      "epoch": 13.499762244412743,
      "grad_norm": 0.2021094262599945,
      "learning_rate": 5.5001587805652586e-05,
      "loss": 0.06,
      "step": 7090
    },
    {
      "epoch": 13.501664289110794,
      "grad_norm": 0.19792579114437103,
      "learning_rate": 5.4995236583042244e-05,
      "loss": 0.0547,
      "step": 7091
    },
    {
      "epoch": 13.503566333808845,
      "grad_norm": 0.20431144535541534,
      "learning_rate": 5.498888536043188e-05,
      "loss": 0.0752,
      "step": 7092
    },
    {
      "epoch": 13.505468378506896,
      "grad_norm": 0.22182844579219818,
      "learning_rate": 5.498253413782153e-05,
      "loss": 0.0724,
      "step": 7093
    },
    {
      "epoch": 13.507370423204945,
      "grad_norm": 0.17189350724220276,
      "learning_rate": 5.4976182915211186e-05,
      "loss": 0.0493,
      "step": 7094
    },
    {
      "epoch": 13.509272467902996,
      "grad_norm": 0.27690860629081726,
      "learning_rate": 5.4969831692600824e-05,
      "loss": 0.0584,
      "step": 7095
    },
    {
      "epoch": 13.511174512601047,
      "grad_norm": 0.15486986935138702,
      "learning_rate": 5.496348046999048e-05,
      "loss": 0.0559,
      "step": 7096
    },
    {
      "epoch": 13.513076557299097,
      "grad_norm": 0.2064022719860077,
      "learning_rate": 5.495712924738012e-05,
      "loss": 0.0689,
      "step": 7097
    },
    {
      "epoch": 13.514978601997147,
      "grad_norm": 0.22928853332996368,
      "learning_rate": 5.4950778024769766e-05,
      "loss": 0.0589,
      "step": 7098
    },
    {
      "epoch": 13.516880646695197,
      "grad_norm": 0.20014768838882446,
      "learning_rate": 5.494442680215942e-05,
      "loss": 0.0652,
      "step": 7099
    },
    {
      "epoch": 13.518782691393248,
      "grad_norm": 0.17018628120422363,
      "learning_rate": 5.493807557954906e-05,
      "loss": 0.0513,
      "step": 7100
    },
    {
      "epoch": 13.520684736091297,
      "grad_norm": 0.23920871317386627,
      "learning_rate": 5.4931724356938715e-05,
      "loss": 0.0442,
      "step": 7101
    },
    {
      "epoch": 13.522586780789348,
      "grad_norm": 0.21291278302669525,
      "learning_rate": 5.492537313432836e-05,
      "loss": 0.0662,
      "step": 7102
    },
    {
      "epoch": 13.5244888254874,
      "grad_norm": 0.17782357335090637,
      "learning_rate": 5.4919021911718005e-05,
      "loss": 0.0644,
      "step": 7103
    },
    {
      "epoch": 13.52639087018545,
      "grad_norm": 0.26954326033592224,
      "learning_rate": 5.491267068910766e-05,
      "loss": 0.0803,
      "step": 7104
    },
    {
      "epoch": 13.5282929148835,
      "grad_norm": 0.14195388555526733,
      "learning_rate": 5.49063194664973e-05,
      "loss": 0.0536,
      "step": 7105
    },
    {
      "epoch": 13.53019495958155,
      "grad_norm": 0.22725562751293182,
      "learning_rate": 5.489996824388695e-05,
      "loss": 0.0708,
      "step": 7106
    },
    {
      "epoch": 13.532097004279601,
      "grad_norm": 0.2652048170566559,
      "learning_rate": 5.48936170212766e-05,
      "loss": 0.0842,
      "step": 7107
    },
    {
      "epoch": 13.533999048977652,
      "grad_norm": 0.1272173821926117,
      "learning_rate": 5.4887265798666244e-05,
      "loss": 0.0599,
      "step": 7108
    },
    {
      "epoch": 13.535901093675701,
      "grad_norm": 0.15636959671974182,
      "learning_rate": 5.4880914576055896e-05,
      "loss": 0.0456,
      "step": 7109
    },
    {
      "epoch": 13.537803138373752,
      "grad_norm": 0.22358985245227814,
      "learning_rate": 5.487456335344554e-05,
      "loss": 0.0626,
      "step": 7110
    },
    {
      "epoch": 13.539705183071803,
      "grad_norm": 0.2860788404941559,
      "learning_rate": 5.4868212130835186e-05,
      "loss": 0.0769,
      "step": 7111
    },
    {
      "epoch": 13.541607227769852,
      "grad_norm": 0.25277575850486755,
      "learning_rate": 5.486186090822484e-05,
      "loss": 0.0626,
      "step": 7112
    },
    {
      "epoch": 13.543509272467903,
      "grad_norm": 0.16722799837589264,
      "learning_rate": 5.485550968561448e-05,
      "loss": 0.0617,
      "step": 7113
    },
    {
      "epoch": 13.545411317165954,
      "grad_norm": 0.22804458439350128,
      "learning_rate": 5.4849158463004135e-05,
      "loss": 0.0553,
      "step": 7114
    },
    {
      "epoch": 13.547313361864004,
      "grad_norm": 0.18392793834209442,
      "learning_rate": 5.484280724039378e-05,
      "loss": 0.0818,
      "step": 7115
    },
    {
      "epoch": 13.549215406562054,
      "grad_norm": 0.13310854136943817,
      "learning_rate": 5.4836456017783425e-05,
      "loss": 0.0487,
      "step": 7116
    },
    {
      "epoch": 13.551117451260104,
      "grad_norm": 0.1897313892841339,
      "learning_rate": 5.483010479517308e-05,
      "loss": 0.0733,
      "step": 7117
    },
    {
      "epoch": 13.553019495958155,
      "grad_norm": 0.15019789338111877,
      "learning_rate": 5.482375357256272e-05,
      "loss": 0.0605,
      "step": 7118
    },
    {
      "epoch": 13.554921540656206,
      "grad_norm": 0.17896315455436707,
      "learning_rate": 5.4817402349952374e-05,
      "loss": 0.0587,
      "step": 7119
    },
    {
      "epoch": 13.556823585354255,
      "grad_norm": 0.16350199282169342,
      "learning_rate": 5.481105112734202e-05,
      "loss": 0.0338,
      "step": 7120
    },
    {
      "epoch": 13.558725630052306,
      "grad_norm": 0.28143972158432007,
      "learning_rate": 5.480469990473166e-05,
      "loss": 0.0888,
      "step": 7121
    },
    {
      "epoch": 13.560627674750357,
      "grad_norm": 0.18174341320991516,
      "learning_rate": 5.4798348682121316e-05,
      "loss": 0.067,
      "step": 7122
    },
    {
      "epoch": 13.562529719448406,
      "grad_norm": 0.12471450120210648,
      "learning_rate": 5.479199745951096e-05,
      "loss": 0.0483,
      "step": 7123
    },
    {
      "epoch": 13.564431764146457,
      "grad_norm": 0.20541246235370636,
      "learning_rate": 5.47856462369006e-05,
      "loss": 0.0655,
      "step": 7124
    },
    {
      "epoch": 13.566333808844508,
      "grad_norm": 0.16435562074184418,
      "learning_rate": 5.477929501429026e-05,
      "loss": 0.0533,
      "step": 7125
    },
    {
      "epoch": 13.568235853542559,
      "grad_norm": 0.14860351383686066,
      "learning_rate": 5.4772943791679896e-05,
      "loss": 0.053,
      "step": 7126
    },
    {
      "epoch": 13.570137898240608,
      "grad_norm": 0.1912887692451477,
      "learning_rate": 5.4766592569069555e-05,
      "loss": 0.0588,
      "step": 7127
    },
    {
      "epoch": 13.572039942938659,
      "grad_norm": 0.17781256139278412,
      "learning_rate": 5.476024134645919e-05,
      "loss": 0.0578,
      "step": 7128
    },
    {
      "epoch": 13.57394198763671,
      "grad_norm": 0.18353131413459778,
      "learning_rate": 5.475389012384884e-05,
      "loss": 0.0493,
      "step": 7129
    },
    {
      "epoch": 13.57584403233476,
      "grad_norm": 0.1685384213924408,
      "learning_rate": 5.474753890123849e-05,
      "loss": 0.0658,
      "step": 7130
    },
    {
      "epoch": 13.57774607703281,
      "grad_norm": 0.18952107429504395,
      "learning_rate": 5.4741187678628135e-05,
      "loss": 0.0557,
      "step": 7131
    },
    {
      "epoch": 13.57964812173086,
      "grad_norm": 0.1624719798564911,
      "learning_rate": 5.4734836456017794e-05,
      "loss": 0.0479,
      "step": 7132
    },
    {
      "epoch": 13.581550166428912,
      "grad_norm": 0.14944607019424438,
      "learning_rate": 5.472848523340743e-05,
      "loss": 0.0606,
      "step": 7133
    },
    {
      "epoch": 13.58345221112696,
      "grad_norm": 0.27262911200523376,
      "learning_rate": 5.472213401079708e-05,
      "loss": 0.0751,
      "step": 7134
    },
    {
      "epoch": 13.585354255825012,
      "grad_norm": 0.26825249195098877,
      "learning_rate": 5.471578278818673e-05,
      "loss": 0.0594,
      "step": 7135
    },
    {
      "epoch": 13.587256300523062,
      "grad_norm": 0.20710812509059906,
      "learning_rate": 5.4709431565576374e-05,
      "loss": 0.0744,
      "step": 7136
    },
    {
      "epoch": 13.589158345221113,
      "grad_norm": 0.27290740609169006,
      "learning_rate": 5.4703080342966026e-05,
      "loss": 0.0731,
      "step": 7137
    },
    {
      "epoch": 13.591060389919162,
      "grad_norm": 0.3027440905570984,
      "learning_rate": 5.469672912035567e-05,
      "loss": 0.0826,
      "step": 7138
    },
    {
      "epoch": 13.592962434617213,
      "grad_norm": 0.13028496503829956,
      "learning_rate": 5.4690377897745316e-05,
      "loss": 0.0417,
      "step": 7139
    },
    {
      "epoch": 13.594864479315264,
      "grad_norm": 0.22668135166168213,
      "learning_rate": 5.468402667513497e-05,
      "loss": 0.0577,
      "step": 7140
    },
    {
      "epoch": 13.596766524013315,
      "grad_norm": 0.20752792060375214,
      "learning_rate": 5.467767545252461e-05,
      "loss": 0.0867,
      "step": 7141
    },
    {
      "epoch": 13.598668568711364,
      "grad_norm": 0.18179014325141907,
      "learning_rate": 5.467132422991426e-05,
      "loss": 0.0604,
      "step": 7142
    },
    {
      "epoch": 13.600570613409415,
      "grad_norm": 0.2199116200208664,
      "learning_rate": 5.466497300730391e-05,
      "loss": 0.0552,
      "step": 7143
    },
    {
      "epoch": 13.602472658107466,
      "grad_norm": 0.36437416076660156,
      "learning_rate": 5.4658621784693555e-05,
      "loss": 0.0819,
      "step": 7144
    },
    {
      "epoch": 13.604374702805515,
      "grad_norm": 0.17574447393417358,
      "learning_rate": 5.4652270562083207e-05,
      "loss": 0.0728,
      "step": 7145
    },
    {
      "epoch": 13.606276747503566,
      "grad_norm": 0.13370124995708466,
      "learning_rate": 5.464591933947285e-05,
      "loss": 0.065,
      "step": 7146
    },
    {
      "epoch": 13.608178792201617,
      "grad_norm": 0.18489646911621094,
      "learning_rate": 5.46395681168625e-05,
      "loss": 0.0566,
      "step": 7147
    },
    {
      "epoch": 13.610080836899668,
      "grad_norm": 0.08928623795509338,
      "learning_rate": 5.463321689425215e-05,
      "loss": 0.0459,
      "step": 7148
    },
    {
      "epoch": 13.611982881597717,
      "grad_norm": 0.11430780589580536,
      "learning_rate": 5.4626865671641794e-05,
      "loss": 0.0611,
      "step": 7149
    },
    {
      "epoch": 13.613884926295768,
      "grad_norm": 0.2574712038040161,
      "learning_rate": 5.4620514449031445e-05,
      "loss": 0.0737,
      "step": 7150
    },
    {
      "epoch": 13.615786970993819,
      "grad_norm": 0.2580837607383728,
      "learning_rate": 5.461416322642109e-05,
      "loss": 0.063,
      "step": 7151
    },
    {
      "epoch": 13.61768901569187,
      "grad_norm": 0.2556533217430115,
      "learning_rate": 5.4607812003810736e-05,
      "loss": 0.0767,
      "step": 7152
    },
    {
      "epoch": 13.619591060389919,
      "grad_norm": 0.2444048970937729,
      "learning_rate": 5.460146078120039e-05,
      "loss": 0.0945,
      "step": 7153
    },
    {
      "epoch": 13.62149310508797,
      "grad_norm": 0.2365688532590866,
      "learning_rate": 5.459510955859003e-05,
      "loss": 0.0624,
      "step": 7154
    },
    {
      "epoch": 13.62339514978602,
      "grad_norm": 0.24796128273010254,
      "learning_rate": 5.4588758335979684e-05,
      "loss": 0.0612,
      "step": 7155
    },
    {
      "epoch": 13.62529719448407,
      "grad_norm": 0.32208365201950073,
      "learning_rate": 5.458240711336933e-05,
      "loss": 0.0815,
      "step": 7156
    },
    {
      "epoch": 13.62719923918212,
      "grad_norm": 0.2308531254529953,
      "learning_rate": 5.457605589075897e-05,
      "loss": 0.0671,
      "step": 7157
    },
    {
      "epoch": 13.629101283880171,
      "grad_norm": 0.13749493658542633,
      "learning_rate": 5.4569704668148626e-05,
      "loss": 0.0526,
      "step": 7158
    },
    {
      "epoch": 13.631003328578222,
      "grad_norm": 0.24286620318889618,
      "learning_rate": 5.4563353445538265e-05,
      "loss": 0.0526,
      "step": 7159
    },
    {
      "epoch": 13.632905373276271,
      "grad_norm": 0.17536072432994843,
      "learning_rate": 5.455700222292791e-05,
      "loss": 0.049,
      "step": 7160
    },
    {
      "epoch": 13.634807417974322,
      "grad_norm": 0.1444406658411026,
      "learning_rate": 5.455065100031757e-05,
      "loss": 0.051,
      "step": 7161
    },
    {
      "epoch": 13.636709462672373,
      "grad_norm": 0.2647935450077057,
      "learning_rate": 5.4544299777707206e-05,
      "loss": 0.0689,
      "step": 7162
    },
    {
      "epoch": 13.638611507370424,
      "grad_norm": 0.14098279178142548,
      "learning_rate": 5.4537948555096865e-05,
      "loss": 0.0579,
      "step": 7163
    },
    {
      "epoch": 13.640513552068473,
      "grad_norm": 0.18277288973331451,
      "learning_rate": 5.4531597332486503e-05,
      "loss": 0.096,
      "step": 7164
    },
    {
      "epoch": 13.642415596766524,
      "grad_norm": 0.17182549834251404,
      "learning_rate": 5.452524610987615e-05,
      "loss": 0.0591,
      "step": 7165
    },
    {
      "epoch": 13.644317641464575,
      "grad_norm": 0.1310691237449646,
      "learning_rate": 5.45188948872658e-05,
      "loss": 0.0724,
      "step": 7166
    },
    {
      "epoch": 13.646219686162624,
      "grad_norm": 0.16267623007297516,
      "learning_rate": 5.4512543664655445e-05,
      "loss": 0.078,
      "step": 7167
    },
    {
      "epoch": 13.648121730860675,
      "grad_norm": 0.15424087643623352,
      "learning_rate": 5.4506192442045104e-05,
      "loss": 0.0704,
      "step": 7168
    },
    {
      "epoch": 13.650023775558726,
      "grad_norm": 0.13465258479118347,
      "learning_rate": 5.449984121943474e-05,
      "loss": 0.0765,
      "step": 7169
    },
    {
      "epoch": 13.651925820256777,
      "grad_norm": 0.19921620190143585,
      "learning_rate": 5.449348999682439e-05,
      "loss": 0.0626,
      "step": 7170
    },
    {
      "epoch": 13.653827864954826,
      "grad_norm": 0.1628495454788208,
      "learning_rate": 5.448713877421404e-05,
      "loss": 0.0812,
      "step": 7171
    },
    {
      "epoch": 13.655729909652877,
      "grad_norm": 0.3022763729095459,
      "learning_rate": 5.4480787551603684e-05,
      "loss": 0.0574,
      "step": 7172
    },
    {
      "epoch": 13.657631954350927,
      "grad_norm": 0.14462687075138092,
      "learning_rate": 5.4474436328993336e-05,
      "loss": 0.0693,
      "step": 7173
    },
    {
      "epoch": 13.659533999048978,
      "grad_norm": 0.24949683248996735,
      "learning_rate": 5.446808510638298e-05,
      "loss": 0.0732,
      "step": 7174
    },
    {
      "epoch": 13.661436043747027,
      "grad_norm": 0.11945445835590363,
      "learning_rate": 5.4461733883772626e-05,
      "loss": 0.0692,
      "step": 7175
    },
    {
      "epoch": 13.663338088445078,
      "grad_norm": 0.22510592639446259,
      "learning_rate": 5.445538266116228e-05,
      "loss": 0.0531,
      "step": 7176
    },
    {
      "epoch": 13.66524013314313,
      "grad_norm": 0.1811244785785675,
      "learning_rate": 5.444903143855192e-05,
      "loss": 0.0602,
      "step": 7177
    },
    {
      "epoch": 13.667142177841178,
      "grad_norm": 0.299470454454422,
      "learning_rate": 5.444268021594157e-05,
      "loss": 0.056,
      "step": 7178
    },
    {
      "epoch": 13.66904422253923,
      "grad_norm": 0.15481477975845337,
      "learning_rate": 5.443632899333122e-05,
      "loss": 0.046,
      "step": 7179
    },
    {
      "epoch": 13.67094626723728,
      "grad_norm": 0.15388630330562592,
      "learning_rate": 5.4429977770720865e-05,
      "loss": 0.0484,
      "step": 7180
    },
    {
      "epoch": 13.672848311935331,
      "grad_norm": 0.25386807322502136,
      "learning_rate": 5.442362654811052e-05,
      "loss": 0.0654,
      "step": 7181
    },
    {
      "epoch": 13.67475035663338,
      "grad_norm": 0.18215690553188324,
      "learning_rate": 5.441727532550016e-05,
      "loss": 0.0591,
      "step": 7182
    },
    {
      "epoch": 13.676652401331431,
      "grad_norm": 0.1830330491065979,
      "learning_rate": 5.441092410288981e-05,
      "loss": 0.0577,
      "step": 7183
    },
    {
      "epoch": 13.678554446029482,
      "grad_norm": 0.21742892265319824,
      "learning_rate": 5.440457288027946e-05,
      "loss": 0.0697,
      "step": 7184
    },
    {
      "epoch": 13.680456490727533,
      "grad_norm": 0.34482452273368835,
      "learning_rate": 5.4398221657669104e-05,
      "loss": 0.0628,
      "step": 7185
    },
    {
      "epoch": 13.682358535425582,
      "grad_norm": 0.1803099662065506,
      "learning_rate": 5.4391870435058756e-05,
      "loss": 0.0586,
      "step": 7186
    },
    {
      "epoch": 13.684260580123633,
      "grad_norm": 0.2071414440870285,
      "learning_rate": 5.43855192124484e-05,
      "loss": 0.0537,
      "step": 7187
    },
    {
      "epoch": 13.686162624821684,
      "grad_norm": 0.14682170748710632,
      "learning_rate": 5.437916798983804e-05,
      "loss": 0.0621,
      "step": 7188
    },
    {
      "epoch": 13.688064669519733,
      "grad_norm": 0.19556795060634613,
      "learning_rate": 5.43728167672277e-05,
      "loss": 0.0524,
      "step": 7189
    },
    {
      "epoch": 13.689966714217784,
      "grad_norm": 0.2657491862773895,
      "learning_rate": 5.436646554461734e-05,
      "loss": 0.0598,
      "step": 7190
    },
    {
      "epoch": 13.691868758915835,
      "grad_norm": 0.28120002150535583,
      "learning_rate": 5.4360114322006995e-05,
      "loss": 0.0694,
      "step": 7191
    },
    {
      "epoch": 13.693770803613885,
      "grad_norm": 0.1710984855890274,
      "learning_rate": 5.435376309939664e-05,
      "loss": 0.0557,
      "step": 7192
    },
    {
      "epoch": 13.695672848311935,
      "grad_norm": 0.1776714026927948,
      "learning_rate": 5.434741187678628e-05,
      "loss": 0.0666,
      "step": 7193
    },
    {
      "epoch": 13.697574893009985,
      "grad_norm": 0.1420143097639084,
      "learning_rate": 5.434106065417594e-05,
      "loss": 0.0581,
      "step": 7194
    },
    {
      "epoch": 13.699476937708036,
      "grad_norm": 0.2275141328573227,
      "learning_rate": 5.4334709431565575e-05,
      "loss": 0.0704,
      "step": 7195
    },
    {
      "epoch": 13.701378982406087,
      "grad_norm": 0.20450417697429657,
      "learning_rate": 5.432835820895522e-05,
      "loss": 0.0861,
      "step": 7196
    },
    {
      "epoch": 13.703281027104136,
      "grad_norm": 0.2755592167377472,
      "learning_rate": 5.432200698634488e-05,
      "loss": 0.0699,
      "step": 7197
    },
    {
      "epoch": 13.705183071802187,
      "grad_norm": 0.1973126381635666,
      "learning_rate": 5.431565576373452e-05,
      "loss": 0.0798,
      "step": 7198
    },
    {
      "epoch": 13.707085116500238,
      "grad_norm": 0.3487046957015991,
      "learning_rate": 5.4309304541124176e-05,
      "loss": 0.0843,
      "step": 7199
    },
    {
      "epoch": 13.708987161198289,
      "grad_norm": 0.2203672230243683,
      "learning_rate": 5.4302953318513814e-05,
      "loss": 0.0599,
      "step": 7200
    },
    {
      "epoch": 13.710889205896338,
      "grad_norm": 0.29556646943092346,
      "learning_rate": 5.429660209590346e-05,
      "loss": 0.076,
      "step": 7201
    },
    {
      "epoch": 13.712791250594389,
      "grad_norm": 0.16539618372917175,
      "learning_rate": 5.429025087329311e-05,
      "loss": 0.0591,
      "step": 7202
    },
    {
      "epoch": 13.71469329529244,
      "grad_norm": 0.2217624932527542,
      "learning_rate": 5.4283899650682756e-05,
      "loss": 0.0483,
      "step": 7203
    },
    {
      "epoch": 13.716595339990489,
      "grad_norm": 0.25081562995910645,
      "learning_rate": 5.4277548428072414e-05,
      "loss": 0.0707,
      "step": 7204
    },
    {
      "epoch": 13.71849738468854,
      "grad_norm": 0.17247477173805237,
      "learning_rate": 5.427119720546205e-05,
      "loss": 0.0644,
      "step": 7205
    },
    {
      "epoch": 13.72039942938659,
      "grad_norm": 0.11861587315797806,
      "learning_rate": 5.42648459828517e-05,
      "loss": 0.0671,
      "step": 7206
    },
    {
      "epoch": 13.722301474084642,
      "grad_norm": 0.20958885550498962,
      "learning_rate": 5.425849476024135e-05,
      "loss": 0.0687,
      "step": 7207
    },
    {
      "epoch": 13.72420351878269,
      "grad_norm": 0.20735900104045868,
      "learning_rate": 5.4252143537630995e-05,
      "loss": 0.0612,
      "step": 7208
    },
    {
      "epoch": 13.726105563480742,
      "grad_norm": 0.16221395134925842,
      "learning_rate": 5.4245792315020647e-05,
      "loss": 0.0706,
      "step": 7209
    },
    {
      "epoch": 13.728007608178793,
      "grad_norm": 0.24386169016361237,
      "learning_rate": 5.423944109241029e-05,
      "loss": 0.0941,
      "step": 7210
    },
    {
      "epoch": 13.729909652876843,
      "grad_norm": 0.12547758221626282,
      "learning_rate": 5.423308986979994e-05,
      "loss": 0.0663,
      "step": 7211
    },
    {
      "epoch": 13.731811697574893,
      "grad_norm": 0.21748408675193787,
      "learning_rate": 5.422673864718959e-05,
      "loss": 0.0669,
      "step": 7212
    },
    {
      "epoch": 13.733713742272943,
      "grad_norm": 0.3474538028240204,
      "learning_rate": 5.4220387424579234e-05,
      "loss": 0.0571,
      "step": 7213
    },
    {
      "epoch": 13.735615786970994,
      "grad_norm": 0.20235182344913483,
      "learning_rate": 5.421403620196888e-05,
      "loss": 0.0567,
      "step": 7214
    },
    {
      "epoch": 13.737517831669045,
      "grad_norm": 0.19568586349487305,
      "learning_rate": 5.420768497935853e-05,
      "loss": 0.0629,
      "step": 7215
    },
    {
      "epoch": 13.739419876367094,
      "grad_norm": 0.238690584897995,
      "learning_rate": 5.4201333756748176e-05,
      "loss": 0.0619,
      "step": 7216
    },
    {
      "epoch": 13.741321921065145,
      "grad_norm": 0.0995822325348854,
      "learning_rate": 5.419498253413783e-05,
      "loss": 0.0699,
      "step": 7217
    },
    {
      "epoch": 13.743223965763196,
      "grad_norm": 0.18771526217460632,
      "learning_rate": 5.418863131152747e-05,
      "loss": 0.0622,
      "step": 7218
    },
    {
      "epoch": 13.745126010461245,
      "grad_norm": 0.2403172105550766,
      "learning_rate": 5.418228008891712e-05,
      "loss": 0.061,
      "step": 7219
    },
    {
      "epoch": 13.747028055159296,
      "grad_norm": 0.3270457983016968,
      "learning_rate": 5.417592886630677e-05,
      "loss": 0.0787,
      "step": 7220
    },
    {
      "epoch": 13.748930099857347,
      "grad_norm": 0.2265048325061798,
      "learning_rate": 5.4169577643696414e-05,
      "loss": 0.06,
      "step": 7221
    },
    {
      "epoch": 13.750832144555398,
      "grad_norm": 0.18946008384227753,
      "learning_rate": 5.4163226421086066e-05,
      "loss": 0.0538,
      "step": 7222
    },
    {
      "epoch": 13.752734189253447,
      "grad_norm": 0.22583557665348053,
      "learning_rate": 5.415687519847571e-05,
      "loss": 0.0624,
      "step": 7223
    },
    {
      "epoch": 13.754636233951498,
      "grad_norm": 0.2988790273666382,
      "learning_rate": 5.415052397586535e-05,
      "loss": 0.0755,
      "step": 7224
    },
    {
      "epoch": 13.756538278649549,
      "grad_norm": 0.24329178035259247,
      "learning_rate": 5.414417275325501e-05,
      "loss": 0.049,
      "step": 7225
    },
    {
      "epoch": 13.7584403233476,
      "grad_norm": 0.13913269340991974,
      "learning_rate": 5.413782153064465e-05,
      "loss": 0.0712,
      "step": 7226
    },
    {
      "epoch": 13.760342368045649,
      "grad_norm": 0.23898306488990784,
      "learning_rate": 5.4131470308034305e-05,
      "loss": 0.0507,
      "step": 7227
    },
    {
      "epoch": 13.7622444127437,
      "grad_norm": 0.14138242602348328,
      "learning_rate": 5.412511908542395e-05,
      "loss": 0.0596,
      "step": 7228
    },
    {
      "epoch": 13.76414645744175,
      "grad_norm": 0.20261605083942413,
      "learning_rate": 5.411876786281359e-05,
      "loss": 0.0539,
      "step": 7229
    },
    {
      "epoch": 13.7660485021398,
      "grad_norm": 0.16496528685092926,
      "learning_rate": 5.411241664020325e-05,
      "loss": 0.0649,
      "step": 7230
    },
    {
      "epoch": 13.76795054683785,
      "grad_norm": 0.3304959535598755,
      "learning_rate": 5.4106065417592885e-05,
      "loss": 0.0644,
      "step": 7231
    },
    {
      "epoch": 13.769852591535901,
      "grad_norm": 0.2712630033493042,
      "learning_rate": 5.409971419498253e-05,
      "loss": 0.0699,
      "step": 7232
    },
    {
      "epoch": 13.771754636233952,
      "grad_norm": 0.21757693588733673,
      "learning_rate": 5.409336297237219e-05,
      "loss": 0.0727,
      "step": 7233
    },
    {
      "epoch": 13.773656680932001,
      "grad_norm": 0.1744980663061142,
      "learning_rate": 5.408701174976183e-05,
      "loss": 0.037,
      "step": 7234
    },
    {
      "epoch": 13.775558725630052,
      "grad_norm": 0.23627837002277374,
      "learning_rate": 5.4080660527151486e-05,
      "loss": 0.0585,
      "step": 7235
    },
    {
      "epoch": 13.777460770328103,
      "grad_norm": 0.19850653409957886,
      "learning_rate": 5.4074309304541124e-05,
      "loss": 0.0629,
      "step": 7236
    },
    {
      "epoch": 13.779362815026154,
      "grad_norm": 0.18019446730613708,
      "learning_rate": 5.406795808193077e-05,
      "loss": 0.0442,
      "step": 7237
    },
    {
      "epoch": 13.781264859724203,
      "grad_norm": 0.26067209243774414,
      "learning_rate": 5.406160685932042e-05,
      "loss": 0.064,
      "step": 7238
    },
    {
      "epoch": 13.783166904422254,
      "grad_norm": 0.27970629930496216,
      "learning_rate": 5.4055255636710066e-05,
      "loss": 0.0782,
      "step": 7239
    },
    {
      "epoch": 13.785068949120305,
      "grad_norm": 0.2066178172826767,
      "learning_rate": 5.404890441409972e-05,
      "loss": 0.0663,
      "step": 7240
    },
    {
      "epoch": 13.786970993818354,
      "grad_norm": 0.18839718401432037,
      "learning_rate": 5.404255319148936e-05,
      "loss": 0.0667,
      "step": 7241
    },
    {
      "epoch": 13.788873038516405,
      "grad_norm": 0.274984747171402,
      "learning_rate": 5.403620196887901e-05,
      "loss": 0.0826,
      "step": 7242
    },
    {
      "epoch": 13.790775083214456,
      "grad_norm": 0.1409456431865692,
      "learning_rate": 5.402985074626866e-05,
      "loss": 0.0455,
      "step": 7243
    },
    {
      "epoch": 13.792677127912507,
      "grad_norm": 0.2696741819381714,
      "learning_rate": 5.4023499523658305e-05,
      "loss": 0.063,
      "step": 7244
    },
    {
      "epoch": 13.794579172610556,
      "grad_norm": 0.21434366703033447,
      "learning_rate": 5.401714830104796e-05,
      "loss": 0.0543,
      "step": 7245
    },
    {
      "epoch": 13.796481217308607,
      "grad_norm": 0.20650018751621246,
      "learning_rate": 5.40107970784376e-05,
      "loss": 0.0513,
      "step": 7246
    },
    {
      "epoch": 13.798383262006658,
      "grad_norm": 0.18018142879009247,
      "learning_rate": 5.400444585582725e-05,
      "loss": 0.0466,
      "step": 7247
    },
    {
      "epoch": 13.800285306704708,
      "grad_norm": 0.14728957414627075,
      "learning_rate": 5.39980946332169e-05,
      "loss": 0.0455,
      "step": 7248
    },
    {
      "epoch": 13.802187351402758,
      "grad_norm": 0.18369527161121368,
      "learning_rate": 5.3991743410606544e-05,
      "loss": 0.049,
      "step": 7249
    },
    {
      "epoch": 13.804089396100808,
      "grad_norm": 0.15343110263347626,
      "learning_rate": 5.398539218799619e-05,
      "loss": 0.044,
      "step": 7250
    },
    {
      "epoch": 13.80599144079886,
      "grad_norm": 0.19930998980998993,
      "learning_rate": 5.397904096538584e-05,
      "loss": 0.0623,
      "step": 7251
    },
    {
      "epoch": 13.807893485496908,
      "grad_norm": 0.15696053206920624,
      "learning_rate": 5.3972689742775486e-05,
      "loss": 0.0719,
      "step": 7252
    },
    {
      "epoch": 13.80979553019496,
      "grad_norm": 0.19954028725624084,
      "learning_rate": 5.396633852016514e-05,
      "loss": 0.0411,
      "step": 7253
    },
    {
      "epoch": 13.81169757489301,
      "grad_norm": 0.3664581775665283,
      "learning_rate": 5.395998729755478e-05,
      "loss": 0.1245,
      "step": 7254
    },
    {
      "epoch": 13.813599619591061,
      "grad_norm": 0.2724677622318268,
      "learning_rate": 5.395363607494443e-05,
      "loss": 0.0596,
      "step": 7255
    },
    {
      "epoch": 13.81550166428911,
      "grad_norm": 0.2526918053627014,
      "learning_rate": 5.394728485233408e-05,
      "loss": 0.0606,
      "step": 7256
    },
    {
      "epoch": 13.817403708987161,
      "grad_norm": 0.2746874988079071,
      "learning_rate": 5.3940933629723725e-05,
      "loss": 0.1144,
      "step": 7257
    },
    {
      "epoch": 13.819305753685212,
      "grad_norm": 0.2281782627105713,
      "learning_rate": 5.393458240711338e-05,
      "loss": 0.0913,
      "step": 7258
    },
    {
      "epoch": 13.821207798383263,
      "grad_norm": 0.2360704243183136,
      "learning_rate": 5.392823118450302e-05,
      "loss": 0.0608,
      "step": 7259
    },
    {
      "epoch": 13.823109843081312,
      "grad_norm": 0.23050999641418457,
      "learning_rate": 5.392187996189266e-05,
      "loss": 0.0475,
      "step": 7260
    },
    {
      "epoch": 13.825011887779363,
      "grad_norm": 0.16015152633190155,
      "learning_rate": 5.391552873928232e-05,
      "loss": 0.0437,
      "step": 7261
    },
    {
      "epoch": 13.826913932477414,
      "grad_norm": 0.12719005346298218,
      "learning_rate": 5.3909177516671964e-05,
      "loss": 0.0702,
      "step": 7262
    },
    {
      "epoch": 13.828815977175463,
      "grad_norm": 0.21993082761764526,
      "learning_rate": 5.3902826294061616e-05,
      "loss": 0.0671,
      "step": 7263
    },
    {
      "epoch": 13.830718021873514,
      "grad_norm": 0.20380781590938568,
      "learning_rate": 5.389647507145126e-05,
      "loss": 0.055,
      "step": 7264
    },
    {
      "epoch": 13.832620066571565,
      "grad_norm": 0.29961729049682617,
      "learning_rate": 5.38901238488409e-05,
      "loss": 0.0689,
      "step": 7265
    },
    {
      "epoch": 13.834522111269616,
      "grad_norm": 0.2058323621749878,
      "learning_rate": 5.388377262623056e-05,
      "loss": 0.0671,
      "step": 7266
    },
    {
      "epoch": 13.836424155967665,
      "grad_norm": 0.33653509616851807,
      "learning_rate": 5.3877421403620196e-05,
      "loss": 0.0624,
      "step": 7267
    },
    {
      "epoch": 13.838326200665716,
      "grad_norm": 0.13654020428657532,
      "learning_rate": 5.387107018100984e-05,
      "loss": 0.0493,
      "step": 7268
    },
    {
      "epoch": 13.840228245363766,
      "grad_norm": 0.31805330514907837,
      "learning_rate": 5.386471895839949e-05,
      "loss": 0.0672,
      "step": 7269
    },
    {
      "epoch": 13.842130290061817,
      "grad_norm": 0.1884007602930069,
      "learning_rate": 5.385836773578914e-05,
      "loss": 0.0721,
      "step": 7270
    },
    {
      "epoch": 13.844032334759866,
      "grad_norm": 0.2302660197019577,
      "learning_rate": 5.3852016513178796e-05,
      "loss": 0.0655,
      "step": 7271
    },
    {
      "epoch": 13.845934379457917,
      "grad_norm": 0.11893751472234726,
      "learning_rate": 5.3845665290568435e-05,
      "loss": 0.05,
      "step": 7272
    },
    {
      "epoch": 13.847836424155968,
      "grad_norm": 0.18107381463050842,
      "learning_rate": 5.383931406795808e-05,
      "loss": 0.0645,
      "step": 7273
    },
    {
      "epoch": 13.849738468854017,
      "grad_norm": 0.2402561604976654,
      "learning_rate": 5.383296284534773e-05,
      "loss": 0.0692,
      "step": 7274
    },
    {
      "epoch": 13.851640513552068,
      "grad_norm": 0.1792139858007431,
      "learning_rate": 5.382661162273738e-05,
      "loss": 0.0536,
      "step": 7275
    },
    {
      "epoch": 13.853542558250119,
      "grad_norm": 0.2682904005050659,
      "learning_rate": 5.382026040012703e-05,
      "loss": 0.0549,
      "step": 7276
    },
    {
      "epoch": 13.85544460294817,
      "grad_norm": 0.2266082614660263,
      "learning_rate": 5.3813909177516674e-05,
      "loss": 0.069,
      "step": 7277
    },
    {
      "epoch": 13.857346647646219,
      "grad_norm": 0.2881041467189789,
      "learning_rate": 5.380755795490632e-05,
      "loss": 0.0659,
      "step": 7278
    },
    {
      "epoch": 13.85924869234427,
      "grad_norm": 0.30093881487846375,
      "learning_rate": 5.380120673229597e-05,
      "loss": 0.0783,
      "step": 7279
    },
    {
      "epoch": 13.86115073704232,
      "grad_norm": 0.2137882113456726,
      "learning_rate": 5.3794855509685616e-05,
      "loss": 0.0688,
      "step": 7280
    },
    {
      "epoch": 13.863052781740372,
      "grad_norm": 0.2526339888572693,
      "learning_rate": 5.378850428707527e-05,
      "loss": 0.0603,
      "step": 7281
    },
    {
      "epoch": 13.86495482643842,
      "grad_norm": 0.20394502580165863,
      "learning_rate": 5.378215306446491e-05,
      "loss": 0.0677,
      "step": 7282
    },
    {
      "epoch": 13.866856871136472,
      "grad_norm": 0.15528029203414917,
      "learning_rate": 5.377580184185456e-05,
      "loss": 0.0619,
      "step": 7283
    },
    {
      "epoch": 13.868758915834523,
      "grad_norm": 0.21263182163238525,
      "learning_rate": 5.376945061924421e-05,
      "loss": 0.0735,
      "step": 7284
    },
    {
      "epoch": 13.870660960532572,
      "grad_norm": 0.2879989743232727,
      "learning_rate": 5.3763099396633854e-05,
      "loss": 0.0747,
      "step": 7285
    },
    {
      "epoch": 13.872563005230623,
      "grad_norm": 0.25435084104537964,
      "learning_rate": 5.37567481740235e-05,
      "loss": 0.058,
      "step": 7286
    },
    {
      "epoch": 13.874465049928673,
      "grad_norm": 0.15193717181682587,
      "learning_rate": 5.375039695141315e-05,
      "loss": 0.0609,
      "step": 7287
    },
    {
      "epoch": 13.876367094626724,
      "grad_norm": 0.20760442316532135,
      "learning_rate": 5.3744045728802796e-05,
      "loss": 0.0512,
      "step": 7288
    },
    {
      "epoch": 13.878269139324773,
      "grad_norm": 0.16267850995063782,
      "learning_rate": 5.373769450619245e-05,
      "loss": 0.0602,
      "step": 7289
    },
    {
      "epoch": 13.880171184022824,
      "grad_norm": 0.2610892951488495,
      "learning_rate": 5.373134328358209e-05,
      "loss": 0.0633,
      "step": 7290
    },
    {
      "epoch": 13.882073228720875,
      "grad_norm": 0.22512303292751312,
      "learning_rate": 5.372499206097173e-05,
      "loss": 0.0598,
      "step": 7291
    },
    {
      "epoch": 13.883975273418926,
      "grad_norm": 0.23303070664405823,
      "learning_rate": 5.371864083836139e-05,
      "loss": 0.0702,
      "step": 7292
    },
    {
      "epoch": 13.885877318116975,
      "grad_norm": 0.19913730025291443,
      "learning_rate": 5.3712289615751035e-05,
      "loss": 0.0662,
      "step": 7293
    },
    {
      "epoch": 13.887779362815026,
      "grad_norm": 0.30238837003707886,
      "learning_rate": 5.370593839314069e-05,
      "loss": 0.0923,
      "step": 7294
    },
    {
      "epoch": 13.889681407513077,
      "grad_norm": 0.22719161212444305,
      "learning_rate": 5.369958717053033e-05,
      "loss": 0.0577,
      "step": 7295
    },
    {
      "epoch": 13.891583452211126,
      "grad_norm": 0.19003400206565857,
      "learning_rate": 5.369323594791997e-05,
      "loss": 0.06,
      "step": 7296
    },
    {
      "epoch": 13.893485496909177,
      "grad_norm": 0.20794767141342163,
      "learning_rate": 5.368688472530963e-05,
      "loss": 0.0596,
      "step": 7297
    },
    {
      "epoch": 13.895387541607228,
      "grad_norm": 0.14656628668308258,
      "learning_rate": 5.368053350269927e-05,
      "loss": 0.0594,
      "step": 7298
    },
    {
      "epoch": 13.897289586305279,
      "grad_norm": 0.1808067113161087,
      "learning_rate": 5.3674182280088926e-05,
      "loss": 0.0743,
      "step": 7299
    },
    {
      "epoch": 13.899191631003328,
      "grad_norm": 0.27814120054244995,
      "learning_rate": 5.366783105747857e-05,
      "loss": 0.0865,
      "step": 7300
    },
    {
      "epoch": 13.901093675701379,
      "grad_norm": 0.17773757874965668,
      "learning_rate": 5.366147983486821e-05,
      "loss": 0.0633,
      "step": 7301
    },
    {
      "epoch": 13.90299572039943,
      "grad_norm": 0.171430766582489,
      "learning_rate": 5.365512861225787e-05,
      "loss": 0.0538,
      "step": 7302
    },
    {
      "epoch": 13.90489776509748,
      "grad_norm": 0.16702894866466522,
      "learning_rate": 5.3648777389647506e-05,
      "loss": 0.0557,
      "step": 7303
    },
    {
      "epoch": 13.90679980979553,
      "grad_norm": 0.241920605301857,
      "learning_rate": 5.364242616703715e-05,
      "loss": 0.0567,
      "step": 7304
    },
    {
      "epoch": 13.90870185449358,
      "grad_norm": 0.1949584186077118,
      "learning_rate": 5.36360749444268e-05,
      "loss": 0.0855,
      "step": 7305
    },
    {
      "epoch": 13.910603899191631,
      "grad_norm": 0.17606988549232483,
      "learning_rate": 5.362972372181645e-05,
      "loss": 0.0715,
      "step": 7306
    },
    {
      "epoch": 13.91250594388968,
      "grad_norm": 0.2018020749092102,
      "learning_rate": 5.362337249920611e-05,
      "loss": 0.0672,
      "step": 7307
    },
    {
      "epoch": 13.914407988587731,
      "grad_norm": 0.20741713047027588,
      "learning_rate": 5.3617021276595745e-05,
      "loss": 0.067,
      "step": 7308
    },
    {
      "epoch": 13.916310033285782,
      "grad_norm": 0.22029127180576324,
      "learning_rate": 5.361067005398539e-05,
      "loss": 0.0438,
      "step": 7309
    },
    {
      "epoch": 13.918212077983833,
      "grad_norm": 0.13996699452400208,
      "learning_rate": 5.360431883137504e-05,
      "loss": 0.0601,
      "step": 7310
    },
    {
      "epoch": 13.920114122681882,
      "grad_norm": 0.23882152140140533,
      "learning_rate": 5.359796760876469e-05,
      "loss": 0.0614,
      "step": 7311
    },
    {
      "epoch": 13.922016167379933,
      "grad_norm": 0.3080902099609375,
      "learning_rate": 5.359161638615434e-05,
      "loss": 0.0838,
      "step": 7312
    },
    {
      "epoch": 13.923918212077984,
      "grad_norm": 0.24677343666553497,
      "learning_rate": 5.3585265163543984e-05,
      "loss": 0.0611,
      "step": 7313
    },
    {
      "epoch": 13.925820256776035,
      "grad_norm": 0.4021768569946289,
      "learning_rate": 5.357891394093363e-05,
      "loss": 0.0867,
      "step": 7314
    },
    {
      "epoch": 13.927722301474084,
      "grad_norm": 0.23158122599124908,
      "learning_rate": 5.357256271832328e-05,
      "loss": 0.0553,
      "step": 7315
    },
    {
      "epoch": 13.929624346172135,
      "grad_norm": 0.2736777067184448,
      "learning_rate": 5.3566211495712926e-05,
      "loss": 0.0649,
      "step": 7316
    },
    {
      "epoch": 13.931526390870186,
      "grad_norm": 0.12983861565589905,
      "learning_rate": 5.355986027310258e-05,
      "loss": 0.0553,
      "step": 7317
    },
    {
      "epoch": 13.933428435568235,
      "grad_norm": 0.20654582977294922,
      "learning_rate": 5.355350905049222e-05,
      "loss": 0.0512,
      "step": 7318
    },
    {
      "epoch": 13.935330480266286,
      "grad_norm": 0.2009374499320984,
      "learning_rate": 5.354715782788187e-05,
      "loss": 0.0513,
      "step": 7319
    },
    {
      "epoch": 13.937232524964337,
      "grad_norm": 0.2418518364429474,
      "learning_rate": 5.354080660527152e-05,
      "loss": 0.0585,
      "step": 7320
    },
    {
      "epoch": 13.939134569662388,
      "grad_norm": 0.17909492552280426,
      "learning_rate": 5.3534455382661165e-05,
      "loss": 0.0595,
      "step": 7321
    },
    {
      "epoch": 13.941036614360437,
      "grad_norm": 0.2896093726158142,
      "learning_rate": 5.352810416005081e-05,
      "loss": 0.1046,
      "step": 7322
    },
    {
      "epoch": 13.942938659058488,
      "grad_norm": 0.3157602846622467,
      "learning_rate": 5.352175293744046e-05,
      "loss": 0.0576,
      "step": 7323
    },
    {
      "epoch": 13.944840703756539,
      "grad_norm": 0.2978362441062927,
      "learning_rate": 5.351540171483011e-05,
      "loss": 0.0701,
      "step": 7324
    },
    {
      "epoch": 13.94674274845459,
      "grad_norm": 0.13338571786880493,
      "learning_rate": 5.350905049221976e-05,
      "loss": 0.0548,
      "step": 7325
    },
    {
      "epoch": 13.948644793152638,
      "grad_norm": 0.5071760416030884,
      "learning_rate": 5.3502699269609404e-05,
      "loss": 0.0801,
      "step": 7326
    },
    {
      "epoch": 13.95054683785069,
      "grad_norm": 0.20635323226451874,
      "learning_rate": 5.349634804699904e-05,
      "loss": 0.0495,
      "step": 7327
    },
    {
      "epoch": 13.95244888254874,
      "grad_norm": 0.3134024441242218,
      "learning_rate": 5.34899968243887e-05,
      "loss": 0.0615,
      "step": 7328
    },
    {
      "epoch": 13.95435092724679,
      "grad_norm": 0.16929863393306732,
      "learning_rate": 5.3483645601778346e-05,
      "loss": 0.0657,
      "step": 7329
    },
    {
      "epoch": 13.95625297194484,
      "grad_norm": 0.23720704019069672,
      "learning_rate": 5.3477294379168e-05,
      "loss": 0.0664,
      "step": 7330
    },
    {
      "epoch": 13.958155016642891,
      "grad_norm": 0.30032214522361755,
      "learning_rate": 5.347094315655764e-05,
      "loss": 0.0661,
      "step": 7331
    },
    {
      "epoch": 13.960057061340942,
      "grad_norm": 0.212426096200943,
      "learning_rate": 5.346459193394728e-05,
      "loss": 0.0686,
      "step": 7332
    },
    {
      "epoch": 13.961959106038991,
      "grad_norm": 0.3285209536552429,
      "learning_rate": 5.345824071133694e-05,
      "loss": 0.0782,
      "step": 7333
    },
    {
      "epoch": 13.963861150737042,
      "grad_norm": 0.19874542951583862,
      "learning_rate": 5.345188948872658e-05,
      "loss": 0.0994,
      "step": 7334
    },
    {
      "epoch": 13.965763195435093,
      "grad_norm": 0.1890346109867096,
      "learning_rate": 5.3445538266116236e-05,
      "loss": 0.0763,
      "step": 7335
    },
    {
      "epoch": 13.967665240133144,
      "grad_norm": 0.20329201221466064,
      "learning_rate": 5.343918704350588e-05,
      "loss": 0.0609,
      "step": 7336
    },
    {
      "epoch": 13.969567284831193,
      "grad_norm": 0.20286491513252258,
      "learning_rate": 5.343283582089552e-05,
      "loss": 0.0491,
      "step": 7337
    },
    {
      "epoch": 13.971469329529244,
      "grad_norm": 0.1681714653968811,
      "learning_rate": 5.342648459828518e-05,
      "loss": 0.0587,
      "step": 7338
    },
    {
      "epoch": 13.973371374227295,
      "grad_norm": 0.23216994106769562,
      "learning_rate": 5.342013337567482e-05,
      "loss": 0.0584,
      "step": 7339
    },
    {
      "epoch": 13.975273418925344,
      "grad_norm": 0.1894179880619049,
      "learning_rate": 5.341378215306446e-05,
      "loss": 0.0499,
      "step": 7340
    },
    {
      "epoch": 13.977175463623395,
      "grad_norm": 0.2528207302093506,
      "learning_rate": 5.3407430930454114e-05,
      "loss": 0.0671,
      "step": 7341
    },
    {
      "epoch": 13.979077508321446,
      "grad_norm": 0.21460577845573425,
      "learning_rate": 5.340107970784376e-05,
      "loss": 0.0538,
      "step": 7342
    },
    {
      "epoch": 13.980979553019496,
      "grad_norm": 0.23168377578258514,
      "learning_rate": 5.339472848523342e-05,
      "loss": 0.066,
      "step": 7343
    },
    {
      "epoch": 13.982881597717546,
      "grad_norm": 0.2557571828365326,
      "learning_rate": 5.3388377262623056e-05,
      "loss": 0.069,
      "step": 7344
    },
    {
      "epoch": 13.984783642415596,
      "grad_norm": 0.28882935643196106,
      "learning_rate": 5.33820260400127e-05,
      "loss": 0.0541,
      "step": 7345
    },
    {
      "epoch": 13.986685687113647,
      "grad_norm": 0.13144026696681976,
      "learning_rate": 5.337567481740235e-05,
      "loss": 0.0678,
      "step": 7346
    },
    {
      "epoch": 13.988587731811698,
      "grad_norm": 0.17657329142093658,
      "learning_rate": 5.3369323594792e-05,
      "loss": 0.0612,
      "step": 7347
    },
    {
      "epoch": 13.990489776509747,
      "grad_norm": 0.1429637223482132,
      "learning_rate": 5.336297237218165e-05,
      "loss": 0.0561,
      "step": 7348
    },
    {
      "epoch": 13.992391821207798,
      "grad_norm": 0.12323079258203506,
      "learning_rate": 5.3356621149571294e-05,
      "loss": 0.059,
      "step": 7349
    },
    {
      "epoch": 13.99429386590585,
      "grad_norm": 0.17069973051548004,
      "learning_rate": 5.335026992696094e-05,
      "loss": 0.0675,
      "step": 7350
    },
    {
      "epoch": 13.9961959106039,
      "grad_norm": 0.16346120834350586,
      "learning_rate": 5.334391870435059e-05,
      "loss": 0.0456,
      "step": 7351
    },
    {
      "epoch": 13.99809795530195,
      "grad_norm": 0.1862858533859253,
      "learning_rate": 5.3337567481740236e-05,
      "loss": 0.0584,
      "step": 7352
    },
    {
      "epoch": 14.0,
      "grad_norm": 0.2782989740371704,
      "learning_rate": 5.333121625912989e-05,
      "loss": 0.0682,
      "step": 7353
    },
    {
      "epoch": 14.00190204469805,
      "grad_norm": 0.2528998851776123,
      "learning_rate": 5.332486503651953e-05,
      "loss": 0.0594,
      "step": 7354
    },
    {
      "epoch": 14.0038040893961,
      "grad_norm": 0.19271168112754822,
      "learning_rate": 5.331851381390918e-05,
      "loss": 0.0616,
      "step": 7355
    },
    {
      "epoch": 14.00570613409415,
      "grad_norm": 0.14756593108177185,
      "learning_rate": 5.331216259129883e-05,
      "loss": 0.0566,
      "step": 7356
    },
    {
      "epoch": 14.007608178792202,
      "grad_norm": 0.2509566843509674,
      "learning_rate": 5.3305811368688475e-05,
      "loss": 0.0683,
      "step": 7357
    },
    {
      "epoch": 14.009510223490253,
      "grad_norm": 0.11840216815471649,
      "learning_rate": 5.329946014607812e-05,
      "loss": 0.0585,
      "step": 7358
    },
    {
      "epoch": 14.011412268188302,
      "grad_norm": 0.15667672455310822,
      "learning_rate": 5.329310892346777e-05,
      "loss": 0.0677,
      "step": 7359
    },
    {
      "epoch": 14.013314312886353,
      "grad_norm": 0.11608525365591049,
      "learning_rate": 5.328675770085742e-05,
      "loss": 0.0521,
      "step": 7360
    },
    {
      "epoch": 14.015216357584404,
      "grad_norm": 0.09074059873819351,
      "learning_rate": 5.328040647824707e-05,
      "loss": 0.0405,
      "step": 7361
    },
    {
      "epoch": 14.017118402282454,
      "grad_norm": 0.20417606830596924,
      "learning_rate": 5.3274055255636714e-05,
      "loss": 0.0486,
      "step": 7362
    },
    {
      "epoch": 14.019020446980504,
      "grad_norm": 0.15287773311138153,
      "learning_rate": 5.326770403302635e-05,
      "loss": 0.0691,
      "step": 7363
    },
    {
      "epoch": 14.020922491678554,
      "grad_norm": 0.2926754057407379,
      "learning_rate": 5.326135281041601e-05,
      "loss": 0.0764,
      "step": 7364
    },
    {
      "epoch": 14.022824536376605,
      "grad_norm": 0.16718968749046326,
      "learning_rate": 5.3255001587805656e-05,
      "loss": 0.0601,
      "step": 7365
    },
    {
      "epoch": 14.024726581074654,
      "grad_norm": 0.18837334215641022,
      "learning_rate": 5.324865036519531e-05,
      "loss": 0.054,
      "step": 7366
    },
    {
      "epoch": 14.026628625772705,
      "grad_norm": 0.1734861582517624,
      "learning_rate": 5.324229914258495e-05,
      "loss": 0.07,
      "step": 7367
    },
    {
      "epoch": 14.028530670470756,
      "grad_norm": 0.38013237714767456,
      "learning_rate": 5.323594791997459e-05,
      "loss": 0.0718,
      "step": 7368
    },
    {
      "epoch": 14.030432715168807,
      "grad_norm": 0.19275639951229095,
      "learning_rate": 5.322959669736425e-05,
      "loss": 0.071,
      "step": 7369
    },
    {
      "epoch": 14.032334759866856,
      "grad_norm": 0.09969597309827805,
      "learning_rate": 5.322324547475389e-05,
      "loss": 0.0414,
      "step": 7370
    },
    {
      "epoch": 14.034236804564907,
      "grad_norm": 0.10100686550140381,
      "learning_rate": 5.321689425214355e-05,
      "loss": 0.0508,
      "step": 7371
    },
    {
      "epoch": 14.036138849262958,
      "grad_norm": 0.11903412640094757,
      "learning_rate": 5.3210543029533185e-05,
      "loss": 0.0839,
      "step": 7372
    },
    {
      "epoch": 14.038040893961009,
      "grad_norm": 0.14304274320602417,
      "learning_rate": 5.320419180692283e-05,
      "loss": 0.0638,
      "step": 7373
    },
    {
      "epoch": 14.039942938659058,
      "grad_norm": 0.24439594149589539,
      "learning_rate": 5.319784058431249e-05,
      "loss": 0.0555,
      "step": 7374
    },
    {
      "epoch": 14.041844983357109,
      "grad_norm": 0.2599331736564636,
      "learning_rate": 5.319148936170213e-05,
      "loss": 0.0546,
      "step": 7375
    },
    {
      "epoch": 14.04374702805516,
      "grad_norm": 0.12374287098646164,
      "learning_rate": 5.318513813909177e-05,
      "loss": 0.0649,
      "step": 7376
    },
    {
      "epoch": 14.045649072753209,
      "grad_norm": 0.21068157255649567,
      "learning_rate": 5.3178786916481424e-05,
      "loss": 0.0727,
      "step": 7377
    },
    {
      "epoch": 14.04755111745126,
      "grad_norm": 0.19516895711421967,
      "learning_rate": 5.317243569387107e-05,
      "loss": 0.0606,
      "step": 7378
    },
    {
      "epoch": 14.04945316214931,
      "grad_norm": 0.27725622057914734,
      "learning_rate": 5.316608447126072e-05,
      "loss": 0.0722,
      "step": 7379
    },
    {
      "epoch": 14.051355206847362,
      "grad_norm": 0.1234855204820633,
      "learning_rate": 5.3159733248650366e-05,
      "loss": 0.0543,
      "step": 7380
    },
    {
      "epoch": 14.05325725154541,
      "grad_norm": 0.23500555753707886,
      "learning_rate": 5.315338202604001e-05,
      "loss": 0.0493,
      "step": 7381
    },
    {
      "epoch": 14.055159296243461,
      "grad_norm": 0.1624249815940857,
      "learning_rate": 5.314703080342966e-05,
      "loss": 0.0442,
      "step": 7382
    },
    {
      "epoch": 14.057061340941512,
      "grad_norm": 0.2046501189470291,
      "learning_rate": 5.314067958081931e-05,
      "loss": 0.0673,
      "step": 7383
    },
    {
      "epoch": 14.058963385639563,
      "grad_norm": 0.1192425861954689,
      "learning_rate": 5.313432835820896e-05,
      "loss": 0.0319,
      "step": 7384
    },
    {
      "epoch": 14.060865430337612,
      "grad_norm": 0.1112857535481453,
      "learning_rate": 5.3127977135598605e-05,
      "loss": 0.0332,
      "step": 7385
    },
    {
      "epoch": 14.062767475035663,
      "grad_norm": 0.26829591393470764,
      "learning_rate": 5.312162591298825e-05,
      "loss": 0.0663,
      "step": 7386
    },
    {
      "epoch": 14.064669519733714,
      "grad_norm": 0.1901654750108719,
      "learning_rate": 5.31152746903779e-05,
      "loss": 0.0653,
      "step": 7387
    },
    {
      "epoch": 14.066571564431765,
      "grad_norm": 0.14392554759979248,
      "learning_rate": 5.310892346776755e-05,
      "loss": 0.0616,
      "step": 7388
    },
    {
      "epoch": 14.068473609129814,
      "grad_norm": 0.144301638007164,
      "learning_rate": 5.31025722451572e-05,
      "loss": 0.0503,
      "step": 7389
    },
    {
      "epoch": 14.070375653827865,
      "grad_norm": 0.2335568219423294,
      "learning_rate": 5.3096221022546844e-05,
      "loss": 0.0583,
      "step": 7390
    },
    {
      "epoch": 14.072277698525916,
      "grad_norm": 0.164510577917099,
      "learning_rate": 5.308986979993649e-05,
      "loss": 0.0557,
      "step": 7391
    },
    {
      "epoch": 14.074179743223965,
      "grad_norm": 0.3761489987373352,
      "learning_rate": 5.308351857732614e-05,
      "loss": 0.0658,
      "step": 7392
    },
    {
      "epoch": 14.076081787922016,
      "grad_norm": 0.2669321596622467,
      "learning_rate": 5.3077167354715786e-05,
      "loss": 0.0635,
      "step": 7393
    },
    {
      "epoch": 14.077983832620067,
      "grad_norm": 0.21432504057884216,
      "learning_rate": 5.307081613210543e-05,
      "loss": 0.0519,
      "step": 7394
    },
    {
      "epoch": 14.079885877318118,
      "grad_norm": 0.316077321767807,
      "learning_rate": 5.306446490949508e-05,
      "loss": 0.0809,
      "step": 7395
    },
    {
      "epoch": 14.081787922016167,
      "grad_norm": 0.14039736986160278,
      "learning_rate": 5.305811368688473e-05,
      "loss": 0.047,
      "step": 7396
    },
    {
      "epoch": 14.083689966714218,
      "grad_norm": 0.1503455489873886,
      "learning_rate": 5.305176246427438e-05,
      "loss": 0.0539,
      "step": 7397
    },
    {
      "epoch": 14.085592011412269,
      "grad_norm": 0.16759897768497467,
      "learning_rate": 5.3045411241664025e-05,
      "loss": 0.0611,
      "step": 7398
    },
    {
      "epoch": 14.08749405611032,
      "grad_norm": 0.13561630249023438,
      "learning_rate": 5.303906001905366e-05,
      "loss": 0.0541,
      "step": 7399
    },
    {
      "epoch": 14.089396100808369,
      "grad_norm": 0.17430227994918823,
      "learning_rate": 5.303270879644332e-05,
      "loss": 0.054,
      "step": 7400
    },
    {
      "epoch": 14.09129814550642,
      "grad_norm": 0.16151241958141327,
      "learning_rate": 5.302635757383296e-05,
      "loss": 0.0471,
      "step": 7401
    },
    {
      "epoch": 14.09320019020447,
      "grad_norm": 0.13555926084518433,
      "learning_rate": 5.302000635122262e-05,
      "loss": 0.0454,
      "step": 7402
    },
    {
      "epoch": 14.09510223490252,
      "grad_norm": 0.19882944226264954,
      "learning_rate": 5.3013655128612264e-05,
      "loss": 0.0769,
      "step": 7403
    },
    {
      "epoch": 14.09700427960057,
      "grad_norm": 0.19104546308517456,
      "learning_rate": 5.30073039060019e-05,
      "loss": 0.0748,
      "step": 7404
    },
    {
      "epoch": 14.098906324298621,
      "grad_norm": 0.24949337542057037,
      "learning_rate": 5.300095268339156e-05,
      "loss": 0.0753,
      "step": 7405
    },
    {
      "epoch": 14.100808368996672,
      "grad_norm": 0.22298850119113922,
      "learning_rate": 5.29946014607812e-05,
      "loss": 0.0556,
      "step": 7406
    },
    {
      "epoch": 14.102710413694721,
      "grad_norm": 0.12767939269542694,
      "learning_rate": 5.298825023817086e-05,
      "loss": 0.0413,
      "step": 7407
    },
    {
      "epoch": 14.104612458392772,
      "grad_norm": 0.3507814407348633,
      "learning_rate": 5.2981899015560496e-05,
      "loss": 0.0677,
      "step": 7408
    },
    {
      "epoch": 14.106514503090823,
      "grad_norm": 0.21291838586330414,
      "learning_rate": 5.297554779295014e-05,
      "loss": 0.0709,
      "step": 7409
    },
    {
      "epoch": 14.108416547788874,
      "grad_norm": 0.20955125987529755,
      "learning_rate": 5.29691965703398e-05,
      "loss": 0.0555,
      "step": 7410
    },
    {
      "epoch": 14.110318592486923,
      "grad_norm": 0.38880980014801025,
      "learning_rate": 5.296284534772944e-05,
      "loss": 0.0605,
      "step": 7411
    },
    {
      "epoch": 14.112220637184974,
      "grad_norm": 0.16443461179733276,
      "learning_rate": 5.295649412511908e-05,
      "loss": 0.0902,
      "step": 7412
    },
    {
      "epoch": 14.114122681883025,
      "grad_norm": 0.11160407215356827,
      "learning_rate": 5.2950142902508735e-05,
      "loss": 0.0431,
      "step": 7413
    },
    {
      "epoch": 14.116024726581074,
      "grad_norm": 0.26912933588027954,
      "learning_rate": 5.294379167989838e-05,
      "loss": 0.0733,
      "step": 7414
    },
    {
      "epoch": 14.117926771279125,
      "grad_norm": 0.11232420057058334,
      "learning_rate": 5.293744045728803e-05,
      "loss": 0.0342,
      "step": 7415
    },
    {
      "epoch": 14.119828815977176,
      "grad_norm": 0.13175956904888153,
      "learning_rate": 5.2931089234677677e-05,
      "loss": 0.0538,
      "step": 7416
    },
    {
      "epoch": 14.121730860675227,
      "grad_norm": 0.1904328614473343,
      "learning_rate": 5.292473801206732e-05,
      "loss": 0.0576,
      "step": 7417
    },
    {
      "epoch": 14.123632905373276,
      "grad_norm": 0.13842704892158508,
      "learning_rate": 5.2918386789456973e-05,
      "loss": 0.036,
      "step": 7418
    },
    {
      "epoch": 14.125534950071327,
      "grad_norm": 0.1916932314634323,
      "learning_rate": 5.291203556684662e-05,
      "loss": 0.0411,
      "step": 7419
    },
    {
      "epoch": 14.127436994769377,
      "grad_norm": 0.1814490705728531,
      "learning_rate": 5.290568434423627e-05,
      "loss": 0.0365,
      "step": 7420
    },
    {
      "epoch": 14.129339039467428,
      "grad_norm": 0.24279388785362244,
      "learning_rate": 5.2899333121625915e-05,
      "loss": 0.0803,
      "step": 7421
    },
    {
      "epoch": 14.131241084165477,
      "grad_norm": 0.2318277359008789,
      "learning_rate": 5.289298189901556e-05,
      "loss": 0.1126,
      "step": 7422
    },
    {
      "epoch": 14.133143128863528,
      "grad_norm": 0.17640984058380127,
      "learning_rate": 5.288663067640521e-05,
      "loss": 0.0591,
      "step": 7423
    },
    {
      "epoch": 14.13504517356158,
      "grad_norm": 0.1813698261976242,
      "learning_rate": 5.288027945379486e-05,
      "loss": 0.0864,
      "step": 7424
    },
    {
      "epoch": 14.136947218259628,
      "grad_norm": 0.23784920573234558,
      "learning_rate": 5.287392823118451e-05,
      "loss": 0.0632,
      "step": 7425
    },
    {
      "epoch": 14.13884926295768,
      "grad_norm": 0.24212495982646942,
      "learning_rate": 5.2867577008574154e-05,
      "loss": 0.0776,
      "step": 7426
    },
    {
      "epoch": 14.14075130765573,
      "grad_norm": 0.10199500620365143,
      "learning_rate": 5.28612257859638e-05,
      "loss": 0.0574,
      "step": 7427
    },
    {
      "epoch": 14.142653352353781,
      "grad_norm": 0.07228594273328781,
      "learning_rate": 5.285487456335345e-05,
      "loss": 0.0407,
      "step": 7428
    },
    {
      "epoch": 14.14455539705183,
      "grad_norm": 0.11618991196155548,
      "learning_rate": 5.2848523340743096e-05,
      "loss": 0.0706,
      "step": 7429
    },
    {
      "epoch": 14.146457441749881,
      "grad_norm": 0.2242993712425232,
      "learning_rate": 5.2842172118132735e-05,
      "loss": 0.0634,
      "step": 7430
    },
    {
      "epoch": 14.148359486447932,
      "grad_norm": 0.19708840548992157,
      "learning_rate": 5.283582089552239e-05,
      "loss": 0.0505,
      "step": 7431
    },
    {
      "epoch": 14.150261531145983,
      "grad_norm": 0.2558425962924957,
      "learning_rate": 5.282946967291204e-05,
      "loss": 0.084,
      "step": 7432
    },
    {
      "epoch": 14.152163575844032,
      "grad_norm": 0.12093882262706757,
      "learning_rate": 5.282311845030169e-05,
      "loss": 0.0448,
      "step": 7433
    },
    {
      "epoch": 14.154065620542083,
      "grad_norm": 0.18094058334827423,
      "learning_rate": 5.2816767227691335e-05,
      "loss": 0.0795,
      "step": 7434
    },
    {
      "epoch": 14.155967665240134,
      "grad_norm": 0.20837660133838654,
      "learning_rate": 5.2810416005080973e-05,
      "loss": 0.0533,
      "step": 7435
    },
    {
      "epoch": 14.157869709938183,
      "grad_norm": 0.20709973573684692,
      "learning_rate": 5.280406478247063e-05,
      "loss": 0.0697,
      "step": 7436
    },
    {
      "epoch": 14.159771754636234,
      "grad_norm": 0.2028302401304245,
      "learning_rate": 5.279771355986027e-05,
      "loss": 0.0565,
      "step": 7437
    },
    {
      "epoch": 14.161673799334284,
      "grad_norm": 0.15181556344032288,
      "learning_rate": 5.279136233724993e-05,
      "loss": 0.0526,
      "step": 7438
    },
    {
      "epoch": 14.163575844032335,
      "grad_norm": 0.16445064544677734,
      "learning_rate": 5.2785011114639574e-05,
      "loss": 0.0502,
      "step": 7439
    },
    {
      "epoch": 14.165477888730384,
      "grad_norm": 0.26812055706977844,
      "learning_rate": 5.277865989202921e-05,
      "loss": 0.0699,
      "step": 7440
    },
    {
      "epoch": 14.167379933428435,
      "grad_norm": 0.16392607986927032,
      "learning_rate": 5.277230866941887e-05,
      "loss": 0.0524,
      "step": 7441
    },
    {
      "epoch": 14.169281978126486,
      "grad_norm": 0.1482769399881363,
      "learning_rate": 5.276595744680851e-05,
      "loss": 0.0901,
      "step": 7442
    },
    {
      "epoch": 14.171184022824537,
      "grad_norm": 0.38180655241012573,
      "learning_rate": 5.275960622419817e-05,
      "loss": 0.0734,
      "step": 7443
    },
    {
      "epoch": 14.173086067522586,
      "grad_norm": 0.13449139893054962,
      "learning_rate": 5.2753255001587806e-05,
      "loss": 0.0537,
      "step": 7444
    },
    {
      "epoch": 14.174988112220637,
      "grad_norm": 0.21906720101833344,
      "learning_rate": 5.274690377897745e-05,
      "loss": 0.0648,
      "step": 7445
    },
    {
      "epoch": 14.176890156918688,
      "grad_norm": 0.23385153710842133,
      "learning_rate": 5.274055255636711e-05,
      "loss": 0.0568,
      "step": 7446
    },
    {
      "epoch": 14.178792201616737,
      "grad_norm": 0.18561255931854248,
      "learning_rate": 5.273420133375675e-05,
      "loss": 0.0571,
      "step": 7447
    },
    {
      "epoch": 14.180694246314788,
      "grad_norm": 0.356733500957489,
      "learning_rate": 5.272785011114639e-05,
      "loss": 0.0734,
      "step": 7448
    },
    {
      "epoch": 14.182596291012839,
      "grad_norm": 0.2717702090740204,
      "learning_rate": 5.2721498888536045e-05,
      "loss": 0.0531,
      "step": 7449
    },
    {
      "epoch": 14.18449833571089,
      "grad_norm": 0.12378834187984467,
      "learning_rate": 5.271514766592569e-05,
      "loss": 0.0587,
      "step": 7450
    },
    {
      "epoch": 14.186400380408939,
      "grad_norm": 0.22417086362838745,
      "learning_rate": 5.270879644331534e-05,
      "loss": 0.0912,
      "step": 7451
    },
    {
      "epoch": 14.18830242510699,
      "grad_norm": 0.13545849919319153,
      "learning_rate": 5.270244522070499e-05,
      "loss": 0.0547,
      "step": 7452
    },
    {
      "epoch": 14.19020446980504,
      "grad_norm": 0.28296709060668945,
      "learning_rate": 5.269609399809463e-05,
      "loss": 0.0721,
      "step": 7453
    },
    {
      "epoch": 14.192106514503092,
      "grad_norm": 0.17048925161361694,
      "learning_rate": 5.2689742775484284e-05,
      "loss": 0.0699,
      "step": 7454
    },
    {
      "epoch": 14.19400855920114,
      "grad_norm": 0.16482289135456085,
      "learning_rate": 5.268339155287393e-05,
      "loss": 0.0499,
      "step": 7455
    },
    {
      "epoch": 14.195910603899192,
      "grad_norm": 0.241130992770195,
      "learning_rate": 5.267704033026358e-05,
      "loss": 0.0873,
      "step": 7456
    },
    {
      "epoch": 14.197812648597242,
      "grad_norm": 0.15253259241580963,
      "learning_rate": 5.2670689107653226e-05,
      "loss": 0.053,
      "step": 7457
    },
    {
      "epoch": 14.199714693295292,
      "grad_norm": 0.19944600760936737,
      "learning_rate": 5.266433788504287e-05,
      "loss": 0.0746,
      "step": 7458
    },
    {
      "epoch": 14.201616737993342,
      "grad_norm": 0.17144067585468292,
      "learning_rate": 5.265798666243252e-05,
      "loss": 0.0691,
      "step": 7459
    },
    {
      "epoch": 14.203518782691393,
      "grad_norm": 0.20575584471225739,
      "learning_rate": 5.265163543982217e-05,
      "loss": 0.0449,
      "step": 7460
    },
    {
      "epoch": 14.205420827389444,
      "grad_norm": 0.11989787220954895,
      "learning_rate": 5.264528421721182e-05,
      "loss": 0.0442,
      "step": 7461
    },
    {
      "epoch": 14.207322872087493,
      "grad_norm": 0.19378036260604858,
      "learning_rate": 5.2638932994601465e-05,
      "loss": 0.0527,
      "step": 7462
    },
    {
      "epoch": 14.209224916785544,
      "grad_norm": 0.1528719812631607,
      "learning_rate": 5.263258177199111e-05,
      "loss": 0.0606,
      "step": 7463
    },
    {
      "epoch": 14.211126961483595,
      "grad_norm": 0.18818438053131104,
      "learning_rate": 5.262623054938076e-05,
      "loss": 0.0509,
      "step": 7464
    },
    {
      "epoch": 14.213029006181646,
      "grad_norm": 0.14130130410194397,
      "learning_rate": 5.261987932677041e-05,
      "loss": 0.0517,
      "step": 7465
    },
    {
      "epoch": 14.214931050879695,
      "grad_norm": 0.1270495057106018,
      "learning_rate": 5.2613528104160045e-05,
      "loss": 0.0651,
      "step": 7466
    },
    {
      "epoch": 14.216833095577746,
      "grad_norm": 0.22663429379463196,
      "learning_rate": 5.2607176881549704e-05,
      "loss": 0.0492,
      "step": 7467
    },
    {
      "epoch": 14.218735140275797,
      "grad_norm": 0.18520976603031158,
      "learning_rate": 5.260082565893935e-05,
      "loss": 0.076,
      "step": 7468
    },
    {
      "epoch": 14.220637184973846,
      "grad_norm": 0.14399489760398865,
      "learning_rate": 5.2594474436329e-05,
      "loss": 0.0581,
      "step": 7469
    },
    {
      "epoch": 14.222539229671897,
      "grad_norm": 0.12707483768463135,
      "learning_rate": 5.2588123213718646e-05,
      "loss": 0.0513,
      "step": 7470
    },
    {
      "epoch": 14.224441274369948,
      "grad_norm": 0.1769390106201172,
      "learning_rate": 5.2581771991108284e-05,
      "loss": 0.0819,
      "step": 7471
    },
    {
      "epoch": 14.226343319067999,
      "grad_norm": 0.17762137949466705,
      "learning_rate": 5.257542076849794e-05,
      "loss": 0.0649,
      "step": 7472
    },
    {
      "epoch": 14.228245363766048,
      "grad_norm": 0.2458849549293518,
      "learning_rate": 5.256906954588758e-05,
      "loss": 0.0641,
      "step": 7473
    },
    {
      "epoch": 14.230147408464099,
      "grad_norm": 0.12928763031959534,
      "learning_rate": 5.256271832327724e-05,
      "loss": 0.0564,
      "step": 7474
    },
    {
      "epoch": 14.23204945316215,
      "grad_norm": 0.11910723894834518,
      "learning_rate": 5.2556367100666884e-05,
      "loss": 0.064,
      "step": 7475
    },
    {
      "epoch": 14.2339514978602,
      "grad_norm": 0.14299996197223663,
      "learning_rate": 5.255001587805652e-05,
      "loss": 0.0482,
      "step": 7476
    },
    {
      "epoch": 14.23585354255825,
      "grad_norm": 0.36283624172210693,
      "learning_rate": 5.254366465544618e-05,
      "loss": 0.0642,
      "step": 7477
    },
    {
      "epoch": 14.2377555872563,
      "grad_norm": 0.19579343497753143,
      "learning_rate": 5.253731343283582e-05,
      "loss": 0.0679,
      "step": 7478
    },
    {
      "epoch": 14.239657631954351,
      "grad_norm": 0.12536804378032684,
      "learning_rate": 5.253096221022548e-05,
      "loss": 0.0553,
      "step": 7479
    },
    {
      "epoch": 14.241559676652402,
      "grad_norm": 0.16742166876792908,
      "learning_rate": 5.2524610987615117e-05,
      "loss": 0.0534,
      "step": 7480
    },
    {
      "epoch": 14.243461721350451,
      "grad_norm": 0.32988205552101135,
      "learning_rate": 5.251825976500476e-05,
      "loss": 0.072,
      "step": 7481
    },
    {
      "epoch": 14.245363766048502,
      "grad_norm": 0.19991889595985413,
      "learning_rate": 5.2511908542394413e-05,
      "loss": 0.0632,
      "step": 7482
    },
    {
      "epoch": 14.247265810746553,
      "grad_norm": 0.23099440336227417,
      "learning_rate": 5.250555731978406e-05,
      "loss": 0.0702,
      "step": 7483
    },
    {
      "epoch": 14.249167855444602,
      "grad_norm": 0.20775353908538818,
      "learning_rate": 5.2499206097173704e-05,
      "loss": 0.0734,
      "step": 7484
    },
    {
      "epoch": 14.251069900142653,
      "grad_norm": 0.20609457790851593,
      "learning_rate": 5.2492854874563355e-05,
      "loss": 0.0596,
      "step": 7485
    },
    {
      "epoch": 14.252971944840704,
      "grad_norm": 0.22942540049552917,
      "learning_rate": 5.2486503651953e-05,
      "loss": 0.0551,
      "step": 7486
    },
    {
      "epoch": 14.254873989538755,
      "grad_norm": 0.19878560304641724,
      "learning_rate": 5.248015242934265e-05,
      "loss": 0.0576,
      "step": 7487
    },
    {
      "epoch": 14.256776034236804,
      "grad_norm": 0.19331955909729004,
      "learning_rate": 5.24738012067323e-05,
      "loss": 0.0868,
      "step": 7488
    },
    {
      "epoch": 14.258678078934855,
      "grad_norm": 0.24519440531730652,
      "learning_rate": 5.246744998412194e-05,
      "loss": 0.0642,
      "step": 7489
    },
    {
      "epoch": 14.260580123632906,
      "grad_norm": 0.23775117099285126,
      "learning_rate": 5.2461098761511594e-05,
      "loss": 0.0739,
      "step": 7490
    },
    {
      "epoch": 14.262482168330957,
      "grad_norm": 0.17239496111869812,
      "learning_rate": 5.245474753890124e-05,
      "loss": 0.0525,
      "step": 7491
    },
    {
      "epoch": 14.264384213029006,
      "grad_norm": 0.09015030413866043,
      "learning_rate": 5.244839631629089e-05,
      "loss": 0.0641,
      "step": 7492
    },
    {
      "epoch": 14.266286257727057,
      "grad_norm": 0.14684611558914185,
      "learning_rate": 5.2442045093680536e-05,
      "loss": 0.0573,
      "step": 7493
    },
    {
      "epoch": 14.268188302425107,
      "grad_norm": 0.20221078395843506,
      "learning_rate": 5.243569387107018e-05,
      "loss": 0.0613,
      "step": 7494
    },
    {
      "epoch": 14.270090347123157,
      "grad_norm": 0.38689568638801575,
      "learning_rate": 5.242934264845983e-05,
      "loss": 0.0707,
      "step": 7495
    },
    {
      "epoch": 14.271992391821207,
      "grad_norm": 0.17096443474292755,
      "learning_rate": 5.242299142584948e-05,
      "loss": 0.0508,
      "step": 7496
    },
    {
      "epoch": 14.273894436519258,
      "grad_norm": 0.14185024797916412,
      "learning_rate": 5.241664020323913e-05,
      "loss": 0.0541,
      "step": 7497
    },
    {
      "epoch": 14.27579648121731,
      "grad_norm": 0.18281078338623047,
      "learning_rate": 5.2410288980628775e-05,
      "loss": 0.0711,
      "step": 7498
    },
    {
      "epoch": 14.277698525915358,
      "grad_norm": 0.17065037786960602,
      "learning_rate": 5.240393775801842e-05,
      "loss": 0.0425,
      "step": 7499
    },
    {
      "epoch": 14.27960057061341,
      "grad_norm": 0.240870401263237,
      "learning_rate": 5.239758653540807e-05,
      "loss": 0.0556,
      "step": 7500
    },
    {
      "epoch": 14.28150261531146,
      "grad_norm": 0.14593961834907532,
      "learning_rate": 5.239123531279772e-05,
      "loss": 0.0551,
      "step": 7501
    },
    {
      "epoch": 14.283404660009511,
      "grad_norm": 0.21113112568855286,
      "learning_rate": 5.2384884090187355e-05,
      "loss": 0.0401,
      "step": 7502
    },
    {
      "epoch": 14.28530670470756,
      "grad_norm": 0.24402189254760742,
      "learning_rate": 5.2378532867577014e-05,
      "loss": 0.0692,
      "step": 7503
    },
    {
      "epoch": 14.287208749405611,
      "grad_norm": 0.290025532245636,
      "learning_rate": 5.237218164496666e-05,
      "loss": 0.0754,
      "step": 7504
    },
    {
      "epoch": 14.289110794103662,
      "grad_norm": 0.21898028254508972,
      "learning_rate": 5.236583042235631e-05,
      "loss": 0.0673,
      "step": 7505
    },
    {
      "epoch": 14.291012838801711,
      "grad_norm": 0.2041853815317154,
      "learning_rate": 5.2359479199745956e-05,
      "loss": 0.054,
      "step": 7506
    },
    {
      "epoch": 14.292914883499762,
      "grad_norm": 0.2848202586174011,
      "learning_rate": 5.2353127977135594e-05,
      "loss": 0.0649,
      "step": 7507
    },
    {
      "epoch": 14.294816928197813,
      "grad_norm": 0.13204796612262726,
      "learning_rate": 5.234677675452525e-05,
      "loss": 0.0733,
      "step": 7508
    },
    {
      "epoch": 14.296718972895864,
      "grad_norm": 0.14004924893379211,
      "learning_rate": 5.234042553191489e-05,
      "loss": 0.0585,
      "step": 7509
    },
    {
      "epoch": 14.298621017593913,
      "grad_norm": 0.1074732318520546,
      "learning_rate": 5.233407430930455e-05,
      "loss": 0.0531,
      "step": 7510
    },
    {
      "epoch": 14.300523062291964,
      "grad_norm": 0.2174833118915558,
      "learning_rate": 5.232772308669419e-05,
      "loss": 0.0768,
      "step": 7511
    },
    {
      "epoch": 14.302425106990015,
      "grad_norm": 0.2654950022697449,
      "learning_rate": 5.232137186408383e-05,
      "loss": 0.0579,
      "step": 7512
    },
    {
      "epoch": 14.304327151688065,
      "grad_norm": 0.14409427344799042,
      "learning_rate": 5.231502064147349e-05,
      "loss": 0.0472,
      "step": 7513
    },
    {
      "epoch": 14.306229196386115,
      "grad_norm": 0.12296228855848312,
      "learning_rate": 5.230866941886313e-05,
      "loss": 0.0449,
      "step": 7514
    },
    {
      "epoch": 14.308131241084165,
      "grad_norm": 0.14399056136608124,
      "learning_rate": 5.230231819625279e-05,
      "loss": 0.0451,
      "step": 7515
    },
    {
      "epoch": 14.310033285782216,
      "grad_norm": 0.17021867632865906,
      "learning_rate": 5.229596697364243e-05,
      "loss": 0.0513,
      "step": 7516
    },
    {
      "epoch": 14.311935330480265,
      "grad_norm": 0.15062189102172852,
      "learning_rate": 5.228961575103207e-05,
      "loss": 0.0484,
      "step": 7517
    },
    {
      "epoch": 14.313837375178316,
      "grad_norm": 0.18984819948673248,
      "learning_rate": 5.2283264528421724e-05,
      "loss": 0.0725,
      "step": 7518
    },
    {
      "epoch": 14.315739419876367,
      "grad_norm": 0.15147118270397186,
      "learning_rate": 5.227691330581137e-05,
      "loss": 0.0559,
      "step": 7519
    },
    {
      "epoch": 14.317641464574418,
      "grad_norm": 0.23060442507266998,
      "learning_rate": 5.2270562083201014e-05,
      "loss": 0.0501,
      "step": 7520
    },
    {
      "epoch": 14.319543509272467,
      "grad_norm": 0.14676730334758759,
      "learning_rate": 5.2264210860590666e-05,
      "loss": 0.0599,
      "step": 7521
    },
    {
      "epoch": 14.321445553970518,
      "grad_norm": 0.21464918553829193,
      "learning_rate": 5.225785963798031e-05,
      "loss": 0.0643,
      "step": 7522
    },
    {
      "epoch": 14.323347598668569,
      "grad_norm": 0.2580917775630951,
      "learning_rate": 5.225150841536996e-05,
      "loss": 0.067,
      "step": 7523
    },
    {
      "epoch": 14.32524964336662,
      "grad_norm": 0.23342084884643555,
      "learning_rate": 5.224515719275961e-05,
      "loss": 0.0746,
      "step": 7524
    },
    {
      "epoch": 14.327151688064669,
      "grad_norm": 0.1253678947687149,
      "learning_rate": 5.223880597014925e-05,
      "loss": 0.0436,
      "step": 7525
    },
    {
      "epoch": 14.32905373276272,
      "grad_norm": 0.20070791244506836,
      "learning_rate": 5.2232454747538905e-05,
      "loss": 0.0628,
      "step": 7526
    },
    {
      "epoch": 14.33095577746077,
      "grad_norm": 0.32024890184402466,
      "learning_rate": 5.222610352492855e-05,
      "loss": 0.0637,
      "step": 7527
    },
    {
      "epoch": 14.332857822158822,
      "grad_norm": 0.17475807666778564,
      "learning_rate": 5.22197523023182e-05,
      "loss": 0.0512,
      "step": 7528
    },
    {
      "epoch": 14.33475986685687,
      "grad_norm": 0.2347768247127533,
      "learning_rate": 5.221340107970785e-05,
      "loss": 0.0553,
      "step": 7529
    },
    {
      "epoch": 14.336661911554922,
      "grad_norm": 0.17313428223133087,
      "learning_rate": 5.220704985709749e-05,
      "loss": 0.0519,
      "step": 7530
    },
    {
      "epoch": 14.338563956252973,
      "grad_norm": 0.1019507348537445,
      "learning_rate": 5.2200698634487144e-05,
      "loss": 0.0738,
      "step": 7531
    },
    {
      "epoch": 14.340466000951022,
      "grad_norm": 0.18478825688362122,
      "learning_rate": 5.219434741187679e-05,
      "loss": 0.0468,
      "step": 7532
    },
    {
      "epoch": 14.342368045649073,
      "grad_norm": 0.33729398250579834,
      "learning_rate": 5.218799618926644e-05,
      "loss": 0.069,
      "step": 7533
    },
    {
      "epoch": 14.344270090347123,
      "grad_norm": 0.18752019107341766,
      "learning_rate": 5.2181644966656086e-05,
      "loss": 0.0614,
      "step": 7534
    },
    {
      "epoch": 14.346172135045174,
      "grad_norm": 0.10709106922149658,
      "learning_rate": 5.217529374404573e-05,
      "loss": 0.0418,
      "step": 7535
    },
    {
      "epoch": 14.348074179743223,
      "grad_norm": 0.2352564036846161,
      "learning_rate": 5.216894252143538e-05,
      "loss": 0.0649,
      "step": 7536
    },
    {
      "epoch": 14.349976224441274,
      "grad_norm": 0.2340080589056015,
      "learning_rate": 5.216259129882503e-05,
      "loss": 0.051,
      "step": 7537
    },
    {
      "epoch": 14.351878269139325,
      "grad_norm": 0.23616208136081696,
      "learning_rate": 5.2156240076214666e-05,
      "loss": 0.0653,
      "step": 7538
    },
    {
      "epoch": 14.353780313837376,
      "grad_norm": 0.1928998827934265,
      "learning_rate": 5.2149888853604324e-05,
      "loss": 0.0623,
      "step": 7539
    },
    {
      "epoch": 14.355682358535425,
      "grad_norm": 0.17965523898601532,
      "learning_rate": 5.214353763099396e-05,
      "loss": 0.0604,
      "step": 7540
    },
    {
      "epoch": 14.357584403233476,
      "grad_norm": 0.15935584902763367,
      "learning_rate": 5.213718640838362e-05,
      "loss": 0.0581,
      "step": 7541
    },
    {
      "epoch": 14.359486447931527,
      "grad_norm": 0.11977368593215942,
      "learning_rate": 5.2130835185773266e-05,
      "loss": 0.0569,
      "step": 7542
    },
    {
      "epoch": 14.361388492629576,
      "grad_norm": 0.11723438650369644,
      "learning_rate": 5.2124483963162905e-05,
      "loss": 0.0492,
      "step": 7543
    },
    {
      "epoch": 14.363290537327627,
      "grad_norm": 0.21849237382411957,
      "learning_rate": 5.211813274055256e-05,
      "loss": 0.0542,
      "step": 7544
    },
    {
      "epoch": 14.365192582025678,
      "grad_norm": 0.14745013415813446,
      "learning_rate": 5.21117815179422e-05,
      "loss": 0.0627,
      "step": 7545
    },
    {
      "epoch": 14.367094626723729,
      "grad_norm": 0.24485322833061218,
      "learning_rate": 5.210543029533186e-05,
      "loss": 0.0658,
      "step": 7546
    },
    {
      "epoch": 14.368996671421778,
      "grad_norm": 0.2918044626712799,
      "learning_rate": 5.20990790727215e-05,
      "loss": 0.0848,
      "step": 7547
    },
    {
      "epoch": 14.370898716119829,
      "grad_norm": 0.13040199875831604,
      "learning_rate": 5.2092727850111144e-05,
      "loss": 0.0835,
      "step": 7548
    },
    {
      "epoch": 14.37280076081788,
      "grad_norm": 0.22240321338176727,
      "learning_rate": 5.20863766275008e-05,
      "loss": 0.0753,
      "step": 7549
    },
    {
      "epoch": 14.37470280551593,
      "grad_norm": 0.1147005707025528,
      "learning_rate": 5.208002540489044e-05,
      "loss": 0.0646,
      "step": 7550
    },
    {
      "epoch": 14.37660485021398,
      "grad_norm": 0.17129144072532654,
      "learning_rate": 5.20736741822801e-05,
      "loss": 0.0577,
      "step": 7551
    },
    {
      "epoch": 14.37850689491203,
      "grad_norm": 0.2121477574110031,
      "learning_rate": 5.206732295966974e-05,
      "loss": 0.0514,
      "step": 7552
    },
    {
      "epoch": 14.380408939610081,
      "grad_norm": 0.29183366894721985,
      "learning_rate": 5.206097173705938e-05,
      "loss": 0.0778,
      "step": 7553
    },
    {
      "epoch": 14.38231098430813,
      "grad_norm": 0.2168087661266327,
      "learning_rate": 5.2054620514449034e-05,
      "loss": 0.0673,
      "step": 7554
    },
    {
      "epoch": 14.384213029006181,
      "grad_norm": 0.29269078373908997,
      "learning_rate": 5.204826929183868e-05,
      "loss": 0.0895,
      "step": 7555
    },
    {
      "epoch": 14.386115073704232,
      "grad_norm": 0.08655479550361633,
      "learning_rate": 5.2041918069228324e-05,
      "loss": 0.0473,
      "step": 7556
    },
    {
      "epoch": 14.388017118402283,
      "grad_norm": 0.2127612829208374,
      "learning_rate": 5.2035566846617976e-05,
      "loss": 0.0725,
      "step": 7557
    },
    {
      "epoch": 14.389919163100332,
      "grad_norm": 0.16400641202926636,
      "learning_rate": 5.202921562400762e-05,
      "loss": 0.0633,
      "step": 7558
    },
    {
      "epoch": 14.391821207798383,
      "grad_norm": 0.30889225006103516,
      "learning_rate": 5.202286440139727e-05,
      "loss": 0.0838,
      "step": 7559
    },
    {
      "epoch": 14.393723252496434,
      "grad_norm": 0.2411080151796341,
      "learning_rate": 5.201651317878692e-05,
      "loss": 0.062,
      "step": 7560
    },
    {
      "epoch": 14.395625297194485,
      "grad_norm": 0.14351652562618256,
      "learning_rate": 5.201016195617656e-05,
      "loss": 0.038,
      "step": 7561
    },
    {
      "epoch": 14.397527341892534,
      "grad_norm": 0.27990370988845825,
      "learning_rate": 5.2003810733566215e-05,
      "loss": 0.0642,
      "step": 7562
    },
    {
      "epoch": 14.399429386590585,
      "grad_norm": 0.22874848544597626,
      "learning_rate": 5.199745951095586e-05,
      "loss": 0.0659,
      "step": 7563
    },
    {
      "epoch": 14.401331431288636,
      "grad_norm": 0.1129019483923912,
      "learning_rate": 5.199110828834551e-05,
      "loss": 0.0415,
      "step": 7564
    },
    {
      "epoch": 14.403233475986685,
      "grad_norm": 0.25754669308662415,
      "learning_rate": 5.198475706573516e-05,
      "loss": 0.0684,
      "step": 7565
    },
    {
      "epoch": 14.405135520684736,
      "grad_norm": 0.19301319122314453,
      "learning_rate": 5.19784058431248e-05,
      "loss": 0.0739,
      "step": 7566
    },
    {
      "epoch": 14.407037565382787,
      "grad_norm": 0.27009403705596924,
      "learning_rate": 5.1972054620514454e-05,
      "loss": 0.0858,
      "step": 7567
    },
    {
      "epoch": 14.408939610080838,
      "grad_norm": 0.17593978345394135,
      "learning_rate": 5.19657033979041e-05,
      "loss": 0.053,
      "step": 7568
    },
    {
      "epoch": 14.410841654778887,
      "grad_norm": 0.14010390639305115,
      "learning_rate": 5.195935217529375e-05,
      "loss": 0.0478,
      "step": 7569
    },
    {
      "epoch": 14.412743699476938,
      "grad_norm": 0.2071506828069687,
      "learning_rate": 5.1953000952683396e-05,
      "loss": 0.0533,
      "step": 7570
    },
    {
      "epoch": 14.414645744174988,
      "grad_norm": 0.13032926619052887,
      "learning_rate": 5.194664973007304e-05,
      "loss": 0.0418,
      "step": 7571
    },
    {
      "epoch": 14.41654778887304,
      "grad_norm": 0.10803449153900146,
      "learning_rate": 5.194029850746269e-05,
      "loss": 0.0495,
      "step": 7572
    },
    {
      "epoch": 14.418449833571088,
      "grad_norm": 0.30941590666770935,
      "learning_rate": 5.193394728485234e-05,
      "loss": 0.0693,
      "step": 7573
    },
    {
      "epoch": 14.42035187826914,
      "grad_norm": 0.13794384896755219,
      "learning_rate": 5.1927596062241976e-05,
      "loss": 0.066,
      "step": 7574
    },
    {
      "epoch": 14.42225392296719,
      "grad_norm": 0.10141877830028534,
      "learning_rate": 5.1921244839631635e-05,
      "loss": 0.0432,
      "step": 7575
    },
    {
      "epoch": 14.42415596766524,
      "grad_norm": 0.2614104747772217,
      "learning_rate": 5.191489361702127e-05,
      "loss": 0.0686,
      "step": 7576
    },
    {
      "epoch": 14.42605801236329,
      "grad_norm": 0.2342391312122345,
      "learning_rate": 5.190854239441093e-05,
      "loss": 0.0546,
      "step": 7577
    },
    {
      "epoch": 14.427960057061341,
      "grad_norm": 0.2741243839263916,
      "learning_rate": 5.190219117180058e-05,
      "loss": 0.0693,
      "step": 7578
    },
    {
      "epoch": 14.429862101759392,
      "grad_norm": 0.14258788526058197,
      "learning_rate": 5.1895839949190215e-05,
      "loss": 0.0844,
      "step": 7579
    },
    {
      "epoch": 14.431764146457441,
      "grad_norm": 0.19403590261936188,
      "learning_rate": 5.1889488726579874e-05,
      "loss": 0.049,
      "step": 7580
    },
    {
      "epoch": 14.433666191155492,
      "grad_norm": 0.10535039007663727,
      "learning_rate": 5.188313750396951e-05,
      "loss": 0.0468,
      "step": 7581
    },
    {
      "epoch": 14.435568235853543,
      "grad_norm": 0.27161017060279846,
      "learning_rate": 5.187678628135917e-05,
      "loss": 0.0733,
      "step": 7582
    },
    {
      "epoch": 14.437470280551594,
      "grad_norm": 0.1395357847213745,
      "learning_rate": 5.187043505874881e-05,
      "loss": 0.0582,
      "step": 7583
    },
    {
      "epoch": 14.439372325249643,
      "grad_norm": 0.12294315546751022,
      "learning_rate": 5.1864083836138454e-05,
      "loss": 0.0635,
      "step": 7584
    },
    {
      "epoch": 14.441274369947694,
      "grad_norm": 0.24253630638122559,
      "learning_rate": 5.185773261352811e-05,
      "loss": 0.0739,
      "step": 7585
    },
    {
      "epoch": 14.443176414645745,
      "grad_norm": 0.1546248495578766,
      "learning_rate": 5.185138139091775e-05,
      "loss": 0.065,
      "step": 7586
    },
    {
      "epoch": 14.445078459343794,
      "grad_norm": 0.15066717565059662,
      "learning_rate": 5.184503016830741e-05,
      "loss": 0.0621,
      "step": 7587
    },
    {
      "epoch": 14.446980504041845,
      "grad_norm": 0.1923573911190033,
      "learning_rate": 5.183867894569705e-05,
      "loss": 0.049,
      "step": 7588
    },
    {
      "epoch": 14.448882548739896,
      "grad_norm": 0.17305736243724823,
      "learning_rate": 5.183232772308669e-05,
      "loss": 0.0524,
      "step": 7589
    },
    {
      "epoch": 14.450784593437946,
      "grad_norm": 0.17465534806251526,
      "learning_rate": 5.1825976500476345e-05,
      "loss": 0.0536,
      "step": 7590
    },
    {
      "epoch": 14.452686638135996,
      "grad_norm": 0.15740615129470825,
      "learning_rate": 5.181962527786599e-05,
      "loss": 0.0552,
      "step": 7591
    },
    {
      "epoch": 14.454588682834046,
      "grad_norm": 0.16666673123836517,
      "learning_rate": 5.1813274055255635e-05,
      "loss": 0.0615,
      "step": 7592
    },
    {
      "epoch": 14.456490727532097,
      "grad_norm": 0.18835856020450592,
      "learning_rate": 5.180692283264529e-05,
      "loss": 0.0486,
      "step": 7593
    },
    {
      "epoch": 14.458392772230148,
      "grad_norm": 0.18493612110614777,
      "learning_rate": 5.180057161003493e-05,
      "loss": 0.0487,
      "step": 7594
    },
    {
      "epoch": 14.460294816928197,
      "grad_norm": 0.15935729444026947,
      "learning_rate": 5.1794220387424584e-05,
      "loss": 0.0569,
      "step": 7595
    },
    {
      "epoch": 14.462196861626248,
      "grad_norm": 0.1528821438550949,
      "learning_rate": 5.178786916481423e-05,
      "loss": 0.0562,
      "step": 7596
    },
    {
      "epoch": 14.464098906324299,
      "grad_norm": 0.11653603613376617,
      "learning_rate": 5.1781517942203874e-05,
      "loss": 0.0651,
      "step": 7597
    },
    {
      "epoch": 14.466000951022348,
      "grad_norm": 0.16568215191364288,
      "learning_rate": 5.1775166719593526e-05,
      "loss": 0.0581,
      "step": 7598
    },
    {
      "epoch": 14.467902995720399,
      "grad_norm": 0.27015024423599243,
      "learning_rate": 5.176881549698317e-05,
      "loss": 0.0625,
      "step": 7599
    },
    {
      "epoch": 14.46980504041845,
      "grad_norm": 0.16504183411598206,
      "learning_rate": 5.176246427437282e-05,
      "loss": 0.0597,
      "step": 7600
    },
    {
      "epoch": 14.4717070851165,
      "grad_norm": 0.1252370923757553,
      "learning_rate": 5.175611305176247e-05,
      "loss": 0.0489,
      "step": 7601
    },
    {
      "epoch": 14.47360912981455,
      "grad_norm": 0.22102156281471252,
      "learning_rate": 5.174976182915211e-05,
      "loss": 0.071,
      "step": 7602
    },
    {
      "epoch": 14.4755111745126,
      "grad_norm": 0.13126230239868164,
      "learning_rate": 5.1743410606541765e-05,
      "loss": 0.0585,
      "step": 7603
    },
    {
      "epoch": 14.477413219210652,
      "grad_norm": 0.17570582032203674,
      "learning_rate": 5.173705938393141e-05,
      "loss": 0.0693,
      "step": 7604
    },
    {
      "epoch": 14.479315263908703,
      "grad_norm": 0.20770826935768127,
      "learning_rate": 5.173070816132106e-05,
      "loss": 0.0689,
      "step": 7605
    },
    {
      "epoch": 14.481217308606752,
      "grad_norm": 0.36989128589630127,
      "learning_rate": 5.1724356938710706e-05,
      "loss": 0.0721,
      "step": 7606
    },
    {
      "epoch": 14.483119353304803,
      "grad_norm": 0.1427404284477234,
      "learning_rate": 5.171800571610035e-05,
      "loss": 0.0496,
      "step": 7607
    },
    {
      "epoch": 14.485021398002853,
      "grad_norm": 0.2404860109090805,
      "learning_rate": 5.171165449349e-05,
      "loss": 0.0785,
      "step": 7608
    },
    {
      "epoch": 14.486923442700903,
      "grad_norm": 0.18874046206474304,
      "learning_rate": 5.170530327087965e-05,
      "loss": 0.0505,
      "step": 7609
    },
    {
      "epoch": 14.488825487398953,
      "grad_norm": 0.17922984063625336,
      "learning_rate": 5.169895204826929e-05,
      "loss": 0.052,
      "step": 7610
    },
    {
      "epoch": 14.490727532097004,
      "grad_norm": 0.1344066709280014,
      "learning_rate": 5.1692600825658945e-05,
      "loss": 0.0449,
      "step": 7611
    },
    {
      "epoch": 14.492629576795055,
      "grad_norm": 0.271064817905426,
      "learning_rate": 5.1686249603048584e-05,
      "loss": 0.0413,
      "step": 7612
    },
    {
      "epoch": 14.494531621493104,
      "grad_norm": 0.29216963052749634,
      "learning_rate": 5.167989838043824e-05,
      "loss": 0.0618,
      "step": 7613
    },
    {
      "epoch": 14.496433666191155,
      "grad_norm": 0.297928124666214,
      "learning_rate": 5.167354715782789e-05,
      "loss": 0.0727,
      "step": 7614
    },
    {
      "epoch": 14.498335710889206,
      "grad_norm": 0.1421874612569809,
      "learning_rate": 5.1667195935217526e-05,
      "loss": 0.0647,
      "step": 7615
    },
    {
      "epoch": 14.500237755587257,
      "grad_norm": 0.17354488372802734,
      "learning_rate": 5.1660844712607184e-05,
      "loss": 0.0541,
      "step": 7616
    },
    {
      "epoch": 14.502139800285306,
      "grad_norm": 0.22090236842632294,
      "learning_rate": 5.165449348999682e-05,
      "loss": 0.0583,
      "step": 7617
    },
    {
      "epoch": 14.504041844983357,
      "grad_norm": 0.2978653907775879,
      "learning_rate": 5.164814226738648e-05,
      "loss": 0.0683,
      "step": 7618
    },
    {
      "epoch": 14.505943889681408,
      "grad_norm": 0.11627073585987091,
      "learning_rate": 5.164179104477612e-05,
      "loss": 0.0542,
      "step": 7619
    },
    {
      "epoch": 14.507845934379457,
      "grad_norm": 0.24588021636009216,
      "learning_rate": 5.1635439822165764e-05,
      "loss": 0.0701,
      "step": 7620
    },
    {
      "epoch": 14.509747979077508,
      "grad_norm": 0.16642743349075317,
      "learning_rate": 5.1629088599555416e-05,
      "loss": 0.0641,
      "step": 7621
    },
    {
      "epoch": 14.511650023775559,
      "grad_norm": 0.16874666512012482,
      "learning_rate": 5.162273737694506e-05,
      "loss": 0.0705,
      "step": 7622
    },
    {
      "epoch": 14.51355206847361,
      "grad_norm": 0.09476254135370255,
      "learning_rate": 5.161638615433472e-05,
      "loss": 0.0532,
      "step": 7623
    },
    {
      "epoch": 14.515454113171659,
      "grad_norm": 0.39242368936538696,
      "learning_rate": 5.161003493172436e-05,
      "loss": 0.0843,
      "step": 7624
    },
    {
      "epoch": 14.51735615786971,
      "grad_norm": 0.22530074417591095,
      "learning_rate": 5.1603683709114e-05,
      "loss": 0.0662,
      "step": 7625
    },
    {
      "epoch": 14.51925820256776,
      "grad_norm": 0.19328154623508453,
      "learning_rate": 5.1597332486503655e-05,
      "loss": 0.0612,
      "step": 7626
    },
    {
      "epoch": 14.521160247265811,
      "grad_norm": 0.2952597737312317,
      "learning_rate": 5.15909812638933e-05,
      "loss": 0.0762,
      "step": 7627
    },
    {
      "epoch": 14.52306229196386,
      "grad_norm": 0.11258865147829056,
      "learning_rate": 5.1584630041282945e-05,
      "loss": 0.0689,
      "step": 7628
    },
    {
      "epoch": 14.524964336661911,
      "grad_norm": 0.12267016619443893,
      "learning_rate": 5.15782788186726e-05,
      "loss": 0.0461,
      "step": 7629
    },
    {
      "epoch": 14.526866381359962,
      "grad_norm": 0.13929691910743713,
      "learning_rate": 5.157192759606224e-05,
      "loss": 0.0439,
      "step": 7630
    },
    {
      "epoch": 14.528768426058011,
      "grad_norm": 0.17429441213607788,
      "learning_rate": 5.1565576373451894e-05,
      "loss": 0.073,
      "step": 7631
    },
    {
      "epoch": 14.530670470756062,
      "grad_norm": 0.15891039371490479,
      "learning_rate": 5.155922515084154e-05,
      "loss": 0.0536,
      "step": 7632
    },
    {
      "epoch": 14.532572515454113,
      "grad_norm": 0.155000239610672,
      "learning_rate": 5.1552873928231184e-05,
      "loss": 0.0483,
      "step": 7633
    },
    {
      "epoch": 14.534474560152164,
      "grad_norm": 0.16273196041584015,
      "learning_rate": 5.1546522705620836e-05,
      "loss": 0.0548,
      "step": 7634
    },
    {
      "epoch": 14.536376604850213,
      "grad_norm": 0.19954794645309448,
      "learning_rate": 5.154017148301048e-05,
      "loss": 0.0578,
      "step": 7635
    },
    {
      "epoch": 14.538278649548264,
      "grad_norm": 0.2829608917236328,
      "learning_rate": 5.153382026040013e-05,
      "loss": 0.071,
      "step": 7636
    },
    {
      "epoch": 14.540180694246315,
      "grad_norm": 0.3026480972766876,
      "learning_rate": 5.152746903778978e-05,
      "loss": 0.0726,
      "step": 7637
    },
    {
      "epoch": 14.542082738944366,
      "grad_norm": 0.22207795083522797,
      "learning_rate": 5.152111781517942e-05,
      "loss": 0.0604,
      "step": 7638
    },
    {
      "epoch": 14.543984783642415,
      "grad_norm": 0.19725818932056427,
      "learning_rate": 5.1514766592569075e-05,
      "loss": 0.0538,
      "step": 7639
    },
    {
      "epoch": 14.545886828340466,
      "grad_norm": 0.31464138627052307,
      "learning_rate": 5.150841536995872e-05,
      "loss": 0.0738,
      "step": 7640
    },
    {
      "epoch": 14.547788873038517,
      "grad_norm": 0.11473199725151062,
      "learning_rate": 5.150206414734837e-05,
      "loss": 0.0624,
      "step": 7641
    },
    {
      "epoch": 14.549690917736568,
      "grad_norm": 0.23509199917316437,
      "learning_rate": 5.149571292473802e-05,
      "loss": 0.0675,
      "step": 7642
    },
    {
      "epoch": 14.551592962434617,
      "grad_norm": 0.0961693674325943,
      "learning_rate": 5.148936170212766e-05,
      "loss": 0.0411,
      "step": 7643
    },
    {
      "epoch": 14.553495007132668,
      "grad_norm": 0.14130799472332,
      "learning_rate": 5.1483010479517314e-05,
      "loss": 0.0665,
      "step": 7644
    },
    {
      "epoch": 14.555397051830719,
      "grad_norm": 0.16295112669467926,
      "learning_rate": 5.147665925690696e-05,
      "loss": 0.0545,
      "step": 7645
    },
    {
      "epoch": 14.557299096528768,
      "grad_norm": 0.19805748760700226,
      "learning_rate": 5.14703080342966e-05,
      "loss": 0.037,
      "step": 7646
    },
    {
      "epoch": 14.559201141226819,
      "grad_norm": 0.24183066189289093,
      "learning_rate": 5.1463956811686256e-05,
      "loss": 0.0659,
      "step": 7647
    },
    {
      "epoch": 14.56110318592487,
      "grad_norm": 0.14304210245609283,
      "learning_rate": 5.1457605589075894e-05,
      "loss": 0.0733,
      "step": 7648
    },
    {
      "epoch": 14.56300523062292,
      "grad_norm": 0.08641630411148071,
      "learning_rate": 5.145125436646555e-05,
      "loss": 0.0552,
      "step": 7649
    },
    {
      "epoch": 14.56490727532097,
      "grad_norm": 0.18774019181728363,
      "learning_rate": 5.144490314385519e-05,
      "loss": 0.0556,
      "step": 7650
    },
    {
      "epoch": 14.56680932001902,
      "grad_norm": 0.1754588782787323,
      "learning_rate": 5.1438551921244836e-05,
      "loss": 0.069,
      "step": 7651
    },
    {
      "epoch": 14.568711364717071,
      "grad_norm": 0.20813584327697754,
      "learning_rate": 5.1432200698634495e-05,
      "loss": 0.0744,
      "step": 7652
    },
    {
      "epoch": 14.570613409415122,
      "grad_norm": 0.22223742306232452,
      "learning_rate": 5.142584947602413e-05,
      "loss": 0.0642,
      "step": 7653
    },
    {
      "epoch": 14.572515454113171,
      "grad_norm": 0.2160334438085556,
      "learning_rate": 5.141949825341379e-05,
      "loss": 0.0521,
      "step": 7654
    },
    {
      "epoch": 14.574417498811222,
      "grad_norm": 0.37020984292030334,
      "learning_rate": 5.141314703080343e-05,
      "loss": 0.0743,
      "step": 7655
    },
    {
      "epoch": 14.576319543509273,
      "grad_norm": 0.18275423347949982,
      "learning_rate": 5.1406795808193075e-05,
      "loss": 0.0774,
      "step": 7656
    },
    {
      "epoch": 14.578221588207322,
      "grad_norm": 0.13438047468662262,
      "learning_rate": 5.140044458558273e-05,
      "loss": 0.0702,
      "step": 7657
    },
    {
      "epoch": 14.580123632905373,
      "grad_norm": 0.1055048257112503,
      "learning_rate": 5.139409336297237e-05,
      "loss": 0.0482,
      "step": 7658
    },
    {
      "epoch": 14.582025677603424,
      "grad_norm": 0.1880851536989212,
      "learning_rate": 5.138774214036203e-05,
      "loss": 0.0708,
      "step": 7659
    },
    {
      "epoch": 14.583927722301475,
      "grad_norm": 0.26509201526641846,
      "learning_rate": 5.138139091775167e-05,
      "loss": 0.0708,
      "step": 7660
    },
    {
      "epoch": 14.585829766999524,
      "grad_norm": 0.26800549030303955,
      "learning_rate": 5.1375039695141314e-05,
      "loss": 0.0592,
      "step": 7661
    },
    {
      "epoch": 14.587731811697575,
      "grad_norm": 0.16615542769432068,
      "learning_rate": 5.1368688472530966e-05,
      "loss": 0.0532,
      "step": 7662
    },
    {
      "epoch": 14.589633856395626,
      "grad_norm": 0.29129597544670105,
      "learning_rate": 5.136233724992061e-05,
      "loss": 0.0684,
      "step": 7663
    },
    {
      "epoch": 14.591535901093676,
      "grad_norm": 0.274124413728714,
      "learning_rate": 5.1355986027310256e-05,
      "loss": 0.0641,
      "step": 7664
    },
    {
      "epoch": 14.593437945791726,
      "grad_norm": 0.20047277212142944,
      "learning_rate": 5.134963480469991e-05,
      "loss": 0.0563,
      "step": 7665
    },
    {
      "epoch": 14.595339990489776,
      "grad_norm": 0.2891108989715576,
      "learning_rate": 5.134328358208955e-05,
      "loss": 0.0593,
      "step": 7666
    },
    {
      "epoch": 14.597242035187827,
      "grad_norm": 0.14130154252052307,
      "learning_rate": 5.1336932359479205e-05,
      "loss": 0.0878,
      "step": 7667
    },
    {
      "epoch": 14.599144079885878,
      "grad_norm": 0.13794413208961487,
      "learning_rate": 5.133058113686885e-05,
      "loss": 0.0724,
      "step": 7668
    },
    {
      "epoch": 14.601046124583927,
      "grad_norm": 0.08242025971412659,
      "learning_rate": 5.1324229914258495e-05,
      "loss": 0.0497,
      "step": 7669
    },
    {
      "epoch": 14.602948169281978,
      "grad_norm": 0.15497428178787231,
      "learning_rate": 5.1317878691648147e-05,
      "loss": 0.0648,
      "step": 7670
    },
    {
      "epoch": 14.60485021398003,
      "grad_norm": 0.1163494810461998,
      "learning_rate": 5.131152746903779e-05,
      "loss": 0.0691,
      "step": 7671
    },
    {
      "epoch": 14.606752258678078,
      "grad_norm": 0.1807105541229248,
      "learning_rate": 5.1305176246427443e-05,
      "loss": 0.0554,
      "step": 7672
    },
    {
      "epoch": 14.60865430337613,
      "grad_norm": 0.19974341988563538,
      "learning_rate": 5.129882502381709e-05,
      "loss": 0.0519,
      "step": 7673
    },
    {
      "epoch": 14.61055634807418,
      "grad_norm": 0.30416256189346313,
      "learning_rate": 5.1292473801206734e-05,
      "loss": 0.0661,
      "step": 7674
    },
    {
      "epoch": 14.612458392772231,
      "grad_norm": 0.1608065962791443,
      "learning_rate": 5.1286122578596385e-05,
      "loss": 0.0703,
      "step": 7675
    },
    {
      "epoch": 14.61436043747028,
      "grad_norm": 0.2223321944475174,
      "learning_rate": 5.127977135598603e-05,
      "loss": 0.0649,
      "step": 7676
    },
    {
      "epoch": 14.616262482168331,
      "grad_norm": 0.19103401899337769,
      "learning_rate": 5.127342013337568e-05,
      "loss": 0.0378,
      "step": 7677
    },
    {
      "epoch": 14.618164526866382,
      "grad_norm": 0.20109812915325165,
      "learning_rate": 5.126706891076533e-05,
      "loss": 0.0795,
      "step": 7678
    },
    {
      "epoch": 14.620066571564433,
      "grad_norm": 0.09337878972291946,
      "learning_rate": 5.1260717688154966e-05,
      "loss": 0.0605,
      "step": 7679
    },
    {
      "epoch": 14.621968616262482,
      "grad_norm": 0.20119719207286835,
      "learning_rate": 5.1254366465544624e-05,
      "loss": 0.0554,
      "step": 7680
    },
    {
      "epoch": 14.623870660960533,
      "grad_norm": 0.1166246086359024,
      "learning_rate": 5.124801524293427e-05,
      "loss": 0.0616,
      "step": 7681
    },
    {
      "epoch": 14.625772705658584,
      "grad_norm": 0.1417192667722702,
      "learning_rate": 5.124166402032391e-05,
      "loss": 0.0679,
      "step": 7682
    },
    {
      "epoch": 14.627674750356633,
      "grad_norm": 0.2099301666021347,
      "learning_rate": 5.1235312797713566e-05,
      "loss": 0.0563,
      "step": 7683
    },
    {
      "epoch": 14.629576795054684,
      "grad_norm": 0.1745438575744629,
      "learning_rate": 5.1228961575103205e-05,
      "loss": 0.0482,
      "step": 7684
    },
    {
      "epoch": 14.631478839752734,
      "grad_norm": 0.1499902904033661,
      "learning_rate": 5.122261035249286e-05,
      "loss": 0.0668,
      "step": 7685
    },
    {
      "epoch": 14.633380884450785,
      "grad_norm": 0.15425054728984833,
      "learning_rate": 5.12162591298825e-05,
      "loss": 0.0446,
      "step": 7686
    },
    {
      "epoch": 14.635282929148834,
      "grad_norm": 0.2625950276851654,
      "learning_rate": 5.1209907907272147e-05,
      "loss": 0.0621,
      "step": 7687
    },
    {
      "epoch": 14.637184973846885,
      "grad_norm": 0.1606355756521225,
      "learning_rate": 5.1203556684661805e-05,
      "loss": 0.0551,
      "step": 7688
    },
    {
      "epoch": 14.639087018544936,
      "grad_norm": 0.16646242141723633,
      "learning_rate": 5.1197205462051443e-05,
      "loss": 0.0491,
      "step": 7689
    },
    {
      "epoch": 14.640989063242987,
      "grad_norm": 0.26278549432754517,
      "learning_rate": 5.11908542394411e-05,
      "loss": 0.0761,
      "step": 7690
    },
    {
      "epoch": 14.642891107941036,
      "grad_norm": 0.11060616374015808,
      "learning_rate": 5.118450301683074e-05,
      "loss": 0.0639,
      "step": 7691
    },
    {
      "epoch": 14.644793152639087,
      "grad_norm": 0.2710588276386261,
      "learning_rate": 5.1178151794220385e-05,
      "loss": 0.0704,
      "step": 7692
    },
    {
      "epoch": 14.646695197337138,
      "grad_norm": 0.2786422371864319,
      "learning_rate": 5.117180057161004e-05,
      "loss": 0.0718,
      "step": 7693
    },
    {
      "epoch": 14.648597242035187,
      "grad_norm": 0.16812217235565186,
      "learning_rate": 5.116544934899968e-05,
      "loss": 0.0705,
      "step": 7694
    },
    {
      "epoch": 14.650499286733238,
      "grad_norm": 0.14156568050384521,
      "learning_rate": 5.115909812638934e-05,
      "loss": 0.0712,
      "step": 7695
    },
    {
      "epoch": 14.652401331431289,
      "grad_norm": 0.1136791855096817,
      "learning_rate": 5.115274690377898e-05,
      "loss": 0.0477,
      "step": 7696
    },
    {
      "epoch": 14.65430337612934,
      "grad_norm": 0.3072340488433838,
      "learning_rate": 5.1146395681168624e-05,
      "loss": 0.0911,
      "step": 7697
    },
    {
      "epoch": 14.656205420827389,
      "grad_norm": 0.2108566164970398,
      "learning_rate": 5.1140044458558276e-05,
      "loss": 0.0518,
      "step": 7698
    },
    {
      "epoch": 14.65810746552544,
      "grad_norm": 0.20374548435211182,
      "learning_rate": 5.113369323594792e-05,
      "loss": 0.0459,
      "step": 7699
    },
    {
      "epoch": 14.66000951022349,
      "grad_norm": 0.20516937971115112,
      "learning_rate": 5.1127342013337566e-05,
      "loss": 0.0601,
      "step": 7700
    },
    {
      "epoch": 14.661911554921542,
      "grad_norm": 0.19593757390975952,
      "learning_rate": 5.112099079072722e-05,
      "loss": 0.0688,
      "step": 7701
    },
    {
      "epoch": 14.66381359961959,
      "grad_norm": 0.13337576389312744,
      "learning_rate": 5.111463956811686e-05,
      "loss": 0.0534,
      "step": 7702
    },
    {
      "epoch": 14.665715644317642,
      "grad_norm": 0.1885136514902115,
      "learning_rate": 5.1108288345506515e-05,
      "loss": 0.0588,
      "step": 7703
    },
    {
      "epoch": 14.667617689015692,
      "grad_norm": 0.2416466921567917,
      "learning_rate": 5.110193712289616e-05,
      "loss": 0.0813,
      "step": 7704
    },
    {
      "epoch": 14.669519733713742,
      "grad_norm": 0.13200879096984863,
      "learning_rate": 5.1095585900285805e-05,
      "loss": 0.063,
      "step": 7705
    },
    {
      "epoch": 14.671421778411792,
      "grad_norm": 0.2144845873117447,
      "learning_rate": 5.108923467767546e-05,
      "loss": 0.0629,
      "step": 7706
    },
    {
      "epoch": 14.673323823109843,
      "grad_norm": 0.12554292380809784,
      "learning_rate": 5.10828834550651e-05,
      "loss": 0.0569,
      "step": 7707
    },
    {
      "epoch": 14.675225867807894,
      "grad_norm": 0.18374310433864594,
      "learning_rate": 5.1076532232454754e-05,
      "loss": 0.0688,
      "step": 7708
    },
    {
      "epoch": 14.677127912505943,
      "grad_norm": 0.17229317128658295,
      "learning_rate": 5.10701810098444e-05,
      "loss": 0.0593,
      "step": 7709
    },
    {
      "epoch": 14.679029957203994,
      "grad_norm": 0.18080811202526093,
      "learning_rate": 5.1063829787234044e-05,
      "loss": 0.0567,
      "step": 7710
    },
    {
      "epoch": 14.680932001902045,
      "grad_norm": 0.19398432970046997,
      "learning_rate": 5.1057478564623696e-05,
      "loss": 0.0641,
      "step": 7711
    },
    {
      "epoch": 14.682834046600096,
      "grad_norm": 0.37756291031837463,
      "learning_rate": 5.105112734201334e-05,
      "loss": 0.0759,
      "step": 7712
    },
    {
      "epoch": 14.684736091298145,
      "grad_norm": 0.1937006264925003,
      "learning_rate": 5.104477611940299e-05,
      "loss": 0.0867,
      "step": 7713
    },
    {
      "epoch": 14.686638135996196,
      "grad_norm": 0.20375970005989075,
      "learning_rate": 5.103842489679264e-05,
      "loss": 0.0539,
      "step": 7714
    },
    {
      "epoch": 14.688540180694247,
      "grad_norm": 0.19085298478603363,
      "learning_rate": 5.1032073674182276e-05,
      "loss": 0.0603,
      "step": 7715
    },
    {
      "epoch": 14.690442225392296,
      "grad_norm": 0.24104918539524078,
      "learning_rate": 5.1025722451571935e-05,
      "loss": 0.0486,
      "step": 7716
    },
    {
      "epoch": 14.692344270090347,
      "grad_norm": 0.279154509305954,
      "learning_rate": 5.101937122896158e-05,
      "loss": 0.0793,
      "step": 7717
    },
    {
      "epoch": 14.694246314788398,
      "grad_norm": 0.1371929794549942,
      "learning_rate": 5.101302000635122e-05,
      "loss": 0.0387,
      "step": 7718
    },
    {
      "epoch": 14.696148359486449,
      "grad_norm": 0.22420379519462585,
      "learning_rate": 5.100666878374088e-05,
      "loss": 0.0554,
      "step": 7719
    },
    {
      "epoch": 14.698050404184498,
      "grad_norm": 0.3003183901309967,
      "learning_rate": 5.1000317561130515e-05,
      "loss": 0.0717,
      "step": 7720
    },
    {
      "epoch": 14.699952448882549,
      "grad_norm": 0.18145035207271576,
      "learning_rate": 5.0993966338520174e-05,
      "loss": 0.0686,
      "step": 7721
    },
    {
      "epoch": 14.7018544935806,
      "grad_norm": 0.23177501559257507,
      "learning_rate": 5.098761511590981e-05,
      "loss": 0.0844,
      "step": 7722
    },
    {
      "epoch": 14.70375653827865,
      "grad_norm": 0.213920459151268,
      "learning_rate": 5.098126389329946e-05,
      "loss": 0.0715,
      "step": 7723
    },
    {
      "epoch": 14.7056585829767,
      "grad_norm": 0.2284982055425644,
      "learning_rate": 5.0974912670689116e-05,
      "loss": 0.0515,
      "step": 7724
    },
    {
      "epoch": 14.70756062767475,
      "grad_norm": 0.2825984060764313,
      "learning_rate": 5.0968561448078754e-05,
      "loss": 0.0701,
      "step": 7725
    },
    {
      "epoch": 14.709462672372801,
      "grad_norm": 0.19270162284374237,
      "learning_rate": 5.096221022546841e-05,
      "loss": 0.0891,
      "step": 7726
    },
    {
      "epoch": 14.71136471707085,
      "grad_norm": 0.12928995490074158,
      "learning_rate": 5.095585900285805e-05,
      "loss": 0.054,
      "step": 7727
    },
    {
      "epoch": 14.713266761768901,
      "grad_norm": 0.1467064470052719,
      "learning_rate": 5.0949507780247696e-05,
      "loss": 0.0634,
      "step": 7728
    },
    {
      "epoch": 14.715168806466952,
      "grad_norm": 0.28742527961730957,
      "learning_rate": 5.094315655763735e-05,
      "loss": 0.0631,
      "step": 7729
    },
    {
      "epoch": 14.717070851165003,
      "grad_norm": 0.22021564841270447,
      "learning_rate": 5.093680533502699e-05,
      "loss": 0.0653,
      "step": 7730
    },
    {
      "epoch": 14.718972895863052,
      "grad_norm": 0.21744635701179504,
      "learning_rate": 5.0930454112416645e-05,
      "loss": 0.0552,
      "step": 7731
    },
    {
      "epoch": 14.720874940561103,
      "grad_norm": 0.10791151225566864,
      "learning_rate": 5.092410288980629e-05,
      "loss": 0.0515,
      "step": 7732
    },
    {
      "epoch": 14.722776985259154,
      "grad_norm": 0.16238145530223846,
      "learning_rate": 5.0917751667195935e-05,
      "loss": 0.0537,
      "step": 7733
    },
    {
      "epoch": 14.724679029957205,
      "grad_norm": 0.22757276892662048,
      "learning_rate": 5.0911400444585587e-05,
      "loss": 0.0653,
      "step": 7734
    },
    {
      "epoch": 14.726581074655254,
      "grad_norm": 0.17570172250270844,
      "learning_rate": 5.090504922197523e-05,
      "loss": 0.0597,
      "step": 7735
    },
    {
      "epoch": 14.728483119353305,
      "grad_norm": 0.174399271607399,
      "learning_rate": 5.089869799936488e-05,
      "loss": 0.0593,
      "step": 7736
    },
    {
      "epoch": 14.730385164051356,
      "grad_norm": 0.14591853320598602,
      "learning_rate": 5.089234677675453e-05,
      "loss": 0.0717,
      "step": 7737
    },
    {
      "epoch": 14.732287208749405,
      "grad_norm": 0.17062704265117645,
      "learning_rate": 5.0885995554144174e-05,
      "loss": 0.0635,
      "step": 7738
    },
    {
      "epoch": 14.734189253447456,
      "grad_norm": 0.24444295465946198,
      "learning_rate": 5.0879644331533825e-05,
      "loss": 0.0645,
      "step": 7739
    },
    {
      "epoch": 14.736091298145507,
      "grad_norm": 0.1436753273010254,
      "learning_rate": 5.087329310892347e-05,
      "loss": 0.0571,
      "step": 7740
    },
    {
      "epoch": 14.737993342843557,
      "grad_norm": 0.18597128987312317,
      "learning_rate": 5.0866941886313116e-05,
      "loss": 0.0626,
      "step": 7741
    },
    {
      "epoch": 14.739895387541607,
      "grad_norm": 0.1565275639295578,
      "learning_rate": 5.086059066370277e-05,
      "loss": 0.0503,
      "step": 7742
    },
    {
      "epoch": 14.741797432239657,
      "grad_norm": 0.20083200931549072,
      "learning_rate": 5.085423944109241e-05,
      "loss": 0.0663,
      "step": 7743
    },
    {
      "epoch": 14.743699476937708,
      "grad_norm": 0.17310886085033417,
      "learning_rate": 5.0847888218482064e-05,
      "loss": 0.0729,
      "step": 7744
    },
    {
      "epoch": 14.74560152163576,
      "grad_norm": 0.1712331473827362,
      "learning_rate": 5.084153699587171e-05,
      "loss": 0.0628,
      "step": 7745
    },
    {
      "epoch": 14.747503566333808,
      "grad_norm": 0.29841336607933044,
      "learning_rate": 5.0835185773261354e-05,
      "loss": 0.0576,
      "step": 7746
    },
    {
      "epoch": 14.74940561103186,
      "grad_norm": 0.27211645245552063,
      "learning_rate": 5.0828834550651006e-05,
      "loss": 0.05,
      "step": 7747
    },
    {
      "epoch": 14.75130765572991,
      "grad_norm": 0.1731298565864563,
      "learning_rate": 5.082248332804065e-05,
      "loss": 0.0679,
      "step": 7748
    },
    {
      "epoch": 14.75320970042796,
      "grad_norm": 0.1546994149684906,
      "learning_rate": 5.08161321054303e-05,
      "loss": 0.0518,
      "step": 7749
    },
    {
      "epoch": 14.75511174512601,
      "grad_norm": 0.20409636199474335,
      "learning_rate": 5.080978088281995e-05,
      "loss": 0.0883,
      "step": 7750
    },
    {
      "epoch": 14.757013789824061,
      "grad_norm": 0.194264218211174,
      "learning_rate": 5.0803429660209587e-05,
      "loss": 0.0677,
      "step": 7751
    },
    {
      "epoch": 14.758915834522112,
      "grad_norm": 0.209025040268898,
      "learning_rate": 5.0797078437599245e-05,
      "loss": 0.0578,
      "step": 7752
    },
    {
      "epoch": 14.760817879220161,
      "grad_norm": 0.24536709487438202,
      "learning_rate": 5.079072721498889e-05,
      "loss": 0.051,
      "step": 7753
    },
    {
      "epoch": 14.762719923918212,
      "grad_norm": 0.2700499892234802,
      "learning_rate": 5.078437599237853e-05,
      "loss": 0.0642,
      "step": 7754
    },
    {
      "epoch": 14.764621968616263,
      "grad_norm": 0.16249141097068787,
      "learning_rate": 5.077802476976819e-05,
      "loss": 0.061,
      "step": 7755
    },
    {
      "epoch": 14.766524013314314,
      "grad_norm": 0.10296684503555298,
      "learning_rate": 5.0771673547157825e-05,
      "loss": 0.0474,
      "step": 7756
    },
    {
      "epoch": 14.768426058012363,
      "grad_norm": 0.29506006836891174,
      "learning_rate": 5.0765322324547484e-05,
      "loss": 0.0731,
      "step": 7757
    },
    {
      "epoch": 14.770328102710414,
      "grad_norm": 0.21537403762340546,
      "learning_rate": 5.075897110193712e-05,
      "loss": 0.0649,
      "step": 7758
    },
    {
      "epoch": 14.772230147408465,
      "grad_norm": 0.2675336003303528,
      "learning_rate": 5.075261987932677e-05,
      "loss": 0.0781,
      "step": 7759
    },
    {
      "epoch": 14.774132192106514,
      "grad_norm": 0.14951659739017487,
      "learning_rate": 5.074626865671642e-05,
      "loss": 0.055,
      "step": 7760
    },
    {
      "epoch": 14.776034236804565,
      "grad_norm": 0.13454633951187134,
      "learning_rate": 5.0739917434106064e-05,
      "loss": 0.0498,
      "step": 7761
    },
    {
      "epoch": 14.777936281502615,
      "grad_norm": 0.08589652925729752,
      "learning_rate": 5.073356621149572e-05,
      "loss": 0.0652,
      "step": 7762
    },
    {
      "epoch": 14.779838326200666,
      "grad_norm": 0.1903010755777359,
      "learning_rate": 5.072721498888536e-05,
      "loss": 0.0731,
      "step": 7763
    },
    {
      "epoch": 14.781740370898715,
      "grad_norm": 0.15340054035186768,
      "learning_rate": 5.0720863766275006e-05,
      "loss": 0.0498,
      "step": 7764
    },
    {
      "epoch": 14.783642415596766,
      "grad_norm": 0.1111394464969635,
      "learning_rate": 5.071451254366466e-05,
      "loss": 0.033,
      "step": 7765
    },
    {
      "epoch": 14.785544460294817,
      "grad_norm": 0.1255098134279251,
      "learning_rate": 5.07081613210543e-05,
      "loss": 0.0609,
      "step": 7766
    },
    {
      "epoch": 14.787446504992868,
      "grad_norm": 0.26106560230255127,
      "learning_rate": 5.0701810098443955e-05,
      "loss": 0.0579,
      "step": 7767
    },
    {
      "epoch": 14.789348549690917,
      "grad_norm": 0.23423247039318085,
      "learning_rate": 5.06954588758336e-05,
      "loss": 0.0514,
      "step": 7768
    },
    {
      "epoch": 14.791250594388968,
      "grad_norm": 0.12845613062381744,
      "learning_rate": 5.0689107653223245e-05,
      "loss": 0.055,
      "step": 7769
    },
    {
      "epoch": 14.793152639087019,
      "grad_norm": 0.17294085025787354,
      "learning_rate": 5.06827564306129e-05,
      "loss": 0.066,
      "step": 7770
    },
    {
      "epoch": 14.795054683785068,
      "grad_norm": 0.18765872716903687,
      "learning_rate": 5.067640520800254e-05,
      "loss": 0.0577,
      "step": 7771
    },
    {
      "epoch": 14.796956728483119,
      "grad_norm": 0.17650574445724487,
      "learning_rate": 5.067005398539219e-05,
      "loss": 0.0841,
      "step": 7772
    },
    {
      "epoch": 14.79885877318117,
      "grad_norm": 0.19177024066448212,
      "learning_rate": 5.066370276278184e-05,
      "loss": 0.0681,
      "step": 7773
    },
    {
      "epoch": 14.80076081787922,
      "grad_norm": 0.2309020757675171,
      "learning_rate": 5.0657351540171484e-05,
      "loss": 0.0791,
      "step": 7774
    },
    {
      "epoch": 14.80266286257727,
      "grad_norm": 0.17799893021583557,
      "learning_rate": 5.0651000317561136e-05,
      "loss": 0.0553,
      "step": 7775
    },
    {
      "epoch": 14.80456490727532,
      "grad_norm": 0.25767970085144043,
      "learning_rate": 5.064464909495078e-05,
      "loss": 0.0489,
      "step": 7776
    },
    {
      "epoch": 14.806466951973372,
      "grad_norm": 0.11283757537603378,
      "learning_rate": 5.0638297872340426e-05,
      "loss": 0.0544,
      "step": 7777
    },
    {
      "epoch": 14.808368996671422,
      "grad_norm": 0.15177033841609955,
      "learning_rate": 5.063194664973008e-05,
      "loss": 0.0546,
      "step": 7778
    },
    {
      "epoch": 14.810271041369472,
      "grad_norm": 0.1496088206768036,
      "learning_rate": 5.062559542711972e-05,
      "loss": 0.0604,
      "step": 7779
    },
    {
      "epoch": 14.812173086067522,
      "grad_norm": 0.21631433069705963,
      "learning_rate": 5.0619244204509375e-05,
      "loss": 0.0499,
      "step": 7780
    },
    {
      "epoch": 14.814075130765573,
      "grad_norm": 0.22041334211826324,
      "learning_rate": 5.061289298189902e-05,
      "loss": 0.0393,
      "step": 7781
    },
    {
      "epoch": 14.815977175463622,
      "grad_norm": 0.18054498732089996,
      "learning_rate": 5.060654175928866e-05,
      "loss": 0.0626,
      "step": 7782
    },
    {
      "epoch": 14.817879220161673,
      "grad_norm": 0.22049325704574585,
      "learning_rate": 5.060019053667832e-05,
      "loss": 0.0625,
      "step": 7783
    },
    {
      "epoch": 14.819781264859724,
      "grad_norm": 0.19691422581672668,
      "learning_rate": 5.059383931406796e-05,
      "loss": 0.0509,
      "step": 7784
    },
    {
      "epoch": 14.821683309557775,
      "grad_norm": 0.20794346928596497,
      "learning_rate": 5.0587488091457614e-05,
      "loss": 0.0625,
      "step": 7785
    },
    {
      "epoch": 14.823585354255824,
      "grad_norm": 0.18965652585029602,
      "learning_rate": 5.058113686884726e-05,
      "loss": 0.0545,
      "step": 7786
    },
    {
      "epoch": 14.825487398953875,
      "grad_norm": 0.13831253349781036,
      "learning_rate": 5.05747856462369e-05,
      "loss": 0.0694,
      "step": 7787
    },
    {
      "epoch": 14.827389443651926,
      "grad_norm": 0.34300696849823,
      "learning_rate": 5.0568434423626556e-05,
      "loss": 0.0767,
      "step": 7788
    },
    {
      "epoch": 14.829291488349977,
      "grad_norm": 0.2049691379070282,
      "learning_rate": 5.0562083201016194e-05,
      "loss": 0.0651,
      "step": 7789
    },
    {
      "epoch": 14.831193533048026,
      "grad_norm": 0.11393442749977112,
      "learning_rate": 5.055573197840584e-05,
      "loss": 0.0624,
      "step": 7790
    },
    {
      "epoch": 14.833095577746077,
      "grad_norm": 0.17523103952407837,
      "learning_rate": 5.05493807557955e-05,
      "loss": 0.0665,
      "step": 7791
    },
    {
      "epoch": 14.834997622444128,
      "grad_norm": 0.24641826748847961,
      "learning_rate": 5.0543029533185136e-05,
      "loss": 0.0644,
      "step": 7792
    },
    {
      "epoch": 14.836899667142179,
      "grad_norm": 0.24968832731246948,
      "learning_rate": 5.0536678310574794e-05,
      "loss": 0.0612,
      "step": 7793
    },
    {
      "epoch": 14.838801711840228,
      "grad_norm": 0.18396158516407013,
      "learning_rate": 5.053032708796443e-05,
      "loss": 0.0635,
      "step": 7794
    },
    {
      "epoch": 14.840703756538279,
      "grad_norm": 0.21605823934078217,
      "learning_rate": 5.052397586535408e-05,
      "loss": 0.0754,
      "step": 7795
    },
    {
      "epoch": 14.84260580123633,
      "grad_norm": 0.08425106108188629,
      "learning_rate": 5.051762464274373e-05,
      "loss": 0.054,
      "step": 7796
    },
    {
      "epoch": 14.844507845934379,
      "grad_norm": 0.13125087320804596,
      "learning_rate": 5.0511273420133375e-05,
      "loss": 0.0476,
      "step": 7797
    },
    {
      "epoch": 14.84640989063243,
      "grad_norm": 0.33811628818511963,
      "learning_rate": 5.050492219752303e-05,
      "loss": 0.0699,
      "step": 7798
    },
    {
      "epoch": 14.84831193533048,
      "grad_norm": 0.16803185641765594,
      "learning_rate": 5.049857097491267e-05,
      "loss": 0.0601,
      "step": 7799
    },
    {
      "epoch": 14.850213980028531,
      "grad_norm": 0.16089555621147156,
      "learning_rate": 5.049221975230232e-05,
      "loss": 0.0646,
      "step": 7800
    },
    {
      "epoch": 14.85211602472658,
      "grad_norm": 0.1432085484266281,
      "learning_rate": 5.048586852969197e-05,
      "loss": 0.0605,
      "step": 7801
    },
    {
      "epoch": 14.854018069424631,
      "grad_norm": 0.1569204181432724,
      "learning_rate": 5.0479517307081614e-05,
      "loss": 0.056,
      "step": 7802
    },
    {
      "epoch": 14.855920114122682,
      "grad_norm": 0.28553643822669983,
      "learning_rate": 5.0473166084471265e-05,
      "loss": 0.062,
      "step": 7803
    },
    {
      "epoch": 14.857822158820733,
      "grad_norm": 0.1551143079996109,
      "learning_rate": 5.046681486186091e-05,
      "loss": 0.0694,
      "step": 7804
    },
    {
      "epoch": 14.859724203518782,
      "grad_norm": 0.21383574604988098,
      "learning_rate": 5.0460463639250556e-05,
      "loss": 0.0762,
      "step": 7805
    },
    {
      "epoch": 14.861626248216833,
      "grad_norm": 0.14724215865135193,
      "learning_rate": 5.045411241664021e-05,
      "loss": 0.0593,
      "step": 7806
    },
    {
      "epoch": 14.863528292914884,
      "grad_norm": 0.1819068044424057,
      "learning_rate": 5.044776119402985e-05,
      "loss": 0.0648,
      "step": 7807
    },
    {
      "epoch": 14.865430337612933,
      "grad_norm": 0.2799537479877472,
      "learning_rate": 5.04414099714195e-05,
      "loss": 0.0537,
      "step": 7808
    },
    {
      "epoch": 14.867332382310984,
      "grad_norm": 0.2654231786727905,
      "learning_rate": 5.043505874880915e-05,
      "loss": 0.0573,
      "step": 7809
    },
    {
      "epoch": 14.869234427009035,
      "grad_norm": 0.2610599994659424,
      "learning_rate": 5.0428707526198794e-05,
      "loss": 0.065,
      "step": 7810
    },
    {
      "epoch": 14.871136471707086,
      "grad_norm": 0.19513475894927979,
      "learning_rate": 5.0422356303588446e-05,
      "loss": 0.0696,
      "step": 7811
    },
    {
      "epoch": 14.873038516405135,
      "grad_norm": 0.1932954043149948,
      "learning_rate": 5.041600508097809e-05,
      "loss": 0.0477,
      "step": 7812
    },
    {
      "epoch": 14.874940561103186,
      "grad_norm": 0.19325312972068787,
      "learning_rate": 5.0409653858367736e-05,
      "loss": 0.0793,
      "step": 7813
    },
    {
      "epoch": 14.876842605801237,
      "grad_norm": 0.14803767204284668,
      "learning_rate": 5.040330263575739e-05,
      "loss": 0.057,
      "step": 7814
    },
    {
      "epoch": 14.878744650499288,
      "grad_norm": 0.18834733963012695,
      "learning_rate": 5.039695141314703e-05,
      "loss": 0.0967,
      "step": 7815
    },
    {
      "epoch": 14.880646695197337,
      "grad_norm": 0.21755129098892212,
      "learning_rate": 5.0390600190536685e-05,
      "loss": 0.0525,
      "step": 7816
    },
    {
      "epoch": 14.882548739895388,
      "grad_norm": 0.12757804989814758,
      "learning_rate": 5.038424896792633e-05,
      "loss": 0.0598,
      "step": 7817
    },
    {
      "epoch": 14.884450784593438,
      "grad_norm": 0.29833969473838806,
      "learning_rate": 5.037789774531597e-05,
      "loss": 0.0691,
      "step": 7818
    },
    {
      "epoch": 14.88635282929149,
      "grad_norm": 0.20505289733409882,
      "learning_rate": 5.037154652270563e-05,
      "loss": 0.0533,
      "step": 7819
    },
    {
      "epoch": 14.888254873989538,
      "grad_norm": 0.25543922185897827,
      "learning_rate": 5.036519530009527e-05,
      "loss": 0.0661,
      "step": 7820
    },
    {
      "epoch": 14.89015691868759,
      "grad_norm": 0.3089730441570282,
      "learning_rate": 5.0358844077484924e-05,
      "loss": 0.0728,
      "step": 7821
    },
    {
      "epoch": 14.89205896338564,
      "grad_norm": 0.14130769670009613,
      "learning_rate": 5.035249285487457e-05,
      "loss": 0.0575,
      "step": 7822
    },
    {
      "epoch": 14.89396100808369,
      "grad_norm": 0.24397386610507965,
      "learning_rate": 5.034614163226421e-05,
      "loss": 0.0647,
      "step": 7823
    },
    {
      "epoch": 14.89586305278174,
      "grad_norm": 0.18317031860351562,
      "learning_rate": 5.0339790409653866e-05,
      "loss": 0.0417,
      "step": 7824
    },
    {
      "epoch": 14.897765097479791,
      "grad_norm": 0.2685723602771759,
      "learning_rate": 5.0333439187043504e-05,
      "loss": 0.095,
      "step": 7825
    },
    {
      "epoch": 14.899667142177842,
      "grad_norm": 0.3372928500175476,
      "learning_rate": 5.032708796443315e-05,
      "loss": 0.0587,
      "step": 7826
    },
    {
      "epoch": 14.901569186875891,
      "grad_norm": 0.4275154769420624,
      "learning_rate": 5.032073674182281e-05,
      "loss": 0.0754,
      "step": 7827
    },
    {
      "epoch": 14.903471231573942,
      "grad_norm": 0.15616847574710846,
      "learning_rate": 5.0314385519212446e-05,
      "loss": 0.0596,
      "step": 7828
    },
    {
      "epoch": 14.905373276271993,
      "grad_norm": 0.12086091190576553,
      "learning_rate": 5.0308034296602105e-05,
      "loss": 0.0386,
      "step": 7829
    },
    {
      "epoch": 14.907275320970044,
      "grad_norm": 0.16133488714694977,
      "learning_rate": 5.030168307399174e-05,
      "loss": 0.0453,
      "step": 7830
    },
    {
      "epoch": 14.909177365668093,
      "grad_norm": 0.43575403094291687,
      "learning_rate": 5.029533185138139e-05,
      "loss": 0.0724,
      "step": 7831
    },
    {
      "epoch": 14.911079410366144,
      "grad_norm": 0.4218147099018097,
      "learning_rate": 5.028898062877104e-05,
      "loss": 0.0725,
      "step": 7832
    },
    {
      "epoch": 14.912981455064195,
      "grad_norm": 0.2845557630062103,
      "learning_rate": 5.0282629406160685e-05,
      "loss": 0.0721,
      "step": 7833
    },
    {
      "epoch": 14.914883499762244,
      "grad_norm": 0.16808871924877167,
      "learning_rate": 5.0276278183550344e-05,
      "loss": 0.0776,
      "step": 7834
    },
    {
      "epoch": 14.916785544460295,
      "grad_norm": 0.3421449661254883,
      "learning_rate": 5.026992696093998e-05,
      "loss": 0.0713,
      "step": 7835
    },
    {
      "epoch": 14.918687589158345,
      "grad_norm": 0.22598189115524292,
      "learning_rate": 5.026357573832963e-05,
      "loss": 0.0606,
      "step": 7836
    },
    {
      "epoch": 14.920589633856396,
      "grad_norm": 0.2761426568031311,
      "learning_rate": 5.025722451571928e-05,
      "loss": 0.0753,
      "step": 7837
    },
    {
      "epoch": 14.922491678554445,
      "grad_norm": 0.2742650806903839,
      "learning_rate": 5.0250873293108924e-05,
      "loss": 0.1,
      "step": 7838
    },
    {
      "epoch": 14.924393723252496,
      "grad_norm": 0.12055804580450058,
      "learning_rate": 5.0244522070498576e-05,
      "loss": 0.0651,
      "step": 7839
    },
    {
      "epoch": 14.926295767950547,
      "grad_norm": 0.1872662454843521,
      "learning_rate": 5.023817084788822e-05,
      "loss": 0.056,
      "step": 7840
    },
    {
      "epoch": 14.928197812648598,
      "grad_norm": 0.19098106026649475,
      "learning_rate": 5.0231819625277866e-05,
      "loss": 0.0653,
      "step": 7841
    },
    {
      "epoch": 14.930099857346647,
      "grad_norm": 0.18531818687915802,
      "learning_rate": 5.022546840266752e-05,
      "loss": 0.0553,
      "step": 7842
    },
    {
      "epoch": 14.932001902044698,
      "grad_norm": 0.48915934562683105,
      "learning_rate": 5.021911718005716e-05,
      "loss": 0.076,
      "step": 7843
    },
    {
      "epoch": 14.933903946742749,
      "grad_norm": 0.18385091423988342,
      "learning_rate": 5.021276595744681e-05,
      "loss": 0.1163,
      "step": 7844
    },
    {
      "epoch": 14.935805991440798,
      "grad_norm": 0.19059492647647858,
      "learning_rate": 5.020641473483646e-05,
      "loss": 0.0706,
      "step": 7845
    },
    {
      "epoch": 14.937708036138849,
      "grad_norm": 0.17051951587200165,
      "learning_rate": 5.0200063512226105e-05,
      "loss": 0.0592,
      "step": 7846
    },
    {
      "epoch": 14.9396100808369,
      "grad_norm": 0.13963809609413147,
      "learning_rate": 5.019371228961576e-05,
      "loss": 0.038,
      "step": 7847
    },
    {
      "epoch": 14.94151212553495,
      "grad_norm": 0.16215941309928894,
      "learning_rate": 5.01873610670054e-05,
      "loss": 0.0542,
      "step": 7848
    },
    {
      "epoch": 14.943414170233,
      "grad_norm": 0.25809141993522644,
      "learning_rate": 5.018100984439505e-05,
      "loss": 0.095,
      "step": 7849
    },
    {
      "epoch": 14.94531621493105,
      "grad_norm": 0.16990099847316742,
      "learning_rate": 5.01746586217847e-05,
      "loss": 0.0739,
      "step": 7850
    },
    {
      "epoch": 14.947218259629102,
      "grad_norm": 0.22535528242588043,
      "learning_rate": 5.0168307399174344e-05,
      "loss": 0.065,
      "step": 7851
    },
    {
      "epoch": 14.949120304327153,
      "grad_norm": 0.13045746088027954,
      "learning_rate": 5.0161956176563996e-05,
      "loss": 0.0592,
      "step": 7852
    },
    {
      "epoch": 14.951022349025202,
      "grad_norm": 0.14321979880332947,
      "learning_rate": 5.015560495395364e-05,
      "loss": 0.0344,
      "step": 7853
    },
    {
      "epoch": 14.952924393723253,
      "grad_norm": 0.20798403024673462,
      "learning_rate": 5.014925373134328e-05,
      "loss": 0.0543,
      "step": 7854
    },
    {
      "epoch": 14.954826438421303,
      "grad_norm": 0.35874882340431213,
      "learning_rate": 5.014290250873294e-05,
      "loss": 0.0715,
      "step": 7855
    },
    {
      "epoch": 14.956728483119353,
      "grad_norm": 0.2704923748970032,
      "learning_rate": 5.013655128612258e-05,
      "loss": 0.0597,
      "step": 7856
    },
    {
      "epoch": 14.958630527817403,
      "grad_norm": 0.18080563843250275,
      "learning_rate": 5.0130200063512235e-05,
      "loss": 0.0629,
      "step": 7857
    },
    {
      "epoch": 14.960532572515454,
      "grad_norm": 0.3157869577407837,
      "learning_rate": 5.012384884090188e-05,
      "loss": 0.0878,
      "step": 7858
    },
    {
      "epoch": 14.962434617213505,
      "grad_norm": 0.3106921315193176,
      "learning_rate": 5.011749761829152e-05,
      "loss": 0.0566,
      "step": 7859
    },
    {
      "epoch": 14.964336661911554,
      "grad_norm": 0.20733977854251862,
      "learning_rate": 5.0111146395681176e-05,
      "loss": 0.0507,
      "step": 7860
    },
    {
      "epoch": 14.966238706609605,
      "grad_norm": 0.14127065241336823,
      "learning_rate": 5.0104795173070815e-05,
      "loss": 0.0591,
      "step": 7861
    },
    {
      "epoch": 14.968140751307656,
      "grad_norm": 0.1346455067396164,
      "learning_rate": 5.009844395046046e-05,
      "loss": 0.0573,
      "step": 7862
    },
    {
      "epoch": 14.970042796005707,
      "grad_norm": 0.18170590698719025,
      "learning_rate": 5.009209272785012e-05,
      "loss": 0.0428,
      "step": 7863
    },
    {
      "epoch": 14.971944840703756,
      "grad_norm": 0.32324326038360596,
      "learning_rate": 5.008574150523976e-05,
      "loss": 0.0801,
      "step": 7864
    },
    {
      "epoch": 14.973846885401807,
      "grad_norm": 0.1589668244123459,
      "learning_rate": 5.0079390282629415e-05,
      "loss": 0.0513,
      "step": 7865
    },
    {
      "epoch": 14.975748930099858,
      "grad_norm": 0.1323167234659195,
      "learning_rate": 5.0073039060019054e-05,
      "loss": 0.0581,
      "step": 7866
    },
    {
      "epoch": 14.977650974797907,
      "grad_norm": 0.28893229365348816,
      "learning_rate": 5.00666878374087e-05,
      "loss": 0.0721,
      "step": 7867
    },
    {
      "epoch": 14.979553019495958,
      "grad_norm": 0.2315034717321396,
      "learning_rate": 5.006033661479835e-05,
      "loss": 0.0652,
      "step": 7868
    },
    {
      "epoch": 14.981455064194009,
      "grad_norm": 0.2682128846645355,
      "learning_rate": 5.0053985392187996e-05,
      "loss": 0.07,
      "step": 7869
    },
    {
      "epoch": 14.98335710889206,
      "grad_norm": 0.1930389106273651,
      "learning_rate": 5.004763416957765e-05,
      "loss": 0.0521,
      "step": 7870
    },
    {
      "epoch": 14.985259153590109,
      "grad_norm": 0.13474808633327484,
      "learning_rate": 5.004128294696729e-05,
      "loss": 0.0716,
      "step": 7871
    },
    {
      "epoch": 14.98716119828816,
      "grad_norm": 0.1883508861064911,
      "learning_rate": 5.003493172435694e-05,
      "loss": 0.0628,
      "step": 7872
    },
    {
      "epoch": 14.98906324298621,
      "grad_norm": 0.1612243503332138,
      "learning_rate": 5.002858050174659e-05,
      "loss": 0.0585,
      "step": 7873
    },
    {
      "epoch": 14.990965287684261,
      "grad_norm": 0.3346334397792816,
      "learning_rate": 5.0022229279136235e-05,
      "loss": 0.0723,
      "step": 7874
    },
    {
      "epoch": 14.99286733238231,
      "grad_norm": 0.26011380553245544,
      "learning_rate": 5.0015878056525886e-05,
      "loss": 0.0682,
      "step": 7875
    },
    {
      "epoch": 14.994769377080361,
      "grad_norm": 0.23277519643306732,
      "learning_rate": 5.000952683391553e-05,
      "loss": 0.0716,
      "step": 7876
    },
    {
      "epoch": 14.996671421778412,
      "grad_norm": 0.12851858139038086,
      "learning_rate": 5.0003175611305176e-05,
      "loss": 0.0756,
      "step": 7877
    },
    {
      "epoch": 14.998573466476461,
      "grad_norm": 0.16410908102989197,
      "learning_rate": 4.999682438869482e-05,
      "loss": 0.0611,
      "step": 7878
    },
    {
      "epoch": 15.000475511174512,
      "grad_norm": 0.3350670635700226,
      "learning_rate": 4.999047316608447e-05,
      "loss": 0.0883,
      "step": 7879
    },
    {
      "epoch": 15.002377555872563,
      "grad_norm": 0.26460301876068115,
      "learning_rate": 4.998412194347412e-05,
      "loss": 0.0813,
      "step": 7880
    },
    {
      "epoch": 15.004279600570614,
      "grad_norm": 0.14847742021083832,
      "learning_rate": 4.997777072086377e-05,
      "loss": 0.0532,
      "step": 7881
    },
    {
      "epoch": 15.006181645268663,
      "grad_norm": 0.25994306802749634,
      "learning_rate": 4.9971419498253415e-05,
      "loss": 0.0624,
      "step": 7882
    },
    {
      "epoch": 15.008083689966714,
      "grad_norm": 0.10206472128629684,
      "learning_rate": 4.996506827564306e-05,
      "loss": 0.0463,
      "step": 7883
    },
    {
      "epoch": 15.009985734664765,
      "grad_norm": 0.10814502090215683,
      "learning_rate": 4.995871705303271e-05,
      "loss": 0.0604,
      "step": 7884
    },
    {
      "epoch": 15.011887779362816,
      "grad_norm": 0.23959451913833618,
      "learning_rate": 4.995236583042236e-05,
      "loss": 0.0629,
      "step": 7885
    },
    {
      "epoch": 15.013789824060865,
      "grad_norm": 0.1656973958015442,
      "learning_rate": 4.994601460781201e-05,
      "loss": 0.0659,
      "step": 7886
    },
    {
      "epoch": 15.015691868758916,
      "grad_norm": 0.18013009428977966,
      "learning_rate": 4.9939663385201654e-05,
      "loss": 0.0403,
      "step": 7887
    },
    {
      "epoch": 15.017593913456967,
      "grad_norm": 0.19583438336849213,
      "learning_rate": 4.99333121625913e-05,
      "loss": 0.0482,
      "step": 7888
    },
    {
      "epoch": 15.019495958155016,
      "grad_norm": 0.07162532955408096,
      "learning_rate": 4.992696093998095e-05,
      "loss": 0.0537,
      "step": 7889
    },
    {
      "epoch": 15.021398002853067,
      "grad_norm": 0.1543794721364975,
      "learning_rate": 4.9920609717370596e-05,
      "loss": 0.0523,
      "step": 7890
    },
    {
      "epoch": 15.023300047551118,
      "grad_norm": 0.22192519903182983,
      "learning_rate": 4.991425849476024e-05,
      "loss": 0.071,
      "step": 7891
    },
    {
      "epoch": 15.025202092249168,
      "grad_norm": 0.13602390885353088,
      "learning_rate": 4.9907907272149886e-05,
      "loss": 0.0617,
      "step": 7892
    },
    {
      "epoch": 15.027104136947218,
      "grad_norm": 0.1319774091243744,
      "learning_rate": 4.990155604953954e-05,
      "loss": 0.0683,
      "step": 7893
    },
    {
      "epoch": 15.029006181645268,
      "grad_norm": 0.2293819636106491,
      "learning_rate": 4.989520482692919e-05,
      "loss": 0.0622,
      "step": 7894
    },
    {
      "epoch": 15.03090822634332,
      "grad_norm": 0.15792877972126007,
      "learning_rate": 4.9888853604318835e-05,
      "loss": 0.0599,
      "step": 7895
    },
    {
      "epoch": 15.03281027104137,
      "grad_norm": 0.13609178364276886,
      "learning_rate": 4.988250238170848e-05,
      "loss": 0.046,
      "step": 7896
    },
    {
      "epoch": 15.03471231573942,
      "grad_norm": 0.1835666298866272,
      "learning_rate": 4.9876151159098125e-05,
      "loss": 0.056,
      "step": 7897
    },
    {
      "epoch": 15.03661436043747,
      "grad_norm": 0.10541585832834244,
      "learning_rate": 4.986979993648778e-05,
      "loss": 0.0478,
      "step": 7898
    },
    {
      "epoch": 15.038516405135521,
      "grad_norm": 0.1154666617512703,
      "learning_rate": 4.986344871387742e-05,
      "loss": 0.0488,
      "step": 7899
    },
    {
      "epoch": 15.04041844983357,
      "grad_norm": 0.2595585584640503,
      "learning_rate": 4.985709749126707e-05,
      "loss": 0.0593,
      "step": 7900
    },
    {
      "epoch": 15.042320494531621,
      "grad_norm": 0.22063757479190826,
      "learning_rate": 4.985074626865672e-05,
      "loss": 0.0664,
      "step": 7901
    },
    {
      "epoch": 15.044222539229672,
      "grad_norm": 0.1932714283466339,
      "learning_rate": 4.9844395046046364e-05,
      "loss": 0.0789,
      "step": 7902
    },
    {
      "epoch": 15.046124583927723,
      "grad_norm": 0.1839333027601242,
      "learning_rate": 4.9838043823436016e-05,
      "loss": 0.0453,
      "step": 7903
    },
    {
      "epoch": 15.048026628625772,
      "grad_norm": 0.10126302391290665,
      "learning_rate": 4.983169260082566e-05,
      "loss": 0.0509,
      "step": 7904
    },
    {
      "epoch": 15.049928673323823,
      "grad_norm": 0.07132267951965332,
      "learning_rate": 4.9825341378215306e-05,
      "loss": 0.0514,
      "step": 7905
    },
    {
      "epoch": 15.051830718021874,
      "grad_norm": 0.2043127417564392,
      "learning_rate": 4.981899015560496e-05,
      "loss": 0.0846,
      "step": 7906
    },
    {
      "epoch": 15.053732762719925,
      "grad_norm": 0.11934273689985275,
      "learning_rate": 4.98126389329946e-05,
      "loss": 0.0515,
      "step": 7907
    },
    {
      "epoch": 15.055634807417974,
      "grad_norm": 0.14380639791488647,
      "learning_rate": 4.9806287710384255e-05,
      "loss": 0.0551,
      "step": 7908
    },
    {
      "epoch": 15.057536852116025,
      "grad_norm": 0.13530582189559937,
      "learning_rate": 4.979993648777389e-05,
      "loss": 0.0457,
      "step": 7909
    },
    {
      "epoch": 15.059438896814076,
      "grad_norm": 0.18853814899921417,
      "learning_rate": 4.9793585265163545e-05,
      "loss": 0.0644,
      "step": 7910
    },
    {
      "epoch": 15.061340941512125,
      "grad_norm": 0.21197612583637238,
      "learning_rate": 4.97872340425532e-05,
      "loss": 0.0527,
      "step": 7911
    },
    {
      "epoch": 15.063242986210176,
      "grad_norm": 0.11743347346782684,
      "learning_rate": 4.978088281994284e-05,
      "loss": 0.0429,
      "step": 7912
    },
    {
      "epoch": 15.065145030908226,
      "grad_norm": 0.2725050151348114,
      "learning_rate": 4.9774531597332494e-05,
      "loss": 0.0604,
      "step": 7913
    },
    {
      "epoch": 15.067047075606277,
      "grad_norm": 0.16864003241062164,
      "learning_rate": 4.976818037472213e-05,
      "loss": 0.0524,
      "step": 7914
    },
    {
      "epoch": 15.068949120304326,
      "grad_norm": 0.0748053789138794,
      "learning_rate": 4.9761829152111784e-05,
      "loss": 0.0272,
      "step": 7915
    },
    {
      "epoch": 15.070851165002377,
      "grad_norm": 0.14363184571266174,
      "learning_rate": 4.975547792950143e-05,
      "loss": 0.0596,
      "step": 7916
    },
    {
      "epoch": 15.072753209700428,
      "grad_norm": 0.15277619659900665,
      "learning_rate": 4.974912670689108e-05,
      "loss": 0.063,
      "step": 7917
    },
    {
      "epoch": 15.074655254398479,
      "grad_norm": 0.2070416361093521,
      "learning_rate": 4.9742775484280726e-05,
      "loss": 0.0563,
      "step": 7918
    },
    {
      "epoch": 15.076557299096528,
      "grad_norm": 0.19699743390083313,
      "learning_rate": 4.973642426167037e-05,
      "loss": 0.0574,
      "step": 7919
    },
    {
      "epoch": 15.078459343794579,
      "grad_norm": 0.15078143775463104,
      "learning_rate": 4.973007303906002e-05,
      "loss": 0.0576,
      "step": 7920
    },
    {
      "epoch": 15.08036138849263,
      "grad_norm": 0.1462477743625641,
      "learning_rate": 4.972372181644967e-05,
      "loss": 0.0408,
      "step": 7921
    },
    {
      "epoch": 15.08226343319068,
      "grad_norm": 0.23796884715557098,
      "learning_rate": 4.971737059383932e-05,
      "loss": 0.0638,
      "step": 7922
    },
    {
      "epoch": 15.08416547788873,
      "grad_norm": 0.10526379942893982,
      "learning_rate": 4.9711019371228965e-05,
      "loss": 0.0419,
      "step": 7923
    },
    {
      "epoch": 15.08606752258678,
      "grad_norm": 0.23168741166591644,
      "learning_rate": 4.970466814861861e-05,
      "loss": 0.0708,
      "step": 7924
    },
    {
      "epoch": 15.087969567284832,
      "grad_norm": 0.13456536829471588,
      "learning_rate": 4.969831692600826e-05,
      "loss": 0.0524,
      "step": 7925
    },
    {
      "epoch": 15.08987161198288,
      "grad_norm": 0.16766615211963654,
      "learning_rate": 4.969196570339791e-05,
      "loss": 0.0508,
      "step": 7926
    },
    {
      "epoch": 15.091773656680932,
      "grad_norm": 0.20241063833236694,
      "learning_rate": 4.968561448078755e-05,
      "loss": 0.0604,
      "step": 7927
    },
    {
      "epoch": 15.093675701378983,
      "grad_norm": 0.14176540076732635,
      "learning_rate": 4.96792632581772e-05,
      "loss": 0.0463,
      "step": 7928
    },
    {
      "epoch": 15.095577746077034,
      "grad_norm": 0.19786062836647034,
      "learning_rate": 4.967291203556685e-05,
      "loss": 0.0519,
      "step": 7929
    },
    {
      "epoch": 15.097479790775083,
      "grad_norm": 0.18619826436042786,
      "learning_rate": 4.96665608129565e-05,
      "loss": 0.0574,
      "step": 7930
    },
    {
      "epoch": 15.099381835473134,
      "grad_norm": 0.17755287885665894,
      "learning_rate": 4.9660209590346146e-05,
      "loss": 0.0739,
      "step": 7931
    },
    {
      "epoch": 15.101283880171184,
      "grad_norm": 0.18910397589206696,
      "learning_rate": 4.965385836773579e-05,
      "loss": 0.0707,
      "step": 7932
    },
    {
      "epoch": 15.103185924869235,
      "grad_norm": 0.16724318265914917,
      "learning_rate": 4.9647507145125436e-05,
      "loss": 0.049,
      "step": 7933
    },
    {
      "epoch": 15.105087969567284,
      "grad_norm": 0.20525459945201874,
      "learning_rate": 4.964115592251509e-05,
      "loss": 0.0656,
      "step": 7934
    },
    {
      "epoch": 15.106990014265335,
      "grad_norm": 0.14179904758930206,
      "learning_rate": 4.963480469990473e-05,
      "loss": 0.0757,
      "step": 7935
    },
    {
      "epoch": 15.108892058963386,
      "grad_norm": 0.1320018470287323,
      "learning_rate": 4.962845347729438e-05,
      "loss": 0.0577,
      "step": 7936
    },
    {
      "epoch": 15.110794103661435,
      "grad_norm": 0.11483266949653625,
      "learning_rate": 4.962210225468403e-05,
      "loss": 0.07,
      "step": 7937
    },
    {
      "epoch": 15.112696148359486,
      "grad_norm": 0.13130733370780945,
      "learning_rate": 4.9615751032073675e-05,
      "loss": 0.0609,
      "step": 7938
    },
    {
      "epoch": 15.114598193057537,
      "grad_norm": 0.14458687603473663,
      "learning_rate": 4.9609399809463326e-05,
      "loss": 0.0496,
      "step": 7939
    },
    {
      "epoch": 15.116500237755588,
      "grad_norm": 0.1600898802280426,
      "learning_rate": 4.960304858685297e-05,
      "loss": 0.0697,
      "step": 7940
    },
    {
      "epoch": 15.118402282453637,
      "grad_norm": 0.09131373465061188,
      "learning_rate": 4.9596697364242617e-05,
      "loss": 0.0401,
      "step": 7941
    },
    {
      "epoch": 15.120304327151688,
      "grad_norm": 0.20547789335250854,
      "learning_rate": 4.959034614163227e-05,
      "loss": 0.0555,
      "step": 7942
    },
    {
      "epoch": 15.122206371849739,
      "grad_norm": 0.15025202929973602,
      "learning_rate": 4.9583994919021913e-05,
      "loss": 0.0505,
      "step": 7943
    },
    {
      "epoch": 15.12410841654779,
      "grad_norm": 0.06455320864915848,
      "learning_rate": 4.9577643696411565e-05,
      "loss": 0.0634,
      "step": 7944
    },
    {
      "epoch": 15.126010461245839,
      "grad_norm": 0.12017594277858734,
      "learning_rate": 4.9571292473801204e-05,
      "loss": 0.0486,
      "step": 7945
    },
    {
      "epoch": 15.12791250594389,
      "grad_norm": 0.18531164526939392,
      "learning_rate": 4.9564941251190855e-05,
      "loss": 0.0552,
      "step": 7946
    },
    {
      "epoch": 15.12981455064194,
      "grad_norm": 0.07348939776420593,
      "learning_rate": 4.95585900285805e-05,
      "loss": 0.0596,
      "step": 7947
    },
    {
      "epoch": 15.13171659533999,
      "grad_norm": 0.3035662770271301,
      "learning_rate": 4.955223880597015e-05,
      "loss": 0.074,
      "step": 7948
    },
    {
      "epoch": 15.13361864003804,
      "grad_norm": 0.14867135882377625,
      "learning_rate": 4.9545887583359804e-05,
      "loss": 0.0563,
      "step": 7949
    },
    {
      "epoch": 15.135520684736091,
      "grad_norm": 0.1565437614917755,
      "learning_rate": 4.953953636074944e-05,
      "loss": 0.059,
      "step": 7950
    },
    {
      "epoch": 15.137422729434142,
      "grad_norm": 0.14548735320568085,
      "learning_rate": 4.9533185138139094e-05,
      "loss": 0.0612,
      "step": 7951
    },
    {
      "epoch": 15.139324774132191,
      "grad_norm": 0.1496451199054718,
      "learning_rate": 4.952683391552874e-05,
      "loss": 0.062,
      "step": 7952
    },
    {
      "epoch": 15.141226818830242,
      "grad_norm": 0.1259707808494568,
      "learning_rate": 4.952048269291839e-05,
      "loss": 0.0517,
      "step": 7953
    },
    {
      "epoch": 15.143128863528293,
      "grad_norm": 0.19889900088310242,
      "learning_rate": 4.9514131470308036e-05,
      "loss": 0.0544,
      "step": 7954
    },
    {
      "epoch": 15.145030908226344,
      "grad_norm": 0.1751696914434433,
      "learning_rate": 4.950778024769768e-05,
      "loss": 0.0821,
      "step": 7955
    },
    {
      "epoch": 15.146932952924393,
      "grad_norm": 0.17510373890399933,
      "learning_rate": 4.950142902508733e-05,
      "loss": 0.0643,
      "step": 7956
    },
    {
      "epoch": 15.148834997622444,
      "grad_norm": 0.11843284219503403,
      "learning_rate": 4.949507780247698e-05,
      "loss": 0.0621,
      "step": 7957
    },
    {
      "epoch": 15.150737042320495,
      "grad_norm": 0.1398295760154724,
      "learning_rate": 4.948872657986663e-05,
      "loss": 0.0846,
      "step": 7958
    },
    {
      "epoch": 15.152639087018544,
      "grad_norm": 0.19116874039173126,
      "learning_rate": 4.9482375357256275e-05,
      "loss": 0.0497,
      "step": 7959
    },
    {
      "epoch": 15.154541131716595,
      "grad_norm": 0.1140001118183136,
      "learning_rate": 4.947602413464592e-05,
      "loss": 0.0514,
      "step": 7960
    },
    {
      "epoch": 15.156443176414646,
      "grad_norm": 0.22718724608421326,
      "learning_rate": 4.946967291203557e-05,
      "loss": 0.0547,
      "step": 7961
    },
    {
      "epoch": 15.158345221112697,
      "grad_norm": 0.20831739902496338,
      "learning_rate": 4.946332168942522e-05,
      "loss": 0.0428,
      "step": 7962
    },
    {
      "epoch": 15.160247265810746,
      "grad_norm": 0.16769562661647797,
      "learning_rate": 4.945697046681486e-05,
      "loss": 0.0577,
      "step": 7963
    },
    {
      "epoch": 15.162149310508797,
      "grad_norm": 0.2267942875623703,
      "learning_rate": 4.945061924420451e-05,
      "loss": 0.0531,
      "step": 7964
    },
    {
      "epoch": 15.164051355206848,
      "grad_norm": 0.166763037443161,
      "learning_rate": 4.944426802159416e-05,
      "loss": 0.0585,
      "step": 7965
    },
    {
      "epoch": 15.165953399904899,
      "grad_norm": 0.13724614679813385,
      "learning_rate": 4.943791679898381e-05,
      "loss": 0.0518,
      "step": 7966
    },
    {
      "epoch": 15.167855444602948,
      "grad_norm": 0.1825903356075287,
      "learning_rate": 4.9431565576373456e-05,
      "loss": 0.0529,
      "step": 7967
    },
    {
      "epoch": 15.169757489300999,
      "grad_norm": 0.12881925702095032,
      "learning_rate": 4.94252143537631e-05,
      "loss": 0.0609,
      "step": 7968
    },
    {
      "epoch": 15.17165953399905,
      "grad_norm": 0.16033561527729034,
      "learning_rate": 4.9418863131152746e-05,
      "loss": 0.0783,
      "step": 7969
    },
    {
      "epoch": 15.173561578697099,
      "grad_norm": 0.14918583631515503,
      "learning_rate": 4.94125119085424e-05,
      "loss": 0.059,
      "step": 7970
    },
    {
      "epoch": 15.17546362339515,
      "grad_norm": 0.1096171960234642,
      "learning_rate": 4.940616068593204e-05,
      "loss": 0.0643,
      "step": 7971
    },
    {
      "epoch": 15.1773656680932,
      "grad_norm": 0.12285967171192169,
      "learning_rate": 4.939980946332169e-05,
      "loss": 0.0533,
      "step": 7972
    },
    {
      "epoch": 15.179267712791251,
      "grad_norm": 0.20297501981258392,
      "learning_rate": 4.939345824071134e-05,
      "loss": 0.0588,
      "step": 7973
    },
    {
      "epoch": 15.1811697574893,
      "grad_norm": 0.1980786919593811,
      "learning_rate": 4.9387107018100985e-05,
      "loss": 0.0553,
      "step": 7974
    },
    {
      "epoch": 15.183071802187351,
      "grad_norm": 0.2785045802593231,
      "learning_rate": 4.938075579549064e-05,
      "loss": 0.0489,
      "step": 7975
    },
    {
      "epoch": 15.184973846885402,
      "grad_norm": 0.1846664994955063,
      "learning_rate": 4.937440457288028e-05,
      "loss": 0.0403,
      "step": 7976
    },
    {
      "epoch": 15.186875891583453,
      "grad_norm": 0.27118930220603943,
      "learning_rate": 4.936805335026993e-05,
      "loss": 0.0696,
      "step": 7977
    },
    {
      "epoch": 15.188777936281502,
      "grad_norm": 0.16328345239162445,
      "learning_rate": 4.936170212765958e-05,
      "loss": 0.0848,
      "step": 7978
    },
    {
      "epoch": 15.190679980979553,
      "grad_norm": 0.1698354184627533,
      "learning_rate": 4.9355350905049224e-05,
      "loss": 0.0666,
      "step": 7979
    },
    {
      "epoch": 15.192582025677604,
      "grad_norm": 0.15171551704406738,
      "learning_rate": 4.9348999682438876e-05,
      "loss": 0.0768,
      "step": 7980
    },
    {
      "epoch": 15.194484070375653,
      "grad_norm": 0.22728462517261505,
      "learning_rate": 4.9342648459828514e-05,
      "loss": 0.069,
      "step": 7981
    },
    {
      "epoch": 15.196386115073704,
      "grad_norm": 0.08936372399330139,
      "learning_rate": 4.9336297237218166e-05,
      "loss": 0.0534,
      "step": 7982
    },
    {
      "epoch": 15.198288159771755,
      "grad_norm": 0.24935117363929749,
      "learning_rate": 4.932994601460781e-05,
      "loss": 0.0567,
      "step": 7983
    },
    {
      "epoch": 15.200190204469806,
      "grad_norm": 0.1630091816186905,
      "learning_rate": 4.932359479199746e-05,
      "loss": 0.0627,
      "step": 7984
    },
    {
      "epoch": 15.202092249167855,
      "grad_norm": 0.1229170486330986,
      "learning_rate": 4.9317243569387115e-05,
      "loss": 0.0909,
      "step": 7985
    },
    {
      "epoch": 15.203994293865906,
      "grad_norm": 0.16645725071430206,
      "learning_rate": 4.931089234677675e-05,
      "loss": 0.0611,
      "step": 7986
    },
    {
      "epoch": 15.205896338563957,
      "grad_norm": 0.12142518907785416,
      "learning_rate": 4.9304541124166405e-05,
      "loss": 0.0431,
      "step": 7987
    },
    {
      "epoch": 15.207798383262007,
      "grad_norm": 0.34056565165519714,
      "learning_rate": 4.929818990155605e-05,
      "loss": 0.0636,
      "step": 7988
    },
    {
      "epoch": 15.209700427960057,
      "grad_norm": 0.09468982368707657,
      "learning_rate": 4.92918386789457e-05,
      "loss": 0.0418,
      "step": 7989
    },
    {
      "epoch": 15.211602472658107,
      "grad_norm": 0.14799001812934875,
      "learning_rate": 4.928548745633535e-05,
      "loss": 0.0498,
      "step": 7990
    },
    {
      "epoch": 15.213504517356158,
      "grad_norm": 0.25312280654907227,
      "learning_rate": 4.927913623372499e-05,
      "loss": 0.0479,
      "step": 7991
    },
    {
      "epoch": 15.21540656205421,
      "grad_norm": 0.19187159836292267,
      "learning_rate": 4.9272785011114644e-05,
      "loss": 0.0559,
      "step": 7992
    },
    {
      "epoch": 15.217308606752258,
      "grad_norm": 0.200269877910614,
      "learning_rate": 4.926643378850429e-05,
      "loss": 0.0629,
      "step": 7993
    },
    {
      "epoch": 15.21921065145031,
      "grad_norm": 0.15146945416927338,
      "learning_rate": 4.926008256589394e-05,
      "loss": 0.0495,
      "step": 7994
    },
    {
      "epoch": 15.22111269614836,
      "grad_norm": 0.1498185396194458,
      "learning_rate": 4.9253731343283586e-05,
      "loss": 0.0619,
      "step": 7995
    },
    {
      "epoch": 15.22301474084641,
      "grad_norm": 0.16426420211791992,
      "learning_rate": 4.924738012067323e-05,
      "loss": 0.0545,
      "step": 7996
    },
    {
      "epoch": 15.22491678554446,
      "grad_norm": 0.2681984007358551,
      "learning_rate": 4.924102889806288e-05,
      "loss": 0.0562,
      "step": 7997
    },
    {
      "epoch": 15.226818830242511,
      "grad_norm": 0.17494945228099823,
      "learning_rate": 4.923467767545253e-05,
      "loss": 0.1054,
      "step": 7998
    },
    {
      "epoch": 15.228720874940562,
      "grad_norm": 0.08447415381669998,
      "learning_rate": 4.922832645284217e-05,
      "loss": 0.0652,
      "step": 7999
    },
    {
      "epoch": 15.230622919638611,
      "grad_norm": 0.17718258500099182,
      "learning_rate": 4.922197523023182e-05,
      "loss": 0.045,
      "step": 8000
    },
    {
      "epoch": 15.232524964336662,
      "grad_norm": 0.1426248401403427,
      "learning_rate": 4.921562400762147e-05,
      "loss": 0.0454,
      "step": 8001
    },
    {
      "epoch": 15.234427009034713,
      "grad_norm": 0.14476601779460907,
      "learning_rate": 4.9209272785011115e-05,
      "loss": 0.0538,
      "step": 8002
    },
    {
      "epoch": 15.236329053732764,
      "grad_norm": 0.14318709075450897,
      "learning_rate": 4.9202921562400766e-05,
      "loss": 0.0609,
      "step": 8003
    },
    {
      "epoch": 15.238231098430813,
      "grad_norm": 0.18595252931118011,
      "learning_rate": 4.919657033979041e-05,
      "loss": 0.0583,
      "step": 8004
    },
    {
      "epoch": 15.240133143128864,
      "grad_norm": 0.11332142353057861,
      "learning_rate": 4.9190219117180057e-05,
      "loss": 0.0462,
      "step": 8005
    },
    {
      "epoch": 15.242035187826914,
      "grad_norm": 0.1512480229139328,
      "learning_rate": 4.918386789456971e-05,
      "loss": 0.0585,
      "step": 8006
    },
    {
      "epoch": 15.243937232524964,
      "grad_norm": 0.09532251209020615,
      "learning_rate": 4.9177516671959353e-05,
      "loss": 0.0483,
      "step": 8007
    },
    {
      "epoch": 15.245839277223014,
      "grad_norm": 0.1605871617794037,
      "learning_rate": 4.9171165449349e-05,
      "loss": 0.0598,
      "step": 8008
    },
    {
      "epoch": 15.247741321921065,
      "grad_norm": 0.22983098030090332,
      "learning_rate": 4.916481422673865e-05,
      "loss": 0.0743,
      "step": 8009
    },
    {
      "epoch": 15.249643366619116,
      "grad_norm": 0.17360562086105347,
      "learning_rate": 4.9158463004128295e-05,
      "loss": 0.049,
      "step": 8010
    },
    {
      "epoch": 15.251545411317165,
      "grad_norm": 0.2593297064304352,
      "learning_rate": 4.915211178151795e-05,
      "loss": 0.0762,
      "step": 8011
    },
    {
      "epoch": 15.253447456015216,
      "grad_norm": 0.1943679302930832,
      "learning_rate": 4.914576055890759e-05,
      "loss": 0.0656,
      "step": 8012
    },
    {
      "epoch": 15.255349500713267,
      "grad_norm": 0.10750841349363327,
      "learning_rate": 4.913940933629724e-05,
      "loss": 0.0611,
      "step": 8013
    },
    {
      "epoch": 15.257251545411318,
      "grad_norm": 0.2768067419528961,
      "learning_rate": 4.913305811368689e-05,
      "loss": 0.0768,
      "step": 8014
    },
    {
      "epoch": 15.259153590109367,
      "grad_norm": 0.12465161085128784,
      "learning_rate": 4.9126706891076534e-05,
      "loss": 0.0536,
      "step": 8015
    },
    {
      "epoch": 15.261055634807418,
      "grad_norm": 0.07539068162441254,
      "learning_rate": 4.9120355668466186e-05,
      "loss": 0.0594,
      "step": 8016
    },
    {
      "epoch": 15.262957679505469,
      "grad_norm": 0.13377007842063904,
      "learning_rate": 4.9114004445855824e-05,
      "loss": 0.0507,
      "step": 8017
    },
    {
      "epoch": 15.264859724203518,
      "grad_norm": 0.11560075730085373,
      "learning_rate": 4.9107653223245476e-05,
      "loss": 0.0618,
      "step": 8018
    },
    {
      "epoch": 15.266761768901569,
      "grad_norm": 0.08740713447332382,
      "learning_rate": 4.910130200063512e-05,
      "loss": 0.0544,
      "step": 8019
    },
    {
      "epoch": 15.26866381359962,
      "grad_norm": 0.23278751969337463,
      "learning_rate": 4.909495077802477e-05,
      "loss": 0.0591,
      "step": 8020
    },
    {
      "epoch": 15.27056585829767,
      "grad_norm": 0.2627377510070801,
      "learning_rate": 4.9088599555414425e-05,
      "loss": 0.0638,
      "step": 8021
    },
    {
      "epoch": 15.27246790299572,
      "grad_norm": 0.1532537043094635,
      "learning_rate": 4.908224833280406e-05,
      "loss": 0.0487,
      "step": 8022
    },
    {
      "epoch": 15.27436994769377,
      "grad_norm": 0.11168943345546722,
      "learning_rate": 4.9075897110193715e-05,
      "loss": 0.0487,
      "step": 8023
    },
    {
      "epoch": 15.276271992391822,
      "grad_norm": 0.24777059257030487,
      "learning_rate": 4.906954588758336e-05,
      "loss": 0.0719,
      "step": 8024
    },
    {
      "epoch": 15.278174037089872,
      "grad_norm": 0.15639886260032654,
      "learning_rate": 4.906319466497301e-05,
      "loss": 0.0588,
      "step": 8025
    },
    {
      "epoch": 15.280076081787922,
      "grad_norm": 0.12868893146514893,
      "learning_rate": 4.905684344236266e-05,
      "loss": 0.0494,
      "step": 8026
    },
    {
      "epoch": 15.281978126485972,
      "grad_norm": 0.1784045547246933,
      "learning_rate": 4.90504922197523e-05,
      "loss": 0.0617,
      "step": 8027
    },
    {
      "epoch": 15.283880171184023,
      "grad_norm": 0.13852459192276,
      "learning_rate": 4.9044140997141954e-05,
      "loss": 0.0685,
      "step": 8028
    },
    {
      "epoch": 15.285782215882072,
      "grad_norm": 0.09702124446630478,
      "learning_rate": 4.90377897745316e-05,
      "loss": 0.0598,
      "step": 8029
    },
    {
      "epoch": 15.287684260580123,
      "grad_norm": 0.23854145407676697,
      "learning_rate": 4.903143855192125e-05,
      "loss": 0.0529,
      "step": 8030
    },
    {
      "epoch": 15.289586305278174,
      "grad_norm": 0.1306440234184265,
      "learning_rate": 4.902508732931089e-05,
      "loss": 0.0705,
      "step": 8031
    },
    {
      "epoch": 15.291488349976225,
      "grad_norm": 0.17986299097537994,
      "learning_rate": 4.901873610670054e-05,
      "loss": 0.0456,
      "step": 8032
    },
    {
      "epoch": 15.293390394674274,
      "grad_norm": 0.26708826422691345,
      "learning_rate": 4.901238488409019e-05,
      "loss": 0.0586,
      "step": 8033
    },
    {
      "epoch": 15.295292439372325,
      "grad_norm": 0.18249185383319855,
      "learning_rate": 4.900603366147984e-05,
      "loss": 0.058,
      "step": 8034
    },
    {
      "epoch": 15.297194484070376,
      "grad_norm": 0.16154305636882782,
      "learning_rate": 4.899968243886948e-05,
      "loss": 0.0452,
      "step": 8035
    },
    {
      "epoch": 15.299096528768427,
      "grad_norm": 0.1726539582014084,
      "learning_rate": 4.899333121625913e-05,
      "loss": 0.0538,
      "step": 8036
    },
    {
      "epoch": 15.300998573466476,
      "grad_norm": 0.20630763471126556,
      "learning_rate": 4.898697999364878e-05,
      "loss": 0.067,
      "step": 8037
    },
    {
      "epoch": 15.302900618164527,
      "grad_norm": 0.1309700757265091,
      "learning_rate": 4.8980628771038425e-05,
      "loss": 0.051,
      "step": 8038
    },
    {
      "epoch": 15.304802662862578,
      "grad_norm": 0.10560029000043869,
      "learning_rate": 4.897427754842808e-05,
      "loss": 0.057,
      "step": 8039
    },
    {
      "epoch": 15.306704707560627,
      "grad_norm": 0.0757644772529602,
      "learning_rate": 4.896792632581772e-05,
      "loss": 0.0491,
      "step": 8040
    },
    {
      "epoch": 15.308606752258678,
      "grad_norm": 0.21504676342010498,
      "learning_rate": 4.896157510320737e-05,
      "loss": 0.0705,
      "step": 8041
    },
    {
      "epoch": 15.310508796956729,
      "grad_norm": 0.06593248248100281,
      "learning_rate": 4.895522388059702e-05,
      "loss": 0.0298,
      "step": 8042
    },
    {
      "epoch": 15.31241084165478,
      "grad_norm": 0.13523049652576447,
      "learning_rate": 4.8948872657986664e-05,
      "loss": 0.0665,
      "step": 8043
    },
    {
      "epoch": 15.314312886352829,
      "grad_norm": 0.1855146586894989,
      "learning_rate": 4.894252143537631e-05,
      "loss": 0.0624,
      "step": 8044
    },
    {
      "epoch": 15.31621493105088,
      "grad_norm": 0.11019548773765564,
      "learning_rate": 4.893617021276596e-05,
      "loss": 0.0486,
      "step": 8045
    },
    {
      "epoch": 15.31811697574893,
      "grad_norm": 0.17465780675411224,
      "learning_rate": 4.8929818990155606e-05,
      "loss": 0.048,
      "step": 8046
    },
    {
      "epoch": 15.320019020446981,
      "grad_norm": 0.1513073593378067,
      "learning_rate": 4.892346776754526e-05,
      "loss": 0.058,
      "step": 8047
    },
    {
      "epoch": 15.32192106514503,
      "grad_norm": 0.13659310340881348,
      "learning_rate": 4.89171165449349e-05,
      "loss": 0.0431,
      "step": 8048
    },
    {
      "epoch": 15.323823109843081,
      "grad_norm": 0.15410743653774261,
      "learning_rate": 4.891076532232455e-05,
      "loss": 0.0499,
      "step": 8049
    },
    {
      "epoch": 15.325725154541132,
      "grad_norm": 0.3651988208293915,
      "learning_rate": 4.89044140997142e-05,
      "loss": 0.0694,
      "step": 8050
    },
    {
      "epoch": 15.327627199239181,
      "grad_norm": 0.2175668627023697,
      "learning_rate": 4.8898062877103845e-05,
      "loss": 0.0703,
      "step": 8051
    },
    {
      "epoch": 15.329529243937232,
      "grad_norm": 0.33220532536506653,
      "learning_rate": 4.8891711654493497e-05,
      "loss": 0.0674,
      "step": 8052
    },
    {
      "epoch": 15.331431288635283,
      "grad_norm": 0.19759467244148254,
      "learning_rate": 4.8885360431883135e-05,
      "loss": 0.0684,
      "step": 8053
    },
    {
      "epoch": 15.333333333333334,
      "grad_norm": 0.20885705947875977,
      "learning_rate": 4.887900920927279e-05,
      "loss": 0.0602,
      "step": 8054
    },
    {
      "epoch": 15.335235378031383,
      "grad_norm": 0.3209524154663086,
      "learning_rate": 4.887265798666243e-05,
      "loss": 0.0743,
      "step": 8055
    },
    {
      "epoch": 15.337137422729434,
      "grad_norm": 0.12623853981494904,
      "learning_rate": 4.8866306764052084e-05,
      "loss": 0.0669,
      "step": 8056
    },
    {
      "epoch": 15.339039467427485,
      "grad_norm": 0.06359723210334778,
      "learning_rate": 4.885995554144173e-05,
      "loss": 0.0423,
      "step": 8057
    },
    {
      "epoch": 15.340941512125536,
      "grad_norm": 0.12852895259857178,
      "learning_rate": 4.8853604318831374e-05,
      "loss": 0.0558,
      "step": 8058
    },
    {
      "epoch": 15.342843556823585,
      "grad_norm": 0.10829676687717438,
      "learning_rate": 4.8847253096221026e-05,
      "loss": 0.0444,
      "step": 8059
    },
    {
      "epoch": 15.344745601521636,
      "grad_norm": 0.17335738241672516,
      "learning_rate": 4.884090187361067e-05,
      "loss": 0.0437,
      "step": 8060
    },
    {
      "epoch": 15.346647646219687,
      "grad_norm": 0.13086184859275818,
      "learning_rate": 4.883455065100032e-05,
      "loss": 0.0584,
      "step": 8061
    },
    {
      "epoch": 15.348549690917736,
      "grad_norm": 0.1777004450559616,
      "learning_rate": 4.882819942838997e-05,
      "loss": 0.048,
      "step": 8062
    },
    {
      "epoch": 15.350451735615787,
      "grad_norm": 0.12129975855350494,
      "learning_rate": 4.882184820577961e-05,
      "loss": 0.052,
      "step": 8063
    },
    {
      "epoch": 15.352353780313837,
      "grad_norm": 0.1653653234243393,
      "learning_rate": 4.8815496983169264e-05,
      "loss": 0.0477,
      "step": 8064
    },
    {
      "epoch": 15.354255825011888,
      "grad_norm": 0.07352624088525772,
      "learning_rate": 4.880914576055891e-05,
      "loss": 0.0561,
      "step": 8065
    },
    {
      "epoch": 15.356157869709937,
      "grad_norm": 0.21875706315040588,
      "learning_rate": 4.880279453794856e-05,
      "loss": 0.0555,
      "step": 8066
    },
    {
      "epoch": 15.358059914407988,
      "grad_norm": 0.12528656423091888,
      "learning_rate": 4.87964433153382e-05,
      "loss": 0.0596,
      "step": 8067
    },
    {
      "epoch": 15.35996195910604,
      "grad_norm": 0.08361958712339401,
      "learning_rate": 4.879009209272785e-05,
      "loss": 0.0494,
      "step": 8068
    },
    {
      "epoch": 15.36186400380409,
      "grad_norm": 0.13227033615112305,
      "learning_rate": 4.87837408701175e-05,
      "loss": 0.0487,
      "step": 8069
    },
    {
      "epoch": 15.36376604850214,
      "grad_norm": 0.21816705167293549,
      "learning_rate": 4.877738964750715e-05,
      "loss": 0.0605,
      "step": 8070
    },
    {
      "epoch": 15.36566809320019,
      "grad_norm": 0.1681966334581375,
      "learning_rate": 4.8771038424896793e-05,
      "loss": 0.0535,
      "step": 8071
    },
    {
      "epoch": 15.367570137898241,
      "grad_norm": 0.11082461476325989,
      "learning_rate": 4.876468720228644e-05,
      "loss": 0.0536,
      "step": 8072
    },
    {
      "epoch": 15.36947218259629,
      "grad_norm": 0.1294272541999817,
      "learning_rate": 4.875833597967609e-05,
      "loss": 0.0673,
      "step": 8073
    },
    {
      "epoch": 15.371374227294341,
      "grad_norm": 0.14270833134651184,
      "learning_rate": 4.8751984757065735e-05,
      "loss": 0.0466,
      "step": 8074
    },
    {
      "epoch": 15.373276271992392,
      "grad_norm": 0.20790541172027588,
      "learning_rate": 4.874563353445539e-05,
      "loss": 0.0535,
      "step": 8075
    },
    {
      "epoch": 15.375178316690443,
      "grad_norm": 0.1547214835882187,
      "learning_rate": 4.873928231184503e-05,
      "loss": 0.0605,
      "step": 8076
    },
    {
      "epoch": 15.377080361388492,
      "grad_norm": 0.1377478688955307,
      "learning_rate": 4.873293108923468e-05,
      "loss": 0.059,
      "step": 8077
    },
    {
      "epoch": 15.378982406086543,
      "grad_norm": 0.23683491349220276,
      "learning_rate": 4.872657986662433e-05,
      "loss": 0.062,
      "step": 8078
    },
    {
      "epoch": 15.380884450784594,
      "grad_norm": 0.14311161637306213,
      "learning_rate": 4.8720228644013974e-05,
      "loss": 0.0723,
      "step": 8079
    },
    {
      "epoch": 15.382786495482645,
      "grad_norm": 0.13526876270771027,
      "learning_rate": 4.871387742140362e-05,
      "loss": 0.0622,
      "step": 8080
    },
    {
      "epoch": 15.384688540180694,
      "grad_norm": 0.1406906694173813,
      "learning_rate": 4.870752619879327e-05,
      "loss": 0.0476,
      "step": 8081
    },
    {
      "epoch": 15.386590584878745,
      "grad_norm": 0.11449279636144638,
      "learning_rate": 4.8701174976182916e-05,
      "loss": 0.0657,
      "step": 8082
    },
    {
      "epoch": 15.388492629576795,
      "grad_norm": 0.1769600659608841,
      "learning_rate": 4.869482375357257e-05,
      "loss": 0.0764,
      "step": 8083
    },
    {
      "epoch": 15.390394674274846,
      "grad_norm": 0.1439645141363144,
      "learning_rate": 4.868847253096221e-05,
      "loss": 0.0559,
      "step": 8084
    },
    {
      "epoch": 15.392296718972895,
      "grad_norm": 0.2625383734703064,
      "learning_rate": 4.868212130835186e-05,
      "loss": 0.0797,
      "step": 8085
    },
    {
      "epoch": 15.394198763670946,
      "grad_norm": 0.15457496047019958,
      "learning_rate": 4.86757700857415e-05,
      "loss": 0.0652,
      "step": 8086
    },
    {
      "epoch": 15.396100808368997,
      "grad_norm": 0.07629223167896271,
      "learning_rate": 4.8669418863131155e-05,
      "loss": 0.0427,
      "step": 8087
    },
    {
      "epoch": 15.398002853067046,
      "grad_norm": 0.09746023267507553,
      "learning_rate": 4.866306764052081e-05,
      "loss": 0.0361,
      "step": 8088
    },
    {
      "epoch": 15.399904897765097,
      "grad_norm": 0.11538923531770706,
      "learning_rate": 4.8656716417910445e-05,
      "loss": 0.0504,
      "step": 8089
    },
    {
      "epoch": 15.401806942463148,
      "grad_norm": 0.16864782571792603,
      "learning_rate": 4.86503651953001e-05,
      "loss": 0.0672,
      "step": 8090
    },
    {
      "epoch": 15.403708987161199,
      "grad_norm": 0.23421289026737213,
      "learning_rate": 4.864401397268974e-05,
      "loss": 0.0648,
      "step": 8091
    },
    {
      "epoch": 15.405611031859248,
      "grad_norm": 0.20838795602321625,
      "learning_rate": 4.8637662750079394e-05,
      "loss": 0.0694,
      "step": 8092
    },
    {
      "epoch": 15.407513076557299,
      "grad_norm": 0.24948616325855255,
      "learning_rate": 4.863131152746904e-05,
      "loss": 0.0723,
      "step": 8093
    },
    {
      "epoch": 15.40941512125535,
      "grad_norm": 0.10712212324142456,
      "learning_rate": 4.8624960304858684e-05,
      "loss": 0.0559,
      "step": 8094
    },
    {
      "epoch": 15.4113171659534,
      "grad_norm": 0.18813996016979218,
      "learning_rate": 4.8618609082248336e-05,
      "loss": 0.0636,
      "step": 8095
    },
    {
      "epoch": 15.41321921065145,
      "grad_norm": 0.10939308255910873,
      "learning_rate": 4.861225785963798e-05,
      "loss": 0.069,
      "step": 8096
    },
    {
      "epoch": 15.4151212553495,
      "grad_norm": 0.24954181909561157,
      "learning_rate": 4.860590663702763e-05,
      "loss": 0.0737,
      "step": 8097
    },
    {
      "epoch": 15.417023300047552,
      "grad_norm": 0.20011039078235626,
      "learning_rate": 4.859955541441728e-05,
      "loss": 0.0501,
      "step": 8098
    },
    {
      "epoch": 15.4189253447456,
      "grad_norm": 0.2183552086353302,
      "learning_rate": 4.859320419180692e-05,
      "loss": 0.0688,
      "step": 8099
    },
    {
      "epoch": 15.420827389443652,
      "grad_norm": 0.0974702537059784,
      "learning_rate": 4.8586852969196575e-05,
      "loss": 0.0511,
      "step": 8100
    },
    {
      "epoch": 15.422729434141702,
      "grad_norm": 0.3261789083480835,
      "learning_rate": 4.858050174658622e-05,
      "loss": 0.0713,
      "step": 8101
    },
    {
      "epoch": 15.424631478839753,
      "grad_norm": 0.1563624143600464,
      "learning_rate": 4.857415052397587e-05,
      "loss": 0.0435,
      "step": 8102
    },
    {
      "epoch": 15.426533523537802,
      "grad_norm": 0.34335121512413025,
      "learning_rate": 4.856779930136551e-05,
      "loss": 0.0841,
      "step": 8103
    },
    {
      "epoch": 15.428435568235853,
      "grad_norm": 0.1576901227235794,
      "learning_rate": 4.856144807875516e-05,
      "loss": 0.0632,
      "step": 8104
    },
    {
      "epoch": 15.430337612933904,
      "grad_norm": 0.31260785460472107,
      "learning_rate": 4.8555096856144814e-05,
      "loss": 0.0439,
      "step": 8105
    },
    {
      "epoch": 15.432239657631955,
      "grad_norm": 0.1318841129541397,
      "learning_rate": 4.854874563353446e-05,
      "loss": 0.0638,
      "step": 8106
    },
    {
      "epoch": 15.434141702330004,
      "grad_norm": 0.143747016787529,
      "learning_rate": 4.8542394410924104e-05,
      "loss": 0.0488,
      "step": 8107
    },
    {
      "epoch": 15.436043747028055,
      "grad_norm": 0.23869341611862183,
      "learning_rate": 4.853604318831375e-05,
      "loss": 0.058,
      "step": 8108
    },
    {
      "epoch": 15.437945791726106,
      "grad_norm": 0.178604394197464,
      "learning_rate": 4.85296919657034e-05,
      "loss": 0.0514,
      "step": 8109
    },
    {
      "epoch": 15.439847836424155,
      "grad_norm": 0.18076927959918976,
      "learning_rate": 4.8523340743093046e-05,
      "loss": 0.0688,
      "step": 8110
    },
    {
      "epoch": 15.441749881122206,
      "grad_norm": 0.2589855194091797,
      "learning_rate": 4.85169895204827e-05,
      "loss": 0.0541,
      "step": 8111
    },
    {
      "epoch": 15.443651925820257,
      "grad_norm": 0.11828634887933731,
      "learning_rate": 4.851063829787234e-05,
      "loss": 0.1125,
      "step": 8112
    },
    {
      "epoch": 15.445553970518308,
      "grad_norm": 0.1342373639345169,
      "learning_rate": 4.850428707526199e-05,
      "loss": 0.0607,
      "step": 8113
    },
    {
      "epoch": 15.447456015216357,
      "grad_norm": 0.12266510725021362,
      "learning_rate": 4.849793585265164e-05,
      "loss": 0.0397,
      "step": 8114
    },
    {
      "epoch": 15.449358059914408,
      "grad_norm": 0.26072725653648376,
      "learning_rate": 4.8491584630041285e-05,
      "loss": 0.0711,
      "step": 8115
    },
    {
      "epoch": 15.451260104612459,
      "grad_norm": 0.06860945373773575,
      "learning_rate": 4.848523340743093e-05,
      "loss": 0.03,
      "step": 8116
    },
    {
      "epoch": 15.45316214931051,
      "grad_norm": 0.09830333292484283,
      "learning_rate": 4.847888218482058e-05,
      "loss": 0.0584,
      "step": 8117
    },
    {
      "epoch": 15.455064194008559,
      "grad_norm": 0.2472551167011261,
      "learning_rate": 4.847253096221023e-05,
      "loss": 0.0856,
      "step": 8118
    },
    {
      "epoch": 15.45696623870661,
      "grad_norm": 0.15534979104995728,
      "learning_rate": 4.846617973959988e-05,
      "loss": 0.0657,
      "step": 8119
    },
    {
      "epoch": 15.45886828340466,
      "grad_norm": 0.3664383590221405,
      "learning_rate": 4.8459828516989524e-05,
      "loss": 0.0638,
      "step": 8120
    },
    {
      "epoch": 15.46077032810271,
      "grad_norm": 0.16775761544704437,
      "learning_rate": 4.845347729437917e-05,
      "loss": 0.0504,
      "step": 8121
    },
    {
      "epoch": 15.46267237280076,
      "grad_norm": 0.1890975385904312,
      "learning_rate": 4.8447126071768814e-05,
      "loss": 0.0579,
      "step": 8122
    },
    {
      "epoch": 15.464574417498811,
      "grad_norm": 0.28263619542121887,
      "learning_rate": 4.8440774849158466e-05,
      "loss": 0.0999,
      "step": 8123
    },
    {
      "epoch": 15.466476462196862,
      "grad_norm": 0.20361609756946564,
      "learning_rate": 4.843442362654812e-05,
      "loss": 0.0689,
      "step": 8124
    },
    {
      "epoch": 15.468378506894911,
      "grad_norm": 0.159327894449234,
      "learning_rate": 4.8428072403937756e-05,
      "loss": 0.0544,
      "step": 8125
    },
    {
      "epoch": 15.470280551592962,
      "grad_norm": 0.07594948261976242,
      "learning_rate": 4.842172118132741e-05,
      "loss": 0.0528,
      "step": 8126
    },
    {
      "epoch": 15.472182596291013,
      "grad_norm": 0.12283579260110855,
      "learning_rate": 4.841536995871705e-05,
      "loss": 0.0477,
      "step": 8127
    },
    {
      "epoch": 15.474084640989064,
      "grad_norm": 0.10897643864154816,
      "learning_rate": 4.8409018736106705e-05,
      "loss": 0.0499,
      "step": 8128
    },
    {
      "epoch": 15.475986685687113,
      "grad_norm": 0.08511614054441452,
      "learning_rate": 4.840266751349635e-05,
      "loss": 0.0664,
      "step": 8129
    },
    {
      "epoch": 15.477888730385164,
      "grad_norm": 0.12299087643623352,
      "learning_rate": 4.8396316290885995e-05,
      "loss": 0.076,
      "step": 8130
    },
    {
      "epoch": 15.479790775083215,
      "grad_norm": 0.20369911193847656,
      "learning_rate": 4.8389965068275646e-05,
      "loss": 0.0533,
      "step": 8131
    },
    {
      "epoch": 15.481692819781266,
      "grad_norm": 0.1380971223115921,
      "learning_rate": 4.838361384566529e-05,
      "loss": 0.0486,
      "step": 8132
    },
    {
      "epoch": 15.483594864479315,
      "grad_norm": 0.15812714397907257,
      "learning_rate": 4.8377262623054943e-05,
      "loss": 0.0564,
      "step": 8133
    },
    {
      "epoch": 15.485496909177366,
      "grad_norm": 0.1546802818775177,
      "learning_rate": 4.837091140044459e-05,
      "loss": 0.0524,
      "step": 8134
    },
    {
      "epoch": 15.487398953875417,
      "grad_norm": 0.156464621424675,
      "learning_rate": 4.8364560177834234e-05,
      "loss": 0.0834,
      "step": 8135
    },
    {
      "epoch": 15.489300998573466,
      "grad_norm": 0.2060707062482834,
      "learning_rate": 4.8358208955223885e-05,
      "loss": 0.0568,
      "step": 8136
    },
    {
      "epoch": 15.491203043271517,
      "grad_norm": 0.25839850306510925,
      "learning_rate": 4.835185773261353e-05,
      "loss": 0.068,
      "step": 8137
    },
    {
      "epoch": 15.493105087969568,
      "grad_norm": 0.2130364328622818,
      "learning_rate": 4.834550651000318e-05,
      "loss": 0.0546,
      "step": 8138
    },
    {
      "epoch": 15.495007132667618,
      "grad_norm": 0.2569848299026489,
      "learning_rate": 4.833915528739282e-05,
      "loss": 0.071,
      "step": 8139
    },
    {
      "epoch": 15.496909177365668,
      "grad_norm": 0.15983575582504272,
      "learning_rate": 4.833280406478247e-05,
      "loss": 0.0456,
      "step": 8140
    },
    {
      "epoch": 15.498811222063718,
      "grad_norm": 0.1227196604013443,
      "learning_rate": 4.832645284217212e-05,
      "loss": 0.0511,
      "step": 8141
    },
    {
      "epoch": 15.50071326676177,
      "grad_norm": 0.13389058411121368,
      "learning_rate": 4.832010161956177e-05,
      "loss": 0.0554,
      "step": 8142
    },
    {
      "epoch": 15.50261531145982,
      "grad_norm": 0.21467959880828857,
      "learning_rate": 4.8313750396951414e-05,
      "loss": 0.0679,
      "step": 8143
    },
    {
      "epoch": 15.50451735615787,
      "grad_norm": 0.13476331532001495,
      "learning_rate": 4.830739917434106e-05,
      "loss": 0.0491,
      "step": 8144
    },
    {
      "epoch": 15.50641940085592,
      "grad_norm": 0.14584288001060486,
      "learning_rate": 4.830104795173071e-05,
      "loss": 0.048,
      "step": 8145
    },
    {
      "epoch": 15.508321445553971,
      "grad_norm": 0.1891852468252182,
      "learning_rate": 4.8294696729120356e-05,
      "loss": 0.0643,
      "step": 8146
    },
    {
      "epoch": 15.51022349025202,
      "grad_norm": 0.1584828644990921,
      "learning_rate": 4.828834550651001e-05,
      "loss": 0.0559,
      "step": 8147
    },
    {
      "epoch": 15.512125534950071,
      "grad_norm": 0.10334078967571259,
      "learning_rate": 4.828199428389965e-05,
      "loss": 0.0486,
      "step": 8148
    },
    {
      "epoch": 15.514027579648122,
      "grad_norm": 0.20074616372585297,
      "learning_rate": 4.82756430612893e-05,
      "loss": 0.08,
      "step": 8149
    },
    {
      "epoch": 15.515929624346173,
      "grad_norm": 0.2370256930589676,
      "learning_rate": 4.826929183867895e-05,
      "loss": 0.087,
      "step": 8150
    },
    {
      "epoch": 15.517831669044222,
      "grad_norm": 0.15915529429912567,
      "learning_rate": 4.8262940616068595e-05,
      "loss": 0.0686,
      "step": 8151
    },
    {
      "epoch": 15.519733713742273,
      "grad_norm": 0.14760880172252655,
      "learning_rate": 4.825658939345824e-05,
      "loss": 0.0624,
      "step": 8152
    },
    {
      "epoch": 15.521635758440324,
      "grad_norm": 0.1157349944114685,
      "learning_rate": 4.825023817084789e-05,
      "loss": 0.0578,
      "step": 8153
    },
    {
      "epoch": 15.523537803138375,
      "grad_norm": 0.1379801630973816,
      "learning_rate": 4.824388694823754e-05,
      "loss": 0.0583,
      "step": 8154
    },
    {
      "epoch": 15.525439847836424,
      "grad_norm": 0.17183905839920044,
      "learning_rate": 4.823753572562719e-05,
      "loss": 0.0585,
      "step": 8155
    },
    {
      "epoch": 15.527341892534475,
      "grad_norm": 0.1357431262731552,
      "learning_rate": 4.8231184503016834e-05,
      "loss": 0.0647,
      "step": 8156
    },
    {
      "epoch": 15.529243937232525,
      "grad_norm": 0.1262008398771286,
      "learning_rate": 4.822483328040648e-05,
      "loss": 0.0488,
      "step": 8157
    },
    {
      "epoch": 15.531145981930575,
      "grad_norm": 0.1317247599363327,
      "learning_rate": 4.8218482057796124e-05,
      "loss": 0.0543,
      "step": 8158
    },
    {
      "epoch": 15.533048026628625,
      "grad_norm": 0.16477461159229279,
      "learning_rate": 4.8212130835185776e-05,
      "loss": 0.0701,
      "step": 8159
    },
    {
      "epoch": 15.534950071326676,
      "grad_norm": 0.2284783273935318,
      "learning_rate": 4.820577961257543e-05,
      "loss": 0.076,
      "step": 8160
    },
    {
      "epoch": 15.536852116024727,
      "grad_norm": 0.10719820857048035,
      "learning_rate": 4.8199428389965066e-05,
      "loss": 0.0649,
      "step": 8161
    },
    {
      "epoch": 15.538754160722776,
      "grad_norm": 0.18569643795490265,
      "learning_rate": 4.819307716735472e-05,
      "loss": 0.09,
      "step": 8162
    },
    {
      "epoch": 15.540656205420827,
      "grad_norm": 0.15722434222698212,
      "learning_rate": 4.818672594474436e-05,
      "loss": 0.0623,
      "step": 8163
    },
    {
      "epoch": 15.542558250118878,
      "grad_norm": 0.19346733391284943,
      "learning_rate": 4.8180374722134015e-05,
      "loss": 0.053,
      "step": 8164
    },
    {
      "epoch": 15.544460294816929,
      "grad_norm": 0.11424779891967773,
      "learning_rate": 4.817402349952366e-05,
      "loss": 0.0659,
      "step": 8165
    },
    {
      "epoch": 15.546362339514978,
      "grad_norm": 0.2630043923854828,
      "learning_rate": 4.8167672276913305e-05,
      "loss": 0.0562,
      "step": 8166
    },
    {
      "epoch": 15.548264384213029,
      "grad_norm": 0.12630194425582886,
      "learning_rate": 4.816132105430296e-05,
      "loss": 0.0722,
      "step": 8167
    },
    {
      "epoch": 15.55016642891108,
      "grad_norm": 0.1517813354730606,
      "learning_rate": 4.81549698316926e-05,
      "loss": 0.0565,
      "step": 8168
    },
    {
      "epoch": 15.552068473609129,
      "grad_norm": 0.11152484267950058,
      "learning_rate": 4.8148618609082254e-05,
      "loss": 0.048,
      "step": 8169
    },
    {
      "epoch": 15.55397051830718,
      "grad_norm": 0.13790766894817352,
      "learning_rate": 4.814226738647189e-05,
      "loss": 0.0543,
      "step": 8170
    },
    {
      "epoch": 15.55587256300523,
      "grad_norm": 0.1103183776140213,
      "learning_rate": 4.8135916163861544e-05,
      "loss": 0.0769,
      "step": 8171
    },
    {
      "epoch": 15.557774607703282,
      "grad_norm": 0.24293473362922668,
      "learning_rate": 4.8129564941251196e-05,
      "loss": 0.0493,
      "step": 8172
    },
    {
      "epoch": 15.55967665240133,
      "grad_norm": 0.17838114500045776,
      "learning_rate": 4.812321371864084e-05,
      "loss": 0.061,
      "step": 8173
    },
    {
      "epoch": 15.561578697099382,
      "grad_norm": 0.272238552570343,
      "learning_rate": 4.811686249603049e-05,
      "loss": 0.0522,
      "step": 8174
    },
    {
      "epoch": 15.563480741797433,
      "grad_norm": 0.18040479719638824,
      "learning_rate": 4.811051127342013e-05,
      "loss": 0.0627,
      "step": 8175
    },
    {
      "epoch": 15.565382786495483,
      "grad_norm": 0.09657354652881622,
      "learning_rate": 4.810416005080978e-05,
      "loss": 0.0507,
      "step": 8176
    },
    {
      "epoch": 15.567284831193533,
      "grad_norm": 0.120743028819561,
      "learning_rate": 4.809780882819943e-05,
      "loss": 0.0567,
      "step": 8177
    },
    {
      "epoch": 15.569186875891583,
      "grad_norm": 0.20059272646903992,
      "learning_rate": 4.809145760558908e-05,
      "loss": 0.0366,
      "step": 8178
    },
    {
      "epoch": 15.571088920589634,
      "grad_norm": 0.13962025940418243,
      "learning_rate": 4.8085106382978725e-05,
      "loss": 0.0628,
      "step": 8179
    },
    {
      "epoch": 15.572990965287683,
      "grad_norm": 0.1431548148393631,
      "learning_rate": 4.807875516036837e-05,
      "loss": 0.0571,
      "step": 8180
    },
    {
      "epoch": 15.574893009985734,
      "grad_norm": 0.17564918100833893,
      "learning_rate": 4.807240393775802e-05,
      "loss": 0.043,
      "step": 8181
    },
    {
      "epoch": 15.576795054683785,
      "grad_norm": 0.1403467208147049,
      "learning_rate": 4.806605271514767e-05,
      "loss": 0.0506,
      "step": 8182
    },
    {
      "epoch": 15.578697099381836,
      "grad_norm": 0.24989385902881622,
      "learning_rate": 4.805970149253732e-05,
      "loss": 0.0742,
      "step": 8183
    },
    {
      "epoch": 15.580599144079885,
      "grad_norm": 0.16122576594352722,
      "learning_rate": 4.8053350269926964e-05,
      "loss": 0.0517,
      "step": 8184
    },
    {
      "epoch": 15.582501188777936,
      "grad_norm": 0.19307172298431396,
      "learning_rate": 4.804699904731661e-05,
      "loss": 0.0508,
      "step": 8185
    },
    {
      "epoch": 15.584403233475987,
      "grad_norm": 0.2435937225818634,
      "learning_rate": 4.804064782470626e-05,
      "loss": 0.071,
      "step": 8186
    },
    {
      "epoch": 15.586305278174038,
      "grad_norm": 0.20309656858444214,
      "learning_rate": 4.8034296602095906e-05,
      "loss": 0.0686,
      "step": 8187
    },
    {
      "epoch": 15.588207322872087,
      "grad_norm": 0.12786833941936493,
      "learning_rate": 4.802794537948555e-05,
      "loss": 0.0542,
      "step": 8188
    },
    {
      "epoch": 15.590109367570138,
      "grad_norm": 0.21476751565933228,
      "learning_rate": 4.80215941568752e-05,
      "loss": 0.063,
      "step": 8189
    },
    {
      "epoch": 15.592011412268189,
      "grad_norm": 0.17351064085960388,
      "learning_rate": 4.801524293426485e-05,
      "loss": 0.0668,
      "step": 8190
    },
    {
      "epoch": 15.593913456966238,
      "grad_norm": 0.12176606804132462,
      "learning_rate": 4.80088917116545e-05,
      "loss": 0.0661,
      "step": 8191
    },
    {
      "epoch": 15.595815501664289,
      "grad_norm": 0.2677692174911499,
      "learning_rate": 4.8002540489044145e-05,
      "loss": 0.0556,
      "step": 8192
    },
    {
      "epoch": 15.59771754636234,
      "grad_norm": 0.23667269945144653,
      "learning_rate": 4.799618926643379e-05,
      "loss": 0.0625,
      "step": 8193
    },
    {
      "epoch": 15.59961959106039,
      "grad_norm": 0.18815861642360687,
      "learning_rate": 4.7989838043823435e-05,
      "loss": 0.0651,
      "step": 8194
    },
    {
      "epoch": 15.60152163575844,
      "grad_norm": 0.11295945942401886,
      "learning_rate": 4.7983486821213087e-05,
      "loss": 0.0463,
      "step": 8195
    },
    {
      "epoch": 15.60342368045649,
      "grad_norm": 0.25460055470466614,
      "learning_rate": 4.797713559860273e-05,
      "loss": 0.1022,
      "step": 8196
    },
    {
      "epoch": 15.605325725154541,
      "grad_norm": 0.15465350449085236,
      "learning_rate": 4.797078437599238e-05,
      "loss": 0.0502,
      "step": 8197
    },
    {
      "epoch": 15.607227769852592,
      "grad_norm": 0.21603916585445404,
      "learning_rate": 4.796443315338203e-05,
      "loss": 0.051,
      "step": 8198
    },
    {
      "epoch": 15.609129814550641,
      "grad_norm": 0.22721683979034424,
      "learning_rate": 4.7958081930771674e-05,
      "loss": 0.0852,
      "step": 8199
    },
    {
      "epoch": 15.611031859248692,
      "grad_norm": 0.11398090422153473,
      "learning_rate": 4.7951730708161325e-05,
      "loss": 0.0467,
      "step": 8200
    },
    {
      "epoch": 15.612933903946743,
      "grad_norm": 0.19684173166751862,
      "learning_rate": 4.794537948555097e-05,
      "loss": 0.0827,
      "step": 8201
    },
    {
      "epoch": 15.614835948644792,
      "grad_norm": 0.12274474650621414,
      "learning_rate": 4.7939028262940616e-05,
      "loss": 0.0463,
      "step": 8202
    },
    {
      "epoch": 15.616737993342843,
      "grad_norm": 0.12376654893159866,
      "learning_rate": 4.793267704033027e-05,
      "loss": 0.0533,
      "step": 8203
    },
    {
      "epoch": 15.618640038040894,
      "grad_norm": 0.17132851481437683,
      "learning_rate": 4.792632581771991e-05,
      "loss": 0.0524,
      "step": 8204
    },
    {
      "epoch": 15.620542082738945,
      "grad_norm": 0.2081228792667389,
      "learning_rate": 4.7919974595109564e-05,
      "loss": 0.0686,
      "step": 8205
    },
    {
      "epoch": 15.622444127436994,
      "grad_norm": 0.18067365884780884,
      "learning_rate": 4.79136233724992e-05,
      "loss": 0.0492,
      "step": 8206
    },
    {
      "epoch": 15.624346172135045,
      "grad_norm": 0.24764224886894226,
      "learning_rate": 4.7907272149888854e-05,
      "loss": 0.0579,
      "step": 8207
    },
    {
      "epoch": 15.626248216833096,
      "grad_norm": 0.2526845932006836,
      "learning_rate": 4.7900920927278506e-05,
      "loss": 0.0663,
      "step": 8208
    },
    {
      "epoch": 15.628150261531147,
      "grad_norm": 0.1451825648546219,
      "learning_rate": 4.789456970466815e-05,
      "loss": 0.0677,
      "step": 8209
    },
    {
      "epoch": 15.630052306229196,
      "grad_norm": 0.1915256679058075,
      "learning_rate": 4.78882184820578e-05,
      "loss": 0.06,
      "step": 8210
    },
    {
      "epoch": 15.631954350927247,
      "grad_norm": 0.27265143394470215,
      "learning_rate": 4.788186725944744e-05,
      "loss": 0.0555,
      "step": 8211
    },
    {
      "epoch": 15.633856395625298,
      "grad_norm": 0.19599300622940063,
      "learning_rate": 4.787551603683709e-05,
      "loss": 0.0525,
      "step": 8212
    },
    {
      "epoch": 15.635758440323347,
      "grad_norm": 0.24216334521770477,
      "learning_rate": 4.786916481422674e-05,
      "loss": 0.0639,
      "step": 8213
    },
    {
      "epoch": 15.637660485021398,
      "grad_norm": 0.22544489800930023,
      "learning_rate": 4.786281359161639e-05,
      "loss": 0.0661,
      "step": 8214
    },
    {
      "epoch": 15.639562529719448,
      "grad_norm": 0.15521033108234406,
      "learning_rate": 4.7856462369006035e-05,
      "loss": 0.0526,
      "step": 8215
    },
    {
      "epoch": 15.6414645744175,
      "grad_norm": 0.20516115427017212,
      "learning_rate": 4.785011114639568e-05,
      "loss": 0.0704,
      "step": 8216
    },
    {
      "epoch": 15.643366619115548,
      "grad_norm": 0.19163091480731964,
      "learning_rate": 4.784375992378533e-05,
      "loss": 0.0649,
      "step": 8217
    },
    {
      "epoch": 15.6452686638136,
      "grad_norm": 0.2147558182477951,
      "learning_rate": 4.783740870117498e-05,
      "loss": 0.0473,
      "step": 8218
    },
    {
      "epoch": 15.64717070851165,
      "grad_norm": 0.1446150541305542,
      "learning_rate": 4.783105747856463e-05,
      "loss": 0.1167,
      "step": 8219
    },
    {
      "epoch": 15.649072753209701,
      "grad_norm": 0.2023557424545288,
      "learning_rate": 4.7824706255954274e-05,
      "loss": 0.0642,
      "step": 8220
    },
    {
      "epoch": 15.65097479790775,
      "grad_norm": 0.1383785754442215,
      "learning_rate": 4.781835503334392e-05,
      "loss": 0.0345,
      "step": 8221
    },
    {
      "epoch": 15.652876842605801,
      "grad_norm": 0.15243691205978394,
      "learning_rate": 4.781200381073357e-05,
      "loss": 0.053,
      "step": 8222
    },
    {
      "epoch": 15.654778887303852,
      "grad_norm": 0.11674341559410095,
      "learning_rate": 4.7805652588123216e-05,
      "loss": 0.0624,
      "step": 8223
    },
    {
      "epoch": 15.656680932001901,
      "grad_norm": 0.2048717737197876,
      "learning_rate": 4.779930136551286e-05,
      "loss": 0.0743,
      "step": 8224
    },
    {
      "epoch": 15.658582976699952,
      "grad_norm": 0.3081852197647095,
      "learning_rate": 4.7792950142902506e-05,
      "loss": 0.0547,
      "step": 8225
    },
    {
      "epoch": 15.660485021398003,
      "grad_norm": 0.20836210250854492,
      "learning_rate": 4.778659892029216e-05,
      "loss": 0.0607,
      "step": 8226
    },
    {
      "epoch": 15.662387066096054,
      "grad_norm": 0.22958265244960785,
      "learning_rate": 4.778024769768181e-05,
      "loss": 0.0693,
      "step": 8227
    },
    {
      "epoch": 15.664289110794103,
      "grad_norm": 0.18431368470191956,
      "learning_rate": 4.7773896475071455e-05,
      "loss": 0.0737,
      "step": 8228
    },
    {
      "epoch": 15.666191155492154,
      "grad_norm": 0.18889397382736206,
      "learning_rate": 4.77675452524611e-05,
      "loss": 0.0923,
      "step": 8229
    },
    {
      "epoch": 15.668093200190205,
      "grad_norm": 0.17911040782928467,
      "learning_rate": 4.7761194029850745e-05,
      "loss": 0.0811,
      "step": 8230
    },
    {
      "epoch": 15.669995244888256,
      "grad_norm": 0.13110655546188354,
      "learning_rate": 4.77548428072404e-05,
      "loss": 0.0603,
      "step": 8231
    },
    {
      "epoch": 15.671897289586305,
      "grad_norm": 0.08929801732301712,
      "learning_rate": 4.774849158463004e-05,
      "loss": 0.043,
      "step": 8232
    },
    {
      "epoch": 15.673799334284356,
      "grad_norm": 0.2299952208995819,
      "learning_rate": 4.774214036201969e-05,
      "loss": 0.0659,
      "step": 8233
    },
    {
      "epoch": 15.675701378982406,
      "grad_norm": 0.3628097176551819,
      "learning_rate": 4.773578913940934e-05,
      "loss": 0.2409,
      "step": 8234
    },
    {
      "epoch": 15.677603423680456,
      "grad_norm": 0.08881036937236786,
      "learning_rate": 4.7729437916798984e-05,
      "loss": 0.0798,
      "step": 8235
    },
    {
      "epoch": 15.679505468378506,
      "grad_norm": 0.14017745852470398,
      "learning_rate": 4.7723086694188636e-05,
      "loss": 0.0633,
      "step": 8236
    },
    {
      "epoch": 15.681407513076557,
      "grad_norm": 0.13978202641010284,
      "learning_rate": 4.771673547157828e-05,
      "loss": 0.0595,
      "step": 8237
    },
    {
      "epoch": 15.683309557774608,
      "grad_norm": 0.28331878781318665,
      "learning_rate": 4.7710384248967926e-05,
      "loss": 0.0717,
      "step": 8238
    },
    {
      "epoch": 15.685211602472657,
      "grad_norm": 0.1580718457698822,
      "learning_rate": 4.770403302635758e-05,
      "loss": 0.0481,
      "step": 8239
    },
    {
      "epoch": 15.687113647170708,
      "grad_norm": 0.2504642605781555,
      "learning_rate": 4.769768180374722e-05,
      "loss": 0.0909,
      "step": 8240
    },
    {
      "epoch": 15.689015691868759,
      "grad_norm": 0.16315074265003204,
      "learning_rate": 4.7691330581136875e-05,
      "loss": 0.0507,
      "step": 8241
    },
    {
      "epoch": 15.69091773656681,
      "grad_norm": 0.28321632742881775,
      "learning_rate": 4.768497935852651e-05,
      "loss": 0.0744,
      "step": 8242
    },
    {
      "epoch": 15.692819781264859,
      "grad_norm": 0.2194773256778717,
      "learning_rate": 4.7678628135916165e-05,
      "loss": 0.0617,
      "step": 8243
    },
    {
      "epoch": 15.69472182596291,
      "grad_norm": 0.17935332655906677,
      "learning_rate": 4.767227691330582e-05,
      "loss": 0.0623,
      "step": 8244
    },
    {
      "epoch": 15.69662387066096,
      "grad_norm": 0.17090705037117004,
      "learning_rate": 4.766592569069546e-05,
      "loss": 0.0495,
      "step": 8245
    },
    {
      "epoch": 15.698525915359012,
      "grad_norm": 0.2691819965839386,
      "learning_rate": 4.7659574468085114e-05,
      "loss": 0.0541,
      "step": 8246
    },
    {
      "epoch": 15.70042796005706,
      "grad_norm": 0.12617425620555878,
      "learning_rate": 4.765322324547475e-05,
      "loss": 0.0743,
      "step": 8247
    },
    {
      "epoch": 15.702330004755112,
      "grad_norm": 0.14223243296146393,
      "learning_rate": 4.7646872022864404e-05,
      "loss": 0.0489,
      "step": 8248
    },
    {
      "epoch": 15.704232049453163,
      "grad_norm": 0.13523830473423004,
      "learning_rate": 4.764052080025405e-05,
      "loss": 0.0681,
      "step": 8249
    },
    {
      "epoch": 15.706134094151212,
      "grad_norm": 0.19689877331256866,
      "learning_rate": 4.76341695776437e-05,
      "loss": 0.0675,
      "step": 8250
    },
    {
      "epoch": 15.708036138849263,
      "grad_norm": 0.313142329454422,
      "learning_rate": 4.7627818355033346e-05,
      "loss": 0.0875,
      "step": 8251
    },
    {
      "epoch": 15.709938183547314,
      "grad_norm": 0.12644127011299133,
      "learning_rate": 4.762146713242299e-05,
      "loss": 0.056,
      "step": 8252
    },
    {
      "epoch": 15.711840228245364,
      "grad_norm": 0.11934950947761536,
      "learning_rate": 4.761511590981264e-05,
      "loss": 0.0783,
      "step": 8253
    },
    {
      "epoch": 15.713742272943414,
      "grad_norm": 0.33159083127975464,
      "learning_rate": 4.760876468720229e-05,
      "loss": 0.0543,
      "step": 8254
    },
    {
      "epoch": 15.715644317641464,
      "grad_norm": 0.18594646453857422,
      "learning_rate": 4.760241346459194e-05,
      "loss": 0.0666,
      "step": 8255
    },
    {
      "epoch": 15.717546362339515,
      "grad_norm": 0.13968947529792786,
      "learning_rate": 4.7596062241981585e-05,
      "loss": 0.0574,
      "step": 8256
    },
    {
      "epoch": 15.719448407037566,
      "grad_norm": 0.18439897894859314,
      "learning_rate": 4.758971101937123e-05,
      "loss": 0.0547,
      "step": 8257
    },
    {
      "epoch": 15.721350451735615,
      "grad_norm": 0.25268253684043884,
      "learning_rate": 4.758335979676088e-05,
      "loss": 0.064,
      "step": 8258
    },
    {
      "epoch": 15.723252496433666,
      "grad_norm": 0.15728792548179626,
      "learning_rate": 4.7577008574150527e-05,
      "loss": 0.073,
      "step": 8259
    },
    {
      "epoch": 15.725154541131717,
      "grad_norm": 0.2027149647474289,
      "learning_rate": 4.757065735154017e-05,
      "loss": 0.0667,
      "step": 8260
    },
    {
      "epoch": 15.727056585829766,
      "grad_norm": 0.12121910601854324,
      "learning_rate": 4.756430612892982e-05,
      "loss": 0.0549,
      "step": 8261
    },
    {
      "epoch": 15.728958630527817,
      "grad_norm": 0.09283791482448578,
      "learning_rate": 4.755795490631947e-05,
      "loss": 0.0552,
      "step": 8262
    },
    {
      "epoch": 15.730860675225868,
      "grad_norm": 0.1861935704946518,
      "learning_rate": 4.755160368370912e-05,
      "loss": 0.0749,
      "step": 8263
    },
    {
      "epoch": 15.732762719923919,
      "grad_norm": 0.14479346573352814,
      "learning_rate": 4.7545252461098765e-05,
      "loss": 0.0667,
      "step": 8264
    },
    {
      "epoch": 15.734664764621968,
      "grad_norm": 0.08699013292789459,
      "learning_rate": 4.753890123848841e-05,
      "loss": 0.0642,
      "step": 8265
    },
    {
      "epoch": 15.736566809320019,
      "grad_norm": 0.11008250713348389,
      "learning_rate": 4.7532550015878056e-05,
      "loss": 0.05,
      "step": 8266
    },
    {
      "epoch": 15.73846885401807,
      "grad_norm": 0.17910899221897125,
      "learning_rate": 4.752619879326771e-05,
      "loss": 0.0991,
      "step": 8267
    },
    {
      "epoch": 15.74037089871612,
      "grad_norm": 0.15026019513607025,
      "learning_rate": 4.751984757065735e-05,
      "loss": 0.0391,
      "step": 8268
    },
    {
      "epoch": 15.74227294341417,
      "grad_norm": 0.2448321282863617,
      "learning_rate": 4.7513496348047e-05,
      "loss": 0.0644,
      "step": 8269
    },
    {
      "epoch": 15.74417498811222,
      "grad_norm": 0.2921144962310791,
      "learning_rate": 4.750714512543665e-05,
      "loss": 0.0709,
      "step": 8270
    },
    {
      "epoch": 15.746077032810271,
      "grad_norm": 0.14855507016181946,
      "learning_rate": 4.7500793902826294e-05,
      "loss": 0.0529,
      "step": 8271
    },
    {
      "epoch": 15.747979077508322,
      "grad_norm": 0.17389892041683197,
      "learning_rate": 4.7494442680215946e-05,
      "loss": 0.0653,
      "step": 8272
    },
    {
      "epoch": 15.749881122206371,
      "grad_norm": 0.204713374376297,
      "learning_rate": 4.748809145760559e-05,
      "loss": 0.0572,
      "step": 8273
    },
    {
      "epoch": 15.751783166904422,
      "grad_norm": 0.22001193463802338,
      "learning_rate": 4.7481740234995236e-05,
      "loss": 0.0698,
      "step": 8274
    },
    {
      "epoch": 15.753685211602473,
      "grad_norm": 0.15850436687469482,
      "learning_rate": 4.747538901238489e-05,
      "loss": 0.0673,
      "step": 8275
    },
    {
      "epoch": 15.755587256300522,
      "grad_norm": 0.11191987991333008,
      "learning_rate": 4.746903778977453e-05,
      "loss": 0.0405,
      "step": 8276
    },
    {
      "epoch": 15.757489300998573,
      "grad_norm": 0.08066005259752274,
      "learning_rate": 4.7462686567164185e-05,
      "loss": 0.0773,
      "step": 8277
    },
    {
      "epoch": 15.759391345696624,
      "grad_norm": 0.11931698024272919,
      "learning_rate": 4.7456335344553823e-05,
      "loss": 0.06,
      "step": 8278
    },
    {
      "epoch": 15.761293390394675,
      "grad_norm": 0.16694439947605133,
      "learning_rate": 4.7449984121943475e-05,
      "loss": 0.0635,
      "step": 8279
    },
    {
      "epoch": 15.763195435092724,
      "grad_norm": 0.2340782880783081,
      "learning_rate": 4.744363289933312e-05,
      "loss": 0.0654,
      "step": 8280
    },
    {
      "epoch": 15.765097479790775,
      "grad_norm": 0.15663287043571472,
      "learning_rate": 4.743728167672277e-05,
      "loss": 0.058,
      "step": 8281
    },
    {
      "epoch": 15.766999524488826,
      "grad_norm": 0.19084382057189941,
      "learning_rate": 4.7430930454112424e-05,
      "loss": 0.0742,
      "step": 8282
    },
    {
      "epoch": 15.768901569186877,
      "grad_norm": 0.21217328310012817,
      "learning_rate": 4.742457923150206e-05,
      "loss": 0.0877,
      "step": 8283
    },
    {
      "epoch": 15.770803613884926,
      "grad_norm": 0.1141515001654625,
      "learning_rate": 4.7418228008891714e-05,
      "loss": 0.0577,
      "step": 8284
    },
    {
      "epoch": 15.772705658582977,
      "grad_norm": 0.26023924350738525,
      "learning_rate": 4.741187678628136e-05,
      "loss": 0.0671,
      "step": 8285
    },
    {
      "epoch": 15.774607703281028,
      "grad_norm": 0.057885292917490005,
      "learning_rate": 4.740552556367101e-05,
      "loss": 0.0479,
      "step": 8286
    },
    {
      "epoch": 15.776509747979077,
      "grad_norm": 0.16264577209949493,
      "learning_rate": 4.7399174341060656e-05,
      "loss": 0.0642,
      "step": 8287
    },
    {
      "epoch": 15.778411792677128,
      "grad_norm": 0.18721486628055573,
      "learning_rate": 4.73928231184503e-05,
      "loss": 0.0642,
      "step": 8288
    },
    {
      "epoch": 15.780313837375179,
      "grad_norm": 0.15313813090324402,
      "learning_rate": 4.738647189583995e-05,
      "loss": 0.0543,
      "step": 8289
    },
    {
      "epoch": 15.78221588207323,
      "grad_norm": 0.13567408919334412,
      "learning_rate": 4.73801206732296e-05,
      "loss": 0.0395,
      "step": 8290
    },
    {
      "epoch": 15.784117926771279,
      "grad_norm": 0.14394468069076538,
      "learning_rate": 4.737376945061925e-05,
      "loss": 0.0514,
      "step": 8291
    },
    {
      "epoch": 15.78601997146933,
      "grad_norm": 0.1596677452325821,
      "learning_rate": 4.7367418228008895e-05,
      "loss": 0.0505,
      "step": 8292
    },
    {
      "epoch": 15.78792201616738,
      "grad_norm": 0.19569194316864014,
      "learning_rate": 4.736106700539854e-05,
      "loss": 0.0541,
      "step": 8293
    },
    {
      "epoch": 15.789824060865431,
      "grad_norm": 0.14746233820915222,
      "learning_rate": 4.735471578278819e-05,
      "loss": 0.0654,
      "step": 8294
    },
    {
      "epoch": 15.79172610556348,
      "grad_norm": 0.13105404376983643,
      "learning_rate": 4.734836456017784e-05,
      "loss": 0.0886,
      "step": 8295
    },
    {
      "epoch": 15.793628150261531,
      "grad_norm": 0.18374143540859222,
      "learning_rate": 4.734201333756748e-05,
      "loss": 0.0623,
      "step": 8296
    },
    {
      "epoch": 15.795530194959582,
      "grad_norm": 0.11379442363977432,
      "learning_rate": 4.733566211495713e-05,
      "loss": 0.0496,
      "step": 8297
    },
    {
      "epoch": 15.797432239657631,
      "grad_norm": 0.20474670827388763,
      "learning_rate": 4.732931089234678e-05,
      "loss": 0.0627,
      "step": 8298
    },
    {
      "epoch": 15.799334284355682,
      "grad_norm": 0.2521989941596985,
      "learning_rate": 4.732295966973643e-05,
      "loss": 0.0693,
      "step": 8299
    },
    {
      "epoch": 15.801236329053733,
      "grad_norm": 0.19155509769916534,
      "learning_rate": 4.7316608447126076e-05,
      "loss": 0.0747,
      "step": 8300
    },
    {
      "epoch": 15.803138373751784,
      "grad_norm": 0.2652791440486908,
      "learning_rate": 4.731025722451572e-05,
      "loss": 0.0554,
      "step": 8301
    },
    {
      "epoch": 15.805040418449833,
      "grad_norm": 0.20905011892318726,
      "learning_rate": 4.7303906001905366e-05,
      "loss": 0.0778,
      "step": 8302
    },
    {
      "epoch": 15.806942463147884,
      "grad_norm": 0.25589483976364136,
      "learning_rate": 4.729755477929502e-05,
      "loss": 0.0563,
      "step": 8303
    },
    {
      "epoch": 15.808844507845935,
      "grad_norm": 0.1645364761352539,
      "learning_rate": 4.729120355668466e-05,
      "loss": 0.0592,
      "step": 8304
    },
    {
      "epoch": 15.810746552543986,
      "grad_norm": 0.12126277387142181,
      "learning_rate": 4.728485233407431e-05,
      "loss": 0.0513,
      "step": 8305
    },
    {
      "epoch": 15.812648597242035,
      "grad_norm": 0.21371135115623474,
      "learning_rate": 4.727850111146396e-05,
      "loss": 0.0693,
      "step": 8306
    },
    {
      "epoch": 15.814550641940086,
      "grad_norm": 0.19225803017616272,
      "learning_rate": 4.7272149888853605e-05,
      "loss": 0.0603,
      "step": 8307
    },
    {
      "epoch": 15.816452686638137,
      "grad_norm": 0.24223649501800537,
      "learning_rate": 4.726579866624326e-05,
      "loss": 0.0554,
      "step": 8308
    },
    {
      "epoch": 15.818354731336186,
      "grad_norm": 0.1132277324795723,
      "learning_rate": 4.72594474436329e-05,
      "loss": 0.0642,
      "step": 8309
    },
    {
      "epoch": 15.820256776034237,
      "grad_norm": 0.2975490391254425,
      "learning_rate": 4.725309622102255e-05,
      "loss": 0.0759,
      "step": 8310
    },
    {
      "epoch": 15.822158820732287,
      "grad_norm": 0.19871625304222107,
      "learning_rate": 4.72467449984122e-05,
      "loss": 0.0671,
      "step": 8311
    },
    {
      "epoch": 15.824060865430338,
      "grad_norm": 0.29103484749794006,
      "learning_rate": 4.7240393775801844e-05,
      "loss": 0.0596,
      "step": 8312
    },
    {
      "epoch": 15.825962910128387,
      "grad_norm": 0.11662924289703369,
      "learning_rate": 4.7234042553191496e-05,
      "loss": 0.064,
      "step": 8313
    },
    {
      "epoch": 15.827864954826438,
      "grad_norm": 0.18209360539913177,
      "learning_rate": 4.7227691330581134e-05,
      "loss": 0.0703,
      "step": 8314
    },
    {
      "epoch": 15.82976699952449,
      "grad_norm": 0.14348343014717102,
      "learning_rate": 4.7221340107970786e-05,
      "loss": 0.0554,
      "step": 8315
    },
    {
      "epoch": 15.83166904422254,
      "grad_norm": 0.23151347041130066,
      "learning_rate": 4.721498888536043e-05,
      "loss": 0.0703,
      "step": 8316
    },
    {
      "epoch": 15.83357108892059,
      "grad_norm": 0.16555550694465637,
      "learning_rate": 4.720863766275008e-05,
      "loss": 0.0641,
      "step": 8317
    },
    {
      "epoch": 15.83547313361864,
      "grad_norm": 0.1724887490272522,
      "learning_rate": 4.7202286440139734e-05,
      "loss": 0.0795,
      "step": 8318
    },
    {
      "epoch": 15.837375178316691,
      "grad_norm": 0.1328611522912979,
      "learning_rate": 4.719593521752937e-05,
      "loss": 0.0583,
      "step": 8319
    },
    {
      "epoch": 15.83927722301474,
      "grad_norm": 0.3197205066680908,
      "learning_rate": 4.7189583994919025e-05,
      "loss": 0.0638,
      "step": 8320
    },
    {
      "epoch": 15.841179267712791,
      "grad_norm": 0.25061967968940735,
      "learning_rate": 4.718323277230867e-05,
      "loss": 0.0644,
      "step": 8321
    },
    {
      "epoch": 15.843081312410842,
      "grad_norm": 0.26406586170196533,
      "learning_rate": 4.717688154969832e-05,
      "loss": 0.0502,
      "step": 8322
    },
    {
      "epoch": 15.844983357108893,
      "grad_norm": 0.15122944116592407,
      "learning_rate": 4.7170530327087967e-05,
      "loss": 0.0611,
      "step": 8323
    },
    {
      "epoch": 15.846885401806942,
      "grad_norm": 0.1766766756772995,
      "learning_rate": 4.716417910447761e-05,
      "loss": 0.0659,
      "step": 8324
    },
    {
      "epoch": 15.848787446504993,
      "grad_norm": 0.20306788384914398,
      "learning_rate": 4.7157827881867263e-05,
      "loss": 0.0619,
      "step": 8325
    },
    {
      "epoch": 15.850689491203044,
      "grad_norm": 0.13225118815898895,
      "learning_rate": 4.715147665925691e-05,
      "loss": 0.0666,
      "step": 8326
    },
    {
      "epoch": 15.852591535901094,
      "grad_norm": 0.19744844734668732,
      "learning_rate": 4.714512543664656e-05,
      "loss": 0.0541,
      "step": 8327
    },
    {
      "epoch": 15.854493580599144,
      "grad_norm": 0.2040027529001236,
      "learning_rate": 4.7138774214036205e-05,
      "loss": 0.0675,
      "step": 8328
    },
    {
      "epoch": 15.856395625297194,
      "grad_norm": 0.28461000323295593,
      "learning_rate": 4.713242299142585e-05,
      "loss": 0.0574,
      "step": 8329
    },
    {
      "epoch": 15.858297669995245,
      "grad_norm": 0.13227978348731995,
      "learning_rate": 4.71260717688155e-05,
      "loss": 0.0588,
      "step": 8330
    },
    {
      "epoch": 15.860199714693294,
      "grad_norm": 0.16374631226062775,
      "learning_rate": 4.711972054620515e-05,
      "loss": 0.0591,
      "step": 8331
    },
    {
      "epoch": 15.862101759391345,
      "grad_norm": 0.22766095399856567,
      "learning_rate": 4.711336932359479e-05,
      "loss": 0.0668,
      "step": 8332
    },
    {
      "epoch": 15.864003804089396,
      "grad_norm": 0.19878962635993958,
      "learning_rate": 4.710701810098444e-05,
      "loss": 0.053,
      "step": 8333
    },
    {
      "epoch": 15.865905848787447,
      "grad_norm": 0.14449188113212585,
      "learning_rate": 4.710066687837409e-05,
      "loss": 0.0646,
      "step": 8334
    },
    {
      "epoch": 15.867807893485496,
      "grad_norm": 0.1568823605775833,
      "learning_rate": 4.7094315655763734e-05,
      "loss": 0.0569,
      "step": 8335
    },
    {
      "epoch": 15.869709938183547,
      "grad_norm": 0.14062386751174927,
      "learning_rate": 4.7087964433153386e-05,
      "loss": 0.0646,
      "step": 8336
    },
    {
      "epoch": 15.871611982881598,
      "grad_norm": 0.17233572900295258,
      "learning_rate": 4.708161321054303e-05,
      "loss": 0.0679,
      "step": 8337
    },
    {
      "epoch": 15.873514027579649,
      "grad_norm": 0.12581130862236023,
      "learning_rate": 4.7075261987932676e-05,
      "loss": 0.0428,
      "step": 8338
    },
    {
      "epoch": 15.875416072277698,
      "grad_norm": 0.161576509475708,
      "learning_rate": 4.706891076532233e-05,
      "loss": 0.0585,
      "step": 8339
    },
    {
      "epoch": 15.877318116975749,
      "grad_norm": 0.22900627553462982,
      "learning_rate": 4.706255954271197e-05,
      "loss": 0.0615,
      "step": 8340
    },
    {
      "epoch": 15.8792201616738,
      "grad_norm": 0.17060215771198273,
      "learning_rate": 4.705620832010162e-05,
      "loss": 0.0928,
      "step": 8341
    },
    {
      "epoch": 15.881122206371849,
      "grad_norm": 0.3081725835800171,
      "learning_rate": 4.704985709749127e-05,
      "loss": 0.0605,
      "step": 8342
    },
    {
      "epoch": 15.8830242510699,
      "grad_norm": 0.24117806553840637,
      "learning_rate": 4.7043505874880915e-05,
      "loss": 0.0799,
      "step": 8343
    },
    {
      "epoch": 15.88492629576795,
      "grad_norm": 0.1827300786972046,
      "learning_rate": 4.703715465227057e-05,
      "loss": 0.077,
      "step": 8344
    },
    {
      "epoch": 15.886828340466002,
      "grad_norm": 0.11569055914878845,
      "learning_rate": 4.703080342966021e-05,
      "loss": 0.0562,
      "step": 8345
    },
    {
      "epoch": 15.88873038516405,
      "grad_norm": 0.17954939603805542,
      "learning_rate": 4.702445220704986e-05,
      "loss": 0.0658,
      "step": 8346
    },
    {
      "epoch": 15.890632429862102,
      "grad_norm": 0.26830634474754333,
      "learning_rate": 4.701810098443951e-05,
      "loss": 0.0567,
      "step": 8347
    },
    {
      "epoch": 15.892534474560152,
      "grad_norm": 0.14093312621116638,
      "learning_rate": 4.7011749761829154e-05,
      "loss": 0.064,
      "step": 8348
    },
    {
      "epoch": 15.894436519258203,
      "grad_norm": 0.12327145040035248,
      "learning_rate": 4.7005398539218806e-05,
      "loss": 0.0519,
      "step": 8349
    },
    {
      "epoch": 15.896338563956252,
      "grad_norm": 0.29526394605636597,
      "learning_rate": 4.6999047316608444e-05,
      "loss": 0.0596,
      "step": 8350
    },
    {
      "epoch": 15.898240608654303,
      "grad_norm": 0.16847003996372223,
      "learning_rate": 4.6992696093998096e-05,
      "loss": 0.0742,
      "step": 8351
    },
    {
      "epoch": 15.900142653352354,
      "grad_norm": 0.10561301559209824,
      "learning_rate": 4.698634487138774e-05,
      "loss": 0.0378,
      "step": 8352
    },
    {
      "epoch": 15.902044698050403,
      "grad_norm": 0.15217702090740204,
      "learning_rate": 4.697999364877739e-05,
      "loss": 0.0701,
      "step": 8353
    },
    {
      "epoch": 15.903946742748454,
      "grad_norm": 0.15947742760181427,
      "learning_rate": 4.6973642426167045e-05,
      "loss": 0.0618,
      "step": 8354
    },
    {
      "epoch": 15.905848787446505,
      "grad_norm": 0.08902820944786072,
      "learning_rate": 4.696729120355668e-05,
      "loss": 0.041,
      "step": 8355
    },
    {
      "epoch": 15.907750832144556,
      "grad_norm": 0.11983463168144226,
      "learning_rate": 4.6960939980946335e-05,
      "loss": 0.0615,
      "step": 8356
    },
    {
      "epoch": 15.909652876842605,
      "grad_norm": 0.3213430941104889,
      "learning_rate": 4.695458875833598e-05,
      "loss": 0.0635,
      "step": 8357
    },
    {
      "epoch": 15.911554921540656,
      "grad_norm": 0.10112504661083221,
      "learning_rate": 4.694823753572563e-05,
      "loss": 0.0647,
      "step": 8358
    },
    {
      "epoch": 15.913456966238707,
      "grad_norm": 0.1713016778230667,
      "learning_rate": 4.694188631311528e-05,
      "loss": 0.0705,
      "step": 8359
    },
    {
      "epoch": 15.915359010936758,
      "grad_norm": 0.18278667330741882,
      "learning_rate": 4.693553509050492e-05,
      "loss": 0.0858,
      "step": 8360
    },
    {
      "epoch": 15.917261055634807,
      "grad_norm": 0.23004773259162903,
      "learning_rate": 4.6929183867894574e-05,
      "loss": 0.0774,
      "step": 8361
    },
    {
      "epoch": 15.919163100332858,
      "grad_norm": 0.13863714039325714,
      "learning_rate": 4.692283264528422e-05,
      "loss": 0.056,
      "step": 8362
    },
    {
      "epoch": 15.921065145030909,
      "grad_norm": 0.17896798253059387,
      "learning_rate": 4.691648142267387e-05,
      "loss": 0.0583,
      "step": 8363
    },
    {
      "epoch": 15.922967189728958,
      "grad_norm": 0.1130218356847763,
      "learning_rate": 4.691013020006351e-05,
      "loss": 0.0527,
      "step": 8364
    },
    {
      "epoch": 15.924869234427009,
      "grad_norm": 0.15107358992099762,
      "learning_rate": 4.690377897745316e-05,
      "loss": 0.0662,
      "step": 8365
    },
    {
      "epoch": 15.92677127912506,
      "grad_norm": 0.1794845461845398,
      "learning_rate": 4.689742775484281e-05,
      "loss": 0.0569,
      "step": 8366
    },
    {
      "epoch": 15.92867332382311,
      "grad_norm": 0.17926360666751862,
      "learning_rate": 4.689107653223246e-05,
      "loss": 0.0607,
      "step": 8367
    },
    {
      "epoch": 15.93057536852116,
      "grad_norm": 0.19702889025211334,
      "learning_rate": 4.68847253096221e-05,
      "loss": 0.0728,
      "step": 8368
    },
    {
      "epoch": 15.93247741321921,
      "grad_norm": 0.13820995390415192,
      "learning_rate": 4.687837408701175e-05,
      "loss": 0.0469,
      "step": 8369
    },
    {
      "epoch": 15.934379457917261,
      "grad_norm": 0.16547589004039764,
      "learning_rate": 4.68720228644014e-05,
      "loss": 0.0393,
      "step": 8370
    },
    {
      "epoch": 15.936281502615312,
      "grad_norm": 0.17976239323616028,
      "learning_rate": 4.6865671641791045e-05,
      "loss": 0.0542,
      "step": 8371
    },
    {
      "epoch": 15.938183547313361,
      "grad_norm": 0.24729850888252258,
      "learning_rate": 4.68593204191807e-05,
      "loss": 0.0729,
      "step": 8372
    },
    {
      "epoch": 15.940085592011412,
      "grad_norm": 0.09225702285766602,
      "learning_rate": 4.685296919657034e-05,
      "loss": 0.0748,
      "step": 8373
    },
    {
      "epoch": 15.941987636709463,
      "grad_norm": 0.27477720379829407,
      "learning_rate": 4.684661797395999e-05,
      "loss": 0.0441,
      "step": 8374
    },
    {
      "epoch": 15.943889681407512,
      "grad_norm": 0.10589617490768433,
      "learning_rate": 4.684026675134964e-05,
      "loss": 0.0509,
      "step": 8375
    },
    {
      "epoch": 15.945791726105563,
      "grad_norm": 0.21618284285068512,
      "learning_rate": 4.6833915528739284e-05,
      "loss": 0.066,
      "step": 8376
    },
    {
      "epoch": 15.947693770803614,
      "grad_norm": 0.11763995885848999,
      "learning_rate": 4.682756430612893e-05,
      "loss": 0.0516,
      "step": 8377
    },
    {
      "epoch": 15.949595815501665,
      "grad_norm": 0.10359404236078262,
      "learning_rate": 4.682121308351858e-05,
      "loss": 0.054,
      "step": 8378
    },
    {
      "epoch": 15.951497860199714,
      "grad_norm": 0.2101137787103653,
      "learning_rate": 4.6814861860908226e-05,
      "loss": 0.0736,
      "step": 8379
    },
    {
      "epoch": 15.953399904897765,
      "grad_norm": 0.14261220395565033,
      "learning_rate": 4.680851063829788e-05,
      "loss": 0.058,
      "step": 8380
    },
    {
      "epoch": 15.955301949595816,
      "grad_norm": 0.18092776834964752,
      "learning_rate": 4.680215941568752e-05,
      "loss": 0.0661,
      "step": 8381
    },
    {
      "epoch": 15.957203994293867,
      "grad_norm": 0.15513984858989716,
      "learning_rate": 4.679580819307717e-05,
      "loss": 0.056,
      "step": 8382
    },
    {
      "epoch": 15.959106038991916,
      "grad_norm": 0.26116275787353516,
      "learning_rate": 4.678945697046682e-05,
      "loss": 0.0685,
      "step": 8383
    },
    {
      "epoch": 15.961008083689967,
      "grad_norm": 0.22781796753406525,
      "learning_rate": 4.6783105747856465e-05,
      "loss": 0.0521,
      "step": 8384
    },
    {
      "epoch": 15.962910128388017,
      "grad_norm": 0.26958703994750977,
      "learning_rate": 4.6776754525246116e-05,
      "loss": 0.0458,
      "step": 8385
    },
    {
      "epoch": 15.964812173086067,
      "grad_norm": 0.14877475798130035,
      "learning_rate": 4.6770403302635755e-05,
      "loss": 0.0831,
      "step": 8386
    },
    {
      "epoch": 15.966714217784117,
      "grad_norm": 0.2953677773475647,
      "learning_rate": 4.676405208002541e-05,
      "loss": 0.0814,
      "step": 8387
    },
    {
      "epoch": 15.968616262482168,
      "grad_norm": 0.12530766427516937,
      "learning_rate": 4.675770085741505e-05,
      "loss": 0.0578,
      "step": 8388
    },
    {
      "epoch": 15.97051830718022,
      "grad_norm": 0.1906973272562027,
      "learning_rate": 4.6751349634804704e-05,
      "loss": 0.063,
      "step": 8389
    },
    {
      "epoch": 15.972420351878268,
      "grad_norm": 0.15123619139194489,
      "learning_rate": 4.674499841219435e-05,
      "loss": 0.0526,
      "step": 8390
    },
    {
      "epoch": 15.97432239657632,
      "grad_norm": 0.2910292148590088,
      "learning_rate": 4.6738647189583994e-05,
      "loss": 0.0715,
      "step": 8391
    },
    {
      "epoch": 15.97622444127437,
      "grad_norm": 0.13333997130393982,
      "learning_rate": 4.6732295966973646e-05,
      "loss": 0.041,
      "step": 8392
    },
    {
      "epoch": 15.978126485972421,
      "grad_norm": 0.16418004035949707,
      "learning_rate": 4.672594474436329e-05,
      "loss": 0.0784,
      "step": 8393
    },
    {
      "epoch": 15.98002853067047,
      "grad_norm": 0.19421927630901337,
      "learning_rate": 4.671959352175294e-05,
      "loss": 0.0501,
      "step": 8394
    },
    {
      "epoch": 15.981930575368521,
      "grad_norm": 0.2116997241973877,
      "learning_rate": 4.671324229914259e-05,
      "loss": 0.0704,
      "step": 8395
    },
    {
      "epoch": 15.983832620066572,
      "grad_norm": 0.12056510895490646,
      "learning_rate": 4.670689107653223e-05,
      "loss": 0.0904,
      "step": 8396
    },
    {
      "epoch": 15.985734664764623,
      "grad_norm": 0.15489526093006134,
      "learning_rate": 4.6700539853921884e-05,
      "loss": 0.0567,
      "step": 8397
    },
    {
      "epoch": 15.987636709462672,
      "grad_norm": 0.25165241956710815,
      "learning_rate": 4.669418863131153e-05,
      "loss": 0.0539,
      "step": 8398
    },
    {
      "epoch": 15.989538754160723,
      "grad_norm": 0.16446714103221893,
      "learning_rate": 4.668783740870118e-05,
      "loss": 0.068,
      "step": 8399
    },
    {
      "epoch": 15.991440798858774,
      "grad_norm": 0.18781155347824097,
      "learning_rate": 4.668148618609082e-05,
      "loss": 0.0507,
      "step": 8400
    },
    {
      "epoch": 15.993342843556823,
      "grad_norm": 0.14649319648742676,
      "learning_rate": 4.667513496348047e-05,
      "loss": 0.0435,
      "step": 8401
    },
    {
      "epoch": 15.995244888254874,
      "grad_norm": 0.11849051713943481,
      "learning_rate": 4.666878374087012e-05,
      "loss": 0.0543,
      "step": 8402
    },
    {
      "epoch": 15.997146932952925,
      "grad_norm": 0.21823644638061523,
      "learning_rate": 4.666243251825977e-05,
      "loss": 0.0669,
      "step": 8403
    },
    {
      "epoch": 15.999048977650975,
      "grad_norm": 0.15036791563034058,
      "learning_rate": 4.6656081295649413e-05,
      "loss": 0.0434,
      "step": 8404
    },
    {
      "epoch": 16.000951022349025,
      "grad_norm": 0.1519092172384262,
      "learning_rate": 4.664973007303906e-05,
      "loss": 0.0591,
      "step": 8405
    },
    {
      "epoch": 16.002853067047077,
      "grad_norm": 0.20130537450313568,
      "learning_rate": 4.664337885042871e-05,
      "loss": 0.0477,
      "step": 8406
    },
    {
      "epoch": 16.004755111745126,
      "grad_norm": 0.1303798258304596,
      "learning_rate": 4.6637027627818355e-05,
      "loss": 0.0667,
      "step": 8407
    },
    {
      "epoch": 16.006657156443175,
      "grad_norm": 0.051126472651958466,
      "learning_rate": 4.663067640520801e-05,
      "loss": 0.0407,
      "step": 8408
    },
    {
      "epoch": 16.008559201141228,
      "grad_norm": 0.10072528570890427,
      "learning_rate": 4.662432518259765e-05,
      "loss": 0.0503,
      "step": 8409
    },
    {
      "epoch": 16.010461245839277,
      "grad_norm": 0.11757353693246841,
      "learning_rate": 4.66179739599873e-05,
      "loss": 0.0611,
      "step": 8410
    },
    {
      "epoch": 16.012363290537326,
      "grad_norm": 0.15045824646949768,
      "learning_rate": 4.661162273737695e-05,
      "loss": 0.051,
      "step": 8411
    },
    {
      "epoch": 16.01426533523538,
      "grad_norm": 0.1323220580816269,
      "learning_rate": 4.6605271514766594e-05,
      "loss": 0.0475,
      "step": 8412
    },
    {
      "epoch": 16.016167379933428,
      "grad_norm": 0.11407540738582611,
      "learning_rate": 4.659892029215624e-05,
      "loss": 0.064,
      "step": 8413
    },
    {
      "epoch": 16.018069424631477,
      "grad_norm": 0.07627911120653152,
      "learning_rate": 4.659256906954589e-05,
      "loss": 0.0529,
      "step": 8414
    },
    {
      "epoch": 16.01997146932953,
      "grad_norm": 0.08195697516202927,
      "learning_rate": 4.6586217846935536e-05,
      "loss": 0.0388,
      "step": 8415
    },
    {
      "epoch": 16.02187351402758,
      "grad_norm": 0.0576738566160202,
      "learning_rate": 4.657986662432519e-05,
      "loss": 0.0462,
      "step": 8416
    },
    {
      "epoch": 16.02377555872563,
      "grad_norm": 0.09754709899425507,
      "learning_rate": 4.657351540171483e-05,
      "loss": 0.0645,
      "step": 8417
    },
    {
      "epoch": 16.02567760342368,
      "grad_norm": 0.09327246248722076,
      "learning_rate": 4.656716417910448e-05,
      "loss": 0.0589,
      "step": 8418
    },
    {
      "epoch": 16.02757964812173,
      "grad_norm": 0.1296929121017456,
      "learning_rate": 4.656081295649412e-05,
      "loss": 0.0468,
      "step": 8419
    },
    {
      "epoch": 16.029481692819783,
      "grad_norm": 0.09633622318506241,
      "learning_rate": 4.6554461733883775e-05,
      "loss": 0.0647,
      "step": 8420
    },
    {
      "epoch": 16.03138373751783,
      "grad_norm": 0.2070033997297287,
      "learning_rate": 4.654811051127343e-05,
      "loss": 0.0568,
      "step": 8421
    },
    {
      "epoch": 16.03328578221588,
      "grad_norm": 0.17608699202537537,
      "learning_rate": 4.6541759288663065e-05,
      "loss": 0.0567,
      "step": 8422
    },
    {
      "epoch": 16.035187826913933,
      "grad_norm": 0.11420154571533203,
      "learning_rate": 4.653540806605272e-05,
      "loss": 0.0836,
      "step": 8423
    },
    {
      "epoch": 16.037089871611983,
      "grad_norm": 0.20214304327964783,
      "learning_rate": 4.652905684344236e-05,
      "loss": 0.0551,
      "step": 8424
    },
    {
      "epoch": 16.03899191631003,
      "grad_norm": 0.3227376341819763,
      "learning_rate": 4.6522705620832014e-05,
      "loss": 0.0788,
      "step": 8425
    },
    {
      "epoch": 16.040893961008084,
      "grad_norm": 0.185341477394104,
      "learning_rate": 4.651635439822166e-05,
      "loss": 0.0539,
      "step": 8426
    },
    {
      "epoch": 16.042796005706133,
      "grad_norm": 0.09545671194791794,
      "learning_rate": 4.6510003175611304e-05,
      "loss": 0.0532,
      "step": 8427
    },
    {
      "epoch": 16.044698050404186,
      "grad_norm": 0.11102737486362457,
      "learning_rate": 4.6503651953000956e-05,
      "loss": 0.0509,
      "step": 8428
    },
    {
      "epoch": 16.046600095102235,
      "grad_norm": 0.1727059930562973,
      "learning_rate": 4.64973007303906e-05,
      "loss": 0.0613,
      "step": 8429
    },
    {
      "epoch": 16.048502139800284,
      "grad_norm": 0.06402069330215454,
      "learning_rate": 4.649094950778025e-05,
      "loss": 0.0619,
      "step": 8430
    },
    {
      "epoch": 16.050404184498337,
      "grad_norm": 0.16342858970165253,
      "learning_rate": 4.64845982851699e-05,
      "loss": 0.063,
      "step": 8431
    },
    {
      "epoch": 16.052306229196386,
      "grad_norm": 0.15879632532596588,
      "learning_rate": 4.647824706255954e-05,
      "loss": 0.0719,
      "step": 8432
    },
    {
      "epoch": 16.054208273894435,
      "grad_norm": 0.07986364513635635,
      "learning_rate": 4.6471895839949195e-05,
      "loss": 0.056,
      "step": 8433
    },
    {
      "epoch": 16.056110318592488,
      "grad_norm": 0.17455364763736725,
      "learning_rate": 4.646554461733884e-05,
      "loss": 0.0622,
      "step": 8434
    },
    {
      "epoch": 16.058012363290537,
      "grad_norm": 0.1904873251914978,
      "learning_rate": 4.645919339472849e-05,
      "loss": 0.0711,
      "step": 8435
    },
    {
      "epoch": 16.059914407988586,
      "grad_norm": 0.2067917138338089,
      "learning_rate": 4.645284217211813e-05,
      "loss": 0.0474,
      "step": 8436
    },
    {
      "epoch": 16.06181645268664,
      "grad_norm": 0.10311882197856903,
      "learning_rate": 4.644649094950778e-05,
      "loss": 0.0616,
      "step": 8437
    },
    {
      "epoch": 16.063718497384688,
      "grad_norm": 0.20031601190567017,
      "learning_rate": 4.644013972689743e-05,
      "loss": 0.0613,
      "step": 8438
    },
    {
      "epoch": 16.06562054208274,
      "grad_norm": 0.1746343970298767,
      "learning_rate": 4.643378850428708e-05,
      "loss": 0.0796,
      "step": 8439
    },
    {
      "epoch": 16.06752258678079,
      "grad_norm": 0.16077488660812378,
      "learning_rate": 4.6427437281676724e-05,
      "loss": 0.0494,
      "step": 8440
    },
    {
      "epoch": 16.06942463147884,
      "grad_norm": 0.2355642169713974,
      "learning_rate": 4.642108605906637e-05,
      "loss": 0.0686,
      "step": 8441
    },
    {
      "epoch": 16.07132667617689,
      "grad_norm": 0.16697682440280914,
      "learning_rate": 4.641473483645602e-05,
      "loss": 0.0622,
      "step": 8442
    },
    {
      "epoch": 16.07322872087494,
      "grad_norm": 0.2681947350502014,
      "learning_rate": 4.6408383613845666e-05,
      "loss": 0.0696,
      "step": 8443
    },
    {
      "epoch": 16.07513076557299,
      "grad_norm": 0.0967094749212265,
      "learning_rate": 4.640203239123532e-05,
      "loss": 0.0476,
      "step": 8444
    },
    {
      "epoch": 16.077032810271042,
      "grad_norm": 0.2911648154258728,
      "learning_rate": 4.639568116862496e-05,
      "loss": 0.077,
      "step": 8445
    },
    {
      "epoch": 16.07893485496909,
      "grad_norm": 0.16672903299331665,
      "learning_rate": 4.638932994601461e-05,
      "loss": 0.0451,
      "step": 8446
    },
    {
      "epoch": 16.08083689966714,
      "grad_norm": 0.15733489394187927,
      "learning_rate": 4.638297872340426e-05,
      "loss": 0.0595,
      "step": 8447
    },
    {
      "epoch": 16.082738944365193,
      "grad_norm": 0.25597983598709106,
      "learning_rate": 4.6376627500793905e-05,
      "loss": 0.0717,
      "step": 8448
    },
    {
      "epoch": 16.084640989063242,
      "grad_norm": 0.13080979883670807,
      "learning_rate": 4.637027627818355e-05,
      "loss": 0.055,
      "step": 8449
    },
    {
      "epoch": 16.086543033761295,
      "grad_norm": 0.23891276121139526,
      "learning_rate": 4.63639250555732e-05,
      "loss": 0.057,
      "step": 8450
    },
    {
      "epoch": 16.088445078459344,
      "grad_norm": 0.13582190871238708,
      "learning_rate": 4.635757383296285e-05,
      "loss": 0.0529,
      "step": 8451
    },
    {
      "epoch": 16.090347123157393,
      "grad_norm": 0.22233571112155914,
      "learning_rate": 4.63512226103525e-05,
      "loss": 0.0618,
      "step": 8452
    },
    {
      "epoch": 16.092249167855446,
      "grad_norm": 0.186944842338562,
      "learning_rate": 4.6344871387742144e-05,
      "loss": 0.0605,
      "step": 8453
    },
    {
      "epoch": 16.094151212553495,
      "grad_norm": 0.13477811217308044,
      "learning_rate": 4.633852016513179e-05,
      "loss": 0.0682,
      "step": 8454
    },
    {
      "epoch": 16.096053257251544,
      "grad_norm": 0.15073803067207336,
      "learning_rate": 4.6332168942521434e-05,
      "loss": 0.0628,
      "step": 8455
    },
    {
      "epoch": 16.097955301949597,
      "grad_norm": 0.25887051224708557,
      "learning_rate": 4.6325817719911086e-05,
      "loss": 0.062,
      "step": 8456
    },
    {
      "epoch": 16.099857346647646,
      "grad_norm": 0.07140891999006271,
      "learning_rate": 4.631946649730074e-05,
      "loss": 0.0562,
      "step": 8457
    },
    {
      "epoch": 16.101759391345695,
      "grad_norm": 0.09805332124233246,
      "learning_rate": 4.6313115274690376e-05,
      "loss": 0.0426,
      "step": 8458
    },
    {
      "epoch": 16.103661436043748,
      "grad_norm": 0.1520451009273529,
      "learning_rate": 4.630676405208003e-05,
      "loss": 0.089,
      "step": 8459
    },
    {
      "epoch": 16.105563480741797,
      "grad_norm": 0.1416059285402298,
      "learning_rate": 4.630041282946967e-05,
      "loss": 0.0585,
      "step": 8460
    },
    {
      "epoch": 16.10746552543985,
      "grad_norm": 0.2078314870595932,
      "learning_rate": 4.6294061606859324e-05,
      "loss": 0.0668,
      "step": 8461
    },
    {
      "epoch": 16.1093675701379,
      "grad_norm": 0.1390911340713501,
      "learning_rate": 4.628771038424897e-05,
      "loss": 0.0512,
      "step": 8462
    },
    {
      "epoch": 16.111269614835948,
      "grad_norm": 0.16498126089572906,
      "learning_rate": 4.6281359161638615e-05,
      "loss": 0.0442,
      "step": 8463
    },
    {
      "epoch": 16.113171659534,
      "grad_norm": 0.15705092251300812,
      "learning_rate": 4.6275007939028266e-05,
      "loss": 0.0542,
      "step": 8464
    },
    {
      "epoch": 16.11507370423205,
      "grad_norm": 0.12232176959514618,
      "learning_rate": 4.626865671641791e-05,
      "loss": 0.0575,
      "step": 8465
    },
    {
      "epoch": 16.1169757489301,
      "grad_norm": 0.10896648466587067,
      "learning_rate": 4.626230549380756e-05,
      "loss": 0.0608,
      "step": 8466
    },
    {
      "epoch": 16.11887779362815,
      "grad_norm": 0.09567489475011826,
      "learning_rate": 4.62559542711972e-05,
      "loss": 0.0568,
      "step": 8467
    },
    {
      "epoch": 16.1207798383262,
      "grad_norm": 0.19636806845664978,
      "learning_rate": 4.6249603048586853e-05,
      "loss": 0.0514,
      "step": 8468
    },
    {
      "epoch": 16.12268188302425,
      "grad_norm": 0.18234984576702118,
      "learning_rate": 4.6243251825976505e-05,
      "loss": 0.0504,
      "step": 8469
    },
    {
      "epoch": 16.124583927722302,
      "grad_norm": 0.15296286344528198,
      "learning_rate": 4.623690060336615e-05,
      "loss": 0.0376,
      "step": 8470
    },
    {
      "epoch": 16.12648597242035,
      "grad_norm": 0.11294987797737122,
      "learning_rate": 4.62305493807558e-05,
      "loss": 0.0489,
      "step": 8471
    },
    {
      "epoch": 16.128388017118404,
      "grad_norm": 0.13020162284374237,
      "learning_rate": 4.622419815814544e-05,
      "loss": 0.0525,
      "step": 8472
    },
    {
      "epoch": 16.130290061816453,
      "grad_norm": 0.13996310532093048,
      "learning_rate": 4.621784693553509e-05,
      "loss": 0.0467,
      "step": 8473
    },
    {
      "epoch": 16.132192106514502,
      "grad_norm": 0.10891363024711609,
      "learning_rate": 4.621149571292474e-05,
      "loss": 0.0486,
      "step": 8474
    },
    {
      "epoch": 16.134094151212555,
      "grad_norm": 0.17132295668125153,
      "learning_rate": 4.620514449031439e-05,
      "loss": 0.0704,
      "step": 8475
    },
    {
      "epoch": 16.135996195910604,
      "grad_norm": 0.18272633850574493,
      "learning_rate": 4.6198793267704034e-05,
      "loss": 0.0472,
      "step": 8476
    },
    {
      "epoch": 16.137898240608653,
      "grad_norm": 0.1485610157251358,
      "learning_rate": 4.619244204509368e-05,
      "loss": 0.0559,
      "step": 8477
    },
    {
      "epoch": 16.139800285306706,
      "grad_norm": 0.13875699043273926,
      "learning_rate": 4.618609082248333e-05,
      "loss": 0.0592,
      "step": 8478
    },
    {
      "epoch": 16.141702330004755,
      "grad_norm": 0.1410178542137146,
      "learning_rate": 4.6179739599872976e-05,
      "loss": 0.0647,
      "step": 8479
    },
    {
      "epoch": 16.143604374702804,
      "grad_norm": 0.230701744556427,
      "learning_rate": 4.617338837726263e-05,
      "loss": 0.0528,
      "step": 8480
    },
    {
      "epoch": 16.145506419400856,
      "grad_norm": 0.23617468774318695,
      "learning_rate": 4.616703715465227e-05,
      "loss": 0.0594,
      "step": 8481
    },
    {
      "epoch": 16.147408464098906,
      "grad_norm": 0.18603870272636414,
      "learning_rate": 4.616068593204192e-05,
      "loss": 0.0433,
      "step": 8482
    },
    {
      "epoch": 16.149310508796958,
      "grad_norm": 0.18782128393650055,
      "learning_rate": 4.615433470943157e-05,
      "loss": 0.0566,
      "step": 8483
    },
    {
      "epoch": 16.151212553495007,
      "grad_norm": 0.25286000967025757,
      "learning_rate": 4.6147983486821215e-05,
      "loss": 0.0785,
      "step": 8484
    },
    {
      "epoch": 16.153114598193056,
      "grad_norm": 0.0904570072889328,
      "learning_rate": 4.614163226421086e-05,
      "loss": 0.0508,
      "step": 8485
    },
    {
      "epoch": 16.15501664289111,
      "grad_norm": 0.2545378506183624,
      "learning_rate": 4.613528104160051e-05,
      "loss": 0.0783,
      "step": 8486
    },
    {
      "epoch": 16.156918687589158,
      "grad_norm": 0.191772922873497,
      "learning_rate": 4.612892981899016e-05,
      "loss": 0.0683,
      "step": 8487
    },
    {
      "epoch": 16.158820732287207,
      "grad_norm": 0.1491425633430481,
      "learning_rate": 4.612257859637981e-05,
      "loss": 0.0737,
      "step": 8488
    },
    {
      "epoch": 16.16072277698526,
      "grad_norm": 0.0845363661646843,
      "learning_rate": 4.6116227373769454e-05,
      "loss": 0.061,
      "step": 8489
    },
    {
      "epoch": 16.16262482168331,
      "grad_norm": 0.24539388716220856,
      "learning_rate": 4.61098761511591e-05,
      "loss": 0.0675,
      "step": 8490
    },
    {
      "epoch": 16.16452686638136,
      "grad_norm": 0.10510896891355515,
      "learning_rate": 4.6103524928548744e-05,
      "loss": 0.0387,
      "step": 8491
    },
    {
      "epoch": 16.16642891107941,
      "grad_norm": 0.13409045338630676,
      "learning_rate": 4.6097173705938396e-05,
      "loss": 0.05,
      "step": 8492
    },
    {
      "epoch": 16.16833095577746,
      "grad_norm": 0.14605560898780823,
      "learning_rate": 4.609082248332804e-05,
      "loss": 0.0747,
      "step": 8493
    },
    {
      "epoch": 16.170233000475513,
      "grad_norm": 0.10427720844745636,
      "learning_rate": 4.6084471260717686e-05,
      "loss": 0.0689,
      "step": 8494
    },
    {
      "epoch": 16.17213504517356,
      "grad_norm": 0.1358361840248108,
      "learning_rate": 4.607812003810734e-05,
      "loss": 0.053,
      "step": 8495
    },
    {
      "epoch": 16.17403708987161,
      "grad_norm": 0.1160115897655487,
      "learning_rate": 4.607176881549698e-05,
      "loss": 0.0422,
      "step": 8496
    },
    {
      "epoch": 16.175939134569663,
      "grad_norm": 0.11981600522994995,
      "learning_rate": 4.6065417592886635e-05,
      "loss": 0.0553,
      "step": 8497
    },
    {
      "epoch": 16.177841179267713,
      "grad_norm": 0.15992315113544464,
      "learning_rate": 4.605906637027628e-05,
      "loss": 0.053,
      "step": 8498
    },
    {
      "epoch": 16.17974322396576,
      "grad_norm": 0.14129142463207245,
      "learning_rate": 4.6052715147665925e-05,
      "loss": 0.0587,
      "step": 8499
    },
    {
      "epoch": 16.181645268663814,
      "grad_norm": 0.12155603617429733,
      "learning_rate": 4.604636392505558e-05,
      "loss": 0.0488,
      "step": 8500
    },
    {
      "epoch": 16.183547313361863,
      "grad_norm": 0.19479072093963623,
      "learning_rate": 4.604001270244522e-05,
      "loss": 0.0592,
      "step": 8501
    },
    {
      "epoch": 16.185449358059916,
      "grad_norm": 0.16191153228282928,
      "learning_rate": 4.6033661479834874e-05,
      "loss": 0.0874,
      "step": 8502
    },
    {
      "epoch": 16.187351402757965,
      "grad_norm": 0.2593137323856354,
      "learning_rate": 4.602731025722451e-05,
      "loss": 0.0732,
      "step": 8503
    },
    {
      "epoch": 16.189253447456014,
      "grad_norm": 0.07688070833683014,
      "learning_rate": 4.6020959034614164e-05,
      "loss": 0.0646,
      "step": 8504
    },
    {
      "epoch": 16.191155492154067,
      "grad_norm": 0.15450550615787506,
      "learning_rate": 4.6014607812003816e-05,
      "loss": 0.0603,
      "step": 8505
    },
    {
      "epoch": 16.193057536852116,
      "grad_norm": 0.31790322065353394,
      "learning_rate": 4.600825658939346e-05,
      "loss": 0.0793,
      "step": 8506
    },
    {
      "epoch": 16.194959581550165,
      "grad_norm": 0.12882187962532043,
      "learning_rate": 4.600190536678311e-05,
      "loss": 0.0524,
      "step": 8507
    },
    {
      "epoch": 16.196861626248218,
      "grad_norm": 0.25104114413261414,
      "learning_rate": 4.599555414417275e-05,
      "loss": 0.0679,
      "step": 8508
    },
    {
      "epoch": 16.198763670946267,
      "grad_norm": 0.13157489895820618,
      "learning_rate": 4.59892029215624e-05,
      "loss": 0.0562,
      "step": 8509
    },
    {
      "epoch": 16.200665715644316,
      "grad_norm": 0.1363213062286377,
      "learning_rate": 4.598285169895205e-05,
      "loss": 0.06,
      "step": 8510
    },
    {
      "epoch": 16.20256776034237,
      "grad_norm": 0.15622791647911072,
      "learning_rate": 4.59765004763417e-05,
      "loss": 0.0722,
      "step": 8511
    },
    {
      "epoch": 16.204469805040418,
      "grad_norm": 0.1738666146993637,
      "learning_rate": 4.5970149253731345e-05,
      "loss": 0.0496,
      "step": 8512
    },
    {
      "epoch": 16.20637184973847,
      "grad_norm": 0.1258222460746765,
      "learning_rate": 4.596379803112099e-05,
      "loss": 0.0528,
      "step": 8513
    },
    {
      "epoch": 16.20827389443652,
      "grad_norm": 0.1232779324054718,
      "learning_rate": 4.595744680851064e-05,
      "loss": 0.0535,
      "step": 8514
    },
    {
      "epoch": 16.21017593913457,
      "grad_norm": 0.1608402281999588,
      "learning_rate": 4.595109558590029e-05,
      "loss": 0.0465,
      "step": 8515
    },
    {
      "epoch": 16.21207798383262,
      "grad_norm": 0.16735756397247314,
      "learning_rate": 4.594474436328994e-05,
      "loss": 0.0594,
      "step": 8516
    },
    {
      "epoch": 16.21398002853067,
      "grad_norm": 0.18840819597244263,
      "learning_rate": 4.5938393140679584e-05,
      "loss": 0.0588,
      "step": 8517
    },
    {
      "epoch": 16.21588207322872,
      "grad_norm": 0.12657614052295685,
      "learning_rate": 4.593204191806923e-05,
      "loss": 0.0669,
      "step": 8518
    },
    {
      "epoch": 16.217784117926772,
      "grad_norm": 0.18073607981204987,
      "learning_rate": 4.592569069545888e-05,
      "loss": 0.0598,
      "step": 8519
    },
    {
      "epoch": 16.21968616262482,
      "grad_norm": 0.14543786644935608,
      "learning_rate": 4.5919339472848526e-05,
      "loss": 0.074,
      "step": 8520
    },
    {
      "epoch": 16.22158820732287,
      "grad_norm": 0.09334524720907211,
      "learning_rate": 4.591298825023817e-05,
      "loss": 0.055,
      "step": 8521
    },
    {
      "epoch": 16.223490252020923,
      "grad_norm": 0.13787609338760376,
      "learning_rate": 4.5906637027627816e-05,
      "loss": 0.0417,
      "step": 8522
    },
    {
      "epoch": 16.225392296718972,
      "grad_norm": 0.204119011759758,
      "learning_rate": 4.590028580501747e-05,
      "loss": 0.0559,
      "step": 8523
    },
    {
      "epoch": 16.227294341417025,
      "grad_norm": 0.12836776673793793,
      "learning_rate": 4.589393458240712e-05,
      "loss": 0.0465,
      "step": 8524
    },
    {
      "epoch": 16.229196386115074,
      "grad_norm": 0.12782390415668488,
      "learning_rate": 4.5887583359796764e-05,
      "loss": 0.0561,
      "step": 8525
    },
    {
      "epoch": 16.231098430813123,
      "grad_norm": 0.2859640419483185,
      "learning_rate": 4.588123213718641e-05,
      "loss": 0.058,
      "step": 8526
    },
    {
      "epoch": 16.233000475511176,
      "grad_norm": 0.11105810105800629,
      "learning_rate": 4.5874880914576055e-05,
      "loss": 0.0688,
      "step": 8527
    },
    {
      "epoch": 16.234902520209225,
      "grad_norm": 0.22457514703273773,
      "learning_rate": 4.5868529691965706e-05,
      "loss": 0.0796,
      "step": 8528
    },
    {
      "epoch": 16.236804564907274,
      "grad_norm": 0.09397421032190323,
      "learning_rate": 4.586217846935535e-05,
      "loss": 0.0595,
      "step": 8529
    },
    {
      "epoch": 16.238706609605327,
      "grad_norm": 0.28852370381355286,
      "learning_rate": 4.5855827246744997e-05,
      "loss": 0.0549,
      "step": 8530
    },
    {
      "epoch": 16.240608654303376,
      "grad_norm": 0.1166771650314331,
      "learning_rate": 4.584947602413465e-05,
      "loss": 0.0493,
      "step": 8531
    },
    {
      "epoch": 16.242510699001425,
      "grad_norm": 0.2033771574497223,
      "learning_rate": 4.5843124801524293e-05,
      "loss": 0.059,
      "step": 8532
    },
    {
      "epoch": 16.244412743699478,
      "grad_norm": 0.10996011644601822,
      "learning_rate": 4.5836773578913945e-05,
      "loss": 0.0445,
      "step": 8533
    },
    {
      "epoch": 16.246314788397527,
      "grad_norm": 0.11977210640907288,
      "learning_rate": 4.583042235630359e-05,
      "loss": 0.0556,
      "step": 8534
    },
    {
      "epoch": 16.24821683309558,
      "grad_norm": 0.2993720769882202,
      "learning_rate": 4.5824071133693235e-05,
      "loss": 0.0709,
      "step": 8535
    },
    {
      "epoch": 16.25011887779363,
      "grad_norm": 0.12694013118743896,
      "learning_rate": 4.581771991108289e-05,
      "loss": 0.0655,
      "step": 8536
    },
    {
      "epoch": 16.252020922491678,
      "grad_norm": 0.1645960509777069,
      "learning_rate": 4.581136868847253e-05,
      "loss": 0.0639,
      "step": 8537
    },
    {
      "epoch": 16.25392296718973,
      "grad_norm": 0.09638939052820206,
      "learning_rate": 4.5805017465862184e-05,
      "loss": 0.0475,
      "step": 8538
    },
    {
      "epoch": 16.25582501188778,
      "grad_norm": 0.20398004353046417,
      "learning_rate": 4.579866624325182e-05,
      "loss": 0.0381,
      "step": 8539
    },
    {
      "epoch": 16.25772705658583,
      "grad_norm": 0.20141464471817017,
      "learning_rate": 4.5792315020641474e-05,
      "loss": 0.0594,
      "step": 8540
    },
    {
      "epoch": 16.25962910128388,
      "grad_norm": 0.16405193507671356,
      "learning_rate": 4.5785963798031126e-05,
      "loss": 0.0421,
      "step": 8541
    },
    {
      "epoch": 16.26153114598193,
      "grad_norm": 0.29997482895851135,
      "learning_rate": 4.577961257542077e-05,
      "loss": 0.0876,
      "step": 8542
    },
    {
      "epoch": 16.26343319067998,
      "grad_norm": 0.06307131052017212,
      "learning_rate": 4.577326135281042e-05,
      "loss": 0.0348,
      "step": 8543
    },
    {
      "epoch": 16.265335235378032,
      "grad_norm": 0.1158008873462677,
      "learning_rate": 4.576691013020006e-05,
      "loss": 0.0291,
      "step": 8544
    },
    {
      "epoch": 16.26723728007608,
      "grad_norm": 0.3139667212963104,
      "learning_rate": 4.576055890758971e-05,
      "loss": 0.0596,
      "step": 8545
    },
    {
      "epoch": 16.269139324774134,
      "grad_norm": 0.3778131306171417,
      "learning_rate": 4.575420768497936e-05,
      "loss": 0.1135,
      "step": 8546
    },
    {
      "epoch": 16.271041369472183,
      "grad_norm": 0.12188387662172318,
      "learning_rate": 4.574785646236901e-05,
      "loss": 0.0578,
      "step": 8547
    },
    {
      "epoch": 16.272943414170232,
      "grad_norm": 0.2640130817890167,
      "learning_rate": 4.5741505239758655e-05,
      "loss": 0.0858,
      "step": 8548
    },
    {
      "epoch": 16.274845458868285,
      "grad_norm": 0.23133710026741028,
      "learning_rate": 4.57351540171483e-05,
      "loss": 0.0729,
      "step": 8549
    },
    {
      "epoch": 16.276747503566334,
      "grad_norm": 0.11715468764305115,
      "learning_rate": 4.572880279453795e-05,
      "loss": 0.0505,
      "step": 8550
    },
    {
      "epoch": 16.278649548264383,
      "grad_norm": 0.23677362501621246,
      "learning_rate": 4.57224515719276e-05,
      "loss": 0.0601,
      "step": 8551
    },
    {
      "epoch": 16.280551592962436,
      "grad_norm": 0.18508802354335785,
      "learning_rate": 4.571610034931725e-05,
      "loss": 0.0696,
      "step": 8552
    },
    {
      "epoch": 16.282453637660485,
      "grad_norm": 0.0752667635679245,
      "learning_rate": 4.5709749126706894e-05,
      "loss": 0.0468,
      "step": 8553
    },
    {
      "epoch": 16.284355682358534,
      "grad_norm": 0.13637211918830872,
      "learning_rate": 4.570339790409654e-05,
      "loss": 0.0423,
      "step": 8554
    },
    {
      "epoch": 16.286257727056586,
      "grad_norm": 0.11643306910991669,
      "learning_rate": 4.569704668148619e-05,
      "loss": 0.0591,
      "step": 8555
    },
    {
      "epoch": 16.288159771754636,
      "grad_norm": 0.1637072116136551,
      "learning_rate": 4.5690695458875836e-05,
      "loss": 0.0689,
      "step": 8556
    },
    {
      "epoch": 16.29006181645269,
      "grad_norm": 0.14512047171592712,
      "learning_rate": 4.568434423626548e-05,
      "loss": 0.0569,
      "step": 8557
    },
    {
      "epoch": 16.291963861150737,
      "grad_norm": 0.18429428339004517,
      "learning_rate": 4.5677993013655126e-05,
      "loss": 0.0767,
      "step": 8558
    },
    {
      "epoch": 16.293865905848786,
      "grad_norm": 0.0927993431687355,
      "learning_rate": 4.567164179104478e-05,
      "loss": 0.0353,
      "step": 8559
    },
    {
      "epoch": 16.29576795054684,
      "grad_norm": 0.08778216689825058,
      "learning_rate": 4.566529056843443e-05,
      "loss": 0.0479,
      "step": 8560
    },
    {
      "epoch": 16.29766999524489,
      "grad_norm": 0.09512235969305038,
      "learning_rate": 4.5658939345824075e-05,
      "loss": 0.0576,
      "step": 8561
    },
    {
      "epoch": 16.299572039942937,
      "grad_norm": 0.20626294612884521,
      "learning_rate": 4.565258812321372e-05,
      "loss": 0.0534,
      "step": 8562
    },
    {
      "epoch": 16.30147408464099,
      "grad_norm": 0.21326245367527008,
      "learning_rate": 4.5646236900603365e-05,
      "loss": 0.0605,
      "step": 8563
    },
    {
      "epoch": 16.30337612933904,
      "grad_norm": 0.17919912934303284,
      "learning_rate": 4.563988567799302e-05,
      "loss": 0.0605,
      "step": 8564
    },
    {
      "epoch": 16.30527817403709,
      "grad_norm": 0.16517671942710876,
      "learning_rate": 4.563353445538266e-05,
      "loss": 0.0455,
      "step": 8565
    },
    {
      "epoch": 16.30718021873514,
      "grad_norm": 0.1184915080666542,
      "learning_rate": 4.562718323277231e-05,
      "loss": 0.0574,
      "step": 8566
    },
    {
      "epoch": 16.30908226343319,
      "grad_norm": 0.065870501101017,
      "learning_rate": 4.562083201016196e-05,
      "loss": 0.0599,
      "step": 8567
    },
    {
      "epoch": 16.310984308131243,
      "grad_norm": 0.17375320196151733,
      "learning_rate": 4.5614480787551604e-05,
      "loss": 0.0594,
      "step": 8568
    },
    {
      "epoch": 16.312886352829292,
      "grad_norm": 0.18376989662647247,
      "learning_rate": 4.5608129564941256e-05,
      "loss": 0.0688,
      "step": 8569
    },
    {
      "epoch": 16.31478839752734,
      "grad_norm": 0.12935777008533478,
      "learning_rate": 4.56017783423309e-05,
      "loss": 0.0479,
      "step": 8570
    },
    {
      "epoch": 16.316690442225394,
      "grad_norm": 0.19566315412521362,
      "learning_rate": 4.5595427119720546e-05,
      "loss": 0.0463,
      "step": 8571
    },
    {
      "epoch": 16.318592486923443,
      "grad_norm": 0.10081527382135391,
      "learning_rate": 4.55890758971102e-05,
      "loss": 0.0389,
      "step": 8572
    },
    {
      "epoch": 16.32049453162149,
      "grad_norm": 0.11026335507631302,
      "learning_rate": 4.558272467449984e-05,
      "loss": 0.0523,
      "step": 8573
    },
    {
      "epoch": 16.322396576319544,
      "grad_norm": 0.1704561710357666,
      "learning_rate": 4.5576373451889495e-05,
      "loss": 0.0661,
      "step": 8574
    },
    {
      "epoch": 16.324298621017594,
      "grad_norm": 0.0993368849158287,
      "learning_rate": 4.557002222927913e-05,
      "loss": 0.059,
      "step": 8575
    },
    {
      "epoch": 16.326200665715643,
      "grad_norm": 0.2205575853586197,
      "learning_rate": 4.5563671006668785e-05,
      "loss": 0.0799,
      "step": 8576
    },
    {
      "epoch": 16.328102710413695,
      "grad_norm": 0.12398331612348557,
      "learning_rate": 4.555731978405843e-05,
      "loss": 0.0684,
      "step": 8577
    },
    {
      "epoch": 16.330004755111744,
      "grad_norm": 0.15090776979923248,
      "learning_rate": 4.555096856144808e-05,
      "loss": 0.0588,
      "step": 8578
    },
    {
      "epoch": 16.331906799809797,
      "grad_norm": 0.2645479142665863,
      "learning_rate": 4.5544617338837734e-05,
      "loss": 0.0968,
      "step": 8579
    },
    {
      "epoch": 16.333808844507846,
      "grad_norm": 0.2185746282339096,
      "learning_rate": 4.553826611622737e-05,
      "loss": 0.0645,
      "step": 8580
    },
    {
      "epoch": 16.335710889205895,
      "grad_norm": 0.09869100898504257,
      "learning_rate": 4.5531914893617024e-05,
      "loss": 0.064,
      "step": 8581
    },
    {
      "epoch": 16.337612933903948,
      "grad_norm": 0.16115908324718475,
      "learning_rate": 4.552556367100667e-05,
      "loss": 0.0526,
      "step": 8582
    },
    {
      "epoch": 16.339514978601997,
      "grad_norm": 0.18371903896331787,
      "learning_rate": 4.551921244839632e-05,
      "loss": 0.0619,
      "step": 8583
    },
    {
      "epoch": 16.341417023300046,
      "grad_norm": 0.16014857590198517,
      "learning_rate": 4.5512861225785966e-05,
      "loss": 0.0611,
      "step": 8584
    },
    {
      "epoch": 16.3433190679981,
      "grad_norm": 0.19126653671264648,
      "learning_rate": 4.550651000317561e-05,
      "loss": 0.0573,
      "step": 8585
    },
    {
      "epoch": 16.345221112696148,
      "grad_norm": 0.11099161207675934,
      "learning_rate": 4.550015878056526e-05,
      "loss": 0.0698,
      "step": 8586
    },
    {
      "epoch": 16.347123157394197,
      "grad_norm": 0.13271382451057434,
      "learning_rate": 4.549380755795491e-05,
      "loss": 0.0433,
      "step": 8587
    },
    {
      "epoch": 16.34902520209225,
      "grad_norm": 0.157912015914917,
      "learning_rate": 4.548745633534456e-05,
      "loss": 0.0679,
      "step": 8588
    },
    {
      "epoch": 16.3509272467903,
      "grad_norm": 0.15022407472133636,
      "learning_rate": 4.5481105112734204e-05,
      "loss": 0.0509,
      "step": 8589
    },
    {
      "epoch": 16.35282929148835,
      "grad_norm": 0.1722324937582016,
      "learning_rate": 4.547475389012385e-05,
      "loss": 0.042,
      "step": 8590
    },
    {
      "epoch": 16.3547313361864,
      "grad_norm": 0.11643511801958084,
      "learning_rate": 4.54684026675135e-05,
      "loss": 0.0602,
      "step": 8591
    },
    {
      "epoch": 16.35663338088445,
      "grad_norm": 0.11521922796964645,
      "learning_rate": 4.5462051444903146e-05,
      "loss": 0.0466,
      "step": 8592
    },
    {
      "epoch": 16.358535425582502,
      "grad_norm": 0.17267794907093048,
      "learning_rate": 4.545570022229279e-05,
      "loss": 0.0437,
      "step": 8593
    },
    {
      "epoch": 16.36043747028055,
      "grad_norm": 0.20885154604911804,
      "learning_rate": 4.5449348999682437e-05,
      "loss": 0.0563,
      "step": 8594
    },
    {
      "epoch": 16.3623395149786,
      "grad_norm": 0.21735620498657227,
      "learning_rate": 4.544299777707209e-05,
      "loss": 0.0589,
      "step": 8595
    },
    {
      "epoch": 16.364241559676653,
      "grad_norm": 0.11824718117713928,
      "learning_rate": 4.543664655446174e-05,
      "loss": 0.0503,
      "step": 8596
    },
    {
      "epoch": 16.366143604374702,
      "grad_norm": 0.16579841077327728,
      "learning_rate": 4.5430295331851385e-05,
      "loss": 0.0692,
      "step": 8597
    },
    {
      "epoch": 16.36804564907275,
      "grad_norm": 0.11211599409580231,
      "learning_rate": 4.542394410924103e-05,
      "loss": 0.075,
      "step": 8598
    },
    {
      "epoch": 16.369947693770804,
      "grad_norm": 0.07448209822177887,
      "learning_rate": 4.5417592886630675e-05,
      "loss": 0.0675,
      "step": 8599
    },
    {
      "epoch": 16.371849738468853,
      "grad_norm": 0.10178573429584503,
      "learning_rate": 4.541124166402033e-05,
      "loss": 0.0556,
      "step": 8600
    },
    {
      "epoch": 16.373751783166906,
      "grad_norm": 0.2198735624551773,
      "learning_rate": 4.540489044140997e-05,
      "loss": 0.0669,
      "step": 8601
    },
    {
      "epoch": 16.375653827864955,
      "grad_norm": 0.20772859454154968,
      "learning_rate": 4.539853921879962e-05,
      "loss": 0.056,
      "step": 8602
    },
    {
      "epoch": 16.377555872563004,
      "grad_norm": 0.23242707550525665,
      "learning_rate": 4.539218799618927e-05,
      "loss": 0.0556,
      "step": 8603
    },
    {
      "epoch": 16.379457917261057,
      "grad_norm": 0.08447839319705963,
      "learning_rate": 4.5385836773578914e-05,
      "loss": 0.0538,
      "step": 8604
    },
    {
      "epoch": 16.381359961959106,
      "grad_norm": 0.15451155602931976,
      "learning_rate": 4.5379485550968566e-05,
      "loss": 0.0444,
      "step": 8605
    },
    {
      "epoch": 16.383262006657155,
      "grad_norm": 0.24637392163276672,
      "learning_rate": 4.537313432835821e-05,
      "loss": 0.0753,
      "step": 8606
    },
    {
      "epoch": 16.385164051355208,
      "grad_norm": 0.11653976887464523,
      "learning_rate": 4.5366783105747856e-05,
      "loss": 0.0495,
      "step": 8607
    },
    {
      "epoch": 16.387066096053257,
      "grad_norm": 0.35665005445480347,
      "learning_rate": 4.536043188313751e-05,
      "loss": 0.0734,
      "step": 8608
    },
    {
      "epoch": 16.388968140751306,
      "grad_norm": 0.09935169667005539,
      "learning_rate": 4.535408066052715e-05,
      "loss": 0.0521,
      "step": 8609
    },
    {
      "epoch": 16.39087018544936,
      "grad_norm": 0.14690108597278595,
      "learning_rate": 4.5347729437916805e-05,
      "loss": 0.0458,
      "step": 8610
    },
    {
      "epoch": 16.392772230147408,
      "grad_norm": 0.14574097096920013,
      "learning_rate": 4.534137821530644e-05,
      "loss": 0.0609,
      "step": 8611
    },
    {
      "epoch": 16.39467427484546,
      "grad_norm": 0.12584878504276276,
      "learning_rate": 4.5335026992696095e-05,
      "loss": 0.0818,
      "step": 8612
    },
    {
      "epoch": 16.39657631954351,
      "grad_norm": 0.17045927047729492,
      "learning_rate": 4.532867577008574e-05,
      "loss": 0.055,
      "step": 8613
    },
    {
      "epoch": 16.39847836424156,
      "grad_norm": 0.1100262850522995,
      "learning_rate": 4.532232454747539e-05,
      "loss": 0.0452,
      "step": 8614
    },
    {
      "epoch": 16.40038040893961,
      "grad_norm": 0.18738383054733276,
      "learning_rate": 4.5315973324865044e-05,
      "loss": 0.0556,
      "step": 8615
    },
    {
      "epoch": 16.40228245363766,
      "grad_norm": 0.15285401046276093,
      "learning_rate": 4.530962210225468e-05,
      "loss": 0.0576,
      "step": 8616
    },
    {
      "epoch": 16.40418449833571,
      "grad_norm": 0.268604576587677,
      "learning_rate": 4.5303270879644334e-05,
      "loss": 0.0649,
      "step": 8617
    },
    {
      "epoch": 16.406086543033762,
      "grad_norm": 0.11927369236946106,
      "learning_rate": 4.529691965703398e-05,
      "loss": 0.0777,
      "step": 8618
    },
    {
      "epoch": 16.40798858773181,
      "grad_norm": 0.2090739905834198,
      "learning_rate": 4.529056843442363e-05,
      "loss": 0.0776,
      "step": 8619
    },
    {
      "epoch": 16.409890632429864,
      "grad_norm": 0.1961270570755005,
      "learning_rate": 4.5284217211813276e-05,
      "loss": 0.0716,
      "step": 8620
    },
    {
      "epoch": 16.411792677127913,
      "grad_norm": 0.18359386920928955,
      "learning_rate": 4.527786598920292e-05,
      "loss": 0.0573,
      "step": 8621
    },
    {
      "epoch": 16.413694721825962,
      "grad_norm": 0.20374822616577148,
      "learning_rate": 4.527151476659257e-05,
      "loss": 0.067,
      "step": 8622
    },
    {
      "epoch": 16.415596766524015,
      "grad_norm": 0.054867297410964966,
      "learning_rate": 4.526516354398222e-05,
      "loss": 0.0575,
      "step": 8623
    },
    {
      "epoch": 16.417498811222064,
      "grad_norm": 0.14874067902565002,
      "learning_rate": 4.525881232137187e-05,
      "loss": 0.0466,
      "step": 8624
    },
    {
      "epoch": 16.419400855920113,
      "grad_norm": 0.15803728997707367,
      "learning_rate": 4.5252461098761515e-05,
      "loss": 0.0604,
      "step": 8625
    },
    {
      "epoch": 16.421302900618166,
      "grad_norm": 0.08036740124225616,
      "learning_rate": 4.524610987615116e-05,
      "loss": 0.031,
      "step": 8626
    },
    {
      "epoch": 16.423204945316215,
      "grad_norm": 0.2112816870212555,
      "learning_rate": 4.523975865354081e-05,
      "loss": 0.0491,
      "step": 8627
    },
    {
      "epoch": 16.425106990014264,
      "grad_norm": 0.13649682700634003,
      "learning_rate": 4.523340743093046e-05,
      "loss": 0.0646,
      "step": 8628
    },
    {
      "epoch": 16.427009034712317,
      "grad_norm": 0.1682991087436676,
      "learning_rate": 4.52270562083201e-05,
      "loss": 0.0595,
      "step": 8629
    },
    {
      "epoch": 16.428911079410366,
      "grad_norm": 0.15135952830314636,
      "learning_rate": 4.522070498570975e-05,
      "loss": 0.0567,
      "step": 8630
    },
    {
      "epoch": 16.43081312410842,
      "grad_norm": 0.25466933846473694,
      "learning_rate": 4.52143537630994e-05,
      "loss": 0.0773,
      "step": 8631
    },
    {
      "epoch": 16.432715168806467,
      "grad_norm": 0.18868578970432281,
      "learning_rate": 4.5208002540489044e-05,
      "loss": 0.0516,
      "step": 8632
    },
    {
      "epoch": 16.434617213504517,
      "grad_norm": 0.09199167042970657,
      "learning_rate": 4.5201651317878696e-05,
      "loss": 0.0461,
      "step": 8633
    },
    {
      "epoch": 16.43651925820257,
      "grad_norm": 0.09501266479492188,
      "learning_rate": 4.519530009526834e-05,
      "loss": 0.0695,
      "step": 8634
    },
    {
      "epoch": 16.43842130290062,
      "grad_norm": 0.08783094584941864,
      "learning_rate": 4.5188948872657986e-05,
      "loss": 0.0584,
      "step": 8635
    },
    {
      "epoch": 16.440323347598667,
      "grad_norm": 0.18721364438533783,
      "learning_rate": 4.518259765004764e-05,
      "loss": 0.0604,
      "step": 8636
    },
    {
      "epoch": 16.44222539229672,
      "grad_norm": 0.18143963813781738,
      "learning_rate": 4.517624642743728e-05,
      "loss": 0.0856,
      "step": 8637
    },
    {
      "epoch": 16.44412743699477,
      "grad_norm": 0.1065838411450386,
      "learning_rate": 4.516989520482693e-05,
      "loss": 0.0616,
      "step": 8638
    },
    {
      "epoch": 16.44602948169282,
      "grad_norm": 0.15803262591362,
      "learning_rate": 4.516354398221658e-05,
      "loss": 0.0434,
      "step": 8639
    },
    {
      "epoch": 16.44793152639087,
      "grad_norm": 0.13359306752681732,
      "learning_rate": 4.5157192759606225e-05,
      "loss": 0.0565,
      "step": 8640
    },
    {
      "epoch": 16.44983357108892,
      "grad_norm": 0.09912649542093277,
      "learning_rate": 4.515084153699588e-05,
      "loss": 0.0541,
      "step": 8641
    },
    {
      "epoch": 16.451735615786973,
      "grad_norm": 0.13819362223148346,
      "learning_rate": 4.514449031438552e-05,
      "loss": 0.0735,
      "step": 8642
    },
    {
      "epoch": 16.453637660485022,
      "grad_norm": 0.1289387047290802,
      "learning_rate": 4.513813909177517e-05,
      "loss": 0.0577,
      "step": 8643
    },
    {
      "epoch": 16.45553970518307,
      "grad_norm": 0.16566374897956848,
      "learning_rate": 4.513178786916482e-05,
      "loss": 0.0636,
      "step": 8644
    },
    {
      "epoch": 16.457441749881124,
      "grad_norm": 0.29767850041389465,
      "learning_rate": 4.5125436646554464e-05,
      "loss": 0.064,
      "step": 8645
    },
    {
      "epoch": 16.459343794579173,
      "grad_norm": 0.07450270652770996,
      "learning_rate": 4.5119085423944116e-05,
      "loss": 0.054,
      "step": 8646
    },
    {
      "epoch": 16.461245839277222,
      "grad_norm": 0.18993811309337616,
      "learning_rate": 4.5112734201333754e-05,
      "loss": 0.0651,
      "step": 8647
    },
    {
      "epoch": 16.463147883975275,
      "grad_norm": 0.2176159918308258,
      "learning_rate": 4.5106382978723406e-05,
      "loss": 0.0601,
      "step": 8648
    },
    {
      "epoch": 16.465049928673324,
      "grad_norm": 0.40817978978157043,
      "learning_rate": 4.510003175611305e-05,
      "loss": 0.0944,
      "step": 8649
    },
    {
      "epoch": 16.466951973371373,
      "grad_norm": 0.13390858471393585,
      "learning_rate": 4.50936805335027e-05,
      "loss": 0.0429,
      "step": 8650
    },
    {
      "epoch": 16.468854018069425,
      "grad_norm": 0.166989266872406,
      "learning_rate": 4.5087329310892354e-05,
      "loss": 0.0493,
      "step": 8651
    },
    {
      "epoch": 16.470756062767475,
      "grad_norm": 0.24092860519886017,
      "learning_rate": 4.508097808828199e-05,
      "loss": 0.0681,
      "step": 8652
    },
    {
      "epoch": 16.472658107465527,
      "grad_norm": 0.11901987344026566,
      "learning_rate": 4.5074626865671645e-05,
      "loss": 0.0632,
      "step": 8653
    },
    {
      "epoch": 16.474560152163576,
      "grad_norm": 0.14405383169651031,
      "learning_rate": 4.506827564306129e-05,
      "loss": 0.0666,
      "step": 8654
    },
    {
      "epoch": 16.476462196861625,
      "grad_norm": 0.07029570639133453,
      "learning_rate": 4.506192442045094e-05,
      "loss": 0.0563,
      "step": 8655
    },
    {
      "epoch": 16.478364241559678,
      "grad_norm": 0.14019885659217834,
      "learning_rate": 4.5055573197840586e-05,
      "loss": 0.0518,
      "step": 8656
    },
    {
      "epoch": 16.480266286257727,
      "grad_norm": 0.12269244343042374,
      "learning_rate": 4.504922197523023e-05,
      "loss": 0.0483,
      "step": 8657
    },
    {
      "epoch": 16.482168330955776,
      "grad_norm": 0.29546546936035156,
      "learning_rate": 4.5042870752619883e-05,
      "loss": 0.0733,
      "step": 8658
    },
    {
      "epoch": 16.48407037565383,
      "grad_norm": 0.12874148786067963,
      "learning_rate": 4.503651953000953e-05,
      "loss": 0.066,
      "step": 8659
    },
    {
      "epoch": 16.485972420351878,
      "grad_norm": 0.22310982644557953,
      "learning_rate": 4.503016830739918e-05,
      "loss": 0.065,
      "step": 8660
    },
    {
      "epoch": 16.487874465049927,
      "grad_norm": 0.12721015512943268,
      "learning_rate": 4.502381708478882e-05,
      "loss": 0.0753,
      "step": 8661
    },
    {
      "epoch": 16.48977650974798,
      "grad_norm": 0.15060150623321533,
      "learning_rate": 4.501746586217847e-05,
      "loss": 0.0525,
      "step": 8662
    },
    {
      "epoch": 16.49167855444603,
      "grad_norm": 0.13536332547664642,
      "learning_rate": 4.501111463956812e-05,
      "loss": 0.0601,
      "step": 8663
    },
    {
      "epoch": 16.49358059914408,
      "grad_norm": 0.1088651642203331,
      "learning_rate": 4.500476341695777e-05,
      "loss": 0.0357,
      "step": 8664
    },
    {
      "epoch": 16.49548264384213,
      "grad_norm": 0.17497827112674713,
      "learning_rate": 4.499841219434741e-05,
      "loss": 0.0709,
      "step": 8665
    },
    {
      "epoch": 16.49738468854018,
      "grad_norm": 0.18341687321662903,
      "learning_rate": 4.499206097173706e-05,
      "loss": 0.0563,
      "step": 8666
    },
    {
      "epoch": 16.499286733238232,
      "grad_norm": 0.11489497125148773,
      "learning_rate": 4.498570974912671e-05,
      "loss": 0.0652,
      "step": 8667
    },
    {
      "epoch": 16.50118877793628,
      "grad_norm": 0.06444443762302399,
      "learning_rate": 4.4979358526516354e-05,
      "loss": 0.0341,
      "step": 8668
    },
    {
      "epoch": 16.50309082263433,
      "grad_norm": 0.13678719103336334,
      "learning_rate": 4.4973007303906006e-05,
      "loss": 0.0571,
      "step": 8669
    },
    {
      "epoch": 16.504992867332383,
      "grad_norm": 0.11370810866355896,
      "learning_rate": 4.496665608129565e-05,
      "loss": 0.0603,
      "step": 8670
    },
    {
      "epoch": 16.506894912030432,
      "grad_norm": 0.1342094987630844,
      "learning_rate": 4.4960304858685296e-05,
      "loss": 0.0576,
      "step": 8671
    },
    {
      "epoch": 16.50879695672848,
      "grad_norm": 0.06139456853270531,
      "learning_rate": 4.495395363607495e-05,
      "loss": 0.0606,
      "step": 8672
    },
    {
      "epoch": 16.510699001426534,
      "grad_norm": 0.1667710840702057,
      "learning_rate": 4.494760241346459e-05,
      "loss": 0.0855,
      "step": 8673
    },
    {
      "epoch": 16.512601046124583,
      "grad_norm": 0.1714620143175125,
      "learning_rate": 4.494125119085424e-05,
      "loss": 0.0632,
      "step": 8674
    },
    {
      "epoch": 16.514503090822636,
      "grad_norm": 0.21141253411769867,
      "learning_rate": 4.493489996824389e-05,
      "loss": 0.0459,
      "step": 8675
    },
    {
      "epoch": 16.516405135520685,
      "grad_norm": 0.19178642332553864,
      "learning_rate": 4.4928548745633535e-05,
      "loss": 0.0448,
      "step": 8676
    },
    {
      "epoch": 16.518307180218734,
      "grad_norm": 0.1269862800836563,
      "learning_rate": 4.492219752302319e-05,
      "loss": 0.0574,
      "step": 8677
    },
    {
      "epoch": 16.520209224916787,
      "grad_norm": 0.31225162744522095,
      "learning_rate": 4.491584630041283e-05,
      "loss": 0.0822,
      "step": 8678
    },
    {
      "epoch": 16.522111269614836,
      "grad_norm": 0.250698983669281,
      "learning_rate": 4.490949507780248e-05,
      "loss": 0.0597,
      "step": 8679
    },
    {
      "epoch": 16.524013314312885,
      "grad_norm": 0.24116088449954987,
      "learning_rate": 4.490314385519213e-05,
      "loss": 0.0481,
      "step": 8680
    },
    {
      "epoch": 16.525915359010938,
      "grad_norm": 0.24615411460399628,
      "learning_rate": 4.4896792632581774e-05,
      "loss": 0.0613,
      "step": 8681
    },
    {
      "epoch": 16.527817403708987,
      "grad_norm": 0.1894298940896988,
      "learning_rate": 4.4890441409971426e-05,
      "loss": 0.0735,
      "step": 8682
    },
    {
      "epoch": 16.529719448407036,
      "grad_norm": 0.09118109941482544,
      "learning_rate": 4.4884090187361064e-05,
      "loss": 0.0493,
      "step": 8683
    },
    {
      "epoch": 16.53162149310509,
      "grad_norm": 0.2347007691860199,
      "learning_rate": 4.4877738964750716e-05,
      "loss": 0.0791,
      "step": 8684
    },
    {
      "epoch": 16.533523537803138,
      "grad_norm": 0.1817847639322281,
      "learning_rate": 4.487138774214036e-05,
      "loss": 0.0587,
      "step": 8685
    },
    {
      "epoch": 16.53542558250119,
      "grad_norm": 0.240115687251091,
      "learning_rate": 4.486503651953001e-05,
      "loss": 0.0673,
      "step": 8686
    },
    {
      "epoch": 16.53732762719924,
      "grad_norm": 0.17856906354427338,
      "learning_rate": 4.485868529691966e-05,
      "loss": 0.0505,
      "step": 8687
    },
    {
      "epoch": 16.53922967189729,
      "grad_norm": 0.15239039063453674,
      "learning_rate": 4.48523340743093e-05,
      "loss": 0.0472,
      "step": 8688
    },
    {
      "epoch": 16.54113171659534,
      "grad_norm": 0.11803076416254044,
      "learning_rate": 4.4845982851698955e-05,
      "loss": 0.0524,
      "step": 8689
    },
    {
      "epoch": 16.54303376129339,
      "grad_norm": 0.11645831167697906,
      "learning_rate": 4.48396316290886e-05,
      "loss": 0.0687,
      "step": 8690
    },
    {
      "epoch": 16.54493580599144,
      "grad_norm": 0.07123297452926636,
      "learning_rate": 4.483328040647825e-05,
      "loss": 0.0347,
      "step": 8691
    },
    {
      "epoch": 16.546837850689492,
      "grad_norm": 0.19448618590831757,
      "learning_rate": 4.48269291838679e-05,
      "loss": 0.071,
      "step": 8692
    },
    {
      "epoch": 16.54873989538754,
      "grad_norm": 0.13725486397743225,
      "learning_rate": 4.482057796125754e-05,
      "loss": 0.0619,
      "step": 8693
    },
    {
      "epoch": 16.55064194008559,
      "grad_norm": 0.17994223535060883,
      "learning_rate": 4.4814226738647194e-05,
      "loss": 0.0602,
      "step": 8694
    },
    {
      "epoch": 16.552543984783643,
      "grad_norm": 0.17840850353240967,
      "learning_rate": 4.480787551603684e-05,
      "loss": 0.0607,
      "step": 8695
    },
    {
      "epoch": 16.554446029481692,
      "grad_norm": 0.16476470232009888,
      "learning_rate": 4.480152429342649e-05,
      "loss": 0.0697,
      "step": 8696
    },
    {
      "epoch": 16.556348074179745,
      "grad_norm": 0.24432149529457092,
      "learning_rate": 4.479517307081613e-05,
      "loss": 0.0793,
      "step": 8697
    },
    {
      "epoch": 16.558250118877794,
      "grad_norm": 0.23164185881614685,
      "learning_rate": 4.478882184820578e-05,
      "loss": 0.066,
      "step": 8698
    },
    {
      "epoch": 16.560152163575843,
      "grad_norm": 0.11395987868309021,
      "learning_rate": 4.478247062559543e-05,
      "loss": 0.0299,
      "step": 8699
    },
    {
      "epoch": 16.562054208273896,
      "grad_norm": 0.11444274336099625,
      "learning_rate": 4.477611940298508e-05,
      "loss": 0.0535,
      "step": 8700
    },
    {
      "epoch": 16.563956252971945,
      "grad_norm": 0.2184406965970993,
      "learning_rate": 4.476976818037472e-05,
      "loss": 0.061,
      "step": 8701
    },
    {
      "epoch": 16.565858297669994,
      "grad_norm": 0.11942858248949051,
      "learning_rate": 4.476341695776437e-05,
      "loss": 0.0583,
      "step": 8702
    },
    {
      "epoch": 16.567760342368047,
      "grad_norm": 0.0957055389881134,
      "learning_rate": 4.475706573515402e-05,
      "loss": 0.0568,
      "step": 8703
    },
    {
      "epoch": 16.569662387066096,
      "grad_norm": 0.27850326895713806,
      "learning_rate": 4.4750714512543665e-05,
      "loss": 0.075,
      "step": 8704
    },
    {
      "epoch": 16.571564431764145,
      "grad_norm": 0.1830519586801529,
      "learning_rate": 4.474436328993332e-05,
      "loss": 0.0564,
      "step": 8705
    },
    {
      "epoch": 16.573466476462198,
      "grad_norm": 0.07941751182079315,
      "learning_rate": 4.473801206732296e-05,
      "loss": 0.0454,
      "step": 8706
    },
    {
      "epoch": 16.575368521160247,
      "grad_norm": 0.11852432787418365,
      "learning_rate": 4.473166084471261e-05,
      "loss": 0.0622,
      "step": 8707
    },
    {
      "epoch": 16.5772705658583,
      "grad_norm": 0.16959674656391144,
      "learning_rate": 4.472530962210226e-05,
      "loss": 0.0552,
      "step": 8708
    },
    {
      "epoch": 16.57917261055635,
      "grad_norm": 0.1568812131881714,
      "learning_rate": 4.4718958399491904e-05,
      "loss": 0.064,
      "step": 8709
    },
    {
      "epoch": 16.581074655254397,
      "grad_norm": 0.12478756904602051,
      "learning_rate": 4.471260717688155e-05,
      "loss": 0.0668,
      "step": 8710
    },
    {
      "epoch": 16.58297669995245,
      "grad_norm": 0.16158562898635864,
      "learning_rate": 4.47062559542712e-05,
      "loss": 0.0551,
      "step": 8711
    },
    {
      "epoch": 16.5848787446505,
      "grad_norm": 0.17838051915168762,
      "learning_rate": 4.4699904731660846e-05,
      "loss": 0.0702,
      "step": 8712
    },
    {
      "epoch": 16.58678078934855,
      "grad_norm": 0.1448523849248886,
      "learning_rate": 4.46935535090505e-05,
      "loss": 0.0665,
      "step": 8713
    },
    {
      "epoch": 16.5886828340466,
      "grad_norm": 0.21258023381233215,
      "learning_rate": 4.468720228644014e-05,
      "loss": 0.0672,
      "step": 8714
    },
    {
      "epoch": 16.59058487874465,
      "grad_norm": 0.13940654695034027,
      "learning_rate": 4.468085106382979e-05,
      "loss": 0.0519,
      "step": 8715
    },
    {
      "epoch": 16.5924869234427,
      "grad_norm": 0.09483486413955688,
      "learning_rate": 4.467449984121943e-05,
      "loss": 0.0645,
      "step": 8716
    },
    {
      "epoch": 16.594388968140752,
      "grad_norm": 0.16645945608615875,
      "learning_rate": 4.4668148618609085e-05,
      "loss": 0.0899,
      "step": 8717
    },
    {
      "epoch": 16.5962910128388,
      "grad_norm": 0.12020470947027206,
      "learning_rate": 4.4661797395998736e-05,
      "loss": 0.0552,
      "step": 8718
    },
    {
      "epoch": 16.598193057536854,
      "grad_norm": 0.11481377482414246,
      "learning_rate": 4.4655446173388375e-05,
      "loss": 0.0538,
      "step": 8719
    },
    {
      "epoch": 16.600095102234903,
      "grad_norm": 0.18040695786476135,
      "learning_rate": 4.4649094950778027e-05,
      "loss": 0.0635,
      "step": 8720
    },
    {
      "epoch": 16.601997146932952,
      "grad_norm": 0.08994857221841812,
      "learning_rate": 4.464274372816767e-05,
      "loss": 0.0567,
      "step": 8721
    },
    {
      "epoch": 16.603899191631005,
      "grad_norm": 0.15336550772190094,
      "learning_rate": 4.4636392505557323e-05,
      "loss": 0.0671,
      "step": 8722
    },
    {
      "epoch": 16.605801236329054,
      "grad_norm": 0.11266004294157028,
      "learning_rate": 4.463004128294697e-05,
      "loss": 0.062,
      "step": 8723
    },
    {
      "epoch": 16.607703281027103,
      "grad_norm": 0.10975999385118484,
      "learning_rate": 4.4623690060336614e-05,
      "loss": 0.0563,
      "step": 8724
    },
    {
      "epoch": 16.609605325725155,
      "grad_norm": 0.11690010130405426,
      "learning_rate": 4.4617338837726265e-05,
      "loss": 0.0757,
      "step": 8725
    },
    {
      "epoch": 16.611507370423205,
      "grad_norm": 0.19387713074684143,
      "learning_rate": 4.461098761511591e-05,
      "loss": 0.0476,
      "step": 8726
    },
    {
      "epoch": 16.613409415121254,
      "grad_norm": 0.11086385697126389,
      "learning_rate": 4.460463639250556e-05,
      "loss": 0.0356,
      "step": 8727
    },
    {
      "epoch": 16.615311459819306,
      "grad_norm": 0.2539948523044586,
      "learning_rate": 4.459828516989521e-05,
      "loss": 0.0421,
      "step": 8728
    },
    {
      "epoch": 16.617213504517355,
      "grad_norm": 0.17350983619689941,
      "learning_rate": 4.459193394728485e-05,
      "loss": 0.0575,
      "step": 8729
    },
    {
      "epoch": 16.619115549215408,
      "grad_norm": 0.11512711644172668,
      "learning_rate": 4.4585582724674504e-05,
      "loss": 0.0512,
      "step": 8730
    },
    {
      "epoch": 16.621017593913457,
      "grad_norm": 0.23560310900211334,
      "learning_rate": 4.457923150206415e-05,
      "loss": 0.0675,
      "step": 8731
    },
    {
      "epoch": 16.622919638611506,
      "grad_norm": 0.13212914764881134,
      "learning_rate": 4.45728802794538e-05,
      "loss": 0.0645,
      "step": 8732
    },
    {
      "epoch": 16.62482168330956,
      "grad_norm": 0.07807771861553192,
      "learning_rate": 4.456652905684344e-05,
      "loss": 0.0537,
      "step": 8733
    },
    {
      "epoch": 16.626723728007608,
      "grad_norm": 0.1315772980451584,
      "learning_rate": 4.456017783423309e-05,
      "loss": 0.0492,
      "step": 8734
    },
    {
      "epoch": 16.628625772705657,
      "grad_norm": 0.17557771503925323,
      "learning_rate": 4.455382661162274e-05,
      "loss": 0.0486,
      "step": 8735
    },
    {
      "epoch": 16.63052781740371,
      "grad_norm": 0.1737060397863388,
      "learning_rate": 4.454747538901239e-05,
      "loss": 0.0476,
      "step": 8736
    },
    {
      "epoch": 16.63242986210176,
      "grad_norm": 0.11324983835220337,
      "learning_rate": 4.454112416640203e-05,
      "loss": 0.0395,
      "step": 8737
    },
    {
      "epoch": 16.634331906799808,
      "grad_norm": 0.0777667984366417,
      "learning_rate": 4.453477294379168e-05,
      "loss": 0.0482,
      "step": 8738
    },
    {
      "epoch": 16.63623395149786,
      "grad_norm": 0.17832614481449127,
      "learning_rate": 4.452842172118133e-05,
      "loss": 0.0607,
      "step": 8739
    },
    {
      "epoch": 16.63813599619591,
      "grad_norm": 0.16999396681785583,
      "learning_rate": 4.4522070498570975e-05,
      "loss": 0.0543,
      "step": 8740
    },
    {
      "epoch": 16.640038040893963,
      "grad_norm": 0.06686799973249435,
      "learning_rate": 4.451571927596063e-05,
      "loss": 0.0596,
      "step": 8741
    },
    {
      "epoch": 16.64194008559201,
      "grad_norm": 0.15245421230793,
      "learning_rate": 4.450936805335027e-05,
      "loss": 0.0653,
      "step": 8742
    },
    {
      "epoch": 16.64384213029006,
      "grad_norm": 0.10946781933307648,
      "learning_rate": 4.450301683073992e-05,
      "loss": 0.0511,
      "step": 8743
    },
    {
      "epoch": 16.645744174988113,
      "grad_norm": 0.133508563041687,
      "learning_rate": 4.449666560812957e-05,
      "loss": 0.0538,
      "step": 8744
    },
    {
      "epoch": 16.647646219686163,
      "grad_norm": 0.1146819069981575,
      "learning_rate": 4.4490314385519214e-05,
      "loss": 0.1121,
      "step": 8745
    },
    {
      "epoch": 16.64954826438421,
      "grad_norm": 0.10533533990383148,
      "learning_rate": 4.448396316290886e-05,
      "loss": 0.0567,
      "step": 8746
    },
    {
      "epoch": 16.651450309082264,
      "grad_norm": 0.10547687858343124,
      "learning_rate": 4.447761194029851e-05,
      "loss": 0.066,
      "step": 8747
    },
    {
      "epoch": 16.653352353780313,
      "grad_norm": 0.18761403858661652,
      "learning_rate": 4.4471260717688156e-05,
      "loss": 0.0635,
      "step": 8748
    },
    {
      "epoch": 16.655254398478363,
      "grad_norm": 0.11144470423460007,
      "learning_rate": 4.446490949507781e-05,
      "loss": 0.0711,
      "step": 8749
    },
    {
      "epoch": 16.657156443176415,
      "grad_norm": 0.16650743782520294,
      "learning_rate": 4.445855827246745e-05,
      "loss": 0.0465,
      "step": 8750
    },
    {
      "epoch": 16.659058487874464,
      "grad_norm": 0.26453036069869995,
      "learning_rate": 4.44522070498571e-05,
      "loss": 0.0659,
      "step": 8751
    },
    {
      "epoch": 16.660960532572517,
      "grad_norm": 0.27274948358535767,
      "learning_rate": 4.444585582724674e-05,
      "loss": 0.0577,
      "step": 8752
    },
    {
      "epoch": 16.662862577270566,
      "grad_norm": 0.2193809300661087,
      "learning_rate": 4.4439504604636395e-05,
      "loss": 0.0724,
      "step": 8753
    },
    {
      "epoch": 16.664764621968615,
      "grad_norm": 0.15518517792224884,
      "learning_rate": 4.443315338202605e-05,
      "loss": 0.0615,
      "step": 8754
    },
    {
      "epoch": 16.666666666666668,
      "grad_norm": 0.09343301504850388,
      "learning_rate": 4.4426802159415685e-05,
      "loss": 0.0574,
      "step": 8755
    },
    {
      "epoch": 16.668568711364717,
      "grad_norm": 0.14349409937858582,
      "learning_rate": 4.442045093680534e-05,
      "loss": 0.0677,
      "step": 8756
    },
    {
      "epoch": 16.670470756062766,
      "grad_norm": 0.13636383414268494,
      "learning_rate": 4.441409971419498e-05,
      "loss": 0.0767,
      "step": 8757
    },
    {
      "epoch": 16.67237280076082,
      "grad_norm": 0.2968313992023468,
      "learning_rate": 4.4407748491584634e-05,
      "loss": 0.0605,
      "step": 8758
    },
    {
      "epoch": 16.674274845458868,
      "grad_norm": 0.18513229489326477,
      "learning_rate": 4.440139726897428e-05,
      "loss": 0.0675,
      "step": 8759
    },
    {
      "epoch": 16.676176890156917,
      "grad_norm": 0.14191585779190063,
      "learning_rate": 4.4395046046363924e-05,
      "loss": 0.0563,
      "step": 8760
    },
    {
      "epoch": 16.67807893485497,
      "grad_norm": 0.10100386291742325,
      "learning_rate": 4.4388694823753576e-05,
      "loss": 0.0682,
      "step": 8761
    },
    {
      "epoch": 16.67998097955302,
      "grad_norm": 0.16358493268489838,
      "learning_rate": 4.438234360114322e-05,
      "loss": 0.0518,
      "step": 8762
    },
    {
      "epoch": 16.68188302425107,
      "grad_norm": 0.19515903294086456,
      "learning_rate": 4.437599237853287e-05,
      "loss": 0.0757,
      "step": 8763
    },
    {
      "epoch": 16.68378506894912,
      "grad_norm": 0.09387758374214172,
      "learning_rate": 4.436964115592252e-05,
      "loss": 0.0704,
      "step": 8764
    },
    {
      "epoch": 16.68568711364717,
      "grad_norm": 0.17992985248565674,
      "learning_rate": 4.436328993331216e-05,
      "loss": 0.0634,
      "step": 8765
    },
    {
      "epoch": 16.687589158345222,
      "grad_norm": 0.0763043537735939,
      "learning_rate": 4.4356938710701815e-05,
      "loss": 0.0597,
      "step": 8766
    },
    {
      "epoch": 16.68949120304327,
      "grad_norm": 0.14316275715827942,
      "learning_rate": 4.435058748809146e-05,
      "loss": 0.0659,
      "step": 8767
    },
    {
      "epoch": 16.69139324774132,
      "grad_norm": 0.23489262163639069,
      "learning_rate": 4.434423626548111e-05,
      "loss": 0.0625,
      "step": 8768
    },
    {
      "epoch": 16.693295292439373,
      "grad_norm": 0.10423056036233902,
      "learning_rate": 4.433788504287075e-05,
      "loss": 0.0496,
      "step": 8769
    },
    {
      "epoch": 16.695197337137422,
      "grad_norm": 0.16861405968666077,
      "learning_rate": 4.43315338202604e-05,
      "loss": 0.065,
      "step": 8770
    },
    {
      "epoch": 16.69709938183547,
      "grad_norm": 0.1740831732749939,
      "learning_rate": 4.432518259765005e-05,
      "loss": 0.0471,
      "step": 8771
    },
    {
      "epoch": 16.699001426533524,
      "grad_norm": 0.13501209020614624,
      "learning_rate": 4.43188313750397e-05,
      "loss": 0.0473,
      "step": 8772
    },
    {
      "epoch": 16.700903471231573,
      "grad_norm": 0.2359786033630371,
      "learning_rate": 4.4312480152429344e-05,
      "loss": 0.071,
      "step": 8773
    },
    {
      "epoch": 16.702805515929626,
      "grad_norm": 0.11417681723833084,
      "learning_rate": 4.430612892981899e-05,
      "loss": 0.0724,
      "step": 8774
    },
    {
      "epoch": 16.704707560627675,
      "grad_norm": 0.07876874506473541,
      "learning_rate": 4.429977770720864e-05,
      "loss": 0.0417,
      "step": 8775
    },
    {
      "epoch": 16.706609605325724,
      "grad_norm": 0.17278744280338287,
      "learning_rate": 4.4293426484598286e-05,
      "loss": 0.0564,
      "step": 8776
    },
    {
      "epoch": 16.708511650023777,
      "grad_norm": 0.175479918718338,
      "learning_rate": 4.428707526198794e-05,
      "loss": 0.054,
      "step": 8777
    },
    {
      "epoch": 16.710413694721826,
      "grad_norm": 0.16353516280651093,
      "learning_rate": 4.428072403937758e-05,
      "loss": 0.0444,
      "step": 8778
    },
    {
      "epoch": 16.712315739419875,
      "grad_norm": 0.09007339924573898,
      "learning_rate": 4.427437281676723e-05,
      "loss": 0.0549,
      "step": 8779
    },
    {
      "epoch": 16.714217784117928,
      "grad_norm": 0.16510137915611267,
      "learning_rate": 4.426802159415688e-05,
      "loss": 0.0726,
      "step": 8780
    },
    {
      "epoch": 16.716119828815977,
      "grad_norm": 0.2640451490879059,
      "learning_rate": 4.4261670371546525e-05,
      "loss": 0.0745,
      "step": 8781
    },
    {
      "epoch": 16.718021873514026,
      "grad_norm": 0.15771816670894623,
      "learning_rate": 4.425531914893617e-05,
      "loss": 0.0525,
      "step": 8782
    },
    {
      "epoch": 16.71992391821208,
      "grad_norm": 0.11966612190008163,
      "learning_rate": 4.424896792632582e-05,
      "loss": 0.0511,
      "step": 8783
    },
    {
      "epoch": 16.721825962910128,
      "grad_norm": 0.2334906905889511,
      "learning_rate": 4.4242616703715467e-05,
      "loss": 0.0518,
      "step": 8784
    },
    {
      "epoch": 16.72372800760818,
      "grad_norm": 0.20914064347743988,
      "learning_rate": 4.423626548110512e-05,
      "loss": 0.0591,
      "step": 8785
    },
    {
      "epoch": 16.72563005230623,
      "grad_norm": 0.08859287202358246,
      "learning_rate": 4.4229914258494763e-05,
      "loss": 0.0704,
      "step": 8786
    },
    {
      "epoch": 16.72753209700428,
      "grad_norm": 0.10872913897037506,
      "learning_rate": 4.422356303588441e-05,
      "loss": 0.0541,
      "step": 8787
    },
    {
      "epoch": 16.72943414170233,
      "grad_norm": 0.2138252556324005,
      "learning_rate": 4.4217211813274054e-05,
      "loss": 0.0505,
      "step": 8788
    },
    {
      "epoch": 16.73133618640038,
      "grad_norm": 0.0785028338432312,
      "learning_rate": 4.4210860590663705e-05,
      "loss": 0.0542,
      "step": 8789
    },
    {
      "epoch": 16.73323823109843,
      "grad_norm": 0.20003677904605865,
      "learning_rate": 4.420450936805336e-05,
      "loss": 0.1044,
      "step": 8790
    },
    {
      "epoch": 16.735140275796482,
      "grad_norm": 0.0482601672410965,
      "learning_rate": 4.4198158145442996e-05,
      "loss": 0.0593,
      "step": 8791
    },
    {
      "epoch": 16.73704232049453,
      "grad_norm": 0.13578015565872192,
      "learning_rate": 4.419180692283265e-05,
      "loss": 0.0608,
      "step": 8792
    },
    {
      "epoch": 16.73894436519258,
      "grad_norm": 0.15712803602218628,
      "learning_rate": 4.418545570022229e-05,
      "loss": 0.0592,
      "step": 8793
    },
    {
      "epoch": 16.740846409890633,
      "grad_norm": 0.09721986204385757,
      "learning_rate": 4.4179104477611944e-05,
      "loss": 0.0435,
      "step": 8794
    },
    {
      "epoch": 16.742748454588682,
      "grad_norm": 0.13611485064029694,
      "learning_rate": 4.417275325500159e-05,
      "loss": 0.0517,
      "step": 8795
    },
    {
      "epoch": 16.744650499286735,
      "grad_norm": 0.11596285551786423,
      "learning_rate": 4.4166402032391234e-05,
      "loss": 0.0538,
      "step": 8796
    },
    {
      "epoch": 16.746552543984784,
      "grad_norm": 0.1389114111661911,
      "learning_rate": 4.4160050809780886e-05,
      "loss": 0.0677,
      "step": 8797
    },
    {
      "epoch": 16.748454588682833,
      "grad_norm": 0.13880036771297455,
      "learning_rate": 4.415369958717053e-05,
      "loss": 0.0557,
      "step": 8798
    },
    {
      "epoch": 16.750356633380886,
      "grad_norm": 0.17362141609191895,
      "learning_rate": 4.414734836456018e-05,
      "loss": 0.0782,
      "step": 8799
    },
    {
      "epoch": 16.752258678078935,
      "grad_norm": 0.21298301219940186,
      "learning_rate": 4.414099714194982e-05,
      "loss": 0.0537,
      "step": 8800
    },
    {
      "epoch": 16.754160722776984,
      "grad_norm": 0.21551482379436493,
      "learning_rate": 4.413464591933947e-05,
      "loss": 0.0584,
      "step": 8801
    },
    {
      "epoch": 16.756062767475036,
      "grad_norm": 0.06975243985652924,
      "learning_rate": 4.4128294696729125e-05,
      "loss": 0.0457,
      "step": 8802
    },
    {
      "epoch": 16.757964812173086,
      "grad_norm": 0.1738225221633911,
      "learning_rate": 4.412194347411877e-05,
      "loss": 0.0568,
      "step": 8803
    },
    {
      "epoch": 16.759866856871135,
      "grad_norm": 0.17236964404582977,
      "learning_rate": 4.411559225150842e-05,
      "loss": 0.0603,
      "step": 8804
    },
    {
      "epoch": 16.761768901569187,
      "grad_norm": 0.1748720407485962,
      "learning_rate": 4.410924102889806e-05,
      "loss": 0.0505,
      "step": 8805
    },
    {
      "epoch": 16.763670946267236,
      "grad_norm": 0.2570509612560272,
      "learning_rate": 4.410288980628771e-05,
      "loss": 0.0623,
      "step": 8806
    },
    {
      "epoch": 16.76557299096529,
      "grad_norm": 0.18713904917240143,
      "learning_rate": 4.409653858367736e-05,
      "loss": 0.0643,
      "step": 8807
    },
    {
      "epoch": 16.767475035663338,
      "grad_norm": 0.1882544904947281,
      "learning_rate": 4.409018736106701e-05,
      "loss": 0.0557,
      "step": 8808
    },
    {
      "epoch": 16.769377080361387,
      "grad_norm": 0.15929445624351501,
      "learning_rate": 4.4083836138456654e-05,
      "loss": 0.0512,
      "step": 8809
    },
    {
      "epoch": 16.77127912505944,
      "grad_norm": 0.17998144030570984,
      "learning_rate": 4.40774849158463e-05,
      "loss": 0.0656,
      "step": 8810
    },
    {
      "epoch": 16.77318116975749,
      "grad_norm": 0.1303442269563675,
      "learning_rate": 4.407113369323595e-05,
      "loss": 0.0475,
      "step": 8811
    },
    {
      "epoch": 16.775083214455538,
      "grad_norm": 0.4382564425468445,
      "learning_rate": 4.4064782470625596e-05,
      "loss": 0.0731,
      "step": 8812
    },
    {
      "epoch": 16.77698525915359,
      "grad_norm": 0.19651977717876434,
      "learning_rate": 4.405843124801525e-05,
      "loss": 0.0697,
      "step": 8813
    },
    {
      "epoch": 16.77888730385164,
      "grad_norm": 0.21434944868087769,
      "learning_rate": 4.405208002540489e-05,
      "loss": 0.0737,
      "step": 8814
    },
    {
      "epoch": 16.780789348549693,
      "grad_norm": 0.2197565883398056,
      "learning_rate": 4.404572880279454e-05,
      "loss": 0.0952,
      "step": 8815
    },
    {
      "epoch": 16.78269139324774,
      "grad_norm": 0.30654266476631165,
      "learning_rate": 4.403937758018419e-05,
      "loss": 0.0658,
      "step": 8816
    },
    {
      "epoch": 16.78459343794579,
      "grad_norm": 0.1521390974521637,
      "learning_rate": 4.4033026357573835e-05,
      "loss": 0.0463,
      "step": 8817
    },
    {
      "epoch": 16.786495482643844,
      "grad_norm": 0.16150012612342834,
      "learning_rate": 4.402667513496348e-05,
      "loss": 0.0599,
      "step": 8818
    },
    {
      "epoch": 16.788397527341893,
      "grad_norm": 0.17025931179523468,
      "learning_rate": 4.402032391235313e-05,
      "loss": 0.0591,
      "step": 8819
    },
    {
      "epoch": 16.79029957203994,
      "grad_norm": 0.17232930660247803,
      "learning_rate": 4.401397268974278e-05,
      "loss": 0.0683,
      "step": 8820
    },
    {
      "epoch": 16.792201616737994,
      "grad_norm": 0.15758217871189117,
      "learning_rate": 4.400762146713243e-05,
      "loss": 0.0466,
      "step": 8821
    },
    {
      "epoch": 16.794103661436043,
      "grad_norm": 0.08380364626646042,
      "learning_rate": 4.4001270244522074e-05,
      "loss": 0.0478,
      "step": 8822
    },
    {
      "epoch": 16.796005706134093,
      "grad_norm": 0.0926741287112236,
      "learning_rate": 4.399491902191172e-05,
      "loss": 0.08,
      "step": 8823
    },
    {
      "epoch": 16.797907750832145,
      "grad_norm": 0.16255873441696167,
      "learning_rate": 4.3988567799301364e-05,
      "loss": 0.0655,
      "step": 8824
    },
    {
      "epoch": 16.799809795530194,
      "grad_norm": 0.11387108266353607,
      "learning_rate": 4.3982216576691016e-05,
      "loss": 0.0658,
      "step": 8825
    },
    {
      "epoch": 16.801711840228247,
      "grad_norm": 0.2308814972639084,
      "learning_rate": 4.397586535408066e-05,
      "loss": 0.0931,
      "step": 8826
    },
    {
      "epoch": 16.803613884926296,
      "grad_norm": 0.12670093774795532,
      "learning_rate": 4.3969514131470306e-05,
      "loss": 0.0406,
      "step": 8827
    },
    {
      "epoch": 16.805515929624345,
      "grad_norm": 0.2202993482351303,
      "learning_rate": 4.396316290885996e-05,
      "loss": 0.0648,
      "step": 8828
    },
    {
      "epoch": 16.807417974322398,
      "grad_norm": 0.12769080698490143,
      "learning_rate": 4.39568116862496e-05,
      "loss": 0.0633,
      "step": 8829
    },
    {
      "epoch": 16.809320019020447,
      "grad_norm": 0.12309324741363525,
      "learning_rate": 4.3950460463639255e-05,
      "loss": 0.0589,
      "step": 8830
    },
    {
      "epoch": 16.811222063718496,
      "grad_norm": 0.1103903204202652,
      "learning_rate": 4.39441092410289e-05,
      "loss": 0.0483,
      "step": 8831
    },
    {
      "epoch": 16.81312410841655,
      "grad_norm": 0.16271445155143738,
      "learning_rate": 4.3937758018418545e-05,
      "loss": 0.0428,
      "step": 8832
    },
    {
      "epoch": 16.815026153114598,
      "grad_norm": 0.17252840101718903,
      "learning_rate": 4.39314067958082e-05,
      "loss": 0.0532,
      "step": 8833
    },
    {
      "epoch": 16.816928197812647,
      "grad_norm": 0.11162525415420532,
      "learning_rate": 4.392505557319784e-05,
      "loss": 0.0567,
      "step": 8834
    },
    {
      "epoch": 16.8188302425107,
      "grad_norm": 0.12144871801137924,
      "learning_rate": 4.3918704350587494e-05,
      "loss": 0.0568,
      "step": 8835
    },
    {
      "epoch": 16.82073228720875,
      "grad_norm": 0.2525745928287506,
      "learning_rate": 4.391235312797713e-05,
      "loss": 0.0436,
      "step": 8836
    },
    {
      "epoch": 16.8226343319068,
      "grad_norm": 0.08460042625665665,
      "learning_rate": 4.3906001905366784e-05,
      "loss": 0.0541,
      "step": 8837
    },
    {
      "epoch": 16.82453637660485,
      "grad_norm": 0.07870131731033325,
      "learning_rate": 4.3899650682756436e-05,
      "loss": 0.0647,
      "step": 8838
    },
    {
      "epoch": 16.8264384213029,
      "grad_norm": 0.17138586938381195,
      "learning_rate": 4.389329946014608e-05,
      "loss": 0.0707,
      "step": 8839
    },
    {
      "epoch": 16.828340466000952,
      "grad_norm": 0.19924390316009521,
      "learning_rate": 4.388694823753573e-05,
      "loss": 0.0723,
      "step": 8840
    },
    {
      "epoch": 16.830242510699,
      "grad_norm": 0.18340855836868286,
      "learning_rate": 4.388059701492537e-05,
      "loss": 0.0445,
      "step": 8841
    },
    {
      "epoch": 16.83214455539705,
      "grad_norm": 0.149722620844841,
      "learning_rate": 4.387424579231502e-05,
      "loss": 0.0507,
      "step": 8842
    },
    {
      "epoch": 16.834046600095103,
      "grad_norm": 0.14245644211769104,
      "learning_rate": 4.386789456970467e-05,
      "loss": 0.0633,
      "step": 8843
    },
    {
      "epoch": 16.835948644793152,
      "grad_norm": 0.06980764120817184,
      "learning_rate": 4.386154334709432e-05,
      "loss": 0.0588,
      "step": 8844
    },
    {
      "epoch": 16.8378506894912,
      "grad_norm": 0.10202120244503021,
      "learning_rate": 4.3855192124483965e-05,
      "loss": 0.0568,
      "step": 8845
    },
    {
      "epoch": 16.839752734189254,
      "grad_norm": 0.10382965207099915,
      "learning_rate": 4.384884090187361e-05,
      "loss": 0.0569,
      "step": 8846
    },
    {
      "epoch": 16.841654778887303,
      "grad_norm": 0.18479005992412567,
      "learning_rate": 4.384248967926326e-05,
      "loss": 0.0561,
      "step": 8847
    },
    {
      "epoch": 16.843556823585356,
      "grad_norm": 0.16195909678936005,
      "learning_rate": 4.3836138456652907e-05,
      "loss": 0.0675,
      "step": 8848
    },
    {
      "epoch": 16.845458868283405,
      "grad_norm": 0.2066333144903183,
      "learning_rate": 4.382978723404256e-05,
      "loss": 0.0605,
      "step": 8849
    },
    {
      "epoch": 16.847360912981454,
      "grad_norm": 0.13491970300674438,
      "learning_rate": 4.3823436011432204e-05,
      "loss": 0.0415,
      "step": 8850
    },
    {
      "epoch": 16.849262957679507,
      "grad_norm": 0.13810434937477112,
      "learning_rate": 4.381708478882185e-05,
      "loss": 0.0603,
      "step": 8851
    },
    {
      "epoch": 16.851165002377556,
      "grad_norm": 0.07920050621032715,
      "learning_rate": 4.38107335662115e-05,
      "loss": 0.0673,
      "step": 8852
    },
    {
      "epoch": 16.853067047075605,
      "grad_norm": 0.17346441745758057,
      "learning_rate": 4.3804382343601145e-05,
      "loss": 0.0397,
      "step": 8853
    },
    {
      "epoch": 16.854969091773658,
      "grad_norm": 0.08918207138776779,
      "learning_rate": 4.379803112099079e-05,
      "loss": 0.0487,
      "step": 8854
    },
    {
      "epoch": 16.856871136471707,
      "grad_norm": 0.11225546151399612,
      "learning_rate": 4.3791679898380436e-05,
      "loss": 0.0653,
      "step": 8855
    },
    {
      "epoch": 16.858773181169756,
      "grad_norm": 0.20793363451957703,
      "learning_rate": 4.378532867577009e-05,
      "loss": 0.0732,
      "step": 8856
    },
    {
      "epoch": 16.86067522586781,
      "grad_norm": 0.16826403141021729,
      "learning_rate": 4.377897745315974e-05,
      "loss": 0.0785,
      "step": 8857
    },
    {
      "epoch": 16.862577270565858,
      "grad_norm": 0.12270570546388626,
      "learning_rate": 4.3772626230549384e-05,
      "loss": 0.0544,
      "step": 8858
    },
    {
      "epoch": 16.86447931526391,
      "grad_norm": 0.11855387687683105,
      "learning_rate": 4.376627500793903e-05,
      "loss": 0.0541,
      "step": 8859
    },
    {
      "epoch": 16.86638135996196,
      "grad_norm": 0.1780882179737091,
      "learning_rate": 4.3759923785328674e-05,
      "loss": 0.0596,
      "step": 8860
    },
    {
      "epoch": 16.86828340466001,
      "grad_norm": 0.0805153101682663,
      "learning_rate": 4.3753572562718326e-05,
      "loss": 0.0437,
      "step": 8861
    },
    {
      "epoch": 16.87018544935806,
      "grad_norm": 0.2657288908958435,
      "learning_rate": 4.374722134010797e-05,
      "loss": 0.0819,
      "step": 8862
    },
    {
      "epoch": 16.87208749405611,
      "grad_norm": 0.14114326238632202,
      "learning_rate": 4.3740870117497616e-05,
      "loss": 0.0635,
      "step": 8863
    },
    {
      "epoch": 16.87398953875416,
      "grad_norm": 0.12601017951965332,
      "learning_rate": 4.373451889488727e-05,
      "loss": 0.0566,
      "step": 8864
    },
    {
      "epoch": 16.875891583452212,
      "grad_norm": 0.18008174002170563,
      "learning_rate": 4.372816767227691e-05,
      "loss": 0.0634,
      "step": 8865
    },
    {
      "epoch": 16.87779362815026,
      "grad_norm": 0.125730499625206,
      "learning_rate": 4.3721816449666565e-05,
      "loss": 0.0577,
      "step": 8866
    },
    {
      "epoch": 16.87969567284831,
      "grad_norm": 0.05357951298356056,
      "learning_rate": 4.371546522705621e-05,
      "loss": 0.0569,
      "step": 8867
    },
    {
      "epoch": 16.881597717546363,
      "grad_norm": 0.06639115512371063,
      "learning_rate": 4.3709114004445855e-05,
      "loss": 0.0472,
      "step": 8868
    },
    {
      "epoch": 16.883499762244412,
      "grad_norm": 0.10009011626243591,
      "learning_rate": 4.370276278183551e-05,
      "loss": 0.0421,
      "step": 8869
    },
    {
      "epoch": 16.885401806942465,
      "grad_norm": 0.19268018007278442,
      "learning_rate": 4.369641155922515e-05,
      "loss": 0.0648,
      "step": 8870
    },
    {
      "epoch": 16.887303851640514,
      "grad_norm": 0.24461723864078522,
      "learning_rate": 4.3690060336614804e-05,
      "loss": 0.0732,
      "step": 8871
    },
    {
      "epoch": 16.889205896338563,
      "grad_norm": 0.13241729140281677,
      "learning_rate": 4.368370911400444e-05,
      "loss": 0.0392,
      "step": 8872
    },
    {
      "epoch": 16.891107941036616,
      "grad_norm": 0.1267736554145813,
      "learning_rate": 4.3677357891394094e-05,
      "loss": 0.0747,
      "step": 8873
    },
    {
      "epoch": 16.893009985734665,
      "grad_norm": 0.2197117805480957,
      "learning_rate": 4.3671006668783746e-05,
      "loss": 0.0591,
      "step": 8874
    },
    {
      "epoch": 16.894912030432714,
      "grad_norm": 0.2949211001396179,
      "learning_rate": 4.366465544617339e-05,
      "loss": 0.0731,
      "step": 8875
    },
    {
      "epoch": 16.896814075130766,
      "grad_norm": 0.13164544105529785,
      "learning_rate": 4.365830422356304e-05,
      "loss": 0.0595,
      "step": 8876
    },
    {
      "epoch": 16.898716119828816,
      "grad_norm": 0.12970279157161713,
      "learning_rate": 4.365195300095268e-05,
      "loss": 0.0346,
      "step": 8877
    },
    {
      "epoch": 16.900618164526865,
      "grad_norm": 0.07170283794403076,
      "learning_rate": 4.364560177834233e-05,
      "loss": 0.0243,
      "step": 8878
    },
    {
      "epoch": 16.902520209224917,
      "grad_norm": 0.14044728875160217,
      "learning_rate": 4.363925055573198e-05,
      "loss": 0.06,
      "step": 8879
    },
    {
      "epoch": 16.904422253922966,
      "grad_norm": 0.1995735913515091,
      "learning_rate": 4.363289933312163e-05,
      "loss": 0.0554,
      "step": 8880
    },
    {
      "epoch": 16.90632429862102,
      "grad_norm": 0.08931490033864975,
      "learning_rate": 4.3626548110511275e-05,
      "loss": 0.0509,
      "step": 8881
    },
    {
      "epoch": 16.90822634331907,
      "grad_norm": 0.17955714464187622,
      "learning_rate": 4.362019688790092e-05,
      "loss": 0.0846,
      "step": 8882
    },
    {
      "epoch": 16.910128388017117,
      "grad_norm": 0.17314022779464722,
      "learning_rate": 4.361384566529057e-05,
      "loss": 0.0589,
      "step": 8883
    },
    {
      "epoch": 16.91203043271517,
      "grad_norm": 0.172648623585701,
      "learning_rate": 4.360749444268022e-05,
      "loss": 0.0612,
      "step": 8884
    },
    {
      "epoch": 16.91393247741322,
      "grad_norm": 0.3400585353374481,
      "learning_rate": 4.360114322006987e-05,
      "loss": 0.0662,
      "step": 8885
    },
    {
      "epoch": 16.91583452211127,
      "grad_norm": 0.23420454561710358,
      "learning_rate": 4.3594791997459514e-05,
      "loss": 0.0595,
      "step": 8886
    },
    {
      "epoch": 16.91773656680932,
      "grad_norm": 0.12573030591011047,
      "learning_rate": 4.358844077484916e-05,
      "loss": 0.05,
      "step": 8887
    },
    {
      "epoch": 16.91963861150737,
      "grad_norm": 0.1461486518383026,
      "learning_rate": 4.358208955223881e-05,
      "loss": 0.0424,
      "step": 8888
    },
    {
      "epoch": 16.92154065620542,
      "grad_norm": 0.16064852476119995,
      "learning_rate": 4.3575738329628456e-05,
      "loss": 0.0652,
      "step": 8889
    },
    {
      "epoch": 16.923442700903472,
      "grad_norm": 0.060332030057907104,
      "learning_rate": 4.35693871070181e-05,
      "loss": 0.0545,
      "step": 8890
    },
    {
      "epoch": 16.92534474560152,
      "grad_norm": 0.23492804169654846,
      "learning_rate": 4.3563035884407746e-05,
      "loss": 0.0675,
      "step": 8891
    },
    {
      "epoch": 16.927246790299574,
      "grad_norm": 0.17212089896202087,
      "learning_rate": 4.35566846617974e-05,
      "loss": 0.0675,
      "step": 8892
    },
    {
      "epoch": 16.929148834997623,
      "grad_norm": 0.15113942325115204,
      "learning_rate": 4.355033343918705e-05,
      "loss": 0.0411,
      "step": 8893
    },
    {
      "epoch": 16.931050879695672,
      "grad_norm": 0.1071106493473053,
      "learning_rate": 4.3543982216576695e-05,
      "loss": 0.0544,
      "step": 8894
    },
    {
      "epoch": 16.932952924393724,
      "grad_norm": 0.11964888870716095,
      "learning_rate": 4.353763099396634e-05,
      "loss": 0.0593,
      "step": 8895
    },
    {
      "epoch": 16.934854969091774,
      "grad_norm": 0.12841713428497314,
      "learning_rate": 4.3531279771355985e-05,
      "loss": 0.0521,
      "step": 8896
    },
    {
      "epoch": 16.936757013789823,
      "grad_norm": 0.10380429774522781,
      "learning_rate": 4.352492854874564e-05,
      "loss": 0.0459,
      "step": 8897
    },
    {
      "epoch": 16.938659058487875,
      "grad_norm": 0.1602270007133484,
      "learning_rate": 4.351857732613528e-05,
      "loss": 0.0603,
      "step": 8898
    },
    {
      "epoch": 16.940561103185924,
      "grad_norm": 0.11509580910205841,
      "learning_rate": 4.351222610352493e-05,
      "loss": 0.0603,
      "step": 8899
    },
    {
      "epoch": 16.942463147883974,
      "grad_norm": 0.26956668496131897,
      "learning_rate": 4.350587488091458e-05,
      "loss": 0.0732,
      "step": 8900
    },
    {
      "epoch": 16.944365192582026,
      "grad_norm": 0.0985705703496933,
      "learning_rate": 4.3499523658304224e-05,
      "loss": 0.0743,
      "step": 8901
    },
    {
      "epoch": 16.946267237280075,
      "grad_norm": 0.08205585181713104,
      "learning_rate": 4.3493172435693876e-05,
      "loss": 0.0305,
      "step": 8902
    },
    {
      "epoch": 16.948169281978128,
      "grad_norm": 0.1018296554684639,
      "learning_rate": 4.348682121308352e-05,
      "loss": 0.0466,
      "step": 8903
    },
    {
      "epoch": 16.950071326676177,
      "grad_norm": 0.13760146498680115,
      "learning_rate": 4.3480469990473166e-05,
      "loss": 0.0462,
      "step": 8904
    },
    {
      "epoch": 16.951973371374226,
      "grad_norm": 0.11483436077833176,
      "learning_rate": 4.347411876786282e-05,
      "loss": 0.0583,
      "step": 8905
    },
    {
      "epoch": 16.95387541607228,
      "grad_norm": 0.18903884291648865,
      "learning_rate": 4.346776754525246e-05,
      "loss": 0.0436,
      "step": 8906
    },
    {
      "epoch": 16.955777460770328,
      "grad_norm": 0.10095207393169403,
      "learning_rate": 4.3461416322642115e-05,
      "loss": 0.0559,
      "step": 8907
    },
    {
      "epoch": 16.957679505468377,
      "grad_norm": 0.0754164457321167,
      "learning_rate": 4.345506510003175e-05,
      "loss": 0.0699,
      "step": 8908
    },
    {
      "epoch": 16.95958155016643,
      "grad_norm": 0.18949873745441437,
      "learning_rate": 4.3448713877421405e-05,
      "loss": 0.0498,
      "step": 8909
    },
    {
      "epoch": 16.96148359486448,
      "grad_norm": 0.16381996870040894,
      "learning_rate": 4.344236265481105e-05,
      "loss": 0.0825,
      "step": 8910
    },
    {
      "epoch": 16.96338563956253,
      "grad_norm": 0.23997418582439423,
      "learning_rate": 4.34360114322007e-05,
      "loss": 0.0599,
      "step": 8911
    },
    {
      "epoch": 16.96528768426058,
      "grad_norm": 0.07024307548999786,
      "learning_rate": 4.3429660209590353e-05,
      "loss": 0.0377,
      "step": 8912
    },
    {
      "epoch": 16.96718972895863,
      "grad_norm": 0.20647969841957092,
      "learning_rate": 4.342330898697999e-05,
      "loss": 0.051,
      "step": 8913
    },
    {
      "epoch": 16.969091773656682,
      "grad_norm": 0.10372774302959442,
      "learning_rate": 4.3416957764369644e-05,
      "loss": 0.0407,
      "step": 8914
    },
    {
      "epoch": 16.97099381835473,
      "grad_norm": 0.0911819338798523,
      "learning_rate": 4.341060654175929e-05,
      "loss": 0.0469,
      "step": 8915
    },
    {
      "epoch": 16.97289586305278,
      "grad_norm": 0.1432979255914688,
      "learning_rate": 4.340425531914894e-05,
      "loss": 0.0522,
      "step": 8916
    },
    {
      "epoch": 16.974797907750833,
      "grad_norm": 0.131869375705719,
      "learning_rate": 4.3397904096538586e-05,
      "loss": 0.0785,
      "step": 8917
    },
    {
      "epoch": 16.976699952448882,
      "grad_norm": 0.2788298428058624,
      "learning_rate": 4.339155287392823e-05,
      "loss": 0.051,
      "step": 8918
    },
    {
      "epoch": 16.97860199714693,
      "grad_norm": 0.16402453184127808,
      "learning_rate": 4.338520165131788e-05,
      "loss": 0.0806,
      "step": 8919
    },
    {
      "epoch": 16.980504041844984,
      "grad_norm": 0.1125316321849823,
      "learning_rate": 4.337885042870753e-05,
      "loss": 0.0777,
      "step": 8920
    },
    {
      "epoch": 16.982406086543033,
      "grad_norm": 0.09106596559286118,
      "learning_rate": 4.337249920609718e-05,
      "loss": 0.0425,
      "step": 8921
    },
    {
      "epoch": 16.984308131241086,
      "grad_norm": 0.13478001952171326,
      "learning_rate": 4.3366147983486824e-05,
      "loss": 0.0396,
      "step": 8922
    },
    {
      "epoch": 16.986210175939135,
      "grad_norm": 0.19370102882385254,
      "learning_rate": 4.335979676087647e-05,
      "loss": 0.0477,
      "step": 8923
    },
    {
      "epoch": 16.988112220637184,
      "grad_norm": 0.1331225037574768,
      "learning_rate": 4.335344553826612e-05,
      "loss": 0.0764,
      "step": 8924
    },
    {
      "epoch": 16.990014265335237,
      "grad_norm": 0.16695334017276764,
      "learning_rate": 4.3347094315655766e-05,
      "loss": 0.0376,
      "step": 8925
    },
    {
      "epoch": 16.991916310033286,
      "grad_norm": 0.13133040070533752,
      "learning_rate": 4.334074309304541e-05,
      "loss": 0.0462,
      "step": 8926
    },
    {
      "epoch": 16.993818354731335,
      "grad_norm": 0.09299250692129135,
      "learning_rate": 4.3334391870435056e-05,
      "loss": 0.0557,
      "step": 8927
    },
    {
      "epoch": 16.995720399429388,
      "grad_norm": 0.11600998789072037,
      "learning_rate": 4.332804064782471e-05,
      "loss": 0.0669,
      "step": 8928
    },
    {
      "epoch": 16.997622444127437,
      "grad_norm": 0.1207716166973114,
      "learning_rate": 4.332168942521436e-05,
      "loss": 0.0536,
      "step": 8929
    },
    {
      "epoch": 16.999524488825486,
      "grad_norm": 0.16815738379955292,
      "learning_rate": 4.3315338202604005e-05,
      "loss": 0.069,
      "step": 8930
    },
    {
      "epoch": 17.00142653352354,
      "grad_norm": 0.1072743833065033,
      "learning_rate": 4.330898697999365e-05,
      "loss": 0.0607,
      "step": 8931
    },
    {
      "epoch": 17.003328578221588,
      "grad_norm": 0.08723615109920502,
      "learning_rate": 4.3302635757383295e-05,
      "loss": 0.0505,
      "step": 8932
    },
    {
      "epoch": 17.00523062291964,
      "grad_norm": 0.22191157937049866,
      "learning_rate": 4.329628453477295e-05,
      "loss": 0.054,
      "step": 8933
    },
    {
      "epoch": 17.00713266761769,
      "grad_norm": 0.10282241553068161,
      "learning_rate": 4.328993331216259e-05,
      "loss": 0.0503,
      "step": 8934
    },
    {
      "epoch": 17.00903471231574,
      "grad_norm": 0.11215320974588394,
      "learning_rate": 4.328358208955224e-05,
      "loss": 0.057,
      "step": 8935
    },
    {
      "epoch": 17.01093675701379,
      "grad_norm": 0.09615226835012436,
      "learning_rate": 4.327723086694189e-05,
      "loss": 0.058,
      "step": 8936
    },
    {
      "epoch": 17.01283880171184,
      "grad_norm": 0.08828164637088776,
      "learning_rate": 4.3270879644331534e-05,
      "loss": 0.0526,
      "step": 8937
    },
    {
      "epoch": 17.01474084640989,
      "grad_norm": 0.0911029502749443,
      "learning_rate": 4.3264528421721186e-05,
      "loss": 0.0547,
      "step": 8938
    },
    {
      "epoch": 17.016642891107942,
      "grad_norm": 0.04437772557139397,
      "learning_rate": 4.325817719911083e-05,
      "loss": 0.0471,
      "step": 8939
    },
    {
      "epoch": 17.01854493580599,
      "grad_norm": 0.11651269346475601,
      "learning_rate": 4.3251825976500476e-05,
      "loss": 0.052,
      "step": 8940
    },
    {
      "epoch": 17.02044698050404,
      "grad_norm": 0.16090331971645355,
      "learning_rate": 4.324547475389013e-05,
      "loss": 0.0876,
      "step": 8941
    },
    {
      "epoch": 17.022349025202093,
      "grad_norm": 0.0497426874935627,
      "learning_rate": 4.323912353127977e-05,
      "loss": 0.0602,
      "step": 8942
    },
    {
      "epoch": 17.024251069900142,
      "grad_norm": 0.15628616511821747,
      "learning_rate": 4.3232772308669425e-05,
      "loss": 0.0632,
      "step": 8943
    },
    {
      "epoch": 17.026153114598195,
      "grad_norm": 0.08810368925333023,
      "learning_rate": 4.322642108605906e-05,
      "loss": 0.0391,
      "step": 8944
    },
    {
      "epoch": 17.028055159296244,
      "grad_norm": 0.21196871995925903,
      "learning_rate": 4.3220069863448715e-05,
      "loss": 0.0584,
      "step": 8945
    },
    {
      "epoch": 17.029957203994293,
      "grad_norm": 0.05883145332336426,
      "learning_rate": 4.321371864083836e-05,
      "loss": 0.0464,
      "step": 8946
    },
    {
      "epoch": 17.031859248692346,
      "grad_norm": 0.18730250000953674,
      "learning_rate": 4.320736741822801e-05,
      "loss": 0.062,
      "step": 8947
    },
    {
      "epoch": 17.033761293390395,
      "grad_norm": 0.16714142262935638,
      "learning_rate": 4.3201016195617664e-05,
      "loss": 0.0442,
      "step": 8948
    },
    {
      "epoch": 17.035663338088444,
      "grad_norm": 0.114844910800457,
      "learning_rate": 4.31946649730073e-05,
      "loss": 0.0503,
      "step": 8949
    },
    {
      "epoch": 17.037565382786497,
      "grad_norm": 0.25291353464126587,
      "learning_rate": 4.3188313750396954e-05,
      "loss": 0.0736,
      "step": 8950
    },
    {
      "epoch": 17.039467427484546,
      "grad_norm": 0.1454509198665619,
      "learning_rate": 4.31819625277866e-05,
      "loss": 0.0613,
      "step": 8951
    },
    {
      "epoch": 17.041369472182595,
      "grad_norm": 0.10287206619977951,
      "learning_rate": 4.317561130517625e-05,
      "loss": 0.0581,
      "step": 8952
    },
    {
      "epoch": 17.043271516880647,
      "grad_norm": 0.12564142048358917,
      "learning_rate": 4.3169260082565896e-05,
      "loss": 0.05,
      "step": 8953
    },
    {
      "epoch": 17.045173561578697,
      "grad_norm": 0.070656418800354,
      "learning_rate": 4.316290885995554e-05,
      "loss": 0.0506,
      "step": 8954
    },
    {
      "epoch": 17.04707560627675,
      "grad_norm": 0.06617604941129684,
      "learning_rate": 4.315655763734519e-05,
      "loss": 0.0458,
      "step": 8955
    },
    {
      "epoch": 17.0489776509748,
      "grad_norm": 0.14419837296009064,
      "learning_rate": 4.315020641473484e-05,
      "loss": 0.0745,
      "step": 8956
    },
    {
      "epoch": 17.050879695672847,
      "grad_norm": 0.18425264954566956,
      "learning_rate": 4.314385519212449e-05,
      "loss": 0.0544,
      "step": 8957
    },
    {
      "epoch": 17.0527817403709,
      "grad_norm": 0.12209388613700867,
      "learning_rate": 4.313750396951413e-05,
      "loss": 0.065,
      "step": 8958
    },
    {
      "epoch": 17.05468378506895,
      "grad_norm": 0.2421053647994995,
      "learning_rate": 4.313115274690378e-05,
      "loss": 0.0572,
      "step": 8959
    },
    {
      "epoch": 17.056585829767,
      "grad_norm": 0.14239047467708588,
      "learning_rate": 4.312480152429343e-05,
      "loss": 0.059,
      "step": 8960
    },
    {
      "epoch": 17.05848787446505,
      "grad_norm": 0.20295925438404083,
      "learning_rate": 4.311845030168308e-05,
      "loss": 0.0644,
      "step": 8961
    },
    {
      "epoch": 17.0603899191631,
      "grad_norm": 0.08817723393440247,
      "learning_rate": 4.311209907907272e-05,
      "loss": 0.0383,
      "step": 8962
    },
    {
      "epoch": 17.06229196386115,
      "grad_norm": 0.14773733913898468,
      "learning_rate": 4.310574785646237e-05,
      "loss": 0.0546,
      "step": 8963
    },
    {
      "epoch": 17.064194008559202,
      "grad_norm": 0.11199629306793213,
      "learning_rate": 4.309939663385202e-05,
      "loss": 0.0372,
      "step": 8964
    },
    {
      "epoch": 17.06609605325725,
      "grad_norm": 0.10948487371206284,
      "learning_rate": 4.3093045411241664e-05,
      "loss": 0.067,
      "step": 8965
    },
    {
      "epoch": 17.067998097955304,
      "grad_norm": 0.1963731348514557,
      "learning_rate": 4.3086694188631316e-05,
      "loss": 0.05,
      "step": 8966
    },
    {
      "epoch": 17.069900142653353,
      "grad_norm": 0.1474691927433014,
      "learning_rate": 4.308034296602096e-05,
      "loss": 0.0403,
      "step": 8967
    },
    {
      "epoch": 17.071802187351402,
      "grad_norm": 0.20371627807617188,
      "learning_rate": 4.3073991743410606e-05,
      "loss": 0.0659,
      "step": 8968
    },
    {
      "epoch": 17.073704232049455,
      "grad_norm": 0.21123164892196655,
      "learning_rate": 4.306764052080026e-05,
      "loss": 0.0592,
      "step": 8969
    },
    {
      "epoch": 17.075606276747504,
      "grad_norm": 0.14474251866340637,
      "learning_rate": 4.30612892981899e-05,
      "loss": 0.0469,
      "step": 8970
    },
    {
      "epoch": 17.077508321445553,
      "grad_norm": 0.1162516176700592,
      "learning_rate": 4.305493807557955e-05,
      "loss": 0.052,
      "step": 8971
    },
    {
      "epoch": 17.079410366143605,
      "grad_norm": 0.09336848556995392,
      "learning_rate": 4.30485868529692e-05,
      "loss": 0.0456,
      "step": 8972
    },
    {
      "epoch": 17.081312410841655,
      "grad_norm": 0.09849970042705536,
      "learning_rate": 4.3042235630358845e-05,
      "loss": 0.0837,
      "step": 8973
    },
    {
      "epoch": 17.083214455539704,
      "grad_norm": 0.0521208755671978,
      "learning_rate": 4.3035884407748497e-05,
      "loss": 0.0376,
      "step": 8974
    },
    {
      "epoch": 17.085116500237756,
      "grad_norm": 0.14380308985710144,
      "learning_rate": 4.302953318513814e-05,
      "loss": 0.0517,
      "step": 8975
    },
    {
      "epoch": 17.087018544935805,
      "grad_norm": 0.06576641649007797,
      "learning_rate": 4.302318196252779e-05,
      "loss": 0.0569,
      "step": 8976
    },
    {
      "epoch": 17.088920589633858,
      "grad_norm": 0.05938639119267464,
      "learning_rate": 4.301683073991744e-05,
      "loss": 0.06,
      "step": 8977
    },
    {
      "epoch": 17.090822634331907,
      "grad_norm": 0.13843615353107452,
      "learning_rate": 4.3010479517307084e-05,
      "loss": 0.0549,
      "step": 8978
    },
    {
      "epoch": 17.092724679029956,
      "grad_norm": 0.2245621383190155,
      "learning_rate": 4.3004128294696735e-05,
      "loss": 0.0582,
      "step": 8979
    },
    {
      "epoch": 17.09462672372801,
      "grad_norm": 0.08693769574165344,
      "learning_rate": 4.2997777072086374e-05,
      "loss": 0.0704,
      "step": 8980
    },
    {
      "epoch": 17.096528768426058,
      "grad_norm": 0.13792353868484497,
      "learning_rate": 4.2991425849476026e-05,
      "loss": 0.0631,
      "step": 8981
    },
    {
      "epoch": 17.098430813124107,
      "grad_norm": 0.13329333066940308,
      "learning_rate": 4.298507462686567e-05,
      "loss": 0.0518,
      "step": 8982
    },
    {
      "epoch": 17.10033285782216,
      "grad_norm": 0.10913736373186111,
      "learning_rate": 4.297872340425532e-05,
      "loss": 0.0565,
      "step": 8983
    },
    {
      "epoch": 17.10223490252021,
      "grad_norm": 0.1331338733434677,
      "learning_rate": 4.2972372181644974e-05,
      "loss": 0.0566,
      "step": 8984
    },
    {
      "epoch": 17.104136947218258,
      "grad_norm": 0.10640410333871841,
      "learning_rate": 4.296602095903461e-05,
      "loss": 0.0434,
      "step": 8985
    },
    {
      "epoch": 17.10603899191631,
      "grad_norm": 0.11068563908338547,
      "learning_rate": 4.2959669736424264e-05,
      "loss": 0.0588,
      "step": 8986
    },
    {
      "epoch": 17.10794103661436,
      "grad_norm": 0.06091459468007088,
      "learning_rate": 4.295331851381391e-05,
      "loss": 0.0449,
      "step": 8987
    },
    {
      "epoch": 17.109843081312412,
      "grad_norm": 0.22359995543956757,
      "learning_rate": 4.294696729120356e-05,
      "loss": 0.0743,
      "step": 8988
    },
    {
      "epoch": 17.11174512601046,
      "grad_norm": 0.1244855746626854,
      "learning_rate": 4.2940616068593206e-05,
      "loss": 0.0628,
      "step": 8989
    },
    {
      "epoch": 17.11364717070851,
      "grad_norm": 0.27003106474876404,
      "learning_rate": 4.293426484598285e-05,
      "loss": 0.065,
      "step": 8990
    },
    {
      "epoch": 17.115549215406563,
      "grad_norm": 0.08874871581792831,
      "learning_rate": 4.29279136233725e-05,
      "loss": 0.0487,
      "step": 8991
    },
    {
      "epoch": 17.117451260104612,
      "grad_norm": 0.07236160337924957,
      "learning_rate": 4.292156240076215e-05,
      "loss": 0.0586,
      "step": 8992
    },
    {
      "epoch": 17.11935330480266,
      "grad_norm": 0.1268148273229599,
      "learning_rate": 4.29152111781518e-05,
      "loss": 0.0529,
      "step": 8993
    },
    {
      "epoch": 17.121255349500714,
      "grad_norm": 0.11032159626483917,
      "learning_rate": 4.290885995554144e-05,
      "loss": 0.0476,
      "step": 8994
    },
    {
      "epoch": 17.123157394198763,
      "grad_norm": 0.16266649961471558,
      "learning_rate": 4.290250873293109e-05,
      "loss": 0.049,
      "step": 8995
    },
    {
      "epoch": 17.125059438896812,
      "grad_norm": 0.06318007409572601,
      "learning_rate": 4.289615751032074e-05,
      "loss": 0.0723,
      "step": 8996
    },
    {
      "epoch": 17.126961483594865,
      "grad_norm": 0.06825682520866394,
      "learning_rate": 4.288980628771039e-05,
      "loss": 0.0415,
      "step": 8997
    },
    {
      "epoch": 17.128863528292914,
      "grad_norm": 0.22206497192382812,
      "learning_rate": 4.288345506510003e-05,
      "loss": 0.0565,
      "step": 8998
    },
    {
      "epoch": 17.130765572990967,
      "grad_norm": 0.1304975301027298,
      "learning_rate": 4.287710384248968e-05,
      "loss": 0.0485,
      "step": 8999
    },
    {
      "epoch": 17.132667617689016,
      "grad_norm": 0.13121330738067627,
      "learning_rate": 4.287075261987933e-05,
      "loss": 0.0577,
      "step": 9000
    },
    {
      "epoch": 17.134569662387065,
      "grad_norm": 0.1600920408964157,
      "learning_rate": 4.2864401397268974e-05,
      "loss": 0.0535,
      "step": 9001
    },
    {
      "epoch": 17.136471707085118,
      "grad_norm": 0.10079571604728699,
      "learning_rate": 4.2858050174658626e-05,
      "loss": 0.0472,
      "step": 9002
    },
    {
      "epoch": 17.138373751783167,
      "grad_norm": 0.18356472253799438,
      "learning_rate": 4.285169895204827e-05,
      "loss": 0.0735,
      "step": 9003
    },
    {
      "epoch": 17.140275796481216,
      "grad_norm": 0.0878254771232605,
      "learning_rate": 4.2845347729437916e-05,
      "loss": 0.0551,
      "step": 9004
    },
    {
      "epoch": 17.14217784117927,
      "grad_norm": 0.12589243054389954,
      "learning_rate": 4.283899650682757e-05,
      "loss": 0.0628,
      "step": 9005
    },
    {
      "epoch": 17.144079885877318,
      "grad_norm": 0.08670032769441605,
      "learning_rate": 4.283264528421721e-05,
      "loss": 0.0425,
      "step": 9006
    },
    {
      "epoch": 17.145981930575367,
      "grad_norm": 0.22173863649368286,
      "learning_rate": 4.282629406160686e-05,
      "loss": 0.0515,
      "step": 9007
    },
    {
      "epoch": 17.14788397527342,
      "grad_norm": 0.06839998066425323,
      "learning_rate": 4.281994283899651e-05,
      "loss": 0.0623,
      "step": 9008
    },
    {
      "epoch": 17.14978601997147,
      "grad_norm": 0.20462091267108917,
      "learning_rate": 4.2813591616386155e-05,
      "loss": 0.0565,
      "step": 9009
    },
    {
      "epoch": 17.15168806466952,
      "grad_norm": 0.1404951810836792,
      "learning_rate": 4.280724039377581e-05,
      "loss": 0.0629,
      "step": 9010
    },
    {
      "epoch": 17.15359010936757,
      "grad_norm": 0.0752057284116745,
      "learning_rate": 4.280088917116545e-05,
      "loss": 0.0615,
      "step": 9011
    },
    {
      "epoch": 17.15549215406562,
      "grad_norm": 0.09520772099494934,
      "learning_rate": 4.27945379485551e-05,
      "loss": 0.0415,
      "step": 9012
    },
    {
      "epoch": 17.157394198763672,
      "grad_norm": 0.06295536458492279,
      "learning_rate": 4.278818672594474e-05,
      "loss": 0.033,
      "step": 9013
    },
    {
      "epoch": 17.15929624346172,
      "grad_norm": 0.07264743000268936,
      "learning_rate": 4.2781835503334394e-05,
      "loss": 0.0495,
      "step": 9014
    },
    {
      "epoch": 17.16119828815977,
      "grad_norm": 0.1129545047879219,
      "learning_rate": 4.2775484280724046e-05,
      "loss": 0.0591,
      "step": 9015
    },
    {
      "epoch": 17.163100332857823,
      "grad_norm": 0.16535106301307678,
      "learning_rate": 4.2769133058113684e-05,
      "loss": 0.0649,
      "step": 9016
    },
    {
      "epoch": 17.165002377555872,
      "grad_norm": 0.05102641135454178,
      "learning_rate": 4.2762781835503336e-05,
      "loss": 0.0437,
      "step": 9017
    },
    {
      "epoch": 17.16690442225392,
      "grad_norm": 0.08517751842737198,
      "learning_rate": 4.275643061289298e-05,
      "loss": 0.0425,
      "step": 9018
    },
    {
      "epoch": 17.168806466951974,
      "grad_norm": 0.12171544134616852,
      "learning_rate": 4.275007939028263e-05,
      "loss": 0.0652,
      "step": 9019
    },
    {
      "epoch": 17.170708511650023,
      "grad_norm": 0.10095885396003723,
      "learning_rate": 4.274372816767228e-05,
      "loss": 0.0673,
      "step": 9020
    },
    {
      "epoch": 17.172610556348076,
      "grad_norm": 0.14293339848518372,
      "learning_rate": 4.273737694506192e-05,
      "loss": 0.0549,
      "step": 9021
    },
    {
      "epoch": 17.174512601046125,
      "grad_norm": 0.07936553657054901,
      "learning_rate": 4.2731025722451575e-05,
      "loss": 0.0509,
      "step": 9022
    },
    {
      "epoch": 17.176414645744174,
      "grad_norm": 0.06809721887111664,
      "learning_rate": 4.272467449984122e-05,
      "loss": 0.0609,
      "step": 9023
    },
    {
      "epoch": 17.178316690442227,
      "grad_norm": 0.09068653732538223,
      "learning_rate": 4.271832327723087e-05,
      "loss": 0.0588,
      "step": 9024
    },
    {
      "epoch": 17.180218735140276,
      "grad_norm": 0.13408097624778748,
      "learning_rate": 4.271197205462052e-05,
      "loss": 0.0421,
      "step": 9025
    },
    {
      "epoch": 17.182120779838325,
      "grad_norm": 0.13926245272159576,
      "learning_rate": 4.270562083201016e-05,
      "loss": 0.076,
      "step": 9026
    },
    {
      "epoch": 17.184022824536378,
      "grad_norm": 0.08979006856679916,
      "learning_rate": 4.2699269609399814e-05,
      "loss": 0.0573,
      "step": 9027
    },
    {
      "epoch": 17.185924869234427,
      "grad_norm": 0.09927933663129807,
      "learning_rate": 4.269291838678946e-05,
      "loss": 0.0415,
      "step": 9028
    },
    {
      "epoch": 17.187826913932476,
      "grad_norm": 0.05623723194003105,
      "learning_rate": 4.268656716417911e-05,
      "loss": 0.0501,
      "step": 9029
    },
    {
      "epoch": 17.18972895863053,
      "grad_norm": 0.10407004505395889,
      "learning_rate": 4.268021594156875e-05,
      "loss": 0.0669,
      "step": 9030
    },
    {
      "epoch": 17.191631003328578,
      "grad_norm": 0.11468394100666046,
      "learning_rate": 4.26738647189584e-05,
      "loss": 0.0453,
      "step": 9031
    },
    {
      "epoch": 17.19353304802663,
      "grad_norm": 0.052839286625385284,
      "learning_rate": 4.266751349634805e-05,
      "loss": 0.0579,
      "step": 9032
    },
    {
      "epoch": 17.19543509272468,
      "grad_norm": 0.16107793152332306,
      "learning_rate": 4.26611622737377e-05,
      "loss": 0.0561,
      "step": 9033
    },
    {
      "epoch": 17.19733713742273,
      "grad_norm": 0.0660393089056015,
      "learning_rate": 4.265481105112734e-05,
      "loss": 0.0532,
      "step": 9034
    },
    {
      "epoch": 17.19923918212078,
      "grad_norm": 0.11889426410198212,
      "learning_rate": 4.264845982851699e-05,
      "loss": 0.0897,
      "step": 9035
    },
    {
      "epoch": 17.20114122681883,
      "grad_norm": 0.04962438344955444,
      "learning_rate": 4.264210860590664e-05,
      "loss": 0.0757,
      "step": 9036
    },
    {
      "epoch": 17.20304327151688,
      "grad_norm": 0.07310406118631363,
      "learning_rate": 4.2635757383296285e-05,
      "loss": 0.0494,
      "step": 9037
    },
    {
      "epoch": 17.204945316214932,
      "grad_norm": 0.16161684691905975,
      "learning_rate": 4.2629406160685937e-05,
      "loss": 0.0645,
      "step": 9038
    },
    {
      "epoch": 17.20684736091298,
      "grad_norm": 0.0898459404706955,
      "learning_rate": 4.262305493807558e-05,
      "loss": 0.0579,
      "step": 9039
    },
    {
      "epoch": 17.20874940561103,
      "grad_norm": 0.060915347188711166,
      "learning_rate": 4.261670371546523e-05,
      "loss": 0.0446,
      "step": 9040
    },
    {
      "epoch": 17.210651450309083,
      "grad_norm": 0.1457240730524063,
      "learning_rate": 4.261035249285488e-05,
      "loss": 0.0509,
      "step": 9041
    },
    {
      "epoch": 17.212553495007132,
      "grad_norm": 0.05205492302775383,
      "learning_rate": 4.2604001270244524e-05,
      "loss": 0.0309,
      "step": 9042
    },
    {
      "epoch": 17.214455539705185,
      "grad_norm": 0.1869315356016159,
      "learning_rate": 4.259765004763417e-05,
      "loss": 0.0455,
      "step": 9043
    },
    {
      "epoch": 17.216357584403234,
      "grad_norm": 0.19125977158546448,
      "learning_rate": 4.259129882502382e-05,
      "loss": 0.0581,
      "step": 9044
    },
    {
      "epoch": 17.218259629101283,
      "grad_norm": 0.11834566295146942,
      "learning_rate": 4.2584947602413466e-05,
      "loss": 0.1067,
      "step": 9045
    },
    {
      "epoch": 17.220161673799335,
      "grad_norm": 0.07625547796487808,
      "learning_rate": 4.257859637980312e-05,
      "loss": 0.0427,
      "step": 9046
    },
    {
      "epoch": 17.222063718497385,
      "grad_norm": 0.08554868400096893,
      "learning_rate": 4.257224515719276e-05,
      "loss": 0.0529,
      "step": 9047
    },
    {
      "epoch": 17.223965763195434,
      "grad_norm": 0.19139155745506287,
      "learning_rate": 4.256589393458241e-05,
      "loss": 0.0744,
      "step": 9048
    },
    {
      "epoch": 17.225867807893486,
      "grad_norm": 0.13853976130485535,
      "learning_rate": 4.255954271197205e-05,
      "loss": 0.0729,
      "step": 9049
    },
    {
      "epoch": 17.227769852591535,
      "grad_norm": 0.1726534515619278,
      "learning_rate": 4.2553191489361704e-05,
      "loss": 0.0614,
      "step": 9050
    },
    {
      "epoch": 17.229671897289585,
      "grad_norm": 0.10510444641113281,
      "learning_rate": 4.2546840266751356e-05,
      "loss": 0.0501,
      "step": 9051
    },
    {
      "epoch": 17.231573941987637,
      "grad_norm": 0.05099013075232506,
      "learning_rate": 4.2540489044140995e-05,
      "loss": 0.0471,
      "step": 9052
    },
    {
      "epoch": 17.233475986685686,
      "grad_norm": 0.17027002573013306,
      "learning_rate": 4.2534137821530646e-05,
      "loss": 0.0463,
      "step": 9053
    },
    {
      "epoch": 17.23537803138374,
      "grad_norm": 0.08581602573394775,
      "learning_rate": 4.252778659892029e-05,
      "loss": 0.0543,
      "step": 9054
    },
    {
      "epoch": 17.237280076081788,
      "grad_norm": 0.1270342320203781,
      "learning_rate": 4.252143537630994e-05,
      "loss": 0.0445,
      "step": 9055
    },
    {
      "epoch": 17.239182120779837,
      "grad_norm": 0.1030355840921402,
      "learning_rate": 4.251508415369959e-05,
      "loss": 0.0526,
      "step": 9056
    },
    {
      "epoch": 17.24108416547789,
      "grad_norm": 0.18514138460159302,
      "learning_rate": 4.2508732931089233e-05,
      "loss": 0.0531,
      "step": 9057
    },
    {
      "epoch": 17.24298621017594,
      "grad_norm": 0.18377025425434113,
      "learning_rate": 4.2502381708478885e-05,
      "loss": 0.0819,
      "step": 9058
    },
    {
      "epoch": 17.244888254873988,
      "grad_norm": 0.18528705835342407,
      "learning_rate": 4.249603048586853e-05,
      "loss": 0.0575,
      "step": 9059
    },
    {
      "epoch": 17.24679029957204,
      "grad_norm": 0.3869665563106537,
      "learning_rate": 4.248967926325818e-05,
      "loss": 0.0807,
      "step": 9060
    },
    {
      "epoch": 17.24869234427009,
      "grad_norm": 0.2586633861064911,
      "learning_rate": 4.248332804064783e-05,
      "loss": 0.0763,
      "step": 9061
    },
    {
      "epoch": 17.25059438896814,
      "grad_norm": 0.2239067256450653,
      "learning_rate": 4.247697681803747e-05,
      "loss": 0.0719,
      "step": 9062
    },
    {
      "epoch": 17.25249643366619,
      "grad_norm": 0.1460609883069992,
      "learning_rate": 4.2470625595427124e-05,
      "loss": 0.0497,
      "step": 9063
    },
    {
      "epoch": 17.25439847836424,
      "grad_norm": 0.13906528055667877,
      "learning_rate": 4.246427437281677e-05,
      "loss": 0.0651,
      "step": 9064
    },
    {
      "epoch": 17.256300523062293,
      "grad_norm": 0.10617247968912125,
      "learning_rate": 4.245792315020642e-05,
      "loss": 0.0621,
      "step": 9065
    },
    {
      "epoch": 17.258202567760343,
      "grad_norm": 0.05661832168698311,
      "learning_rate": 4.245157192759606e-05,
      "loss": 0.061,
      "step": 9066
    },
    {
      "epoch": 17.26010461245839,
      "grad_norm": 0.2621058523654938,
      "learning_rate": 4.244522070498571e-05,
      "loss": 0.0571,
      "step": 9067
    },
    {
      "epoch": 17.262006657156444,
      "grad_norm": 0.07775229215621948,
      "learning_rate": 4.2438869482375356e-05,
      "loss": 0.039,
      "step": 9068
    },
    {
      "epoch": 17.263908701854493,
      "grad_norm": 0.07392751425504684,
      "learning_rate": 4.243251825976501e-05,
      "loss": 0.0761,
      "step": 9069
    },
    {
      "epoch": 17.265810746552543,
      "grad_norm": 0.1043642908334732,
      "learning_rate": 4.242616703715465e-05,
      "loss": 0.052,
      "step": 9070
    },
    {
      "epoch": 17.267712791250595,
      "grad_norm": 0.2676692306995392,
      "learning_rate": 4.24198158145443e-05,
      "loss": 0.0587,
      "step": 9071
    },
    {
      "epoch": 17.269614835948644,
      "grad_norm": 0.2079627513885498,
      "learning_rate": 4.241346459193395e-05,
      "loss": 0.0574,
      "step": 9072
    },
    {
      "epoch": 17.271516880646693,
      "grad_norm": 0.06418656557798386,
      "learning_rate": 4.2407113369323595e-05,
      "loss": 0.0541,
      "step": 9073
    },
    {
      "epoch": 17.273418925344746,
      "grad_norm": 0.09768824279308319,
      "learning_rate": 4.240076214671325e-05,
      "loss": 0.0499,
      "step": 9074
    },
    {
      "epoch": 17.275320970042795,
      "grad_norm": 0.04967249184846878,
      "learning_rate": 4.239441092410289e-05,
      "loss": 0.0345,
      "step": 9075
    },
    {
      "epoch": 17.277223014740848,
      "grad_norm": 0.2924100160598755,
      "learning_rate": 4.238805970149254e-05,
      "loss": 0.0736,
      "step": 9076
    },
    {
      "epoch": 17.279125059438897,
      "grad_norm": 0.10819932073354721,
      "learning_rate": 4.238170847888219e-05,
      "loss": 0.0488,
      "step": 9077
    },
    {
      "epoch": 17.281027104136946,
      "grad_norm": 0.3450181186199188,
      "learning_rate": 4.2375357256271834e-05,
      "loss": 0.0732,
      "step": 9078
    },
    {
      "epoch": 17.282929148835,
      "grad_norm": 0.14250418543815613,
      "learning_rate": 4.236900603366148e-05,
      "loss": 0.048,
      "step": 9079
    },
    {
      "epoch": 17.284831193533048,
      "grad_norm": 0.08190340548753738,
      "learning_rate": 4.236265481105113e-05,
      "loss": 0.0354,
      "step": 9080
    },
    {
      "epoch": 17.286733238231097,
      "grad_norm": 0.11507826298475266,
      "learning_rate": 4.2356303588440776e-05,
      "loss": 0.0638,
      "step": 9081
    },
    {
      "epoch": 17.28863528292915,
      "grad_norm": 0.11006578058004379,
      "learning_rate": 4.234995236583043e-05,
      "loss": 0.0569,
      "step": 9082
    },
    {
      "epoch": 17.2905373276272,
      "grad_norm": 0.09880606085062027,
      "learning_rate": 4.234360114322007e-05,
      "loss": 0.0535,
      "step": 9083
    },
    {
      "epoch": 17.292439372325248,
      "grad_norm": 0.13434161245822906,
      "learning_rate": 4.233724992060972e-05,
      "loss": 0.0501,
      "step": 9084
    },
    {
      "epoch": 17.2943414170233,
      "grad_norm": 0.08274086564779282,
      "learning_rate": 4.233089869799936e-05,
      "loss": 0.0626,
      "step": 9085
    },
    {
      "epoch": 17.29624346172135,
      "grad_norm": 0.11226063221693039,
      "learning_rate": 4.2324547475389015e-05,
      "loss": 0.0591,
      "step": 9086
    },
    {
      "epoch": 17.298145506419402,
      "grad_norm": 0.14939825236797333,
      "learning_rate": 4.231819625277867e-05,
      "loss": 0.0635,
      "step": 9087
    },
    {
      "epoch": 17.30004755111745,
      "grad_norm": 0.18860632181167603,
      "learning_rate": 4.2311845030168305e-05,
      "loss": 0.0735,
      "step": 9088
    },
    {
      "epoch": 17.3019495958155,
      "grad_norm": 0.05516693368554115,
      "learning_rate": 4.230549380755796e-05,
      "loss": 0.0582,
      "step": 9089
    },
    {
      "epoch": 17.303851640513553,
      "grad_norm": 0.16363638639450073,
      "learning_rate": 4.22991425849476e-05,
      "loss": 0.0537,
      "step": 9090
    },
    {
      "epoch": 17.305753685211602,
      "grad_norm": 0.11976895481348038,
      "learning_rate": 4.2292791362337254e-05,
      "loss": 0.0503,
      "step": 9091
    },
    {
      "epoch": 17.30765572990965,
      "grad_norm": 0.13421820104122162,
      "learning_rate": 4.22864401397269e-05,
      "loss": 0.0905,
      "step": 9092
    },
    {
      "epoch": 17.309557774607704,
      "grad_norm": 0.18360990285873413,
      "learning_rate": 4.2280088917116544e-05,
      "loss": 0.0594,
      "step": 9093
    },
    {
      "epoch": 17.311459819305753,
      "grad_norm": 0.09387153387069702,
      "learning_rate": 4.2273737694506196e-05,
      "loss": 0.0587,
      "step": 9094
    },
    {
      "epoch": 17.313361864003806,
      "grad_norm": 0.060100257396698,
      "learning_rate": 4.226738647189584e-05,
      "loss": 0.0604,
      "step": 9095
    },
    {
      "epoch": 17.315263908701855,
      "grad_norm": 0.16482652723789215,
      "learning_rate": 4.226103524928549e-05,
      "loss": 0.0586,
      "step": 9096
    },
    {
      "epoch": 17.317165953399904,
      "grad_norm": 0.16404573619365692,
      "learning_rate": 4.225468402667513e-05,
      "loss": 0.0553,
      "step": 9097
    },
    {
      "epoch": 17.319067998097957,
      "grad_norm": 0.1388007551431656,
      "learning_rate": 4.224833280406478e-05,
      "loss": 0.0633,
      "step": 9098
    },
    {
      "epoch": 17.320970042796006,
      "grad_norm": 0.04446406289935112,
      "learning_rate": 4.2241981581454435e-05,
      "loss": 0.0445,
      "step": 9099
    },
    {
      "epoch": 17.322872087494055,
      "grad_norm": 0.06030283495783806,
      "learning_rate": 4.223563035884408e-05,
      "loss": 0.0517,
      "step": 9100
    },
    {
      "epoch": 17.324774132192108,
      "grad_norm": 0.1730167716741562,
      "learning_rate": 4.222927913623373e-05,
      "loss": 0.0596,
      "step": 9101
    },
    {
      "epoch": 17.326676176890157,
      "grad_norm": 0.13487887382507324,
      "learning_rate": 4.222292791362337e-05,
      "loss": 0.0545,
      "step": 9102
    },
    {
      "epoch": 17.328578221588206,
      "grad_norm": 0.05873151496052742,
      "learning_rate": 4.221657669101302e-05,
      "loss": 0.0553,
      "step": 9103
    },
    {
      "epoch": 17.33048026628626,
      "grad_norm": 0.1045122891664505,
      "learning_rate": 4.221022546840267e-05,
      "loss": 0.0678,
      "step": 9104
    },
    {
      "epoch": 17.332382310984308,
      "grad_norm": 0.06493869423866272,
      "learning_rate": 4.220387424579232e-05,
      "loss": 0.0556,
      "step": 9105
    },
    {
      "epoch": 17.33428435568236,
      "grad_norm": 0.09537652879953384,
      "learning_rate": 4.2197523023181964e-05,
      "loss": 0.0442,
      "step": 9106
    },
    {
      "epoch": 17.33618640038041,
      "grad_norm": 0.15815818309783936,
      "learning_rate": 4.219117180057161e-05,
      "loss": 0.0654,
      "step": 9107
    },
    {
      "epoch": 17.33808844507846,
      "grad_norm": 0.11845643073320389,
      "learning_rate": 4.218482057796126e-05,
      "loss": 0.0905,
      "step": 9108
    },
    {
      "epoch": 17.33999048977651,
      "grad_norm": 0.06879471242427826,
      "learning_rate": 4.2178469355350906e-05,
      "loss": 0.0449,
      "step": 9109
    },
    {
      "epoch": 17.34189253447456,
      "grad_norm": 0.05066360533237457,
      "learning_rate": 4.217211813274056e-05,
      "loss": 0.0496,
      "step": 9110
    },
    {
      "epoch": 17.34379457917261,
      "grad_norm": 0.041844941675662994,
      "learning_rate": 4.21657669101302e-05,
      "loss": 0.0498,
      "step": 9111
    },
    {
      "epoch": 17.345696623870662,
      "grad_norm": 0.19003881514072418,
      "learning_rate": 4.215941568751985e-05,
      "loss": 0.0543,
      "step": 9112
    },
    {
      "epoch": 17.34759866856871,
      "grad_norm": 0.09213196486234665,
      "learning_rate": 4.21530644649095e-05,
      "loss": 0.0466,
      "step": 9113
    },
    {
      "epoch": 17.34950071326676,
      "grad_norm": 0.14282000064849854,
      "learning_rate": 4.2146713242299144e-05,
      "loss": 0.0569,
      "step": 9114
    },
    {
      "epoch": 17.351402757964813,
      "grad_norm": 0.05803923308849335,
      "learning_rate": 4.214036201968879e-05,
      "loss": 0.037,
      "step": 9115
    },
    {
      "epoch": 17.353304802662862,
      "grad_norm": 0.07661207020282745,
      "learning_rate": 4.213401079707844e-05,
      "loss": 0.0388,
      "step": 9116
    },
    {
      "epoch": 17.355206847360915,
      "grad_norm": 0.05535397306084633,
      "learning_rate": 4.2127659574468086e-05,
      "loss": 0.0336,
      "step": 9117
    },
    {
      "epoch": 17.357108892058964,
      "grad_norm": 0.13280822336673737,
      "learning_rate": 4.212130835185774e-05,
      "loss": 0.0639,
      "step": 9118
    },
    {
      "epoch": 17.359010936757013,
      "grad_norm": 0.12745091319084167,
      "learning_rate": 4.211495712924738e-05,
      "loss": 0.0723,
      "step": 9119
    },
    {
      "epoch": 17.360912981455066,
      "grad_norm": 0.1197303980588913,
      "learning_rate": 4.210860590663703e-05,
      "loss": 0.0559,
      "step": 9120
    },
    {
      "epoch": 17.362815026153115,
      "grad_norm": 0.24809758365154266,
      "learning_rate": 4.2102254684026673e-05,
      "loss": 0.0603,
      "step": 9121
    },
    {
      "epoch": 17.364717070851164,
      "grad_norm": 0.10350430756807327,
      "learning_rate": 4.2095903461416325e-05,
      "loss": 0.0623,
      "step": 9122
    },
    {
      "epoch": 17.366619115549216,
      "grad_norm": 0.1336364895105362,
      "learning_rate": 4.208955223880597e-05,
      "loss": 0.0749,
      "step": 9123
    },
    {
      "epoch": 17.368521160247266,
      "grad_norm": 0.07633499056100845,
      "learning_rate": 4.2083201016195615e-05,
      "loss": 0.0486,
      "step": 9124
    },
    {
      "epoch": 17.370423204945315,
      "grad_norm": 0.06507447361946106,
      "learning_rate": 4.207684979358527e-05,
      "loss": 0.0404,
      "step": 9125
    },
    {
      "epoch": 17.372325249643367,
      "grad_norm": 0.19423268735408783,
      "learning_rate": 4.207049857097491e-05,
      "loss": 0.0531,
      "step": 9126
    },
    {
      "epoch": 17.374227294341416,
      "grad_norm": 0.14882993698120117,
      "learning_rate": 4.2064147348364564e-05,
      "loss": 0.0511,
      "step": 9127
    },
    {
      "epoch": 17.37612933903947,
      "grad_norm": 0.10786750912666321,
      "learning_rate": 4.205779612575421e-05,
      "loss": 0.0769,
      "step": 9128
    },
    {
      "epoch": 17.378031383737518,
      "grad_norm": 0.12144813686609268,
      "learning_rate": 4.2051444903143854e-05,
      "loss": 0.0612,
      "step": 9129
    },
    {
      "epoch": 17.379933428435567,
      "grad_norm": 0.07385139167308807,
      "learning_rate": 4.2045093680533506e-05,
      "loss": 0.0645,
      "step": 9130
    },
    {
      "epoch": 17.38183547313362,
      "grad_norm": 0.12522324919700623,
      "learning_rate": 4.203874245792315e-05,
      "loss": 0.0597,
      "step": 9131
    },
    {
      "epoch": 17.38373751783167,
      "grad_norm": 0.1233920082449913,
      "learning_rate": 4.20323912353128e-05,
      "loss": 0.0779,
      "step": 9132
    },
    {
      "epoch": 17.385639562529718,
      "grad_norm": 0.12006913125514984,
      "learning_rate": 4.202604001270244e-05,
      "loss": 0.0399,
      "step": 9133
    },
    {
      "epoch": 17.38754160722777,
      "grad_norm": 0.14210516214370728,
      "learning_rate": 4.201968879009209e-05,
      "loss": 0.0545,
      "step": 9134
    },
    {
      "epoch": 17.38944365192582,
      "grad_norm": 0.08909688889980316,
      "learning_rate": 4.2013337567481745e-05,
      "loss": 0.0703,
      "step": 9135
    },
    {
      "epoch": 17.39134569662387,
      "grad_norm": 0.06299816071987152,
      "learning_rate": 4.200698634487139e-05,
      "loss": 0.043,
      "step": 9136
    },
    {
      "epoch": 17.39324774132192,
      "grad_norm": 0.16345307230949402,
      "learning_rate": 4.200063512226104e-05,
      "loss": 0.0625,
      "step": 9137
    },
    {
      "epoch": 17.39514978601997,
      "grad_norm": 0.14235182106494904,
      "learning_rate": 4.199428389965068e-05,
      "loss": 0.0742,
      "step": 9138
    },
    {
      "epoch": 17.397051830718024,
      "grad_norm": 0.14397858083248138,
      "learning_rate": 4.198793267704033e-05,
      "loss": 0.069,
      "step": 9139
    },
    {
      "epoch": 17.398953875416073,
      "grad_norm": 0.10805745422840118,
      "learning_rate": 4.198158145442998e-05,
      "loss": 0.0615,
      "step": 9140
    },
    {
      "epoch": 17.40085592011412,
      "grad_norm": 0.19101940095424652,
      "learning_rate": 4.197523023181963e-05,
      "loss": 0.0384,
      "step": 9141
    },
    {
      "epoch": 17.402757964812174,
      "grad_norm": 0.15251103043556213,
      "learning_rate": 4.1968879009209274e-05,
      "loss": 0.0377,
      "step": 9142
    },
    {
      "epoch": 17.404660009510224,
      "grad_norm": 0.11683591455221176,
      "learning_rate": 4.196252778659892e-05,
      "loss": 0.0575,
      "step": 9143
    },
    {
      "epoch": 17.406562054208273,
      "grad_norm": 0.11389310657978058,
      "learning_rate": 4.195617656398857e-05,
      "loss": 0.0598,
      "step": 9144
    },
    {
      "epoch": 17.408464098906325,
      "grad_norm": 0.12031054496765137,
      "learning_rate": 4.1949825341378216e-05,
      "loss": 0.0459,
      "step": 9145
    },
    {
      "epoch": 17.410366143604374,
      "grad_norm": 0.18291601538658142,
      "learning_rate": 4.194347411876787e-05,
      "loss": 0.0553,
      "step": 9146
    },
    {
      "epoch": 17.412268188302424,
      "grad_norm": 0.12735716998577118,
      "learning_rate": 4.193712289615751e-05,
      "loss": 0.0358,
      "step": 9147
    },
    {
      "epoch": 17.414170233000476,
      "grad_norm": 0.15188108384609222,
      "learning_rate": 4.193077167354716e-05,
      "loss": 0.0463,
      "step": 9148
    },
    {
      "epoch": 17.416072277698525,
      "grad_norm": 0.14176872372627258,
      "learning_rate": 4.192442045093681e-05,
      "loss": 0.0633,
      "step": 9149
    },
    {
      "epoch": 17.417974322396578,
      "grad_norm": 0.17167988419532776,
      "learning_rate": 4.1918069228326455e-05,
      "loss": 0.0494,
      "step": 9150
    },
    {
      "epoch": 17.419876367094627,
      "grad_norm": 0.07536623626947403,
      "learning_rate": 4.19117180057161e-05,
      "loss": 0.0484,
      "step": 9151
    },
    {
      "epoch": 17.421778411792676,
      "grad_norm": 0.1476680040359497,
      "learning_rate": 4.1905366783105745e-05,
      "loss": 0.0615,
      "step": 9152
    },
    {
      "epoch": 17.42368045649073,
      "grad_norm": 0.09555692970752716,
      "learning_rate": 4.18990155604954e-05,
      "loss": 0.0685,
      "step": 9153
    },
    {
      "epoch": 17.425582501188778,
      "grad_norm": 0.18903157114982605,
      "learning_rate": 4.189266433788505e-05,
      "loss": 0.0547,
      "step": 9154
    },
    {
      "epoch": 17.427484545886827,
      "grad_norm": 0.13380809128284454,
      "learning_rate": 4.1886313115274694e-05,
      "loss": 0.0475,
      "step": 9155
    },
    {
      "epoch": 17.42938659058488,
      "grad_norm": 0.20205047726631165,
      "learning_rate": 4.187996189266434e-05,
      "loss": 0.064,
      "step": 9156
    },
    {
      "epoch": 17.43128863528293,
      "grad_norm": 0.15946093201637268,
      "learning_rate": 4.1873610670053984e-05,
      "loss": 0.0455,
      "step": 9157
    },
    {
      "epoch": 17.433190679980978,
      "grad_norm": 0.10756520926952362,
      "learning_rate": 4.1867259447443636e-05,
      "loss": 0.0474,
      "step": 9158
    },
    {
      "epoch": 17.43509272467903,
      "grad_norm": 0.07083479315042496,
      "learning_rate": 4.186090822483328e-05,
      "loss": 0.0591,
      "step": 9159
    },
    {
      "epoch": 17.43699476937708,
      "grad_norm": 0.21123982965946198,
      "learning_rate": 4.1854557002222926e-05,
      "loss": 0.0682,
      "step": 9160
    },
    {
      "epoch": 17.438896814075132,
      "grad_norm": 0.15162432193756104,
      "learning_rate": 4.184820577961258e-05,
      "loss": 0.0763,
      "step": 9161
    },
    {
      "epoch": 17.44079885877318,
      "grad_norm": 0.11110099405050278,
      "learning_rate": 4.184185455700222e-05,
      "loss": 0.0495,
      "step": 9162
    },
    {
      "epoch": 17.44270090347123,
      "grad_norm": 0.12019350379705429,
      "learning_rate": 4.1835503334391875e-05,
      "loss": 0.0407,
      "step": 9163
    },
    {
      "epoch": 17.444602948169283,
      "grad_norm": 0.166811004281044,
      "learning_rate": 4.182915211178152e-05,
      "loss": 0.0632,
      "step": 9164
    },
    {
      "epoch": 17.446504992867332,
      "grad_norm": 0.12711088359355927,
      "learning_rate": 4.1822800889171165e-05,
      "loss": 0.0347,
      "step": 9165
    },
    {
      "epoch": 17.44840703756538,
      "grad_norm": 0.09420914947986603,
      "learning_rate": 4.181644966656082e-05,
      "loss": 0.0427,
      "step": 9166
    },
    {
      "epoch": 17.450309082263434,
      "grad_norm": 0.12314459681510925,
      "learning_rate": 4.181009844395046e-05,
      "loss": 0.0591,
      "step": 9167
    },
    {
      "epoch": 17.452211126961483,
      "grad_norm": 0.05265713855624199,
      "learning_rate": 4.1803747221340114e-05,
      "loss": 0.0378,
      "step": 9168
    },
    {
      "epoch": 17.454113171659532,
      "grad_norm": 0.08246375620365143,
      "learning_rate": 4.179739599872975e-05,
      "loss": 0.0409,
      "step": 9169
    },
    {
      "epoch": 17.456015216357585,
      "grad_norm": 0.08644389361143112,
      "learning_rate": 4.1791044776119404e-05,
      "loss": 0.0511,
      "step": 9170
    },
    {
      "epoch": 17.457917261055634,
      "grad_norm": 0.10643388330936432,
      "learning_rate": 4.1784693553509056e-05,
      "loss": 0.0592,
      "step": 9171
    },
    {
      "epoch": 17.459819305753687,
      "grad_norm": 0.1023353561758995,
      "learning_rate": 4.17783423308987e-05,
      "loss": 0.0657,
      "step": 9172
    },
    {
      "epoch": 17.461721350451736,
      "grad_norm": 0.2840350270271301,
      "learning_rate": 4.177199110828835e-05,
      "loss": 0.0902,
      "step": 9173
    },
    {
      "epoch": 17.463623395149785,
      "grad_norm": 0.23073290288448334,
      "learning_rate": 4.176563988567799e-05,
      "loss": 0.0592,
      "step": 9174
    },
    {
      "epoch": 17.465525439847838,
      "grad_norm": 0.15955200791358948,
      "learning_rate": 4.175928866306764e-05,
      "loss": 0.0418,
      "step": 9175
    },
    {
      "epoch": 17.467427484545887,
      "grad_norm": 0.17804458737373352,
      "learning_rate": 4.175293744045729e-05,
      "loss": 0.0592,
      "step": 9176
    },
    {
      "epoch": 17.469329529243936,
      "grad_norm": 0.055416397750377655,
      "learning_rate": 4.174658621784694e-05,
      "loss": 0.0351,
      "step": 9177
    },
    {
      "epoch": 17.47123157394199,
      "grad_norm": 0.10548324137926102,
      "learning_rate": 4.1740234995236585e-05,
      "loss": 0.0484,
      "step": 9178
    },
    {
      "epoch": 17.473133618640038,
      "grad_norm": 0.06586059927940369,
      "learning_rate": 4.173388377262623e-05,
      "loss": 0.0719,
      "step": 9179
    },
    {
      "epoch": 17.475035663338087,
      "grad_norm": 0.10012559592723846,
      "learning_rate": 4.172753255001588e-05,
      "loss": 0.0842,
      "step": 9180
    },
    {
      "epoch": 17.47693770803614,
      "grad_norm": 0.13187265396118164,
      "learning_rate": 4.1721181327405526e-05,
      "loss": 0.0786,
      "step": 9181
    },
    {
      "epoch": 17.47883975273419,
      "grad_norm": 0.08381135761737823,
      "learning_rate": 4.171483010479518e-05,
      "loss": 0.0743,
      "step": 9182
    },
    {
      "epoch": 17.48074179743224,
      "grad_norm": 0.1986529380083084,
      "learning_rate": 4.1708478882184823e-05,
      "loss": 0.0751,
      "step": 9183
    },
    {
      "epoch": 17.48264384213029,
      "grad_norm": 0.12799467146396637,
      "learning_rate": 4.170212765957447e-05,
      "loss": 0.0566,
      "step": 9184
    },
    {
      "epoch": 17.48454588682834,
      "grad_norm": 0.2495906800031662,
      "learning_rate": 4.169577643696412e-05,
      "loss": 0.0542,
      "step": 9185
    },
    {
      "epoch": 17.486447931526392,
      "grad_norm": 0.0946408286690712,
      "learning_rate": 4.1689425214353765e-05,
      "loss": 0.0524,
      "step": 9186
    },
    {
      "epoch": 17.48834997622444,
      "grad_norm": 0.16105763614177704,
      "learning_rate": 4.168307399174341e-05,
      "loss": 0.0692,
      "step": 9187
    },
    {
      "epoch": 17.49025202092249,
      "grad_norm": 0.12365740537643433,
      "learning_rate": 4.1676722769133056e-05,
      "loss": 0.0774,
      "step": 9188
    },
    {
      "epoch": 17.492154065620543,
      "grad_norm": 0.231750950217247,
      "learning_rate": 4.167037154652271e-05,
      "loss": 0.0684,
      "step": 9189
    },
    {
      "epoch": 17.494056110318592,
      "grad_norm": 0.19699423015117645,
      "learning_rate": 4.166402032391236e-05,
      "loss": 0.0652,
      "step": 9190
    },
    {
      "epoch": 17.49595815501664,
      "grad_norm": 0.11165280640125275,
      "learning_rate": 4.1657669101302004e-05,
      "loss": 0.0556,
      "step": 9191
    },
    {
      "epoch": 17.497860199714694,
      "grad_norm": 0.13366112112998962,
      "learning_rate": 4.165131787869165e-05,
      "loss": 0.0587,
      "step": 9192
    },
    {
      "epoch": 17.499762244412743,
      "grad_norm": 0.07594496011734009,
      "learning_rate": 4.1644966656081294e-05,
      "loss": 0.0555,
      "step": 9193
    },
    {
      "epoch": 17.501664289110796,
      "grad_norm": 0.14714041352272034,
      "learning_rate": 4.1638615433470946e-05,
      "loss": 0.073,
      "step": 9194
    },
    {
      "epoch": 17.503566333808845,
      "grad_norm": 0.08447927236557007,
      "learning_rate": 4.163226421086059e-05,
      "loss": 0.0683,
      "step": 9195
    },
    {
      "epoch": 17.505468378506894,
      "grad_norm": 0.09991919994354248,
      "learning_rate": 4.1625912988250236e-05,
      "loss": 0.0686,
      "step": 9196
    },
    {
      "epoch": 17.507370423204947,
      "grad_norm": 0.055409085005521774,
      "learning_rate": 4.161956176563989e-05,
      "loss": 0.0458,
      "step": 9197
    },
    {
      "epoch": 17.509272467902996,
      "grad_norm": 0.13919416069984436,
      "learning_rate": 4.161321054302953e-05,
      "loss": 0.0596,
      "step": 9198
    },
    {
      "epoch": 17.511174512601045,
      "grad_norm": 0.14709287881851196,
      "learning_rate": 4.1606859320419185e-05,
      "loss": 0.0463,
      "step": 9199
    },
    {
      "epoch": 17.513076557299097,
      "grad_norm": 0.06390659511089325,
      "learning_rate": 4.160050809780883e-05,
      "loss": 0.0631,
      "step": 9200
    },
    {
      "epoch": 17.514978601997147,
      "grad_norm": 0.08628946542739868,
      "learning_rate": 4.1594156875198475e-05,
      "loss": 0.0562,
      "step": 9201
    },
    {
      "epoch": 17.516880646695196,
      "grad_norm": 0.06527147442102432,
      "learning_rate": 4.158780565258813e-05,
      "loss": 0.052,
      "step": 9202
    },
    {
      "epoch": 17.51878269139325,
      "grad_norm": 0.1150786429643631,
      "learning_rate": 4.158145442997777e-05,
      "loss": 0.064,
      "step": 9203
    },
    {
      "epoch": 17.520684736091297,
      "grad_norm": 0.17034420371055603,
      "learning_rate": 4.1575103207367424e-05,
      "loss": 0.079,
      "step": 9204
    },
    {
      "epoch": 17.52258678078935,
      "grad_norm": 0.07963655143976212,
      "learning_rate": 4.156875198475706e-05,
      "loss": 0.0377,
      "step": 9205
    },
    {
      "epoch": 17.5244888254874,
      "grad_norm": 0.17314720153808594,
      "learning_rate": 4.1562400762146714e-05,
      "loss": 0.0445,
      "step": 9206
    },
    {
      "epoch": 17.52639087018545,
      "grad_norm": 0.09530575573444366,
      "learning_rate": 4.155604953953636e-05,
      "loss": 0.056,
      "step": 9207
    },
    {
      "epoch": 17.5282929148835,
      "grad_norm": 0.1464606672525406,
      "learning_rate": 4.154969831692601e-05,
      "loss": 0.0531,
      "step": 9208
    },
    {
      "epoch": 17.53019495958155,
      "grad_norm": 0.15889886021614075,
      "learning_rate": 4.154334709431566e-05,
      "loss": 0.05,
      "step": 9209
    },
    {
      "epoch": 17.5320970042796,
      "grad_norm": 0.11242558807134628,
      "learning_rate": 4.15369958717053e-05,
      "loss": 0.0485,
      "step": 9210
    },
    {
      "epoch": 17.533999048977652,
      "grad_norm": 0.05474596098065376,
      "learning_rate": 4.153064464909495e-05,
      "loss": 0.0551,
      "step": 9211
    },
    {
      "epoch": 17.5359010936757,
      "grad_norm": 0.06351004540920258,
      "learning_rate": 4.15242934264846e-05,
      "loss": 0.0738,
      "step": 9212
    },
    {
      "epoch": 17.537803138373754,
      "grad_norm": 0.17597231268882751,
      "learning_rate": 4.151794220387425e-05,
      "loss": 0.044,
      "step": 9213
    },
    {
      "epoch": 17.539705183071803,
      "grad_norm": 0.05791318044066429,
      "learning_rate": 4.1511590981263895e-05,
      "loss": 0.0631,
      "step": 9214
    },
    {
      "epoch": 17.541607227769852,
      "grad_norm": 0.1329626441001892,
      "learning_rate": 4.150523975865354e-05,
      "loss": 0.0715,
      "step": 9215
    },
    {
      "epoch": 17.543509272467904,
      "grad_norm": 0.07686080783605576,
      "learning_rate": 4.149888853604319e-05,
      "loss": 0.0579,
      "step": 9216
    },
    {
      "epoch": 17.545411317165954,
      "grad_norm": 0.14616870880126953,
      "learning_rate": 4.149253731343284e-05,
      "loss": 0.0482,
      "step": 9217
    },
    {
      "epoch": 17.547313361864003,
      "grad_norm": 0.14873604476451874,
      "learning_rate": 4.148618609082249e-05,
      "loss": 0.0594,
      "step": 9218
    },
    {
      "epoch": 17.549215406562055,
      "grad_norm": 0.0986090674996376,
      "learning_rate": 4.1479834868212134e-05,
      "loss": 0.0768,
      "step": 9219
    },
    {
      "epoch": 17.551117451260104,
      "grad_norm": 0.11944249272346497,
      "learning_rate": 4.147348364560178e-05,
      "loss": 0.0506,
      "step": 9220
    },
    {
      "epoch": 17.553019495958154,
      "grad_norm": 0.06361818313598633,
      "learning_rate": 4.146713242299143e-05,
      "loss": 0.0476,
      "step": 9221
    },
    {
      "epoch": 17.554921540656206,
      "grad_norm": 0.1397957056760788,
      "learning_rate": 4.1460781200381076e-05,
      "loss": 0.0636,
      "step": 9222
    },
    {
      "epoch": 17.556823585354255,
      "grad_norm": 0.08839410543441772,
      "learning_rate": 4.145442997777072e-05,
      "loss": 0.0517,
      "step": 9223
    },
    {
      "epoch": 17.558725630052308,
      "grad_norm": 0.24125997722148895,
      "learning_rate": 4.1448078755160366e-05,
      "loss": 0.0421,
      "step": 9224
    },
    {
      "epoch": 17.560627674750357,
      "grad_norm": 0.07652424275875092,
      "learning_rate": 4.144172753255002e-05,
      "loss": 0.0591,
      "step": 9225
    },
    {
      "epoch": 17.562529719448406,
      "grad_norm": 0.08479491621255875,
      "learning_rate": 4.143537630993967e-05,
      "loss": 0.0524,
      "step": 9226
    },
    {
      "epoch": 17.56443176414646,
      "grad_norm": 0.15499165654182434,
      "learning_rate": 4.1429025087329315e-05,
      "loss": 0.0569,
      "step": 9227
    },
    {
      "epoch": 17.566333808844508,
      "grad_norm": 0.1465625762939453,
      "learning_rate": 4.142267386471896e-05,
      "loss": 0.0578,
      "step": 9228
    },
    {
      "epoch": 17.568235853542557,
      "grad_norm": 0.19261913001537323,
      "learning_rate": 4.1416322642108605e-05,
      "loss": 0.049,
      "step": 9229
    },
    {
      "epoch": 17.57013789824061,
      "grad_norm": 0.1542554348707199,
      "learning_rate": 4.140997141949826e-05,
      "loss": 0.0541,
      "step": 9230
    },
    {
      "epoch": 17.57203994293866,
      "grad_norm": 0.12585055828094482,
      "learning_rate": 4.14036201968879e-05,
      "loss": 0.0514,
      "step": 9231
    },
    {
      "epoch": 17.573941987636708,
      "grad_norm": 0.06084693223237991,
      "learning_rate": 4.139726897427755e-05,
      "loss": 0.0485,
      "step": 9232
    },
    {
      "epoch": 17.57584403233476,
      "grad_norm": 0.06361477077007294,
      "learning_rate": 4.13909177516672e-05,
      "loss": 0.0597,
      "step": 9233
    },
    {
      "epoch": 17.57774607703281,
      "grad_norm": 0.14518912136554718,
      "learning_rate": 4.1384566529056844e-05,
      "loss": 0.0767,
      "step": 9234
    },
    {
      "epoch": 17.579648121730862,
      "grad_norm": 0.14766274392604828,
      "learning_rate": 4.1378215306446496e-05,
      "loss": 0.0475,
      "step": 9235
    },
    {
      "epoch": 17.58155016642891,
      "grad_norm": 0.15703773498535156,
      "learning_rate": 4.137186408383614e-05,
      "loss": 0.0614,
      "step": 9236
    },
    {
      "epoch": 17.58345221112696,
      "grad_norm": 0.17378023266792297,
      "learning_rate": 4.1365512861225786e-05,
      "loss": 0.0478,
      "step": 9237
    },
    {
      "epoch": 17.585354255825013,
      "grad_norm": 0.18739935755729675,
      "learning_rate": 4.135916163861544e-05,
      "loss": 0.0629,
      "step": 9238
    },
    {
      "epoch": 17.587256300523062,
      "grad_norm": 0.10931984335184097,
      "learning_rate": 4.135281041600508e-05,
      "loss": 0.0717,
      "step": 9239
    },
    {
      "epoch": 17.58915834522111,
      "grad_norm": 0.23572999238967896,
      "learning_rate": 4.1346459193394734e-05,
      "loss": 0.0562,
      "step": 9240
    },
    {
      "epoch": 17.591060389919164,
      "grad_norm": 0.2369711995124817,
      "learning_rate": 4.134010797078437e-05,
      "loss": 0.0428,
      "step": 9241
    },
    {
      "epoch": 17.592962434617213,
      "grad_norm": 0.14604787528514862,
      "learning_rate": 4.1333756748174025e-05,
      "loss": 0.0603,
      "step": 9242
    },
    {
      "epoch": 17.594864479315262,
      "grad_norm": 0.06450549513101578,
      "learning_rate": 4.132740552556367e-05,
      "loss": 0.0687,
      "step": 9243
    },
    {
      "epoch": 17.596766524013315,
      "grad_norm": 0.10120340436697006,
      "learning_rate": 4.132105430295332e-05,
      "loss": 0.0708,
      "step": 9244
    },
    {
      "epoch": 17.598668568711364,
      "grad_norm": 0.09536679089069366,
      "learning_rate": 4.131470308034297e-05,
      "loss": 0.0538,
      "step": 9245
    },
    {
      "epoch": 17.600570613409417,
      "grad_norm": 0.13788338005542755,
      "learning_rate": 4.130835185773261e-05,
      "loss": 0.0704,
      "step": 9246
    },
    {
      "epoch": 17.602472658107466,
      "grad_norm": 0.1518242061138153,
      "learning_rate": 4.1302000635122263e-05,
      "loss": 0.0709,
      "step": 9247
    },
    {
      "epoch": 17.604374702805515,
      "grad_norm": 0.18178845942020416,
      "learning_rate": 4.129564941251191e-05,
      "loss": 0.0732,
      "step": 9248
    },
    {
      "epoch": 17.606276747503568,
      "grad_norm": 0.06598396599292755,
      "learning_rate": 4.128929818990156e-05,
      "loss": 0.0535,
      "step": 9249
    },
    {
      "epoch": 17.608178792201617,
      "grad_norm": 0.07990466803312302,
      "learning_rate": 4.1282946967291205e-05,
      "loss": 0.0512,
      "step": 9250
    },
    {
      "epoch": 17.610080836899666,
      "grad_norm": 0.07749851047992706,
      "learning_rate": 4.127659574468085e-05,
      "loss": 0.0499,
      "step": 9251
    },
    {
      "epoch": 17.61198288159772,
      "grad_norm": 0.07298611849546432,
      "learning_rate": 4.12702445220705e-05,
      "loss": 0.0527,
      "step": 9252
    },
    {
      "epoch": 17.613884926295768,
      "grad_norm": 0.06433837860822678,
      "learning_rate": 4.126389329946015e-05,
      "loss": 0.0421,
      "step": 9253
    },
    {
      "epoch": 17.615786970993817,
      "grad_norm": 0.13365700840950012,
      "learning_rate": 4.12575420768498e-05,
      "loss": 0.057,
      "step": 9254
    },
    {
      "epoch": 17.61768901569187,
      "grad_norm": 0.16515879333019257,
      "learning_rate": 4.1251190854239444e-05,
      "loss": 0.0356,
      "step": 9255
    },
    {
      "epoch": 17.61959106038992,
      "grad_norm": 0.16000066697597504,
      "learning_rate": 4.124483963162909e-05,
      "loss": 0.0581,
      "step": 9256
    },
    {
      "epoch": 17.62149310508797,
      "grad_norm": 0.09683466702699661,
      "learning_rate": 4.123848840901874e-05,
      "loss": 0.0531,
      "step": 9257
    },
    {
      "epoch": 17.62339514978602,
      "grad_norm": 0.1364901065826416,
      "learning_rate": 4.1232137186408386e-05,
      "loss": 0.066,
      "step": 9258
    },
    {
      "epoch": 17.62529719448407,
      "grad_norm": 0.07379920780658722,
      "learning_rate": 4.122578596379803e-05,
      "loss": 0.0489,
      "step": 9259
    },
    {
      "epoch": 17.627199239182122,
      "grad_norm": 0.13212312757968903,
      "learning_rate": 4.1219434741187676e-05,
      "loss": 0.0741,
      "step": 9260
    },
    {
      "epoch": 17.62910128388017,
      "grad_norm": 0.14788813889026642,
      "learning_rate": 4.121308351857733e-05,
      "loss": 0.0576,
      "step": 9261
    },
    {
      "epoch": 17.63100332857822,
      "grad_norm": 0.0804387703537941,
      "learning_rate": 4.120673229596697e-05,
      "loss": 0.0426,
      "step": 9262
    },
    {
      "epoch": 17.632905373276273,
      "grad_norm": 0.14654266834259033,
      "learning_rate": 4.1200381073356625e-05,
      "loss": 0.0627,
      "step": 9263
    },
    {
      "epoch": 17.634807417974322,
      "grad_norm": 0.08965173363685608,
      "learning_rate": 4.119402985074627e-05,
      "loss": 0.0574,
      "step": 9264
    },
    {
      "epoch": 17.63670946267237,
      "grad_norm": 0.17583389580249786,
      "learning_rate": 4.1187678628135915e-05,
      "loss": 0.0757,
      "step": 9265
    },
    {
      "epoch": 17.638611507370424,
      "grad_norm": 0.06688621640205383,
      "learning_rate": 4.118132740552557e-05,
      "loss": 0.0662,
      "step": 9266
    },
    {
      "epoch": 17.640513552068473,
      "grad_norm": 0.10981274396181107,
      "learning_rate": 4.117497618291521e-05,
      "loss": 0.0439,
      "step": 9267
    },
    {
      "epoch": 17.642415596766526,
      "grad_norm": 0.07107902318239212,
      "learning_rate": 4.116862496030486e-05,
      "loss": 0.0575,
      "step": 9268
    },
    {
      "epoch": 17.644317641464575,
      "grad_norm": 0.11891138553619385,
      "learning_rate": 4.116227373769451e-05,
      "loss": 0.0433,
      "step": 9269
    },
    {
      "epoch": 17.646219686162624,
      "grad_norm": 0.15733720362186432,
      "learning_rate": 4.1155922515084154e-05,
      "loss": 0.0527,
      "step": 9270
    },
    {
      "epoch": 17.648121730860677,
      "grad_norm": 0.12876959145069122,
      "learning_rate": 4.1149571292473806e-05,
      "loss": 0.0607,
      "step": 9271
    },
    {
      "epoch": 17.650023775558726,
      "grad_norm": 0.08375310152769089,
      "learning_rate": 4.114322006986345e-05,
      "loss": 0.06,
      "step": 9272
    },
    {
      "epoch": 17.651925820256775,
      "grad_norm": 0.21883268654346466,
      "learning_rate": 4.1136868847253096e-05,
      "loss": 0.0459,
      "step": 9273
    },
    {
      "epoch": 17.653827864954827,
      "grad_norm": 0.06427763402462006,
      "learning_rate": 4.113051762464275e-05,
      "loss": 0.0428,
      "step": 9274
    },
    {
      "epoch": 17.655729909652877,
      "grad_norm": 0.12494489550590515,
      "learning_rate": 4.112416640203239e-05,
      "loss": 0.0632,
      "step": 9275
    },
    {
      "epoch": 17.657631954350926,
      "grad_norm": 0.125833198428154,
      "learning_rate": 4.1117815179422045e-05,
      "loss": 0.0849,
      "step": 9276
    },
    {
      "epoch": 17.65953399904898,
      "grad_norm": 0.09488727897405624,
      "learning_rate": 4.111146395681168e-05,
      "loss": 0.0638,
      "step": 9277
    },
    {
      "epoch": 17.661436043747027,
      "grad_norm": 0.29696619510650635,
      "learning_rate": 4.1105112734201335e-05,
      "loss": 0.06,
      "step": 9278
    },
    {
      "epoch": 17.66333808844508,
      "grad_norm": 0.14339213073253632,
      "learning_rate": 4.109876151159098e-05,
      "loss": 0.0465,
      "step": 9279
    },
    {
      "epoch": 17.66524013314313,
      "grad_norm": 0.1306292712688446,
      "learning_rate": 4.109241028898063e-05,
      "loss": 0.0431,
      "step": 9280
    },
    {
      "epoch": 17.66714217784118,
      "grad_norm": 0.09760496765375137,
      "learning_rate": 4.1086059066370284e-05,
      "loss": 0.0697,
      "step": 9281
    },
    {
      "epoch": 17.66904422253923,
      "grad_norm": 0.19088013470172882,
      "learning_rate": 4.107970784375992e-05,
      "loss": 0.0653,
      "step": 9282
    },
    {
      "epoch": 17.67094626723728,
      "grad_norm": 0.09736175835132599,
      "learning_rate": 4.1073356621149574e-05,
      "loss": 0.0622,
      "step": 9283
    },
    {
      "epoch": 17.67284831193533,
      "grad_norm": 0.11242244392633438,
      "learning_rate": 4.106700539853922e-05,
      "loss": 0.0884,
      "step": 9284
    },
    {
      "epoch": 17.674750356633382,
      "grad_norm": 0.10703061521053314,
      "learning_rate": 4.106065417592887e-05,
      "loss": 0.0626,
      "step": 9285
    },
    {
      "epoch": 17.67665240133143,
      "grad_norm": 0.16119727492332458,
      "learning_rate": 4.1054302953318516e-05,
      "loss": 0.0697,
      "step": 9286
    },
    {
      "epoch": 17.67855444602948,
      "grad_norm": 0.09668602794408798,
      "learning_rate": 4.104795173070816e-05,
      "loss": 0.0612,
      "step": 9287
    },
    {
      "epoch": 17.680456490727533,
      "grad_norm": 0.1401471346616745,
      "learning_rate": 4.104160050809781e-05,
      "loss": 0.0609,
      "step": 9288
    },
    {
      "epoch": 17.682358535425582,
      "grad_norm": 0.05594095587730408,
      "learning_rate": 4.103524928548746e-05,
      "loss": 0.0497,
      "step": 9289
    },
    {
      "epoch": 17.684260580123635,
      "grad_norm": 0.11976776272058487,
      "learning_rate": 4.102889806287711e-05,
      "loss": 0.0902,
      "step": 9290
    },
    {
      "epoch": 17.686162624821684,
      "grad_norm": 0.211091086268425,
      "learning_rate": 4.102254684026675e-05,
      "loss": 0.0502,
      "step": 9291
    },
    {
      "epoch": 17.688064669519733,
      "grad_norm": 0.14465105533599854,
      "learning_rate": 4.10161956176564e-05,
      "loss": 0.0681,
      "step": 9292
    },
    {
      "epoch": 17.689966714217785,
      "grad_norm": 0.10496453195810318,
      "learning_rate": 4.100984439504605e-05,
      "loss": 0.0595,
      "step": 9293
    },
    {
      "epoch": 17.691868758915835,
      "grad_norm": 0.09541040658950806,
      "learning_rate": 4.10034931724357e-05,
      "loss": 0.0452,
      "step": 9294
    },
    {
      "epoch": 17.693770803613884,
      "grad_norm": 0.43040457367897034,
      "learning_rate": 4.099714194982534e-05,
      "loss": 0.0689,
      "step": 9295
    },
    {
      "epoch": 17.695672848311936,
      "grad_norm": 0.13549309968948364,
      "learning_rate": 4.099079072721499e-05,
      "loss": 0.0734,
      "step": 9296
    },
    {
      "epoch": 17.697574893009985,
      "grad_norm": 0.10545626282691956,
      "learning_rate": 4.098443950460464e-05,
      "loss": 0.0585,
      "step": 9297
    },
    {
      "epoch": 17.699476937708035,
      "grad_norm": 0.1298322230577469,
      "learning_rate": 4.0978088281994284e-05,
      "loss": 0.0664,
      "step": 9298
    },
    {
      "epoch": 17.701378982406087,
      "grad_norm": 0.09501194208860397,
      "learning_rate": 4.0971737059383936e-05,
      "loss": 0.0539,
      "step": 9299
    },
    {
      "epoch": 17.703281027104136,
      "grad_norm": 0.1124708279967308,
      "learning_rate": 4.096538583677358e-05,
      "loss": 0.0644,
      "step": 9300
    },
    {
      "epoch": 17.70518307180219,
      "grad_norm": 0.14791899919509888,
      "learning_rate": 4.0959034614163226e-05,
      "loss": 0.0434,
      "step": 9301
    },
    {
      "epoch": 17.707085116500238,
      "grad_norm": 0.08081545680761337,
      "learning_rate": 4.095268339155288e-05,
      "loss": 0.0393,
      "step": 9302
    },
    {
      "epoch": 17.708987161198287,
      "grad_norm": 0.2517786920070648,
      "learning_rate": 4.094633216894252e-05,
      "loss": 0.0839,
      "step": 9303
    },
    {
      "epoch": 17.71088920589634,
      "grad_norm": 0.10763931274414062,
      "learning_rate": 4.093998094633217e-05,
      "loss": 0.0427,
      "step": 9304
    },
    {
      "epoch": 17.71279125059439,
      "grad_norm": 0.1142471581697464,
      "learning_rate": 4.093362972372182e-05,
      "loss": 0.0615,
      "step": 9305
    },
    {
      "epoch": 17.714693295292438,
      "grad_norm": 0.07118445634841919,
      "learning_rate": 4.0927278501111465e-05,
      "loss": 0.0606,
      "step": 9306
    },
    {
      "epoch": 17.71659533999049,
      "grad_norm": 0.0904897153377533,
      "learning_rate": 4.0920927278501116e-05,
      "loss": 0.0464,
      "step": 9307
    },
    {
      "epoch": 17.71849738468854,
      "grad_norm": 0.203889861702919,
      "learning_rate": 4.091457605589076e-05,
      "loss": 0.1143,
      "step": 9308
    },
    {
      "epoch": 17.72039942938659,
      "grad_norm": 0.17758524417877197,
      "learning_rate": 4.0908224833280407e-05,
      "loss": 0.0507,
      "step": 9309
    },
    {
      "epoch": 17.72230147408464,
      "grad_norm": 0.2005939483642578,
      "learning_rate": 4.090187361067006e-05,
      "loss": 0.0502,
      "step": 9310
    },
    {
      "epoch": 17.72420351878269,
      "grad_norm": 0.06343390792608261,
      "learning_rate": 4.0895522388059703e-05,
      "loss": 0.0507,
      "step": 9311
    },
    {
      "epoch": 17.726105563480743,
      "grad_norm": 0.1854616403579712,
      "learning_rate": 4.0889171165449355e-05,
      "loss": 0.0696,
      "step": 9312
    },
    {
      "epoch": 17.728007608178793,
      "grad_norm": 0.08541128039360046,
      "learning_rate": 4.0882819942838994e-05,
      "loss": 0.0475,
      "step": 9313
    },
    {
      "epoch": 17.72990965287684,
      "grad_norm": 0.12330986559391022,
      "learning_rate": 4.0876468720228645e-05,
      "loss": 0.0585,
      "step": 9314
    },
    {
      "epoch": 17.731811697574894,
      "grad_norm": 0.12485690414905548,
      "learning_rate": 4.087011749761829e-05,
      "loss": 0.0476,
      "step": 9315
    },
    {
      "epoch": 17.733713742272943,
      "grad_norm": 0.08095765858888626,
      "learning_rate": 4.086376627500794e-05,
      "loss": 0.057,
      "step": 9316
    },
    {
      "epoch": 17.735615786970993,
      "grad_norm": 0.1634143441915512,
      "learning_rate": 4.085741505239759e-05,
      "loss": 0.0732,
      "step": 9317
    },
    {
      "epoch": 17.737517831669045,
      "grad_norm": 0.1279916763305664,
      "learning_rate": 4.085106382978723e-05,
      "loss": 0.1143,
      "step": 9318
    },
    {
      "epoch": 17.739419876367094,
      "grad_norm": 0.11108207702636719,
      "learning_rate": 4.0844712607176884e-05,
      "loss": 0.0565,
      "step": 9319
    },
    {
      "epoch": 17.741321921065143,
      "grad_norm": 0.056614622473716736,
      "learning_rate": 4.083836138456653e-05,
      "loss": 0.0423,
      "step": 9320
    },
    {
      "epoch": 17.743223965763196,
      "grad_norm": 0.11688531935214996,
      "learning_rate": 4.083201016195618e-05,
      "loss": 0.0519,
      "step": 9321
    },
    {
      "epoch": 17.745126010461245,
      "grad_norm": 0.11915720999240875,
      "learning_rate": 4.0825658939345826e-05,
      "loss": 0.0385,
      "step": 9322
    },
    {
      "epoch": 17.747028055159298,
      "grad_norm": 0.12320053577423096,
      "learning_rate": 4.081930771673547e-05,
      "loss": 0.0548,
      "step": 9323
    },
    {
      "epoch": 17.748930099857347,
      "grad_norm": 0.11300956457853317,
      "learning_rate": 4.081295649412512e-05,
      "loss": 0.0758,
      "step": 9324
    },
    {
      "epoch": 17.750832144555396,
      "grad_norm": 0.1921488642692566,
      "learning_rate": 4.080660527151477e-05,
      "loss": 0.0541,
      "step": 9325
    },
    {
      "epoch": 17.75273418925345,
      "grad_norm": 0.128909632563591,
      "learning_rate": 4.080025404890442e-05,
      "loss": 0.0445,
      "step": 9326
    },
    {
      "epoch": 17.754636233951498,
      "grad_norm": 0.23205921053886414,
      "learning_rate": 4.079390282629406e-05,
      "loss": 0.0938,
      "step": 9327
    },
    {
      "epoch": 17.756538278649547,
      "grad_norm": 0.15705791115760803,
      "learning_rate": 4.078755160368371e-05,
      "loss": 0.0517,
      "step": 9328
    },
    {
      "epoch": 17.7584403233476,
      "grad_norm": 0.11689947545528412,
      "learning_rate": 4.078120038107336e-05,
      "loss": 0.0541,
      "step": 9329
    },
    {
      "epoch": 17.76034236804565,
      "grad_norm": 0.09988643229007721,
      "learning_rate": 4.077484915846301e-05,
      "loss": 0.0467,
      "step": 9330
    },
    {
      "epoch": 17.762244412743698,
      "grad_norm": 0.09019043296575546,
      "learning_rate": 4.076849793585265e-05,
      "loss": 0.0545,
      "step": 9331
    },
    {
      "epoch": 17.76414645744175,
      "grad_norm": 0.22703583538532257,
      "learning_rate": 4.07621467132423e-05,
      "loss": 0.0712,
      "step": 9332
    },
    {
      "epoch": 17.7660485021398,
      "grad_norm": 0.10877406597137451,
      "learning_rate": 4.075579549063195e-05,
      "loss": 0.0526,
      "step": 9333
    },
    {
      "epoch": 17.767950546837852,
      "grad_norm": 0.11539573222398758,
      "learning_rate": 4.0749444268021594e-05,
      "loss": 0.0718,
      "step": 9334
    },
    {
      "epoch": 17.7698525915359,
      "grad_norm": 0.061567336320877075,
      "learning_rate": 4.0743093045411246e-05,
      "loss": 0.0614,
      "step": 9335
    },
    {
      "epoch": 17.77175463623395,
      "grad_norm": 0.18987290561199188,
      "learning_rate": 4.073674182280089e-05,
      "loss": 0.0762,
      "step": 9336
    },
    {
      "epoch": 17.773656680932003,
      "grad_norm": 0.28458330035209656,
      "learning_rate": 4.0730390600190536e-05,
      "loss": 0.0795,
      "step": 9337
    },
    {
      "epoch": 17.775558725630052,
      "grad_norm": 0.11308696120977402,
      "learning_rate": 4.072403937758019e-05,
      "loss": 0.0466,
      "step": 9338
    },
    {
      "epoch": 17.7774607703281,
      "grad_norm": 0.2028082013130188,
      "learning_rate": 4.071768815496983e-05,
      "loss": 0.0587,
      "step": 9339
    },
    {
      "epoch": 17.779362815026154,
      "grad_norm": 0.17155446112155914,
      "learning_rate": 4.071133693235948e-05,
      "loss": 0.0851,
      "step": 9340
    },
    {
      "epoch": 17.781264859724203,
      "grad_norm": 0.16016116738319397,
      "learning_rate": 4.070498570974913e-05,
      "loss": 0.0902,
      "step": 9341
    },
    {
      "epoch": 17.783166904422252,
      "grad_norm": 0.09539316594600677,
      "learning_rate": 4.0698634487138775e-05,
      "loss": 0.0589,
      "step": 9342
    },
    {
      "epoch": 17.785068949120305,
      "grad_norm": 0.1688692420721054,
      "learning_rate": 4.069228326452843e-05,
      "loss": 0.0627,
      "step": 9343
    },
    {
      "epoch": 17.786970993818354,
      "grad_norm": 0.0660916119813919,
      "learning_rate": 4.068593204191807e-05,
      "loss": 0.048,
      "step": 9344
    },
    {
      "epoch": 17.788873038516407,
      "grad_norm": 0.2147185355424881,
      "learning_rate": 4.067958081930772e-05,
      "loss": 0.058,
      "step": 9345
    },
    {
      "epoch": 17.790775083214456,
      "grad_norm": 0.07245621830224991,
      "learning_rate": 4.067322959669736e-05,
      "loss": 0.0511,
      "step": 9346
    },
    {
      "epoch": 17.792677127912505,
      "grad_norm": 0.12138759344816208,
      "learning_rate": 4.0666878374087014e-05,
      "loss": 0.0457,
      "step": 9347
    },
    {
      "epoch": 17.794579172610558,
      "grad_norm": 0.1689005345106125,
      "learning_rate": 4.0660527151476666e-05,
      "loss": 0.0657,
      "step": 9348
    },
    {
      "epoch": 17.796481217308607,
      "grad_norm": 0.10606629401445389,
      "learning_rate": 4.0654175928866304e-05,
      "loss": 0.0565,
      "step": 9349
    },
    {
      "epoch": 17.798383262006656,
      "grad_norm": 0.34530267119407654,
      "learning_rate": 4.0647824706255956e-05,
      "loss": 0.0664,
      "step": 9350
    },
    {
      "epoch": 17.80028530670471,
      "grad_norm": 0.14835722744464874,
      "learning_rate": 4.06414734836456e-05,
      "loss": 0.062,
      "step": 9351
    },
    {
      "epoch": 17.802187351402758,
      "grad_norm": 0.12061794847249985,
      "learning_rate": 4.063512226103525e-05,
      "loss": 0.0522,
      "step": 9352
    },
    {
      "epoch": 17.804089396100807,
      "grad_norm": 0.15052704513072968,
      "learning_rate": 4.06287710384249e-05,
      "loss": 0.052,
      "step": 9353
    },
    {
      "epoch": 17.80599144079886,
      "grad_norm": 0.12164096534252167,
      "learning_rate": 4.062241981581454e-05,
      "loss": 0.0635,
      "step": 9354
    },
    {
      "epoch": 17.80789348549691,
      "grad_norm": 0.05688644200563431,
      "learning_rate": 4.0616068593204195e-05,
      "loss": 0.0416,
      "step": 9355
    },
    {
      "epoch": 17.80979553019496,
      "grad_norm": 0.10969038307666779,
      "learning_rate": 4.060971737059384e-05,
      "loss": 0.0704,
      "step": 9356
    },
    {
      "epoch": 17.81169757489301,
      "grad_norm": 0.06995966285467148,
      "learning_rate": 4.060336614798349e-05,
      "loss": 0.0733,
      "step": 9357
    },
    {
      "epoch": 17.81359961959106,
      "grad_norm": 0.13542746007442474,
      "learning_rate": 4.059701492537314e-05,
      "loss": 0.0735,
      "step": 9358
    },
    {
      "epoch": 17.815501664289112,
      "grad_norm": 0.1096753180027008,
      "learning_rate": 4.059066370276278e-05,
      "loss": 0.0606,
      "step": 9359
    },
    {
      "epoch": 17.81740370898716,
      "grad_norm": 0.2971826195716858,
      "learning_rate": 4.0584312480152434e-05,
      "loss": 0.0839,
      "step": 9360
    },
    {
      "epoch": 17.81930575368521,
      "grad_norm": 0.17957726120948792,
      "learning_rate": 4.057796125754208e-05,
      "loss": 0.0581,
      "step": 9361
    },
    {
      "epoch": 17.821207798383263,
      "grad_norm": 0.1047806441783905,
      "learning_rate": 4.057161003493173e-05,
      "loss": 0.0479,
      "step": 9362
    },
    {
      "epoch": 17.823109843081312,
      "grad_norm": 0.15887323021888733,
      "learning_rate": 4.056525881232137e-05,
      "loss": 0.0457,
      "step": 9363
    },
    {
      "epoch": 17.82501188777936,
      "grad_norm": 0.15369363129138947,
      "learning_rate": 4.055890758971102e-05,
      "loss": 0.0684,
      "step": 9364
    },
    {
      "epoch": 17.826913932477414,
      "grad_norm": 0.19774940609931946,
      "learning_rate": 4.055255636710067e-05,
      "loss": 0.067,
      "step": 9365
    },
    {
      "epoch": 17.828815977175463,
      "grad_norm": 0.1530078649520874,
      "learning_rate": 4.054620514449032e-05,
      "loss": 0.0652,
      "step": 9366
    },
    {
      "epoch": 17.830718021873516,
      "grad_norm": 0.3438361585140228,
      "learning_rate": 4.053985392187996e-05,
      "loss": 0.0593,
      "step": 9367
    },
    {
      "epoch": 17.832620066571565,
      "grad_norm": 0.16979555785655975,
      "learning_rate": 4.053350269926961e-05,
      "loss": 0.0505,
      "step": 9368
    },
    {
      "epoch": 17.834522111269614,
      "grad_norm": 0.07902833074331284,
      "learning_rate": 4.052715147665926e-05,
      "loss": 0.0478,
      "step": 9369
    },
    {
      "epoch": 17.836424155967666,
      "grad_norm": 0.18646280467510223,
      "learning_rate": 4.0520800254048905e-05,
      "loss": 0.0504,
      "step": 9370
    },
    {
      "epoch": 17.838326200665716,
      "grad_norm": 0.17114807665348053,
      "learning_rate": 4.0514449031438556e-05,
      "loss": 0.0792,
      "step": 9371
    },
    {
      "epoch": 17.840228245363765,
      "grad_norm": 0.18604673445224762,
      "learning_rate": 4.05080978088282e-05,
      "loss": 0.0628,
      "step": 9372
    },
    {
      "epoch": 17.842130290061817,
      "grad_norm": 0.13605882227420807,
      "learning_rate": 4.0501746586217847e-05,
      "loss": 0.0522,
      "step": 9373
    },
    {
      "epoch": 17.844032334759866,
      "grad_norm": 0.11220191419124603,
      "learning_rate": 4.04953953636075e-05,
      "loss": 0.0601,
      "step": 9374
    },
    {
      "epoch": 17.845934379457915,
      "grad_norm": 0.09776952862739563,
      "learning_rate": 4.0489044140997144e-05,
      "loss": 0.0745,
      "step": 9375
    },
    {
      "epoch": 17.847836424155968,
      "grad_norm": 0.15183986723423004,
      "learning_rate": 4.048269291838679e-05,
      "loss": 0.0586,
      "step": 9376
    },
    {
      "epoch": 17.849738468854017,
      "grad_norm": 0.19326747953891754,
      "learning_rate": 4.047634169577644e-05,
      "loss": 0.0676,
      "step": 9377
    },
    {
      "epoch": 17.85164051355207,
      "grad_norm": 0.058094944804906845,
      "learning_rate": 4.0469990473166085e-05,
      "loss": 0.0496,
      "step": 9378
    },
    {
      "epoch": 17.85354255825012,
      "grad_norm": 0.14105942845344543,
      "learning_rate": 4.046363925055574e-05,
      "loss": 0.0498,
      "step": 9379
    },
    {
      "epoch": 17.855444602948168,
      "grad_norm": 0.2387823611497879,
      "learning_rate": 4.045728802794538e-05,
      "loss": 0.0693,
      "step": 9380
    },
    {
      "epoch": 17.85734664764622,
      "grad_norm": 0.10509767383337021,
      "learning_rate": 4.045093680533503e-05,
      "loss": 0.0575,
      "step": 9381
    },
    {
      "epoch": 17.85924869234427,
      "grad_norm": 0.18093208968639374,
      "learning_rate": 4.044458558272467e-05,
      "loss": 0.0415,
      "step": 9382
    },
    {
      "epoch": 17.86115073704232,
      "grad_norm": 0.19939205050468445,
      "learning_rate": 4.0438234360114324e-05,
      "loss": 0.0646,
      "step": 9383
    },
    {
      "epoch": 17.86305278174037,
      "grad_norm": 0.14076243340969086,
      "learning_rate": 4.0431883137503976e-05,
      "loss": 0.0718,
      "step": 9384
    },
    {
      "epoch": 17.86495482643842,
      "grad_norm": 0.20016847550868988,
      "learning_rate": 4.0425531914893614e-05,
      "loss": 0.055,
      "step": 9385
    },
    {
      "epoch": 17.86685687113647,
      "grad_norm": 0.0891173854470253,
      "learning_rate": 4.0419180692283266e-05,
      "loss": 0.057,
      "step": 9386
    },
    {
      "epoch": 17.868758915834523,
      "grad_norm": 0.1260504126548767,
      "learning_rate": 4.041282946967291e-05,
      "loss": 0.059,
      "step": 9387
    },
    {
      "epoch": 17.87066096053257,
      "grad_norm": 0.05717617645859718,
      "learning_rate": 4.040647824706256e-05,
      "loss": 0.0497,
      "step": 9388
    },
    {
      "epoch": 17.872563005230624,
      "grad_norm": 0.06866225600242615,
      "learning_rate": 4.040012702445221e-05,
      "loss": 0.0599,
      "step": 9389
    },
    {
      "epoch": 17.874465049928673,
      "grad_norm": 0.09405122697353363,
      "learning_rate": 4.039377580184185e-05,
      "loss": 0.0536,
      "step": 9390
    },
    {
      "epoch": 17.876367094626723,
      "grad_norm": 0.09114693105220795,
      "learning_rate": 4.0387424579231505e-05,
      "loss": 0.045,
      "step": 9391
    },
    {
      "epoch": 17.878269139324775,
      "grad_norm": 0.12476793676614761,
      "learning_rate": 4.038107335662115e-05,
      "loss": 0.0473,
      "step": 9392
    },
    {
      "epoch": 17.880171184022824,
      "grad_norm": 0.14439664781093597,
      "learning_rate": 4.03747221340108e-05,
      "loss": 0.0608,
      "step": 9393
    },
    {
      "epoch": 17.882073228720873,
      "grad_norm": 0.09330099076032639,
      "learning_rate": 4.036837091140045e-05,
      "loss": 0.0676,
      "step": 9394
    },
    {
      "epoch": 17.883975273418926,
      "grad_norm": 0.10382474213838577,
      "learning_rate": 4.036201968879009e-05,
      "loss": 0.049,
      "step": 9395
    },
    {
      "epoch": 17.885877318116975,
      "grad_norm": 0.22255682945251465,
      "learning_rate": 4.0355668466179744e-05,
      "loss": 0.0505,
      "step": 9396
    },
    {
      "epoch": 17.887779362815024,
      "grad_norm": 0.059081196784973145,
      "learning_rate": 4.034931724356939e-05,
      "loss": 0.0489,
      "step": 9397
    },
    {
      "epoch": 17.889681407513077,
      "grad_norm": 0.11481790244579315,
      "learning_rate": 4.034296602095904e-05,
      "loss": 0.0633,
      "step": 9398
    },
    {
      "epoch": 17.891583452211126,
      "grad_norm": 0.08961722254753113,
      "learning_rate": 4.033661479834868e-05,
      "loss": 0.0376,
      "step": 9399
    },
    {
      "epoch": 17.89348549690918,
      "grad_norm": 0.10771533101797104,
      "learning_rate": 4.033026357573833e-05,
      "loss": 0.0568,
      "step": 9400
    },
    {
      "epoch": 17.895387541607228,
      "grad_norm": 0.1264873445034027,
      "learning_rate": 4.0323912353127976e-05,
      "loss": 0.0677,
      "step": 9401
    },
    {
      "epoch": 17.897289586305277,
      "grad_norm": 0.12841086089611053,
      "learning_rate": 4.031756113051763e-05,
      "loss": 0.0479,
      "step": 9402
    },
    {
      "epoch": 17.89919163100333,
      "grad_norm": 0.09496669471263885,
      "learning_rate": 4.031120990790727e-05,
      "loss": 0.0567,
      "step": 9403
    },
    {
      "epoch": 17.90109367570138,
      "grad_norm": 0.17920030653476715,
      "learning_rate": 4.030485868529692e-05,
      "loss": 0.0715,
      "step": 9404
    },
    {
      "epoch": 17.902995720399428,
      "grad_norm": 0.1478681117296219,
      "learning_rate": 4.029850746268657e-05,
      "loss": 0.0614,
      "step": 9405
    },
    {
      "epoch": 17.90489776509748,
      "grad_norm": 0.22791436314582825,
      "learning_rate": 4.0292156240076215e-05,
      "loss": 0.0608,
      "step": 9406
    },
    {
      "epoch": 17.90679980979553,
      "grad_norm": 0.11851348727941513,
      "learning_rate": 4.028580501746587e-05,
      "loss": 0.0701,
      "step": 9407
    },
    {
      "epoch": 17.908701854493582,
      "grad_norm": 0.06529906392097473,
      "learning_rate": 4.027945379485551e-05,
      "loss": 0.0572,
      "step": 9408
    },
    {
      "epoch": 17.91060389919163,
      "grad_norm": 0.095278300344944,
      "learning_rate": 4.027310257224516e-05,
      "loss": 0.0657,
      "step": 9409
    },
    {
      "epoch": 17.91250594388968,
      "grad_norm": 0.12427160888910294,
      "learning_rate": 4.026675134963481e-05,
      "loss": 0.0339,
      "step": 9410
    },
    {
      "epoch": 17.914407988587733,
      "grad_norm": 0.23034276068210602,
      "learning_rate": 4.0260400127024454e-05,
      "loss": 0.0669,
      "step": 9411
    },
    {
      "epoch": 17.916310033285782,
      "grad_norm": 0.2623104155063629,
      "learning_rate": 4.02540489044141e-05,
      "loss": 0.0885,
      "step": 9412
    },
    {
      "epoch": 17.91821207798383,
      "grad_norm": 0.13643351197242737,
      "learning_rate": 4.024769768180375e-05,
      "loss": 0.0639,
      "step": 9413
    },
    {
      "epoch": 17.920114122681884,
      "grad_norm": 0.20017459988594055,
      "learning_rate": 4.0241346459193396e-05,
      "loss": 0.0479,
      "step": 9414
    },
    {
      "epoch": 17.922016167379933,
      "grad_norm": 0.08508337289094925,
      "learning_rate": 4.023499523658305e-05,
      "loss": 0.0467,
      "step": 9415
    },
    {
      "epoch": 17.923918212077982,
      "grad_norm": 0.09643582999706268,
      "learning_rate": 4.022864401397269e-05,
      "loss": 0.0551,
      "step": 9416
    },
    {
      "epoch": 17.925820256776035,
      "grad_norm": 0.28682664036750793,
      "learning_rate": 4.022229279136234e-05,
      "loss": 0.0963,
      "step": 9417
    },
    {
      "epoch": 17.927722301474084,
      "grad_norm": 0.14447623491287231,
      "learning_rate": 4.021594156875198e-05,
      "loss": 0.0735,
      "step": 9418
    },
    {
      "epoch": 17.929624346172137,
      "grad_norm": 0.04629140719771385,
      "learning_rate": 4.0209590346141635e-05,
      "loss": 0.0339,
      "step": 9419
    },
    {
      "epoch": 17.931526390870186,
      "grad_norm": 0.18401731550693512,
      "learning_rate": 4.020323912353129e-05,
      "loss": 0.0519,
      "step": 9420
    },
    {
      "epoch": 17.933428435568235,
      "grad_norm": 0.14074082672595978,
      "learning_rate": 4.0196887900920925e-05,
      "loss": 0.079,
      "step": 9421
    },
    {
      "epoch": 17.935330480266288,
      "grad_norm": 0.071122907102108,
      "learning_rate": 4.019053667831058e-05,
      "loss": 0.0601,
      "step": 9422
    },
    {
      "epoch": 17.937232524964337,
      "grad_norm": 0.20074374973773956,
      "learning_rate": 4.018418545570022e-05,
      "loss": 0.0644,
      "step": 9423
    },
    {
      "epoch": 17.939134569662386,
      "grad_norm": 0.17692893743515015,
      "learning_rate": 4.0177834233089874e-05,
      "loss": 0.0552,
      "step": 9424
    },
    {
      "epoch": 17.94103661436044,
      "grad_norm": 0.2404235452413559,
      "learning_rate": 4.017148301047952e-05,
      "loss": 0.0498,
      "step": 9425
    },
    {
      "epoch": 17.942938659058488,
      "grad_norm": 0.19352063536643982,
      "learning_rate": 4.0165131787869164e-05,
      "loss": 0.056,
      "step": 9426
    },
    {
      "epoch": 17.944840703756537,
      "grad_norm": 0.17906826734542847,
      "learning_rate": 4.0158780565258816e-05,
      "loss": 0.0584,
      "step": 9427
    },
    {
      "epoch": 17.94674274845459,
      "grad_norm": 0.06080584228038788,
      "learning_rate": 4.015242934264846e-05,
      "loss": 0.0345,
      "step": 9428
    },
    {
      "epoch": 17.94864479315264,
      "grad_norm": 0.1453639268875122,
      "learning_rate": 4.014607812003811e-05,
      "loss": 0.0718,
      "step": 9429
    },
    {
      "epoch": 17.95054683785069,
      "grad_norm": 0.10231205821037292,
      "learning_rate": 4.013972689742775e-05,
      "loss": 0.0571,
      "step": 9430
    },
    {
      "epoch": 17.95244888254874,
      "grad_norm": 0.11176486313343048,
      "learning_rate": 4.01333756748174e-05,
      "loss": 0.0518,
      "step": 9431
    },
    {
      "epoch": 17.95435092724679,
      "grad_norm": 0.2039271742105484,
      "learning_rate": 4.0127024452207055e-05,
      "loss": 0.0782,
      "step": 9432
    },
    {
      "epoch": 17.956252971944842,
      "grad_norm": 0.08733384311199188,
      "learning_rate": 4.01206732295967e-05,
      "loss": 0.0624,
      "step": 9433
    },
    {
      "epoch": 17.95815501664289,
      "grad_norm": 0.13305670022964478,
      "learning_rate": 4.011432200698635e-05,
      "loss": 0.0678,
      "step": 9434
    },
    {
      "epoch": 17.96005706134094,
      "grad_norm": 0.2694123685359955,
      "learning_rate": 4.010797078437599e-05,
      "loss": 0.0662,
      "step": 9435
    },
    {
      "epoch": 17.961959106038993,
      "grad_norm": 0.16008444130420685,
      "learning_rate": 4.010161956176564e-05,
      "loss": 0.0668,
      "step": 9436
    },
    {
      "epoch": 17.963861150737042,
      "grad_norm": 0.09083766490221024,
      "learning_rate": 4.009526833915529e-05,
      "loss": 0.0699,
      "step": 9437
    },
    {
      "epoch": 17.96576319543509,
      "grad_norm": 0.07772954553365707,
      "learning_rate": 4.008891711654494e-05,
      "loss": 0.0796,
      "step": 9438
    },
    {
      "epoch": 17.967665240133144,
      "grad_norm": 0.14468494057655334,
      "learning_rate": 4.0082565893934584e-05,
      "loss": 0.0389,
      "step": 9439
    },
    {
      "epoch": 17.969567284831193,
      "grad_norm": 0.20007829368114471,
      "learning_rate": 4.007621467132423e-05,
      "loss": 0.0711,
      "step": 9440
    },
    {
      "epoch": 17.971469329529246,
      "grad_norm": 0.14036419987678528,
      "learning_rate": 4.006986344871388e-05,
      "loss": 0.0542,
      "step": 9441
    },
    {
      "epoch": 17.973371374227295,
      "grad_norm": 0.13243912160396576,
      "learning_rate": 4.0063512226103526e-05,
      "loss": 0.0605,
      "step": 9442
    },
    {
      "epoch": 17.975273418925344,
      "grad_norm": 0.17457394301891327,
      "learning_rate": 4.005716100349318e-05,
      "loss": 0.048,
      "step": 9443
    },
    {
      "epoch": 17.977175463623396,
      "grad_norm": 0.14571994543075562,
      "learning_rate": 4.005080978088282e-05,
      "loss": 0.0397,
      "step": 9444
    },
    {
      "epoch": 17.979077508321446,
      "grad_norm": 0.1325615793466568,
      "learning_rate": 4.004445855827247e-05,
      "loss": 0.0579,
      "step": 9445
    },
    {
      "epoch": 17.980979553019495,
      "grad_norm": 0.129814013838768,
      "learning_rate": 4.003810733566212e-05,
      "loss": 0.074,
      "step": 9446
    },
    {
      "epoch": 17.982881597717547,
      "grad_norm": 0.11676670610904694,
      "learning_rate": 4.0031756113051764e-05,
      "loss": 0.0544,
      "step": 9447
    },
    {
      "epoch": 17.984783642415596,
      "grad_norm": 0.09288892149925232,
      "learning_rate": 4.002540489044141e-05,
      "loss": 0.0406,
      "step": 9448
    },
    {
      "epoch": 17.986685687113646,
      "grad_norm": 0.15818986296653748,
      "learning_rate": 4.001905366783106e-05,
      "loss": 0.0514,
      "step": 9449
    },
    {
      "epoch": 17.9885877318117,
      "grad_norm": 0.17206180095672607,
      "learning_rate": 4.0012702445220706e-05,
      "loss": 0.1,
      "step": 9450
    },
    {
      "epoch": 17.990489776509747,
      "grad_norm": 0.16787725687026978,
      "learning_rate": 4.000635122261036e-05,
      "loss": 0.0557,
      "step": 9451
    },
    {
      "epoch": 17.9923918212078,
      "grad_norm": 0.18895794451236725,
      "learning_rate": 4e-05,
      "loss": 0.0568,
      "step": 9452
    },
    {
      "epoch": 17.99429386590585,
      "grad_norm": 0.21941068768501282,
      "learning_rate": 3.999364877738965e-05,
      "loss": 0.0464,
      "step": 9453
    },
    {
      "epoch": 17.9961959106039,
      "grad_norm": 0.16816534101963043,
      "learning_rate": 3.9987297554779293e-05,
      "loss": 0.046,
      "step": 9454
    },
    {
      "epoch": 17.99809795530195,
      "grad_norm": 0.16162028908729553,
      "learning_rate": 3.9980946332168945e-05,
      "loss": 0.0777,
      "step": 9455
    },
    {
      "epoch": 18.0,
      "grad_norm": 0.18661636114120483,
      "learning_rate": 3.997459510955859e-05,
      "loss": 0.0686,
      "step": 9456
    },
    {
      "epoch": 18.00190204469805,
      "grad_norm": 0.05548053979873657,
      "learning_rate": 3.9968243886948235e-05,
      "loss": 0.0855,
      "step": 9457
    },
    {
      "epoch": 18.0038040893961,
      "grad_norm": 0.08743029087781906,
      "learning_rate": 3.996189266433789e-05,
      "loss": 0.0389,
      "step": 9458
    },
    {
      "epoch": 18.00570613409415,
      "grad_norm": 0.08235698193311691,
      "learning_rate": 3.995554144172753e-05,
      "loss": 0.074,
      "step": 9459
    },
    {
      "epoch": 18.0076081787922,
      "grad_norm": 0.07210565358400345,
      "learning_rate": 3.9949190219117184e-05,
      "loss": 0.0665,
      "step": 9460
    },
    {
      "epoch": 18.009510223490253,
      "grad_norm": 0.08912031352519989,
      "learning_rate": 3.994283899650683e-05,
      "loss": 0.0453,
      "step": 9461
    },
    {
      "epoch": 18.0114122681883,
      "grad_norm": 0.05706711858510971,
      "learning_rate": 3.9936487773896474e-05,
      "loss": 0.044,
      "step": 9462
    },
    {
      "epoch": 18.013314312886354,
      "grad_norm": 0.129007026553154,
      "learning_rate": 3.9930136551286126e-05,
      "loss": 0.052,
      "step": 9463
    },
    {
      "epoch": 18.015216357584404,
      "grad_norm": 0.043063823133707047,
      "learning_rate": 3.992378532867577e-05,
      "loss": 0.0264,
      "step": 9464
    },
    {
      "epoch": 18.017118402282453,
      "grad_norm": 0.12655483186244965,
      "learning_rate": 3.991743410606542e-05,
      "loss": 0.0494,
      "step": 9465
    },
    {
      "epoch": 18.019020446980505,
      "grad_norm": 0.0626833587884903,
      "learning_rate": 3.991108288345506e-05,
      "loss": 0.0386,
      "step": 9466
    },
    {
      "epoch": 18.020922491678554,
      "grad_norm": 0.10463612526655197,
      "learning_rate": 3.990473166084471e-05,
      "loss": 0.0736,
      "step": 9467
    },
    {
      "epoch": 18.022824536376604,
      "grad_norm": 0.1362827867269516,
      "learning_rate": 3.9898380438234365e-05,
      "loss": 0.0405,
      "step": 9468
    },
    {
      "epoch": 18.024726581074656,
      "grad_norm": 0.07904388010501862,
      "learning_rate": 3.989202921562401e-05,
      "loss": 0.0388,
      "step": 9469
    },
    {
      "epoch": 18.026628625772705,
      "grad_norm": 0.09696925431489944,
      "learning_rate": 3.988567799301366e-05,
      "loss": 0.0374,
      "step": 9470
    },
    {
      "epoch": 18.028530670470754,
      "grad_norm": 0.08310946077108383,
      "learning_rate": 3.98793267704033e-05,
      "loss": 0.0488,
      "step": 9471
    },
    {
      "epoch": 18.030432715168807,
      "grad_norm": 0.06327807903289795,
      "learning_rate": 3.987297554779295e-05,
      "loss": 0.0656,
      "step": 9472
    },
    {
      "epoch": 18.032334759866856,
      "grad_norm": 0.07315284013748169,
      "learning_rate": 3.98666243251826e-05,
      "loss": 0.0439,
      "step": 9473
    },
    {
      "epoch": 18.03423680456491,
      "grad_norm": 0.07850883901119232,
      "learning_rate": 3.986027310257225e-05,
      "loss": 0.0356,
      "step": 9474
    },
    {
      "epoch": 18.036138849262958,
      "grad_norm": 0.03636496514081955,
      "learning_rate": 3.9853921879961894e-05,
      "loss": 0.0411,
      "step": 9475
    },
    {
      "epoch": 18.038040893961007,
      "grad_norm": 0.11127316951751709,
      "learning_rate": 3.984757065735154e-05,
      "loss": 0.0594,
      "step": 9476
    },
    {
      "epoch": 18.03994293865906,
      "grad_norm": 0.15661272406578064,
      "learning_rate": 3.984121943474119e-05,
      "loss": 0.0472,
      "step": 9477
    },
    {
      "epoch": 18.04184498335711,
      "grad_norm": 0.08048074692487717,
      "learning_rate": 3.9834868212130836e-05,
      "loss": 0.0471,
      "step": 9478
    },
    {
      "epoch": 18.043747028055158,
      "grad_norm": 0.0691787526011467,
      "learning_rate": 3.982851698952049e-05,
      "loss": 0.0572,
      "step": 9479
    },
    {
      "epoch": 18.04564907275321,
      "grad_norm": 0.10766416043043137,
      "learning_rate": 3.982216576691013e-05,
      "loss": 0.0494,
      "step": 9480
    },
    {
      "epoch": 18.04755111745126,
      "grad_norm": 0.11665042489767075,
      "learning_rate": 3.981581454429978e-05,
      "loss": 0.0449,
      "step": 9481
    },
    {
      "epoch": 18.04945316214931,
      "grad_norm": 0.1952570229768753,
      "learning_rate": 3.980946332168943e-05,
      "loss": 0.0646,
      "step": 9482
    },
    {
      "epoch": 18.05135520684736,
      "grad_norm": 0.15641002357006073,
      "learning_rate": 3.9803112099079075e-05,
      "loss": 0.0513,
      "step": 9483
    },
    {
      "epoch": 18.05325725154541,
      "grad_norm": 0.08670450747013092,
      "learning_rate": 3.979676087646872e-05,
      "loss": 0.0517,
      "step": 9484
    },
    {
      "epoch": 18.055159296243463,
      "grad_norm": 0.11871020495891571,
      "learning_rate": 3.9790409653858365e-05,
      "loss": 0.052,
      "step": 9485
    },
    {
      "epoch": 18.057061340941512,
      "grad_norm": 0.05478886887431145,
      "learning_rate": 3.978405843124802e-05,
      "loss": 0.0514,
      "step": 9486
    },
    {
      "epoch": 18.05896338563956,
      "grad_norm": 0.15602456033229828,
      "learning_rate": 3.977770720863767e-05,
      "loss": 0.0646,
      "step": 9487
    },
    {
      "epoch": 18.060865430337614,
      "grad_norm": 0.20108024775981903,
      "learning_rate": 3.9771355986027314e-05,
      "loss": 0.073,
      "step": 9488
    },
    {
      "epoch": 18.062767475035663,
      "grad_norm": 0.23799696564674377,
      "learning_rate": 3.976500476341696e-05,
      "loss": 0.0526,
      "step": 9489
    },
    {
      "epoch": 18.064669519733712,
      "grad_norm": 0.11661495268344879,
      "learning_rate": 3.9758653540806604e-05,
      "loss": 0.0729,
      "step": 9490
    },
    {
      "epoch": 18.066571564431765,
      "grad_norm": 0.06901799887418747,
      "learning_rate": 3.9752302318196256e-05,
      "loss": 0.0471,
      "step": 9491
    },
    {
      "epoch": 18.068473609129814,
      "grad_norm": 0.05189414322376251,
      "learning_rate": 3.97459510955859e-05,
      "loss": 0.0488,
      "step": 9492
    },
    {
      "epoch": 18.070375653827863,
      "grad_norm": 0.06394827365875244,
      "learning_rate": 3.9739599872975546e-05,
      "loss": 0.0549,
      "step": 9493
    },
    {
      "epoch": 18.072277698525916,
      "grad_norm": 0.057227928191423416,
      "learning_rate": 3.97332486503652e-05,
      "loss": 0.0508,
      "step": 9494
    },
    {
      "epoch": 18.074179743223965,
      "grad_norm": 0.09823498874902725,
      "learning_rate": 3.972689742775484e-05,
      "loss": 0.0394,
      "step": 9495
    },
    {
      "epoch": 18.076081787922018,
      "grad_norm": 0.1826847344636917,
      "learning_rate": 3.9720546205144495e-05,
      "loss": 0.0597,
      "step": 9496
    },
    {
      "epoch": 18.077983832620067,
      "grad_norm": 0.08450349420309067,
      "learning_rate": 3.971419498253414e-05,
      "loss": 0.0665,
      "step": 9497
    },
    {
      "epoch": 18.079885877318116,
      "grad_norm": 0.05135255306959152,
      "learning_rate": 3.9707843759923785e-05,
      "loss": 0.0418,
      "step": 9498
    },
    {
      "epoch": 18.08178792201617,
      "grad_norm": 0.13653814792633057,
      "learning_rate": 3.9701492537313437e-05,
      "loss": 0.0554,
      "step": 9499
    },
    {
      "epoch": 18.083689966714218,
      "grad_norm": 0.21696190536022186,
      "learning_rate": 3.969514131470308e-05,
      "loss": 0.053,
      "step": 9500
    },
    {
      "epoch": 18.085592011412267,
      "grad_norm": 0.2852039933204651,
      "learning_rate": 3.9688790092092733e-05,
      "loss": 0.0645,
      "step": 9501
    },
    {
      "epoch": 18.08749405611032,
      "grad_norm": 0.2527969479560852,
      "learning_rate": 3.968243886948237e-05,
      "loss": 0.0782,
      "step": 9502
    },
    {
      "epoch": 18.08939610080837,
      "grad_norm": 0.10853327065706253,
      "learning_rate": 3.9676087646872024e-05,
      "loss": 0.0895,
      "step": 9503
    },
    {
      "epoch": 18.091298145506418,
      "grad_norm": 0.08898596465587616,
      "learning_rate": 3.966973642426167e-05,
      "loss": 0.0604,
      "step": 9504
    },
    {
      "epoch": 18.09320019020447,
      "grad_norm": 0.03284698724746704,
      "learning_rate": 3.966338520165132e-05,
      "loss": 0.0402,
      "step": 9505
    },
    {
      "epoch": 18.09510223490252,
      "grad_norm": 0.0882166400551796,
      "learning_rate": 3.965703397904097e-05,
      "loss": 0.0717,
      "step": 9506
    },
    {
      "epoch": 18.097004279600572,
      "grad_norm": 0.06083635240793228,
      "learning_rate": 3.965068275643061e-05,
      "loss": 0.0522,
      "step": 9507
    },
    {
      "epoch": 18.09890632429862,
      "grad_norm": 0.07042375952005386,
      "learning_rate": 3.964433153382026e-05,
      "loss": 0.0435,
      "step": 9508
    },
    {
      "epoch": 18.10080836899667,
      "grad_norm": 0.10826339572668076,
      "learning_rate": 3.963798031120991e-05,
      "loss": 0.0635,
      "step": 9509
    },
    {
      "epoch": 18.102710413694723,
      "grad_norm": 0.09245724976062775,
      "learning_rate": 3.963162908859956e-05,
      "loss": 0.0637,
      "step": 9510
    },
    {
      "epoch": 18.104612458392772,
      "grad_norm": 0.21710948646068573,
      "learning_rate": 3.9625277865989204e-05,
      "loss": 0.0586,
      "step": 9511
    },
    {
      "epoch": 18.10651450309082,
      "grad_norm": 0.10763686895370483,
      "learning_rate": 3.961892664337885e-05,
      "loss": 0.0554,
      "step": 9512
    },
    {
      "epoch": 18.108416547788874,
      "grad_norm": 0.05405883491039276,
      "learning_rate": 3.96125754207685e-05,
      "loss": 0.0436,
      "step": 9513
    },
    {
      "epoch": 18.110318592486923,
      "grad_norm": 0.07853370159864426,
      "learning_rate": 3.9606224198158146e-05,
      "loss": 0.042,
      "step": 9514
    },
    {
      "epoch": 18.112220637184972,
      "grad_norm": 0.07172422111034393,
      "learning_rate": 3.95998729755478e-05,
      "loss": 0.0596,
      "step": 9515
    },
    {
      "epoch": 18.114122681883025,
      "grad_norm": 0.11832434684038162,
      "learning_rate": 3.959352175293744e-05,
      "loss": 0.0605,
      "step": 9516
    },
    {
      "epoch": 18.116024726581074,
      "grad_norm": 0.12276314944028854,
      "learning_rate": 3.958717053032709e-05,
      "loss": 0.0341,
      "step": 9517
    },
    {
      "epoch": 18.117926771279127,
      "grad_norm": 0.057973865419626236,
      "learning_rate": 3.958081930771674e-05,
      "loss": 0.0481,
      "step": 9518
    },
    {
      "epoch": 18.119828815977176,
      "grad_norm": 0.12534184753894806,
      "learning_rate": 3.9574468085106385e-05,
      "loss": 0.0603,
      "step": 9519
    },
    {
      "epoch": 18.121730860675225,
      "grad_norm": 0.12781070172786713,
      "learning_rate": 3.956811686249603e-05,
      "loss": 0.0556,
      "step": 9520
    },
    {
      "epoch": 18.123632905373277,
      "grad_norm": 0.2016114741563797,
      "learning_rate": 3.9561765639885675e-05,
      "loss": 0.0567,
      "step": 9521
    },
    {
      "epoch": 18.125534950071327,
      "grad_norm": 0.15887132287025452,
      "learning_rate": 3.955541441727533e-05,
      "loss": 0.0557,
      "step": 9522
    },
    {
      "epoch": 18.127436994769376,
      "grad_norm": 0.12661483883857727,
      "learning_rate": 3.954906319466498e-05,
      "loss": 0.0605,
      "step": 9523
    },
    {
      "epoch": 18.12933903946743,
      "grad_norm": 0.10252156108617783,
      "learning_rate": 3.9542711972054624e-05,
      "loss": 0.0406,
      "step": 9524
    },
    {
      "epoch": 18.131241084165477,
      "grad_norm": 0.10569868236780167,
      "learning_rate": 3.953636074944427e-05,
      "loss": 0.0643,
      "step": 9525
    },
    {
      "epoch": 18.13314312886353,
      "grad_norm": 0.08536829799413681,
      "learning_rate": 3.9530009526833914e-05,
      "loss": 0.0708,
      "step": 9526
    },
    {
      "epoch": 18.13504517356158,
      "grad_norm": 0.14921991527080536,
      "learning_rate": 3.9523658304223566e-05,
      "loss": 0.0765,
      "step": 9527
    },
    {
      "epoch": 18.13694721825963,
      "grad_norm": 0.19710759818553925,
      "learning_rate": 3.951730708161321e-05,
      "loss": 0.0667,
      "step": 9528
    },
    {
      "epoch": 18.13884926295768,
      "grad_norm": 0.08138823509216309,
      "learning_rate": 3.9510955859002856e-05,
      "loss": 0.0736,
      "step": 9529
    },
    {
      "epoch": 18.14075130765573,
      "grad_norm": 0.06059684976935387,
      "learning_rate": 3.950460463639251e-05,
      "loss": 0.0562,
      "step": 9530
    },
    {
      "epoch": 18.14265335235378,
      "grad_norm": 0.13167749345302582,
      "learning_rate": 3.949825341378215e-05,
      "loss": 0.035,
      "step": 9531
    },
    {
      "epoch": 18.144555397051832,
      "grad_norm": 0.07073647528886795,
      "learning_rate": 3.9491902191171805e-05,
      "loss": 0.0455,
      "step": 9532
    },
    {
      "epoch": 18.14645744174988,
      "grad_norm": 0.10322169214487076,
      "learning_rate": 3.948555096856145e-05,
      "loss": 0.0489,
      "step": 9533
    },
    {
      "epoch": 18.14835948644793,
      "grad_norm": 0.07083230465650558,
      "learning_rate": 3.9479199745951095e-05,
      "loss": 0.0558,
      "step": 9534
    },
    {
      "epoch": 18.150261531145983,
      "grad_norm": 0.08989362418651581,
      "learning_rate": 3.947284852334075e-05,
      "loss": 0.0578,
      "step": 9535
    },
    {
      "epoch": 18.152163575844032,
      "grad_norm": 0.1856735646724701,
      "learning_rate": 3.946649730073039e-05,
      "loss": 0.0562,
      "step": 9536
    },
    {
      "epoch": 18.154065620542085,
      "grad_norm": 0.1071576476097107,
      "learning_rate": 3.9460146078120044e-05,
      "loss": 0.0415,
      "step": 9537
    },
    {
      "epoch": 18.155967665240134,
      "grad_norm": 0.18466047942638397,
      "learning_rate": 3.945379485550968e-05,
      "loss": 0.0639,
      "step": 9538
    },
    {
      "epoch": 18.157869709938183,
      "grad_norm": 0.29862919449806213,
      "learning_rate": 3.9447443632899334e-05,
      "loss": 0.0697,
      "step": 9539
    },
    {
      "epoch": 18.159771754636235,
      "grad_norm": 0.10799165070056915,
      "learning_rate": 3.944109241028898e-05,
      "loss": 0.0372,
      "step": 9540
    },
    {
      "epoch": 18.161673799334284,
      "grad_norm": 0.12296050041913986,
      "learning_rate": 3.943474118767863e-05,
      "loss": 0.0696,
      "step": 9541
    },
    {
      "epoch": 18.163575844032334,
      "grad_norm": 0.19168108701705933,
      "learning_rate": 3.942838996506828e-05,
      "loss": 0.0673,
      "step": 9542
    },
    {
      "epoch": 18.165477888730386,
      "grad_norm": 0.06364916265010834,
      "learning_rate": 3.942203874245792e-05,
      "loss": 0.0381,
      "step": 9543
    },
    {
      "epoch": 18.167379933428435,
      "grad_norm": 0.1005852073431015,
      "learning_rate": 3.941568751984757e-05,
      "loss": 0.0535,
      "step": 9544
    },
    {
      "epoch": 18.169281978126484,
      "grad_norm": 0.10540492832660675,
      "learning_rate": 3.940933629723722e-05,
      "loss": 0.0536,
      "step": 9545
    },
    {
      "epoch": 18.171184022824537,
      "grad_norm": 0.13814863562583923,
      "learning_rate": 3.940298507462687e-05,
      "loss": 0.0619,
      "step": 9546
    },
    {
      "epoch": 18.173086067522586,
      "grad_norm": 0.1335911750793457,
      "learning_rate": 3.9396633852016515e-05,
      "loss": 0.0459,
      "step": 9547
    },
    {
      "epoch": 18.17498811222064,
      "grad_norm": 0.16157962381839752,
      "learning_rate": 3.939028262940616e-05,
      "loss": 0.0548,
      "step": 9548
    },
    {
      "epoch": 18.176890156918688,
      "grad_norm": 0.05575089901685715,
      "learning_rate": 3.938393140679581e-05,
      "loss": 0.0643,
      "step": 9549
    },
    {
      "epoch": 18.178792201616737,
      "grad_norm": 0.10665811598300934,
      "learning_rate": 3.937758018418546e-05,
      "loss": 0.0458,
      "step": 9550
    },
    {
      "epoch": 18.18069424631479,
      "grad_norm": 0.12008770555257797,
      "learning_rate": 3.937122896157511e-05,
      "loss": 0.032,
      "step": 9551
    },
    {
      "epoch": 18.18259629101284,
      "grad_norm": 0.09343229979276657,
      "learning_rate": 3.9364877738964754e-05,
      "loss": 0.0488,
      "step": 9552
    },
    {
      "epoch": 18.184498335710888,
      "grad_norm": 0.08769842237234116,
      "learning_rate": 3.93585265163544e-05,
      "loss": 0.066,
      "step": 9553
    },
    {
      "epoch": 18.18640038040894,
      "grad_norm": 0.07487979531288147,
      "learning_rate": 3.935217529374405e-05,
      "loss": 0.0614,
      "step": 9554
    },
    {
      "epoch": 18.18830242510699,
      "grad_norm": 0.15553294122219086,
      "learning_rate": 3.9345824071133696e-05,
      "loss": 0.0797,
      "step": 9555
    },
    {
      "epoch": 18.19020446980504,
      "grad_norm": 0.19011497497558594,
      "learning_rate": 3.933947284852334e-05,
      "loss": 0.0698,
      "step": 9556
    },
    {
      "epoch": 18.19210651450309,
      "grad_norm": 0.1535508632659912,
      "learning_rate": 3.9333121625912986e-05,
      "loss": 0.0686,
      "step": 9557
    },
    {
      "epoch": 18.19400855920114,
      "grad_norm": 0.18773606419563293,
      "learning_rate": 3.932677040330264e-05,
      "loss": 0.0541,
      "step": 9558
    },
    {
      "epoch": 18.195910603899193,
      "grad_norm": 0.06210647150874138,
      "learning_rate": 3.932041918069228e-05,
      "loss": 0.0367,
      "step": 9559
    },
    {
      "epoch": 18.197812648597242,
      "grad_norm": 0.09794329106807709,
      "learning_rate": 3.9314067958081935e-05,
      "loss": 0.06,
      "step": 9560
    },
    {
      "epoch": 18.19971469329529,
      "grad_norm": 0.0627339705824852,
      "learning_rate": 3.930771673547158e-05,
      "loss": 0.0619,
      "step": 9561
    },
    {
      "epoch": 18.201616737993344,
      "grad_norm": 0.1042981967329979,
      "learning_rate": 3.9301365512861225e-05,
      "loss": 0.0616,
      "step": 9562
    },
    {
      "epoch": 18.203518782691393,
      "grad_norm": 0.05875473469495773,
      "learning_rate": 3.9295014290250877e-05,
      "loss": 0.0597,
      "step": 9563
    },
    {
      "epoch": 18.205420827389442,
      "grad_norm": 0.08454211801290512,
      "learning_rate": 3.928866306764052e-05,
      "loss": 0.0553,
      "step": 9564
    },
    {
      "epoch": 18.207322872087495,
      "grad_norm": 0.052836548537015915,
      "learning_rate": 3.928231184503017e-05,
      "loss": 0.0726,
      "step": 9565
    },
    {
      "epoch": 18.209224916785544,
      "grad_norm": 0.15156498551368713,
      "learning_rate": 3.927596062241982e-05,
      "loss": 0.0887,
      "step": 9566
    },
    {
      "epoch": 18.211126961483593,
      "grad_norm": 0.11899285763502121,
      "learning_rate": 3.9269609399809464e-05,
      "loss": 0.0536,
      "step": 9567
    },
    {
      "epoch": 18.213029006181646,
      "grad_norm": 0.048953719437122345,
      "learning_rate": 3.9263258177199115e-05,
      "loss": 0.0678,
      "step": 9568
    },
    {
      "epoch": 18.214931050879695,
      "grad_norm": 0.1252700537443161,
      "learning_rate": 3.925690695458876e-05,
      "loss": 0.0574,
      "step": 9569
    },
    {
      "epoch": 18.216833095577748,
      "grad_norm": 0.06751298904418945,
      "learning_rate": 3.9250555731978406e-05,
      "loss": 0.0457,
      "step": 9570
    },
    {
      "epoch": 18.218735140275797,
      "grad_norm": 0.0774497240781784,
      "learning_rate": 3.924420450936806e-05,
      "loss": 0.0438,
      "step": 9571
    },
    {
      "epoch": 18.220637184973846,
      "grad_norm": 0.1059066578745842,
      "learning_rate": 3.92378532867577e-05,
      "loss": 0.0509,
      "step": 9572
    },
    {
      "epoch": 18.2225392296719,
      "grad_norm": 0.04546501487493515,
      "learning_rate": 3.9231502064147354e-05,
      "loss": 0.0409,
      "step": 9573
    },
    {
      "epoch": 18.224441274369948,
      "grad_norm": 0.07507932186126709,
      "learning_rate": 3.922515084153699e-05,
      "loss": 0.0461,
      "step": 9574
    },
    {
      "epoch": 18.226343319067997,
      "grad_norm": 0.06796334683895111,
      "learning_rate": 3.9218799618926644e-05,
      "loss": 0.0583,
      "step": 9575
    },
    {
      "epoch": 18.22824536376605,
      "grad_norm": 0.09650825709104538,
      "learning_rate": 3.921244839631629e-05,
      "loss": 0.0627,
      "step": 9576
    },
    {
      "epoch": 18.2301474084641,
      "grad_norm": 0.04558471590280533,
      "learning_rate": 3.920609717370594e-05,
      "loss": 0.047,
      "step": 9577
    },
    {
      "epoch": 18.232049453162148,
      "grad_norm": 0.09572546929121017,
      "learning_rate": 3.919974595109559e-05,
      "loss": 0.0425,
      "step": 9578
    },
    {
      "epoch": 18.2339514978602,
      "grad_norm": 0.14911305904388428,
      "learning_rate": 3.919339472848523e-05,
      "loss": 0.0578,
      "step": 9579
    },
    {
      "epoch": 18.23585354255825,
      "grad_norm": 0.08339757472276688,
      "learning_rate": 3.918704350587488e-05,
      "loss": 0.101,
      "step": 9580
    },
    {
      "epoch": 18.237755587256302,
      "grad_norm": 0.0699375569820404,
      "learning_rate": 3.918069228326453e-05,
      "loss": 0.0612,
      "step": 9581
    },
    {
      "epoch": 18.23965763195435,
      "grad_norm": 0.1783619374036789,
      "learning_rate": 3.917434106065418e-05,
      "loss": 0.0563,
      "step": 9582
    },
    {
      "epoch": 18.2415596766524,
      "grad_norm": 0.06447594612836838,
      "learning_rate": 3.9167989838043825e-05,
      "loss": 0.0614,
      "step": 9583
    },
    {
      "epoch": 18.243461721350453,
      "grad_norm": 0.06985105574131012,
      "learning_rate": 3.916163861543347e-05,
      "loss": 0.0364,
      "step": 9584
    },
    {
      "epoch": 18.245363766048502,
      "grad_norm": 0.1365368813276291,
      "learning_rate": 3.915528739282312e-05,
      "loss": 0.0699,
      "step": 9585
    },
    {
      "epoch": 18.24726581074655,
      "grad_norm": 0.07527942955493927,
      "learning_rate": 3.914893617021277e-05,
      "loss": 0.0385,
      "step": 9586
    },
    {
      "epoch": 18.249167855444604,
      "grad_norm": 0.06135424226522446,
      "learning_rate": 3.914258494760242e-05,
      "loss": 0.0781,
      "step": 9587
    },
    {
      "epoch": 18.251069900142653,
      "grad_norm": 0.04152238368988037,
      "learning_rate": 3.913623372499206e-05,
      "loss": 0.066,
      "step": 9588
    },
    {
      "epoch": 18.252971944840702,
      "grad_norm": 0.12861324846744537,
      "learning_rate": 3.912988250238171e-05,
      "loss": 0.0587,
      "step": 9589
    },
    {
      "epoch": 18.254873989538755,
      "grad_norm": 0.17604462802410126,
      "learning_rate": 3.912353127977136e-05,
      "loss": 0.0682,
      "step": 9590
    },
    {
      "epoch": 18.256776034236804,
      "grad_norm": 0.04712732136249542,
      "learning_rate": 3.9117180057161006e-05,
      "loss": 0.0481,
      "step": 9591
    },
    {
      "epoch": 18.258678078934857,
      "grad_norm": 0.12864311039447784,
      "learning_rate": 3.911082883455065e-05,
      "loss": 0.0647,
      "step": 9592
    },
    {
      "epoch": 18.260580123632906,
      "grad_norm": 0.08881017565727234,
      "learning_rate": 3.9104477611940296e-05,
      "loss": 0.0716,
      "step": 9593
    },
    {
      "epoch": 18.262482168330955,
      "grad_norm": 0.21135424077510834,
      "learning_rate": 3.909812638932995e-05,
      "loss": 0.0576,
      "step": 9594
    },
    {
      "epoch": 18.264384213029007,
      "grad_norm": 0.05159885808825493,
      "learning_rate": 3.909177516671959e-05,
      "loss": 0.0697,
      "step": 9595
    },
    {
      "epoch": 18.266286257727057,
      "grad_norm": 0.05154569074511528,
      "learning_rate": 3.9085423944109245e-05,
      "loss": 0.0645,
      "step": 9596
    },
    {
      "epoch": 18.268188302425106,
      "grad_norm": 0.1738889068365097,
      "learning_rate": 3.907907272149889e-05,
      "loss": 0.0484,
      "step": 9597
    },
    {
      "epoch": 18.27009034712316,
      "grad_norm": 0.11774470657110214,
      "learning_rate": 3.9072721498888535e-05,
      "loss": 0.0434,
      "step": 9598
    },
    {
      "epoch": 18.271992391821207,
      "grad_norm": 0.06924553215503693,
      "learning_rate": 3.906637027627819e-05,
      "loss": 0.0341,
      "step": 9599
    },
    {
      "epoch": 18.273894436519257,
      "grad_norm": 0.15278260409832,
      "learning_rate": 3.906001905366783e-05,
      "loss": 0.0477,
      "step": 9600
    },
    {
      "epoch": 18.27579648121731,
      "grad_norm": 0.036438386887311935,
      "learning_rate": 3.905366783105748e-05,
      "loss": 0.0452,
      "step": 9601
    },
    {
      "epoch": 18.27769852591536,
      "grad_norm": 0.04843940585851669,
      "learning_rate": 3.904731660844713e-05,
      "loss": 0.0548,
      "step": 9602
    },
    {
      "epoch": 18.27960057061341,
      "grad_norm": 0.0953247919678688,
      "learning_rate": 3.9040965385836774e-05,
      "loss": 0.0593,
      "step": 9603
    },
    {
      "epoch": 18.28150261531146,
      "grad_norm": 0.04957016557455063,
      "learning_rate": 3.9034614163226426e-05,
      "loss": 0.0571,
      "step": 9604
    },
    {
      "epoch": 18.28340466000951,
      "grad_norm": 0.07393553107976913,
      "learning_rate": 3.902826294061607e-05,
      "loss": 0.0408,
      "step": 9605
    },
    {
      "epoch": 18.285306704707562,
      "grad_norm": 0.12568874657154083,
      "learning_rate": 3.9021911718005716e-05,
      "loss": 0.0572,
      "step": 9606
    },
    {
      "epoch": 18.28720874940561,
      "grad_norm": 0.19227972626686096,
      "learning_rate": 3.901556049539537e-05,
      "loss": 0.0561,
      "step": 9607
    },
    {
      "epoch": 18.28911079410366,
      "grad_norm": 0.1546197235584259,
      "learning_rate": 3.900920927278501e-05,
      "loss": 0.0544,
      "step": 9608
    },
    {
      "epoch": 18.291012838801713,
      "grad_norm": 0.0979616641998291,
      "learning_rate": 3.9002858050174665e-05,
      "loss": 0.0687,
      "step": 9609
    },
    {
      "epoch": 18.292914883499762,
      "grad_norm": 0.1661892682313919,
      "learning_rate": 3.89965068275643e-05,
      "loss": 0.0571,
      "step": 9610
    },
    {
      "epoch": 18.29481692819781,
      "grad_norm": 0.051106564700603485,
      "learning_rate": 3.8990155604953955e-05,
      "loss": 0.0521,
      "step": 9611
    },
    {
      "epoch": 18.296718972895864,
      "grad_norm": 0.0543074905872345,
      "learning_rate": 3.89838043823436e-05,
      "loss": 0.0588,
      "step": 9612
    },
    {
      "epoch": 18.298621017593913,
      "grad_norm": 0.09150418639183044,
      "learning_rate": 3.897745315973325e-05,
      "loss": 0.0831,
      "step": 9613
    },
    {
      "epoch": 18.300523062291965,
      "grad_norm": 0.08435828238725662,
      "learning_rate": 3.89711019371229e-05,
      "loss": 0.06,
      "step": 9614
    },
    {
      "epoch": 18.302425106990015,
      "grad_norm": 0.07140857726335526,
      "learning_rate": 3.896475071451254e-05,
      "loss": 0.0492,
      "step": 9615
    },
    {
      "epoch": 18.304327151688064,
      "grad_norm": 0.1313878744840622,
      "learning_rate": 3.8958399491902194e-05,
      "loss": 0.0446,
      "step": 9616
    },
    {
      "epoch": 18.306229196386116,
      "grad_norm": 0.16966655850410461,
      "learning_rate": 3.895204826929184e-05,
      "loss": 0.0497,
      "step": 9617
    },
    {
      "epoch": 18.308131241084165,
      "grad_norm": 0.05971246212720871,
      "learning_rate": 3.894569704668149e-05,
      "loss": 0.0466,
      "step": 9618
    },
    {
      "epoch": 18.310033285782215,
      "grad_norm": 0.09588700532913208,
      "learning_rate": 3.8939345824071136e-05,
      "loss": 0.0595,
      "step": 9619
    },
    {
      "epoch": 18.311935330480267,
      "grad_norm": 0.15829181671142578,
      "learning_rate": 3.893299460146078e-05,
      "loss": 0.0684,
      "step": 9620
    },
    {
      "epoch": 18.313837375178316,
      "grad_norm": 0.061553433537483215,
      "learning_rate": 3.892664337885043e-05,
      "loss": 0.0504,
      "step": 9621
    },
    {
      "epoch": 18.315739419876365,
      "grad_norm": 0.1587333083152771,
      "learning_rate": 3.892029215624008e-05,
      "loss": 0.0598,
      "step": 9622
    },
    {
      "epoch": 18.317641464574418,
      "grad_norm": 0.11620922386646271,
      "learning_rate": 3.891394093362973e-05,
      "loss": 0.0547,
      "step": 9623
    },
    {
      "epoch": 18.319543509272467,
      "grad_norm": 0.0708557590842247,
      "learning_rate": 3.890758971101937e-05,
      "loss": 0.0519,
      "step": 9624
    },
    {
      "epoch": 18.32144555397052,
      "grad_norm": 0.10213367640972137,
      "learning_rate": 3.890123848840902e-05,
      "loss": 0.0543,
      "step": 9625
    },
    {
      "epoch": 18.32334759866857,
      "grad_norm": 0.06195381283760071,
      "learning_rate": 3.889488726579867e-05,
      "loss": 0.055,
      "step": 9626
    },
    {
      "epoch": 18.325249643366618,
      "grad_norm": 0.06340436637401581,
      "learning_rate": 3.8888536043188317e-05,
      "loss": 0.0644,
      "step": 9627
    },
    {
      "epoch": 18.32715168806467,
      "grad_norm": 0.5210753679275513,
      "learning_rate": 3.888218482057796e-05,
      "loss": 0.0675,
      "step": 9628
    },
    {
      "epoch": 18.32905373276272,
      "grad_norm": 0.12499348074197769,
      "learning_rate": 3.887583359796761e-05,
      "loss": 0.0557,
      "step": 9629
    },
    {
      "epoch": 18.33095577746077,
      "grad_norm": 0.05937238037586212,
      "learning_rate": 3.886948237535726e-05,
      "loss": 0.06,
      "step": 9630
    },
    {
      "epoch": 18.33285782215882,
      "grad_norm": 0.07244454324245453,
      "learning_rate": 3.8863131152746904e-05,
      "loss": 0.0454,
      "step": 9631
    },
    {
      "epoch": 18.33475986685687,
      "grad_norm": 0.16597986221313477,
      "learning_rate": 3.8856779930136555e-05,
      "loss": 0.0826,
      "step": 9632
    },
    {
      "epoch": 18.33666191155492,
      "grad_norm": 0.09148790687322617,
      "learning_rate": 3.88504287075262e-05,
      "loss": 0.0522,
      "step": 9633
    },
    {
      "epoch": 18.338563956252973,
      "grad_norm": 0.05849875882267952,
      "learning_rate": 3.8844077484915846e-05,
      "loss": 0.051,
      "step": 9634
    },
    {
      "epoch": 18.34046600095102,
      "grad_norm": 0.044688720256090164,
      "learning_rate": 3.88377262623055e-05,
      "loss": 0.0503,
      "step": 9635
    },
    {
      "epoch": 18.342368045649074,
      "grad_norm": 0.06226931884884834,
      "learning_rate": 3.883137503969514e-05,
      "loss": 0.0483,
      "step": 9636
    },
    {
      "epoch": 18.344270090347123,
      "grad_norm": 0.0717255249619484,
      "learning_rate": 3.882502381708479e-05,
      "loss": 0.0649,
      "step": 9637
    },
    {
      "epoch": 18.346172135045173,
      "grad_norm": 0.07575838267803192,
      "learning_rate": 3.881867259447444e-05,
      "loss": 0.0658,
      "step": 9638
    },
    {
      "epoch": 18.348074179743225,
      "grad_norm": 0.1126398965716362,
      "learning_rate": 3.8812321371864084e-05,
      "loss": 0.0519,
      "step": 9639
    },
    {
      "epoch": 18.349976224441274,
      "grad_norm": 0.11585336923599243,
      "learning_rate": 3.8805970149253736e-05,
      "loss": 0.0546,
      "step": 9640
    },
    {
      "epoch": 18.351878269139323,
      "grad_norm": 0.2785167396068573,
      "learning_rate": 3.879961892664338e-05,
      "loss": 0.0624,
      "step": 9641
    },
    {
      "epoch": 18.353780313837376,
      "grad_norm": 0.21943818032741547,
      "learning_rate": 3.8793267704033026e-05,
      "loss": 0.0512,
      "step": 9642
    },
    {
      "epoch": 18.355682358535425,
      "grad_norm": 0.09890948235988617,
      "learning_rate": 3.878691648142267e-05,
      "loss": 0.0527,
      "step": 9643
    },
    {
      "epoch": 18.357584403233474,
      "grad_norm": 0.08608918637037277,
      "learning_rate": 3.878056525881232e-05,
      "loss": 0.0714,
      "step": 9644
    },
    {
      "epoch": 18.359486447931527,
      "grad_norm": 0.21114416420459747,
      "learning_rate": 3.8774214036201975e-05,
      "loss": 0.0473,
      "step": 9645
    },
    {
      "epoch": 18.361388492629576,
      "grad_norm": 0.12862014770507812,
      "learning_rate": 3.8767862813591614e-05,
      "loss": 0.0524,
      "step": 9646
    },
    {
      "epoch": 18.36329053732763,
      "grad_norm": 0.1056724265217781,
      "learning_rate": 3.8761511590981265e-05,
      "loss": 0.0483,
      "step": 9647
    },
    {
      "epoch": 18.365192582025678,
      "grad_norm": 0.05439826101064682,
      "learning_rate": 3.875516036837091e-05,
      "loss": 0.0332,
      "step": 9648
    },
    {
      "epoch": 18.367094626723727,
      "grad_norm": 0.12853200733661652,
      "learning_rate": 3.874880914576056e-05,
      "loss": 0.0461,
      "step": 9649
    },
    {
      "epoch": 18.36899667142178,
      "grad_norm": 0.10463688522577286,
      "learning_rate": 3.874245792315021e-05,
      "loss": 0.0686,
      "step": 9650
    },
    {
      "epoch": 18.37089871611983,
      "grad_norm": 0.10133697092533112,
      "learning_rate": 3.873610670053985e-05,
      "loss": 0.0452,
      "step": 9651
    },
    {
      "epoch": 18.372800760817878,
      "grad_norm": 0.13786332309246063,
      "learning_rate": 3.8729755477929504e-05,
      "loss": 0.0674,
      "step": 9652
    },
    {
      "epoch": 18.37470280551593,
      "grad_norm": 0.1269512176513672,
      "learning_rate": 3.872340425531915e-05,
      "loss": 0.0563,
      "step": 9653
    },
    {
      "epoch": 18.37660485021398,
      "grad_norm": 0.08988190442323685,
      "learning_rate": 3.87170530327088e-05,
      "loss": 0.0461,
      "step": 9654
    },
    {
      "epoch": 18.37850689491203,
      "grad_norm": 0.29547205567359924,
      "learning_rate": 3.8710701810098446e-05,
      "loss": 0.0721,
      "step": 9655
    },
    {
      "epoch": 18.38040893961008,
      "grad_norm": 0.07171815633773804,
      "learning_rate": 3.870435058748809e-05,
      "loss": 0.0729,
      "step": 9656
    },
    {
      "epoch": 18.38231098430813,
      "grad_norm": 0.10650094598531723,
      "learning_rate": 3.869799936487774e-05,
      "loss": 0.0643,
      "step": 9657
    },
    {
      "epoch": 18.384213029006183,
      "grad_norm": 0.0583612322807312,
      "learning_rate": 3.869164814226739e-05,
      "loss": 0.0441,
      "step": 9658
    },
    {
      "epoch": 18.386115073704232,
      "grad_norm": 0.14970900118350983,
      "learning_rate": 3.868529691965704e-05,
      "loss": 0.0553,
      "step": 9659
    },
    {
      "epoch": 18.38801711840228,
      "grad_norm": 0.07470922917127609,
      "learning_rate": 3.867894569704668e-05,
      "loss": 0.055,
      "step": 9660
    },
    {
      "epoch": 18.389919163100334,
      "grad_norm": 0.06020907685160637,
      "learning_rate": 3.867259447443633e-05,
      "loss": 0.0705,
      "step": 9661
    },
    {
      "epoch": 18.391821207798383,
      "grad_norm": 0.12234506756067276,
      "learning_rate": 3.866624325182598e-05,
      "loss": 0.0523,
      "step": 9662
    },
    {
      "epoch": 18.393723252496432,
      "grad_norm": 0.1236337348818779,
      "learning_rate": 3.865989202921563e-05,
      "loss": 0.0653,
      "step": 9663
    },
    {
      "epoch": 18.395625297194485,
      "grad_norm": 0.12261811643838882,
      "learning_rate": 3.865354080660527e-05,
      "loss": 0.0414,
      "step": 9664
    },
    {
      "epoch": 18.397527341892534,
      "grad_norm": 0.07360794395208359,
      "learning_rate": 3.864718958399492e-05,
      "loss": 0.059,
      "step": 9665
    },
    {
      "epoch": 18.399429386590583,
      "grad_norm": 0.07546956092119217,
      "learning_rate": 3.864083836138457e-05,
      "loss": 0.0534,
      "step": 9666
    },
    {
      "epoch": 18.401331431288636,
      "grad_norm": 0.058266524225473404,
      "learning_rate": 3.8634487138774214e-05,
      "loss": 0.0653,
      "step": 9667
    },
    {
      "epoch": 18.403233475986685,
      "grad_norm": 0.21543626487255096,
      "learning_rate": 3.8628135916163866e-05,
      "loss": 0.0642,
      "step": 9668
    },
    {
      "epoch": 18.405135520684738,
      "grad_norm": 0.04944286495447159,
      "learning_rate": 3.862178469355351e-05,
      "loss": 0.0543,
      "step": 9669
    },
    {
      "epoch": 18.407037565382787,
      "grad_norm": 0.0344243198633194,
      "learning_rate": 3.8615433470943156e-05,
      "loss": 0.0378,
      "step": 9670
    },
    {
      "epoch": 18.408939610080836,
      "grad_norm": 0.04360641539096832,
      "learning_rate": 3.860908224833281e-05,
      "loss": 0.0411,
      "step": 9671
    },
    {
      "epoch": 18.41084165477889,
      "grad_norm": 0.0884445533156395,
      "learning_rate": 3.860273102572245e-05,
      "loss": 0.0583,
      "step": 9672
    },
    {
      "epoch": 18.412743699476938,
      "grad_norm": 0.42718735337257385,
      "learning_rate": 3.85963798031121e-05,
      "loss": 0.0634,
      "step": 9673
    },
    {
      "epoch": 18.414645744174987,
      "grad_norm": 0.18887248635292053,
      "learning_rate": 3.859002858050175e-05,
      "loss": 0.0466,
      "step": 9674
    },
    {
      "epoch": 18.41654778887304,
      "grad_norm": 0.11296916007995605,
      "learning_rate": 3.8583677357891395e-05,
      "loss": 0.0456,
      "step": 9675
    },
    {
      "epoch": 18.41844983357109,
      "grad_norm": 0.07237987220287323,
      "learning_rate": 3.857732613528105e-05,
      "loss": 0.0856,
      "step": 9676
    },
    {
      "epoch": 18.420351878269138,
      "grad_norm": 0.08038794994354248,
      "learning_rate": 3.857097491267069e-05,
      "loss": 0.0928,
      "step": 9677
    },
    {
      "epoch": 18.42225392296719,
      "grad_norm": 0.07001104205846786,
      "learning_rate": 3.856462369006034e-05,
      "loss": 0.0416,
      "step": 9678
    },
    {
      "epoch": 18.42415596766524,
      "grad_norm": 0.09842484444379807,
      "learning_rate": 3.855827246744998e-05,
      "loss": 0.056,
      "step": 9679
    },
    {
      "epoch": 18.426058012363292,
      "grad_norm": 0.11220046132802963,
      "learning_rate": 3.8551921244839634e-05,
      "loss": 0.0422,
      "step": 9680
    },
    {
      "epoch": 18.42796005706134,
      "grad_norm": 0.053936492651700974,
      "learning_rate": 3.8545570022229286e-05,
      "loss": 0.044,
      "step": 9681
    },
    {
      "epoch": 18.42986210175939,
      "grad_norm": 0.061404965817928314,
      "learning_rate": 3.8539218799618924e-05,
      "loss": 0.0497,
      "step": 9682
    },
    {
      "epoch": 18.431764146457443,
      "grad_norm": 0.15061938762664795,
      "learning_rate": 3.8532867577008576e-05,
      "loss": 0.0508,
      "step": 9683
    },
    {
      "epoch": 18.433666191155492,
      "grad_norm": 0.13211102783679962,
      "learning_rate": 3.852651635439822e-05,
      "loss": 0.057,
      "step": 9684
    },
    {
      "epoch": 18.43556823585354,
      "grad_norm": 0.0436306893825531,
      "learning_rate": 3.852016513178787e-05,
      "loss": 0.0682,
      "step": 9685
    },
    {
      "epoch": 18.437470280551594,
      "grad_norm": 0.19738776981830597,
      "learning_rate": 3.851381390917752e-05,
      "loss": 0.059,
      "step": 9686
    },
    {
      "epoch": 18.439372325249643,
      "grad_norm": 0.27122023701667786,
      "learning_rate": 3.850746268656716e-05,
      "loss": 0.055,
      "step": 9687
    },
    {
      "epoch": 18.441274369947692,
      "grad_norm": 0.11561478674411774,
      "learning_rate": 3.8501111463956815e-05,
      "loss": 0.0462,
      "step": 9688
    },
    {
      "epoch": 18.443176414645745,
      "grad_norm": 0.09915643185377121,
      "learning_rate": 3.849476024134646e-05,
      "loss": 0.0674,
      "step": 9689
    },
    {
      "epoch": 18.445078459343794,
      "grad_norm": 0.049296531826257706,
      "learning_rate": 3.848840901873611e-05,
      "loss": 0.0515,
      "step": 9690
    },
    {
      "epoch": 18.446980504041846,
      "grad_norm": 0.04804540425539017,
      "learning_rate": 3.848205779612576e-05,
      "loss": 0.0466,
      "step": 9691
    },
    {
      "epoch": 18.448882548739896,
      "grad_norm": 0.17450645565986633,
      "learning_rate": 3.84757065735154e-05,
      "loss": 0.0555,
      "step": 9692
    },
    {
      "epoch": 18.450784593437945,
      "grad_norm": 0.1308811604976654,
      "learning_rate": 3.8469355350905054e-05,
      "loss": 0.0601,
      "step": 9693
    },
    {
      "epoch": 18.452686638135997,
      "grad_norm": 0.09358209371566772,
      "learning_rate": 3.84630041282947e-05,
      "loss": 0.0626,
      "step": 9694
    },
    {
      "epoch": 18.454588682834046,
      "grad_norm": 0.13447386026382446,
      "learning_rate": 3.845665290568435e-05,
      "loss": 0.0655,
      "step": 9695
    },
    {
      "epoch": 18.456490727532096,
      "grad_norm": 0.08530683815479279,
      "learning_rate": 3.845030168307399e-05,
      "loss": 0.0517,
      "step": 9696
    },
    {
      "epoch": 18.458392772230148,
      "grad_norm": 0.038144953548908234,
      "learning_rate": 3.844395046046364e-05,
      "loss": 0.0435,
      "step": 9697
    },
    {
      "epoch": 18.460294816928197,
      "grad_norm": 0.11692409962415695,
      "learning_rate": 3.8437599237853286e-05,
      "loss": 0.0495,
      "step": 9698
    },
    {
      "epoch": 18.46219686162625,
      "grad_norm": 0.07053724676370621,
      "learning_rate": 3.843124801524294e-05,
      "loss": 0.0512,
      "step": 9699
    },
    {
      "epoch": 18.4640989063243,
      "grad_norm": 0.107081338763237,
      "learning_rate": 3.842489679263258e-05,
      "loss": 0.061,
      "step": 9700
    },
    {
      "epoch": 18.466000951022348,
      "grad_norm": 0.07253830879926682,
      "learning_rate": 3.841854557002223e-05,
      "loss": 0.0708,
      "step": 9701
    },
    {
      "epoch": 18.4679029957204,
      "grad_norm": 0.1957426518201828,
      "learning_rate": 3.841219434741188e-05,
      "loss": 0.0759,
      "step": 9702
    },
    {
      "epoch": 18.46980504041845,
      "grad_norm": 0.14733651280403137,
      "learning_rate": 3.8405843124801525e-05,
      "loss": 0.037,
      "step": 9703
    },
    {
      "epoch": 18.4717070851165,
      "grad_norm": 0.1280432641506195,
      "learning_rate": 3.8399491902191176e-05,
      "loss": 0.053,
      "step": 9704
    },
    {
      "epoch": 18.47360912981455,
      "grad_norm": 0.0851169005036354,
      "learning_rate": 3.839314067958082e-05,
      "loss": 0.073,
      "step": 9705
    },
    {
      "epoch": 18.4755111745126,
      "grad_norm": 0.15699921548366547,
      "learning_rate": 3.8386789456970467e-05,
      "loss": 0.0489,
      "step": 9706
    },
    {
      "epoch": 18.47741321921065,
      "grad_norm": 0.24640056490898132,
      "learning_rate": 3.838043823436012e-05,
      "loss": 0.067,
      "step": 9707
    },
    {
      "epoch": 18.479315263908703,
      "grad_norm": 0.0925830528140068,
      "learning_rate": 3.8374087011749763e-05,
      "loss": 0.0458,
      "step": 9708
    },
    {
      "epoch": 18.48121730860675,
      "grad_norm": 0.17469635605812073,
      "learning_rate": 3.836773578913941e-05,
      "loss": 0.0629,
      "step": 9709
    },
    {
      "epoch": 18.483119353304804,
      "grad_norm": 0.10162944346666336,
      "learning_rate": 3.836138456652906e-05,
      "loss": 0.054,
      "step": 9710
    },
    {
      "epoch": 18.485021398002853,
      "grad_norm": 0.1978461891412735,
      "learning_rate": 3.8355033343918705e-05,
      "loss": 0.047,
      "step": 9711
    },
    {
      "epoch": 18.486923442700903,
      "grad_norm": 0.06301341205835342,
      "learning_rate": 3.834868212130836e-05,
      "loss": 0.0588,
      "step": 9712
    },
    {
      "epoch": 18.488825487398955,
      "grad_norm": 0.08585040271282196,
      "learning_rate": 3.8342330898698e-05,
      "loss": 0.0502,
      "step": 9713
    },
    {
      "epoch": 18.490727532097004,
      "grad_norm": 0.05659103021025658,
      "learning_rate": 3.833597967608765e-05,
      "loss": 0.0508,
      "step": 9714
    },
    {
      "epoch": 18.492629576795053,
      "grad_norm": 0.07119178771972656,
      "learning_rate": 3.832962845347729e-05,
      "loss": 0.0379,
      "step": 9715
    },
    {
      "epoch": 18.494531621493106,
      "grad_norm": 0.12380468100309372,
      "learning_rate": 3.8323277230866944e-05,
      "loss": 0.0693,
      "step": 9716
    },
    {
      "epoch": 18.496433666191155,
      "grad_norm": 0.09325138479471207,
      "learning_rate": 3.8316926008256596e-05,
      "loss": 0.0478,
      "step": 9717
    },
    {
      "epoch": 18.498335710889204,
      "grad_norm": 0.04756435379385948,
      "learning_rate": 3.8310574785646234e-05,
      "loss": 0.0549,
      "step": 9718
    },
    {
      "epoch": 18.500237755587257,
      "grad_norm": 0.06491974741220474,
      "learning_rate": 3.8304223563035886e-05,
      "loss": 0.0498,
      "step": 9719
    },
    {
      "epoch": 18.502139800285306,
      "grad_norm": 0.07270771265029907,
      "learning_rate": 3.829787234042553e-05,
      "loss": 0.0391,
      "step": 9720
    },
    {
      "epoch": 18.50404184498336,
      "grad_norm": 0.07048745453357697,
      "learning_rate": 3.829152111781518e-05,
      "loss": 0.0373,
      "step": 9721
    },
    {
      "epoch": 18.505943889681408,
      "grad_norm": 0.1918988972902298,
      "learning_rate": 3.828516989520483e-05,
      "loss": 0.0563,
      "step": 9722
    },
    {
      "epoch": 18.507845934379457,
      "grad_norm": 0.18292328715324402,
      "learning_rate": 3.827881867259447e-05,
      "loss": 0.0504,
      "step": 9723
    },
    {
      "epoch": 18.50974797907751,
      "grad_norm": 0.06293179094791412,
      "learning_rate": 3.8272467449984125e-05,
      "loss": 0.0593,
      "step": 9724
    },
    {
      "epoch": 18.51165002377556,
      "grad_norm": 0.0628834217786789,
      "learning_rate": 3.826611622737377e-05,
      "loss": 0.0526,
      "step": 9725
    },
    {
      "epoch": 18.513552068473608,
      "grad_norm": 0.15109018981456757,
      "learning_rate": 3.825976500476342e-05,
      "loss": 0.0599,
      "step": 9726
    },
    {
      "epoch": 18.51545411317166,
      "grad_norm": 0.0666005089879036,
      "learning_rate": 3.825341378215306e-05,
      "loss": 0.069,
      "step": 9727
    },
    {
      "epoch": 18.51735615786971,
      "grad_norm": 0.1296105533838272,
      "learning_rate": 3.824706255954271e-05,
      "loss": 0.0578,
      "step": 9728
    },
    {
      "epoch": 18.51925820256776,
      "grad_norm": 0.07938151061534882,
      "learning_rate": 3.8240711336932364e-05,
      "loss": 0.0481,
      "step": 9729
    },
    {
      "epoch": 18.52116024726581,
      "grad_norm": 0.20475080609321594,
      "learning_rate": 3.823436011432201e-05,
      "loss": 0.0718,
      "step": 9730
    },
    {
      "epoch": 18.52306229196386,
      "grad_norm": 0.21218127012252808,
      "learning_rate": 3.822800889171166e-05,
      "loss": 0.0447,
      "step": 9731
    },
    {
      "epoch": 18.524964336661913,
      "grad_norm": 0.11543460190296173,
      "learning_rate": 3.82216576691013e-05,
      "loss": 0.0589,
      "step": 9732
    },
    {
      "epoch": 18.526866381359962,
      "grad_norm": 0.10953544080257416,
      "learning_rate": 3.821530644649095e-05,
      "loss": 0.0467,
      "step": 9733
    },
    {
      "epoch": 18.52876842605801,
      "grad_norm": 0.13765722513198853,
      "learning_rate": 3.8208955223880596e-05,
      "loss": 0.0424,
      "step": 9734
    },
    {
      "epoch": 18.530670470756064,
      "grad_norm": 0.0649055689573288,
      "learning_rate": 3.820260400127025e-05,
      "loss": 0.0432,
      "step": 9735
    },
    {
      "epoch": 18.532572515454113,
      "grad_norm": 0.041515689343214035,
      "learning_rate": 3.819625277865989e-05,
      "loss": 0.0477,
      "step": 9736
    },
    {
      "epoch": 18.534474560152162,
      "grad_norm": 0.046874310821294785,
      "learning_rate": 3.818990155604954e-05,
      "loss": 0.051,
      "step": 9737
    },
    {
      "epoch": 18.536376604850215,
      "grad_norm": 0.12375843524932861,
      "learning_rate": 3.818355033343919e-05,
      "loss": 0.0588,
      "step": 9738
    },
    {
      "epoch": 18.538278649548264,
      "grad_norm": 0.06181807816028595,
      "learning_rate": 3.8177199110828835e-05,
      "loss": 0.044,
      "step": 9739
    },
    {
      "epoch": 18.540180694246313,
      "grad_norm": 0.12714730203151703,
      "learning_rate": 3.817084788821849e-05,
      "loss": 0.0484,
      "step": 9740
    },
    {
      "epoch": 18.542082738944366,
      "grad_norm": 0.05887744575738907,
      "learning_rate": 3.816449666560813e-05,
      "loss": 0.0635,
      "step": 9741
    },
    {
      "epoch": 18.543984783642415,
      "grad_norm": 0.06551394611597061,
      "learning_rate": 3.815814544299778e-05,
      "loss": 0.0699,
      "step": 9742
    },
    {
      "epoch": 18.545886828340468,
      "grad_norm": 0.18102426826953888,
      "learning_rate": 3.815179422038743e-05,
      "loss": 0.0656,
      "step": 9743
    },
    {
      "epoch": 18.547788873038517,
      "grad_norm": 0.15179798007011414,
      "learning_rate": 3.8145442997777074e-05,
      "loss": 0.0622,
      "step": 9744
    },
    {
      "epoch": 18.549690917736566,
      "grad_norm": 0.17021843791007996,
      "learning_rate": 3.813909177516672e-05,
      "loss": 0.0572,
      "step": 9745
    },
    {
      "epoch": 18.55159296243462,
      "grad_norm": 0.05107571929693222,
      "learning_rate": 3.813274055255637e-05,
      "loss": 0.0587,
      "step": 9746
    },
    {
      "epoch": 18.553495007132668,
      "grad_norm": 0.09118980914354324,
      "learning_rate": 3.8126389329946016e-05,
      "loss": 0.0641,
      "step": 9747
    },
    {
      "epoch": 18.555397051830717,
      "grad_norm": 0.12549744546413422,
      "learning_rate": 3.812003810733567e-05,
      "loss": 0.0464,
      "step": 9748
    },
    {
      "epoch": 18.55729909652877,
      "grad_norm": 0.05056701973080635,
      "learning_rate": 3.811368688472531e-05,
      "loss": 0.0522,
      "step": 9749
    },
    {
      "epoch": 18.55920114122682,
      "grad_norm": 0.04617400839924812,
      "learning_rate": 3.810733566211496e-05,
      "loss": 0.0492,
      "step": 9750
    },
    {
      "epoch": 18.561103185924868,
      "grad_norm": 0.16483484208583832,
      "learning_rate": 3.81009844395046e-05,
      "loss": 0.0613,
      "step": 9751
    },
    {
      "epoch": 18.56300523062292,
      "grad_norm": 0.07073646783828735,
      "learning_rate": 3.8094633216894255e-05,
      "loss": 0.0484,
      "step": 9752
    },
    {
      "epoch": 18.56490727532097,
      "grad_norm": 0.07303962856531143,
      "learning_rate": 3.80882819942839e-05,
      "loss": 0.0639,
      "step": 9753
    },
    {
      "epoch": 18.566809320019022,
      "grad_norm": 0.07416113466024399,
      "learning_rate": 3.8081930771673545e-05,
      "loss": 0.0531,
      "step": 9754
    },
    {
      "epoch": 18.56871136471707,
      "grad_norm": 0.03861843794584274,
      "learning_rate": 3.80755795490632e-05,
      "loss": 0.0566,
      "step": 9755
    },
    {
      "epoch": 18.57061340941512,
      "grad_norm": 0.15391989052295685,
      "learning_rate": 3.806922832645284e-05,
      "loss": 0.0696,
      "step": 9756
    },
    {
      "epoch": 18.572515454113173,
      "grad_norm": 0.10909277945756912,
      "learning_rate": 3.8062877103842494e-05,
      "loss": 0.0531,
      "step": 9757
    },
    {
      "epoch": 18.574417498811222,
      "grad_norm": 0.24341370165348053,
      "learning_rate": 3.805652588123214e-05,
      "loss": 0.0571,
      "step": 9758
    },
    {
      "epoch": 18.57631954350927,
      "grad_norm": 0.0979224294424057,
      "learning_rate": 3.8050174658621784e-05,
      "loss": 0.0577,
      "step": 9759
    },
    {
      "epoch": 18.578221588207324,
      "grad_norm": 0.13480636477470398,
      "learning_rate": 3.8043823436011436e-05,
      "loss": 0.0605,
      "step": 9760
    },
    {
      "epoch": 18.580123632905373,
      "grad_norm": 0.18941956758499146,
      "learning_rate": 3.803747221340108e-05,
      "loss": 0.0463,
      "step": 9761
    },
    {
      "epoch": 18.582025677603422,
      "grad_norm": 0.049710191786289215,
      "learning_rate": 3.803112099079073e-05,
      "loss": 0.0525,
      "step": 9762
    },
    {
      "epoch": 18.583927722301475,
      "grad_norm": 0.17660582065582275,
      "learning_rate": 3.802476976818037e-05,
      "loss": 0.0711,
      "step": 9763
    },
    {
      "epoch": 18.585829766999524,
      "grad_norm": 0.22386837005615234,
      "learning_rate": 3.801841854557002e-05,
      "loss": 0.072,
      "step": 9764
    },
    {
      "epoch": 18.587731811697576,
      "grad_norm": 0.11074668169021606,
      "learning_rate": 3.8012067322959674e-05,
      "loss": 0.0549,
      "step": 9765
    },
    {
      "epoch": 18.589633856395626,
      "grad_norm": 0.055570632219314575,
      "learning_rate": 3.800571610034932e-05,
      "loss": 0.0565,
      "step": 9766
    },
    {
      "epoch": 18.591535901093675,
      "grad_norm": 0.14890772104263306,
      "learning_rate": 3.799936487773897e-05,
      "loss": 0.0642,
      "step": 9767
    },
    {
      "epoch": 18.593437945791727,
      "grad_norm": 0.15929993987083435,
      "learning_rate": 3.799301365512861e-05,
      "loss": 0.0684,
      "step": 9768
    },
    {
      "epoch": 18.595339990489776,
      "grad_norm": 0.12377023696899414,
      "learning_rate": 3.798666243251826e-05,
      "loss": 0.0611,
      "step": 9769
    },
    {
      "epoch": 18.597242035187826,
      "grad_norm": 0.15041153132915497,
      "learning_rate": 3.7980311209907907e-05,
      "loss": 0.0554,
      "step": 9770
    },
    {
      "epoch": 18.59914407988588,
      "grad_norm": 0.12970572710037231,
      "learning_rate": 3.797395998729756e-05,
      "loss": 0.0572,
      "step": 9771
    },
    {
      "epoch": 18.601046124583927,
      "grad_norm": 0.15677066147327423,
      "learning_rate": 3.7967608764687203e-05,
      "loss": 0.0614,
      "step": 9772
    },
    {
      "epoch": 18.602948169281976,
      "grad_norm": 0.06385092437267303,
      "learning_rate": 3.796125754207685e-05,
      "loss": 0.0552,
      "step": 9773
    },
    {
      "epoch": 18.60485021398003,
      "grad_norm": 0.08093471825122833,
      "learning_rate": 3.79549063194665e-05,
      "loss": 0.0529,
      "step": 9774
    },
    {
      "epoch": 18.60675225867808,
      "grad_norm": 0.09495603293180466,
      "learning_rate": 3.7948555096856145e-05,
      "loss": 0.0734,
      "step": 9775
    },
    {
      "epoch": 18.60865430337613,
      "grad_norm": 0.05810278654098511,
      "learning_rate": 3.79422038742458e-05,
      "loss": 0.0792,
      "step": 9776
    },
    {
      "epoch": 18.61055634807418,
      "grad_norm": 0.09176381677389145,
      "learning_rate": 3.793585265163544e-05,
      "loss": 0.0643,
      "step": 9777
    },
    {
      "epoch": 18.61245839277223,
      "grad_norm": 0.16379646956920624,
      "learning_rate": 3.792950142902509e-05,
      "loss": 0.0446,
      "step": 9778
    },
    {
      "epoch": 18.614360437470282,
      "grad_norm": 0.1531377136707306,
      "learning_rate": 3.792315020641474e-05,
      "loss": 0.0583,
      "step": 9779
    },
    {
      "epoch": 18.61626248216833,
      "grad_norm": 0.09140586107969284,
      "learning_rate": 3.7916798983804384e-05,
      "loss": 0.0493,
      "step": 9780
    },
    {
      "epoch": 18.61816452686638,
      "grad_norm": 0.10958456248044968,
      "learning_rate": 3.791044776119403e-05,
      "loss": 0.0586,
      "step": 9781
    },
    {
      "epoch": 18.620066571564433,
      "grad_norm": 0.19568602740764618,
      "learning_rate": 3.7904096538583674e-05,
      "loss": 0.0589,
      "step": 9782
    },
    {
      "epoch": 18.621968616262482,
      "grad_norm": 0.14151211082935333,
      "learning_rate": 3.7897745315973326e-05,
      "loss": 0.0544,
      "step": 9783
    },
    {
      "epoch": 18.62387066096053,
      "grad_norm": 0.05674430727958679,
      "learning_rate": 3.789139409336298e-05,
      "loss": 0.0442,
      "step": 9784
    },
    {
      "epoch": 18.625772705658584,
      "grad_norm": 0.16973741352558136,
      "learning_rate": 3.788504287075262e-05,
      "loss": 0.0504,
      "step": 9785
    },
    {
      "epoch": 18.627674750356633,
      "grad_norm": 0.24356374144554138,
      "learning_rate": 3.787869164814227e-05,
      "loss": 0.0613,
      "step": 9786
    },
    {
      "epoch": 18.629576795054685,
      "grad_norm": 0.10595981031656265,
      "learning_rate": 3.787234042553191e-05,
      "loss": 0.0528,
      "step": 9787
    },
    {
      "epoch": 18.631478839752734,
      "grad_norm": 0.19772733747959137,
      "learning_rate": 3.7865989202921565e-05,
      "loss": 0.0585,
      "step": 9788
    },
    {
      "epoch": 18.633380884450784,
      "grad_norm": 0.05060689151287079,
      "learning_rate": 3.785963798031121e-05,
      "loss": 0.0452,
      "step": 9789
    },
    {
      "epoch": 18.635282929148836,
      "grad_norm": 0.3703983724117279,
      "learning_rate": 3.7853286757700855e-05,
      "loss": 0.0902,
      "step": 9790
    },
    {
      "epoch": 18.637184973846885,
      "grad_norm": 0.17736543715000153,
      "learning_rate": 3.784693553509051e-05,
      "loss": 0.0503,
      "step": 9791
    },
    {
      "epoch": 18.639087018544934,
      "grad_norm": 0.13419793546199799,
      "learning_rate": 3.784058431248015e-05,
      "loss": 0.0479,
      "step": 9792
    },
    {
      "epoch": 18.640989063242987,
      "grad_norm": 0.08425474166870117,
      "learning_rate": 3.7834233089869804e-05,
      "loss": 0.0586,
      "step": 9793
    },
    {
      "epoch": 18.642891107941036,
      "grad_norm": 0.2423335611820221,
      "learning_rate": 3.782788186725945e-05,
      "loss": 0.0635,
      "step": 9794
    },
    {
      "epoch": 18.644793152639085,
      "grad_norm": 0.09958801418542862,
      "learning_rate": 3.7821530644649094e-05,
      "loss": 0.0569,
      "step": 9795
    },
    {
      "epoch": 18.646695197337138,
      "grad_norm": 0.0595712773501873,
      "learning_rate": 3.7815179422038746e-05,
      "loss": 0.0542,
      "step": 9796
    },
    {
      "epoch": 18.648597242035187,
      "grad_norm": 0.13254064321517944,
      "learning_rate": 3.780882819942839e-05,
      "loss": 0.0587,
      "step": 9797
    },
    {
      "epoch": 18.65049928673324,
      "grad_norm": 0.0418279692530632,
      "learning_rate": 3.780247697681804e-05,
      "loss": 0.0637,
      "step": 9798
    },
    {
      "epoch": 18.65240133143129,
      "grad_norm": 0.1887357085943222,
      "learning_rate": 3.779612575420768e-05,
      "loss": 0.061,
      "step": 9799
    },
    {
      "epoch": 18.654303376129338,
      "grad_norm": 0.07455997169017792,
      "learning_rate": 3.778977453159733e-05,
      "loss": 0.0563,
      "step": 9800
    },
    {
      "epoch": 18.65620542082739,
      "grad_norm": 0.1308373212814331,
      "learning_rate": 3.7783423308986985e-05,
      "loss": 0.0601,
      "step": 9801
    },
    {
      "epoch": 18.65810746552544,
      "grad_norm": 0.13394196331501007,
      "learning_rate": 3.777707208637663e-05,
      "loss": 0.0591,
      "step": 9802
    },
    {
      "epoch": 18.66000951022349,
      "grad_norm": 0.12642307579517365,
      "learning_rate": 3.777072086376628e-05,
      "loss": 0.0466,
      "step": 9803
    },
    {
      "epoch": 18.66191155492154,
      "grad_norm": 0.20655491948127747,
      "learning_rate": 3.776436964115592e-05,
      "loss": 0.0612,
      "step": 9804
    },
    {
      "epoch": 18.66381359961959,
      "grad_norm": 0.045045897364616394,
      "learning_rate": 3.775801841854557e-05,
      "loss": 0.06,
      "step": 9805
    },
    {
      "epoch": 18.665715644317643,
      "grad_norm": 0.16911624372005463,
      "learning_rate": 3.775166719593522e-05,
      "loss": 0.0519,
      "step": 9806
    },
    {
      "epoch": 18.667617689015692,
      "grad_norm": 0.10162369161844254,
      "learning_rate": 3.774531597332487e-05,
      "loss": 0.0657,
      "step": 9807
    },
    {
      "epoch": 18.66951973371374,
      "grad_norm": 0.06274081021547318,
      "learning_rate": 3.7738964750714514e-05,
      "loss": 0.0473,
      "step": 9808
    },
    {
      "epoch": 18.671421778411794,
      "grad_norm": 0.1434098482131958,
      "learning_rate": 3.773261352810416e-05,
      "loss": 0.0669,
      "step": 9809
    },
    {
      "epoch": 18.673323823109843,
      "grad_norm": 0.29461508989334106,
      "learning_rate": 3.772626230549381e-05,
      "loss": 0.053,
      "step": 9810
    },
    {
      "epoch": 18.675225867807892,
      "grad_norm": 0.12209858745336533,
      "learning_rate": 3.7719911082883456e-05,
      "loss": 0.0585,
      "step": 9811
    },
    {
      "epoch": 18.677127912505945,
      "grad_norm": 0.16458265483379364,
      "learning_rate": 3.771355986027311e-05,
      "loss": 0.0582,
      "step": 9812
    },
    {
      "epoch": 18.679029957203994,
      "grad_norm": 0.08619740605354309,
      "learning_rate": 3.770720863766275e-05,
      "loss": 0.0412,
      "step": 9813
    },
    {
      "epoch": 18.680932001902043,
      "grad_norm": 0.08477325737476349,
      "learning_rate": 3.77008574150524e-05,
      "loss": 0.0433,
      "step": 9814
    },
    {
      "epoch": 18.682834046600096,
      "grad_norm": 0.35625943541526794,
      "learning_rate": 3.769450619244205e-05,
      "loss": 0.0609,
      "step": 9815
    },
    {
      "epoch": 18.684736091298145,
      "grad_norm": 0.08370004594326019,
      "learning_rate": 3.7688154969831695e-05,
      "loss": 0.0459,
      "step": 9816
    },
    {
      "epoch": 18.686638135996198,
      "grad_norm": 0.3711557388305664,
      "learning_rate": 3.768180374722134e-05,
      "loss": 0.0722,
      "step": 9817
    },
    {
      "epoch": 18.688540180694247,
      "grad_norm": 0.17637227475643158,
      "learning_rate": 3.7675452524610985e-05,
      "loss": 0.0735,
      "step": 9818
    },
    {
      "epoch": 18.690442225392296,
      "grad_norm": 0.06635082513093948,
      "learning_rate": 3.766910130200064e-05,
      "loss": 0.0682,
      "step": 9819
    },
    {
      "epoch": 18.69234427009035,
      "grad_norm": 0.061188675463199615,
      "learning_rate": 3.766275007939029e-05,
      "loss": 0.0557,
      "step": 9820
    },
    {
      "epoch": 18.694246314788398,
      "grad_norm": 0.06702593713998795,
      "learning_rate": 3.7656398856779934e-05,
      "loss": 0.0534,
      "step": 9821
    },
    {
      "epoch": 18.696148359486447,
      "grad_norm": 0.18100468814373016,
      "learning_rate": 3.765004763416958e-05,
      "loss": 0.0593,
      "step": 9822
    },
    {
      "epoch": 18.6980504041845,
      "grad_norm": 0.09489830583333969,
      "learning_rate": 3.7643696411559224e-05,
      "loss": 0.07,
      "step": 9823
    },
    {
      "epoch": 18.69995244888255,
      "grad_norm": 0.10796690732240677,
      "learning_rate": 3.7637345188948876e-05,
      "loss": 0.0696,
      "step": 9824
    },
    {
      "epoch": 18.701854493580598,
      "grad_norm": 0.09641650319099426,
      "learning_rate": 3.763099396633852e-05,
      "loss": 0.0543,
      "step": 9825
    },
    {
      "epoch": 18.70375653827865,
      "grad_norm": 0.17483478784561157,
      "learning_rate": 3.7624642743728166e-05,
      "loss": 0.0752,
      "step": 9826
    },
    {
      "epoch": 18.7056585829767,
      "grad_norm": 0.07531461119651794,
      "learning_rate": 3.761829152111782e-05,
      "loss": 0.0588,
      "step": 9827
    },
    {
      "epoch": 18.707560627674752,
      "grad_norm": 0.11430740356445312,
      "learning_rate": 3.761194029850746e-05,
      "loss": 0.0515,
      "step": 9828
    },
    {
      "epoch": 18.7094626723728,
      "grad_norm": 0.10490404814481735,
      "learning_rate": 3.7605589075897114e-05,
      "loss": 0.0527,
      "step": 9829
    },
    {
      "epoch": 18.71136471707085,
      "grad_norm": 0.24117016792297363,
      "learning_rate": 3.759923785328676e-05,
      "loss": 0.0681,
      "step": 9830
    },
    {
      "epoch": 18.713266761768903,
      "grad_norm": 0.05267453193664551,
      "learning_rate": 3.7592886630676405e-05,
      "loss": 0.0641,
      "step": 9831
    },
    {
      "epoch": 18.715168806466952,
      "grad_norm": 0.13616354763507843,
      "learning_rate": 3.7586535408066056e-05,
      "loss": 0.0584,
      "step": 9832
    },
    {
      "epoch": 18.717070851165,
      "grad_norm": 0.045377179980278015,
      "learning_rate": 3.75801841854557e-05,
      "loss": 0.0579,
      "step": 9833
    },
    {
      "epoch": 18.718972895863054,
      "grad_norm": 0.30514270067214966,
      "learning_rate": 3.757383296284535e-05,
      "loss": 0.0697,
      "step": 9834
    },
    {
      "epoch": 18.720874940561103,
      "grad_norm": 0.19669856131076813,
      "learning_rate": 3.756748174023499e-05,
      "loss": 0.067,
      "step": 9835
    },
    {
      "epoch": 18.722776985259152,
      "grad_norm": 0.13380081951618195,
      "learning_rate": 3.7561130517624643e-05,
      "loss": 0.0563,
      "step": 9836
    },
    {
      "epoch": 18.724679029957205,
      "grad_norm": 0.0629597082734108,
      "learning_rate": 3.755477929501429e-05,
      "loss": 0.0643,
      "step": 9837
    },
    {
      "epoch": 18.726581074655254,
      "grad_norm": 0.11358380317687988,
      "learning_rate": 3.754842807240394e-05,
      "loss": 0.0594,
      "step": 9838
    },
    {
      "epoch": 18.728483119353307,
      "grad_norm": 0.28481239080429077,
      "learning_rate": 3.754207684979359e-05,
      "loss": 0.0534,
      "step": 9839
    },
    {
      "epoch": 18.730385164051356,
      "grad_norm": 0.4196884334087372,
      "learning_rate": 3.753572562718323e-05,
      "loss": 0.0625,
      "step": 9840
    },
    {
      "epoch": 18.732287208749405,
      "grad_norm": 0.24829143285751343,
      "learning_rate": 3.752937440457288e-05,
      "loss": 0.062,
      "step": 9841
    },
    {
      "epoch": 18.734189253447457,
      "grad_norm": 0.047377172857522964,
      "learning_rate": 3.752302318196253e-05,
      "loss": 0.0363,
      "step": 9842
    },
    {
      "epoch": 18.736091298145507,
      "grad_norm": 0.07733766734600067,
      "learning_rate": 3.751667195935218e-05,
      "loss": 0.0448,
      "step": 9843
    },
    {
      "epoch": 18.737993342843556,
      "grad_norm": 0.1072051152586937,
      "learning_rate": 3.7510320736741824e-05,
      "loss": 0.0521,
      "step": 9844
    },
    {
      "epoch": 18.73989538754161,
      "grad_norm": 0.08531222492456436,
      "learning_rate": 3.750396951413147e-05,
      "loss": 0.0828,
      "step": 9845
    },
    {
      "epoch": 18.741797432239657,
      "grad_norm": 0.091969795525074,
      "learning_rate": 3.749761829152112e-05,
      "loss": 0.0317,
      "step": 9846
    },
    {
      "epoch": 18.743699476937707,
      "grad_norm": 0.061905376613140106,
      "learning_rate": 3.7491267068910766e-05,
      "loss": 0.0319,
      "step": 9847
    },
    {
      "epoch": 18.74560152163576,
      "grad_norm": 0.17943806946277618,
      "learning_rate": 3.748491584630042e-05,
      "loss": 0.0665,
      "step": 9848
    },
    {
      "epoch": 18.74750356633381,
      "grad_norm": 0.06048533692955971,
      "learning_rate": 3.747856462369006e-05,
      "loss": 0.0466,
      "step": 9849
    },
    {
      "epoch": 18.74940561103186,
      "grad_norm": 0.07251735031604767,
      "learning_rate": 3.747221340107971e-05,
      "loss": 0.0579,
      "step": 9850
    },
    {
      "epoch": 18.75130765572991,
      "grad_norm": 0.07588538527488708,
      "learning_rate": 3.746586217846936e-05,
      "loss": 0.0602,
      "step": 9851
    },
    {
      "epoch": 18.75320970042796,
      "grad_norm": 0.2024068981409073,
      "learning_rate": 3.7459510955859005e-05,
      "loss": 0.0633,
      "step": 9852
    },
    {
      "epoch": 18.755111745126012,
      "grad_norm": 0.06142650172114372,
      "learning_rate": 3.745315973324865e-05,
      "loss": 0.0575,
      "step": 9853
    },
    {
      "epoch": 18.75701378982406,
      "grad_norm": 0.11343108862638474,
      "learning_rate": 3.7446808510638295e-05,
      "loss": 0.0657,
      "step": 9854
    },
    {
      "epoch": 18.75891583452211,
      "grad_norm": 0.2155323028564453,
      "learning_rate": 3.744045728802795e-05,
      "loss": 0.0611,
      "step": 9855
    },
    {
      "epoch": 18.760817879220163,
      "grad_norm": 0.11705797165632248,
      "learning_rate": 3.74341060654176e-05,
      "loss": 0.0512,
      "step": 9856
    },
    {
      "epoch": 18.762719923918212,
      "grad_norm": 0.09704883396625519,
      "learning_rate": 3.7427754842807244e-05,
      "loss": 0.0471,
      "step": 9857
    },
    {
      "epoch": 18.76462196861626,
      "grad_norm": 0.08368376642465591,
      "learning_rate": 3.742140362019689e-05,
      "loss": 0.0594,
      "step": 9858
    },
    {
      "epoch": 18.766524013314314,
      "grad_norm": 0.1207922101020813,
      "learning_rate": 3.7415052397586534e-05,
      "loss": 0.042,
      "step": 9859
    },
    {
      "epoch": 18.768426058012363,
      "grad_norm": 0.039832353591918945,
      "learning_rate": 3.7408701174976186e-05,
      "loss": 0.053,
      "step": 9860
    },
    {
      "epoch": 18.770328102710415,
      "grad_norm": 0.045956000685691833,
      "learning_rate": 3.740234995236583e-05,
      "loss": 0.0463,
      "step": 9861
    },
    {
      "epoch": 18.772230147408465,
      "grad_norm": 0.12723292410373688,
      "learning_rate": 3.7395998729755476e-05,
      "loss": 0.0632,
      "step": 9862
    },
    {
      "epoch": 18.774132192106514,
      "grad_norm": 0.04386424273252487,
      "learning_rate": 3.738964750714513e-05,
      "loss": 0.061,
      "step": 9863
    },
    {
      "epoch": 18.776034236804566,
      "grad_norm": 0.12253253161907196,
      "learning_rate": 3.738329628453477e-05,
      "loss": 0.0559,
      "step": 9864
    },
    {
      "epoch": 18.777936281502615,
      "grad_norm": 0.11969088017940521,
      "learning_rate": 3.7376945061924425e-05,
      "loss": 0.069,
      "step": 9865
    },
    {
      "epoch": 18.779838326200665,
      "grad_norm": 0.252768874168396,
      "learning_rate": 3.737059383931407e-05,
      "loss": 0.0745,
      "step": 9866
    },
    {
      "epoch": 18.781740370898717,
      "grad_norm": 0.1536416858434677,
      "learning_rate": 3.7364242616703715e-05,
      "loss": 0.0688,
      "step": 9867
    },
    {
      "epoch": 18.783642415596766,
      "grad_norm": 0.16543470323085785,
      "learning_rate": 3.735789139409337e-05,
      "loss": 0.0532,
      "step": 9868
    },
    {
      "epoch": 18.785544460294815,
      "grad_norm": 0.14163970947265625,
      "learning_rate": 3.735154017148301e-05,
      "loss": 0.0601,
      "step": 9869
    },
    {
      "epoch": 18.787446504992868,
      "grad_norm": 0.11656033247709274,
      "learning_rate": 3.7345188948872664e-05,
      "loss": 0.0668,
      "step": 9870
    },
    {
      "epoch": 18.789348549690917,
      "grad_norm": 0.0778069794178009,
      "learning_rate": 3.73388377262623e-05,
      "loss": 0.0359,
      "step": 9871
    },
    {
      "epoch": 18.79125059438897,
      "grad_norm": 0.2035890817642212,
      "learning_rate": 3.7332486503651954e-05,
      "loss": 0.06,
      "step": 9872
    },
    {
      "epoch": 18.79315263908702,
      "grad_norm": 0.040098268538713455,
      "learning_rate": 3.73261352810416e-05,
      "loss": 0.0392,
      "step": 9873
    },
    {
      "epoch": 18.795054683785068,
      "grad_norm": 0.12130275368690491,
      "learning_rate": 3.731978405843125e-05,
      "loss": 0.053,
      "step": 9874
    },
    {
      "epoch": 18.79695672848312,
      "grad_norm": 0.2116493582725525,
      "learning_rate": 3.73134328358209e-05,
      "loss": 0.0629,
      "step": 9875
    },
    {
      "epoch": 18.79885877318117,
      "grad_norm": 0.19741222262382507,
      "learning_rate": 3.730708161321054e-05,
      "loss": 0.0778,
      "step": 9876
    },
    {
      "epoch": 18.80076081787922,
      "grad_norm": 0.09970089048147202,
      "learning_rate": 3.730073039060019e-05,
      "loss": 0.0457,
      "step": 9877
    },
    {
      "epoch": 18.80266286257727,
      "grad_norm": 0.15264080464839935,
      "learning_rate": 3.729437916798984e-05,
      "loss": 0.0455,
      "step": 9878
    },
    {
      "epoch": 18.80456490727532,
      "grad_norm": 0.1529812514781952,
      "learning_rate": 3.728802794537949e-05,
      "loss": 0.0518,
      "step": 9879
    },
    {
      "epoch": 18.80646695197337,
      "grad_norm": 0.18237809836864471,
      "learning_rate": 3.7281676722769135e-05,
      "loss": 0.0594,
      "step": 9880
    },
    {
      "epoch": 18.808368996671422,
      "grad_norm": 0.23065491020679474,
      "learning_rate": 3.727532550015878e-05,
      "loss": 0.0785,
      "step": 9881
    },
    {
      "epoch": 18.81027104136947,
      "grad_norm": 0.05364197492599487,
      "learning_rate": 3.726897427754843e-05,
      "loss": 0.0384,
      "step": 9882
    },
    {
      "epoch": 18.812173086067524,
      "grad_norm": 0.05554148554801941,
      "learning_rate": 3.726262305493808e-05,
      "loss": 0.057,
      "step": 9883
    },
    {
      "epoch": 18.814075130765573,
      "grad_norm": 0.1097605749964714,
      "learning_rate": 3.725627183232773e-05,
      "loss": 0.0664,
      "step": 9884
    },
    {
      "epoch": 18.815977175463622,
      "grad_norm": 0.05670863762497902,
      "learning_rate": 3.7249920609717374e-05,
      "loss": 0.0581,
      "step": 9885
    },
    {
      "epoch": 18.817879220161675,
      "grad_norm": 0.09237883985042572,
      "learning_rate": 3.724356938710702e-05,
      "loss": 0.0391,
      "step": 9886
    },
    {
      "epoch": 18.819781264859724,
      "grad_norm": 0.12565287947654724,
      "learning_rate": 3.723721816449667e-05,
      "loss": 0.069,
      "step": 9887
    },
    {
      "epoch": 18.821683309557773,
      "grad_norm": 0.24651116132736206,
      "learning_rate": 3.7230866941886316e-05,
      "loss": 0.0576,
      "step": 9888
    },
    {
      "epoch": 18.823585354255826,
      "grad_norm": 0.21622343361377716,
      "learning_rate": 3.722451571927596e-05,
      "loss": 0.0798,
      "step": 9889
    },
    {
      "epoch": 18.825487398953875,
      "grad_norm": 0.12411265820264816,
      "learning_rate": 3.7218164496665606e-05,
      "loss": 0.064,
      "step": 9890
    },
    {
      "epoch": 18.827389443651924,
      "grad_norm": 0.1262037754058838,
      "learning_rate": 3.721181327405526e-05,
      "loss": 0.0737,
      "step": 9891
    },
    {
      "epoch": 18.829291488349977,
      "grad_norm": 0.15461592376232147,
      "learning_rate": 3.72054620514449e-05,
      "loss": 0.0531,
      "step": 9892
    },
    {
      "epoch": 18.831193533048026,
      "grad_norm": 0.598727822303772,
      "learning_rate": 3.7199110828834555e-05,
      "loss": 0.0751,
      "step": 9893
    },
    {
      "epoch": 18.83309557774608,
      "grad_norm": 0.1309041827917099,
      "learning_rate": 3.71927596062242e-05,
      "loss": 0.0494,
      "step": 9894
    },
    {
      "epoch": 18.834997622444128,
      "grad_norm": 0.1532328873872757,
      "learning_rate": 3.7186408383613845e-05,
      "loss": 0.0464,
      "step": 9895
    },
    {
      "epoch": 18.836899667142177,
      "grad_norm": 0.14370164275169373,
      "learning_rate": 3.7180057161003496e-05,
      "loss": 0.0651,
      "step": 9896
    },
    {
      "epoch": 18.83880171184023,
      "grad_norm": 0.14575162529945374,
      "learning_rate": 3.717370593839314e-05,
      "loss": 0.0625,
      "step": 9897
    },
    {
      "epoch": 18.84070375653828,
      "grad_norm": 0.06873523443937302,
      "learning_rate": 3.7167354715782787e-05,
      "loss": 0.0535,
      "step": 9898
    },
    {
      "epoch": 18.842605801236328,
      "grad_norm": 0.10545411705970764,
      "learning_rate": 3.716100349317244e-05,
      "loss": 0.0578,
      "step": 9899
    },
    {
      "epoch": 18.84450784593438,
      "grad_norm": 0.08575357496738434,
      "learning_rate": 3.7154652270562084e-05,
      "loss": 0.053,
      "step": 9900
    },
    {
      "epoch": 18.84640989063243,
      "grad_norm": 0.13706247508525848,
      "learning_rate": 3.7148301047951735e-05,
      "loss": 0.0733,
      "step": 9901
    },
    {
      "epoch": 18.84831193533048,
      "grad_norm": 0.14186565577983856,
      "learning_rate": 3.714194982534138e-05,
      "loss": 0.0816,
      "step": 9902
    },
    {
      "epoch": 18.85021398002853,
      "grad_norm": 0.08031070232391357,
      "learning_rate": 3.7135598602731025e-05,
      "loss": 0.0756,
      "step": 9903
    },
    {
      "epoch": 18.85211602472658,
      "grad_norm": 0.0668841004371643,
      "learning_rate": 3.712924738012068e-05,
      "loss": 0.0619,
      "step": 9904
    },
    {
      "epoch": 18.854018069424633,
      "grad_norm": 0.1793297678232193,
      "learning_rate": 3.712289615751032e-05,
      "loss": 0.0636,
      "step": 9905
    },
    {
      "epoch": 18.855920114122682,
      "grad_norm": 0.1464206874370575,
      "learning_rate": 3.7116544934899974e-05,
      "loss": 0.0615,
      "step": 9906
    },
    {
      "epoch": 18.85782215882073,
      "grad_norm": 0.07394376397132874,
      "learning_rate": 3.711019371228961e-05,
      "loss": 0.05,
      "step": 9907
    },
    {
      "epoch": 18.859724203518784,
      "grad_norm": 0.07852068543434143,
      "learning_rate": 3.7103842489679264e-05,
      "loss": 0.0643,
      "step": 9908
    },
    {
      "epoch": 18.861626248216833,
      "grad_norm": 0.2197633534669876,
      "learning_rate": 3.709749126706891e-05,
      "loss": 0.062,
      "step": 9909
    },
    {
      "epoch": 18.863528292914882,
      "grad_norm": 0.07696213573217392,
      "learning_rate": 3.709114004445856e-05,
      "loss": 0.0222,
      "step": 9910
    },
    {
      "epoch": 18.865430337612935,
      "grad_norm": 0.06208226457238197,
      "learning_rate": 3.708478882184821e-05,
      "loss": 0.0553,
      "step": 9911
    },
    {
      "epoch": 18.867332382310984,
      "grad_norm": 0.16470803320407867,
      "learning_rate": 3.707843759923785e-05,
      "loss": 0.0664,
      "step": 9912
    },
    {
      "epoch": 18.869234427009033,
      "grad_norm": 0.056323740631341934,
      "learning_rate": 3.70720863766275e-05,
      "loss": 0.061,
      "step": 9913
    },
    {
      "epoch": 18.871136471707086,
      "grad_norm": 0.038745343685150146,
      "learning_rate": 3.706573515401715e-05,
      "loss": 0.0502,
      "step": 9914
    },
    {
      "epoch": 18.873038516405135,
      "grad_norm": 0.13453522324562073,
      "learning_rate": 3.70593839314068e-05,
      "loss": 0.0754,
      "step": 9915
    },
    {
      "epoch": 18.874940561103188,
      "grad_norm": 0.1173306480050087,
      "learning_rate": 3.7053032708796445e-05,
      "loss": 0.0605,
      "step": 9916
    },
    {
      "epoch": 18.876842605801237,
      "grad_norm": 0.10980434715747833,
      "learning_rate": 3.704668148618609e-05,
      "loss": 0.0537,
      "step": 9917
    },
    {
      "epoch": 18.878744650499286,
      "grad_norm": 0.19415007531642914,
      "learning_rate": 3.704033026357574e-05,
      "loss": 0.0584,
      "step": 9918
    },
    {
      "epoch": 18.88064669519734,
      "grad_norm": 0.08781924843788147,
      "learning_rate": 3.703397904096539e-05,
      "loss": 0.0435,
      "step": 9919
    },
    {
      "epoch": 18.882548739895388,
      "grad_norm": 0.26694977283477783,
      "learning_rate": 3.702762781835504e-05,
      "loss": 0.0606,
      "step": 9920
    },
    {
      "epoch": 18.884450784593437,
      "grad_norm": 0.14742548763751984,
      "learning_rate": 3.702127659574468e-05,
      "loss": 0.047,
      "step": 9921
    },
    {
      "epoch": 18.88635282929149,
      "grad_norm": 0.04659523442387581,
      "learning_rate": 3.701492537313433e-05,
      "loss": 0.0386,
      "step": 9922
    },
    {
      "epoch": 18.88825487398954,
      "grad_norm": 0.3608655631542206,
      "learning_rate": 3.700857415052398e-05,
      "loss": 0.0619,
      "step": 9923
    },
    {
      "epoch": 18.890156918687588,
      "grad_norm": 0.05233491584658623,
      "learning_rate": 3.7002222927913626e-05,
      "loss": 0.0539,
      "step": 9924
    },
    {
      "epoch": 18.89205896338564,
      "grad_norm": 0.16963283717632294,
      "learning_rate": 3.699587170530327e-05,
      "loss": 0.0695,
      "step": 9925
    },
    {
      "epoch": 18.89396100808369,
      "grad_norm": 0.04924313351511955,
      "learning_rate": 3.6989520482692916e-05,
      "loss": 0.069,
      "step": 9926
    },
    {
      "epoch": 18.895863052781742,
      "grad_norm": 0.1375628262758255,
      "learning_rate": 3.698316926008257e-05,
      "loss": 0.0702,
      "step": 9927
    },
    {
      "epoch": 18.89776509747979,
      "grad_norm": 0.10310663282871246,
      "learning_rate": 3.697681803747221e-05,
      "loss": 0.0487,
      "step": 9928
    },
    {
      "epoch": 18.89966714217784,
      "grad_norm": 0.04487544670701027,
      "learning_rate": 3.6970466814861865e-05,
      "loss": 0.0444,
      "step": 9929
    },
    {
      "epoch": 18.901569186875893,
      "grad_norm": 0.15529364347457886,
      "learning_rate": 3.696411559225151e-05,
      "loss": 0.0508,
      "step": 9930
    },
    {
      "epoch": 18.903471231573942,
      "grad_norm": 0.08870761096477509,
      "learning_rate": 3.6957764369641155e-05,
      "loss": 0.0575,
      "step": 9931
    },
    {
      "epoch": 18.90537327627199,
      "grad_norm": 0.06250828504562378,
      "learning_rate": 3.695141314703081e-05,
      "loss": 0.0562,
      "step": 9932
    },
    {
      "epoch": 18.907275320970044,
      "grad_norm": 0.08538121730089188,
      "learning_rate": 3.694506192442045e-05,
      "loss": 0.0466,
      "step": 9933
    },
    {
      "epoch": 18.909177365668093,
      "grad_norm": 0.0566830113530159,
      "learning_rate": 3.69387107018101e-05,
      "loss": 0.0609,
      "step": 9934
    },
    {
      "epoch": 18.911079410366142,
      "grad_norm": 0.04256056994199753,
      "learning_rate": 3.693235947919975e-05,
      "loss": 0.0433,
      "step": 9935
    },
    {
      "epoch": 18.912981455064195,
      "grad_norm": 0.0648518055677414,
      "learning_rate": 3.6926008256589394e-05,
      "loss": 0.0739,
      "step": 9936
    },
    {
      "epoch": 18.914883499762244,
      "grad_norm": 0.14472368359565735,
      "learning_rate": 3.6919657033979046e-05,
      "loss": 0.0407,
      "step": 9937
    },
    {
      "epoch": 18.916785544460296,
      "grad_norm": 0.14395765960216522,
      "learning_rate": 3.691330581136869e-05,
      "loss": 0.0698,
      "step": 9938
    },
    {
      "epoch": 18.918687589158345,
      "grad_norm": 0.08163261413574219,
      "learning_rate": 3.6906954588758336e-05,
      "loss": 0.0626,
      "step": 9939
    },
    {
      "epoch": 18.920589633856395,
      "grad_norm": 0.12085986137390137,
      "learning_rate": 3.690060336614799e-05,
      "loss": 0.0368,
      "step": 9940
    },
    {
      "epoch": 18.922491678554447,
      "grad_norm": 0.11507006734609604,
      "learning_rate": 3.689425214353763e-05,
      "loss": 0.068,
      "step": 9941
    },
    {
      "epoch": 18.924393723252496,
      "grad_norm": 0.10598558932542801,
      "learning_rate": 3.6887900920927285e-05,
      "loss": 0.0858,
      "step": 9942
    },
    {
      "epoch": 18.926295767950545,
      "grad_norm": 0.13431620597839355,
      "learning_rate": 3.688154969831692e-05,
      "loss": 0.0557,
      "step": 9943
    },
    {
      "epoch": 18.928197812648598,
      "grad_norm": 0.1234978586435318,
      "learning_rate": 3.6875198475706575e-05,
      "loss": 0.0528,
      "step": 9944
    },
    {
      "epoch": 18.930099857346647,
      "grad_norm": 0.34976011514663696,
      "learning_rate": 3.686884725309622e-05,
      "loss": 0.1134,
      "step": 9945
    },
    {
      "epoch": 18.932001902044696,
      "grad_norm": 0.1423528492450714,
      "learning_rate": 3.686249603048587e-05,
      "loss": 0.0527,
      "step": 9946
    },
    {
      "epoch": 18.93390394674275,
      "grad_norm": 0.3800347149372101,
      "learning_rate": 3.685614480787552e-05,
      "loss": 0.0717,
      "step": 9947
    },
    {
      "epoch": 18.935805991440798,
      "grad_norm": 0.09121809899806976,
      "learning_rate": 3.684979358526516e-05,
      "loss": 0.0526,
      "step": 9948
    },
    {
      "epoch": 18.93770803613885,
      "grad_norm": 0.17573939263820648,
      "learning_rate": 3.6843442362654814e-05,
      "loss": 0.0665,
      "step": 9949
    },
    {
      "epoch": 18.9396100808369,
      "grad_norm": 0.09202369302511215,
      "learning_rate": 3.683709114004446e-05,
      "loss": 0.0464,
      "step": 9950
    },
    {
      "epoch": 18.94151212553495,
      "grad_norm": 0.1471950262784958,
      "learning_rate": 3.683073991743411e-05,
      "loss": 0.0745,
      "step": 9951
    },
    {
      "epoch": 18.943414170233,
      "grad_norm": 0.16292431950569153,
      "learning_rate": 3.6824388694823756e-05,
      "loss": 0.0447,
      "step": 9952
    },
    {
      "epoch": 18.94531621493105,
      "grad_norm": 0.1641230583190918,
      "learning_rate": 3.68180374722134e-05,
      "loss": 0.0566,
      "step": 9953
    },
    {
      "epoch": 18.9472182596291,
      "grad_norm": 0.1794688105583191,
      "learning_rate": 3.681168624960305e-05,
      "loss": 0.0677,
      "step": 9954
    },
    {
      "epoch": 18.949120304327153,
      "grad_norm": 0.10057643055915833,
      "learning_rate": 3.68053350269927e-05,
      "loss": 0.056,
      "step": 9955
    },
    {
      "epoch": 18.9510223490252,
      "grad_norm": 0.08977890014648438,
      "learning_rate": 3.679898380438235e-05,
      "loss": 0.0816,
      "step": 9956
    },
    {
      "epoch": 18.95292439372325,
      "grad_norm": 0.1806146204471588,
      "learning_rate": 3.679263258177199e-05,
      "loss": 0.0523,
      "step": 9957
    },
    {
      "epoch": 18.954826438421303,
      "grad_norm": 0.1459428369998932,
      "learning_rate": 3.678628135916164e-05,
      "loss": 0.0549,
      "step": 9958
    },
    {
      "epoch": 18.956728483119353,
      "grad_norm": 0.08471745997667313,
      "learning_rate": 3.677993013655129e-05,
      "loss": 0.064,
      "step": 9959
    },
    {
      "epoch": 18.958630527817405,
      "grad_norm": 0.06056053936481476,
      "learning_rate": 3.6773578913940937e-05,
      "loss": 0.0492,
      "step": 9960
    },
    {
      "epoch": 18.960532572515454,
      "grad_norm": 0.09086448699235916,
      "learning_rate": 3.676722769133058e-05,
      "loss": 0.0583,
      "step": 9961
    },
    {
      "epoch": 18.962434617213503,
      "grad_norm": 0.10881451517343521,
      "learning_rate": 3.676087646872023e-05,
      "loss": 0.0544,
      "step": 9962
    },
    {
      "epoch": 18.964336661911556,
      "grad_norm": 0.1478363275527954,
      "learning_rate": 3.675452524610988e-05,
      "loss": 0.0497,
      "step": 9963
    },
    {
      "epoch": 18.966238706609605,
      "grad_norm": 0.19032606482505798,
      "learning_rate": 3.6748174023499524e-05,
      "loss": 0.0643,
      "step": 9964
    },
    {
      "epoch": 18.968140751307654,
      "grad_norm": 0.13042497634887695,
      "learning_rate": 3.6741822800889175e-05,
      "loss": 0.0437,
      "step": 9965
    },
    {
      "epoch": 18.970042796005707,
      "grad_norm": 0.0954694151878357,
      "learning_rate": 3.673547157827882e-05,
      "loss": 0.0472,
      "step": 9966
    },
    {
      "epoch": 18.971944840703756,
      "grad_norm": 0.06691458821296692,
      "learning_rate": 3.6729120355668466e-05,
      "loss": 0.0587,
      "step": 9967
    },
    {
      "epoch": 18.973846885401805,
      "grad_norm": 0.13316847383975983,
      "learning_rate": 3.672276913305812e-05,
      "loss": 0.0685,
      "step": 9968
    },
    {
      "epoch": 18.975748930099858,
      "grad_norm": 0.21592672169208527,
      "learning_rate": 3.671641791044776e-05,
      "loss": 0.0573,
      "step": 9969
    },
    {
      "epoch": 18.977650974797907,
      "grad_norm": 0.10553871840238571,
      "learning_rate": 3.671006668783741e-05,
      "loss": 0.0456,
      "step": 9970
    },
    {
      "epoch": 18.97955301949596,
      "grad_norm": 0.27601495385169983,
      "learning_rate": 3.670371546522706e-05,
      "loss": 0.068,
      "step": 9971
    },
    {
      "epoch": 18.98145506419401,
      "grad_norm": 0.09027394652366638,
      "learning_rate": 3.6697364242616704e-05,
      "loss": 0.0466,
      "step": 9972
    },
    {
      "epoch": 18.983357108892058,
      "grad_norm": 0.05911664292216301,
      "learning_rate": 3.6691013020006356e-05,
      "loss": 0.0467,
      "step": 9973
    },
    {
      "epoch": 18.98525915359011,
      "grad_norm": 0.10544353723526001,
      "learning_rate": 3.6684661797396e-05,
      "loss": 0.0597,
      "step": 9974
    },
    {
      "epoch": 18.98716119828816,
      "grad_norm": 0.09973032772541046,
      "learning_rate": 3.6678310574785646e-05,
      "loss": 0.0438,
      "step": 9975
    },
    {
      "epoch": 18.98906324298621,
      "grad_norm": 0.24099227786064148,
      "learning_rate": 3.667195935217529e-05,
      "loss": 0.0803,
      "step": 9976
    },
    {
      "epoch": 18.99096528768426,
      "grad_norm": 0.11139773577451706,
      "learning_rate": 3.666560812956494e-05,
      "loss": 0.0556,
      "step": 9977
    },
    {
      "epoch": 18.99286733238231,
      "grad_norm": 0.08617474138736725,
      "learning_rate": 3.6659256906954595e-05,
      "loss": 0.0469,
      "step": 9978
    },
    {
      "epoch": 18.99476937708036,
      "grad_norm": 0.12138453125953674,
      "learning_rate": 3.6652905684344233e-05,
      "loss": 0.0576,
      "step": 9979
    },
    {
      "epoch": 18.996671421778412,
      "grad_norm": 0.10799018293619156,
      "learning_rate": 3.6646554461733885e-05,
      "loss": 0.0406,
      "step": 9980
    },
    {
      "epoch": 18.99857346647646,
      "grad_norm": 0.12339373677968979,
      "learning_rate": 3.664020323912353e-05,
      "loss": 0.0572,
      "step": 9981
    },
    {
      "epoch": 19.000475511174514,
      "grad_norm": 0.39112773537635803,
      "learning_rate": 3.663385201651318e-05,
      "loss": 0.0725,
      "step": 9982
    },
    {
      "epoch": 19.002377555872563,
      "grad_norm": 0.05707325041294098,
      "learning_rate": 3.662750079390283e-05,
      "loss": 0.0684,
      "step": 9983
    },
    {
      "epoch": 19.004279600570612,
      "grad_norm": 0.0484582856297493,
      "learning_rate": 3.662114957129247e-05,
      "loss": 0.0501,
      "step": 9984
    },
    {
      "epoch": 19.006181645268665,
      "grad_norm": 0.20799849927425385,
      "learning_rate": 3.6614798348682124e-05,
      "loss": 0.0806,
      "step": 9985
    },
    {
      "epoch": 19.008083689966714,
      "grad_norm": 0.11114415526390076,
      "learning_rate": 3.660844712607177e-05,
      "loss": 0.0699,
      "step": 9986
    },
    {
      "epoch": 19.009985734664763,
      "grad_norm": 0.06936005502939224,
      "learning_rate": 3.660209590346142e-05,
      "loss": 0.0785,
      "step": 9987
    },
    {
      "epoch": 19.011887779362816,
      "grad_norm": 0.17966769635677338,
      "learning_rate": 3.6595744680851066e-05,
      "loss": 0.0642,
      "step": 9988
    },
    {
      "epoch": 19.013789824060865,
      "grad_norm": 0.057487085461616516,
      "learning_rate": 3.658939345824071e-05,
      "loss": 0.0541,
      "step": 9989
    },
    {
      "epoch": 19.015691868758918,
      "grad_norm": 0.215240940451622,
      "learning_rate": 3.658304223563036e-05,
      "loss": 0.0479,
      "step": 9990
    },
    {
      "epoch": 19.017593913456967,
      "grad_norm": 0.11331439763307571,
      "learning_rate": 3.657669101302001e-05,
      "loss": 0.0383,
      "step": 9991
    },
    {
      "epoch": 19.019495958155016,
      "grad_norm": 0.10905072838068008,
      "learning_rate": 3.657033979040966e-05,
      "loss": 0.0314,
      "step": 9992
    },
    {
      "epoch": 19.02139800285307,
      "grad_norm": 0.08001358062028885,
      "learning_rate": 3.65639885677993e-05,
      "loss": 0.0359,
      "step": 9993
    },
    {
      "epoch": 19.023300047551118,
      "grad_norm": 0.0985516756772995,
      "learning_rate": 3.655763734518895e-05,
      "loss": 0.0513,
      "step": 9994
    },
    {
      "epoch": 19.025202092249167,
      "grad_norm": 0.1317702978849411,
      "learning_rate": 3.65512861225786e-05,
      "loss": 0.0576,
      "step": 9995
    },
    {
      "epoch": 19.02710413694722,
      "grad_norm": 0.33076581358909607,
      "learning_rate": 3.654493489996825e-05,
      "loss": 0.0773,
      "step": 9996
    },
    {
      "epoch": 19.02900618164527,
      "grad_norm": 0.06298033893108368,
      "learning_rate": 3.653858367735789e-05,
      "loss": 0.0953,
      "step": 9997
    },
    {
      "epoch": 19.030908226343318,
      "grad_norm": 0.049048468470573425,
      "learning_rate": 3.653223245474754e-05,
      "loss": 0.0387,
      "step": 9998
    },
    {
      "epoch": 19.03281027104137,
      "grad_norm": 0.14701910316944122,
      "learning_rate": 3.652588123213719e-05,
      "loss": 0.0793,
      "step": 9999
    },
    {
      "epoch": 19.03471231573942,
      "grad_norm": 0.16040882468223572,
      "learning_rate": 3.6519530009526834e-05,
      "loss": 0.0787,
      "step": 10000
    },
    {
      "epoch": 19.036614360437472,
      "grad_norm": 0.08740217983722687,
      "learning_rate": 3.6513178786916486e-05,
      "loss": 0.0637,
      "step": 10001
    },
    {
      "epoch": 19.03851640513552,
      "grad_norm": 0.20253409445285797,
      "learning_rate": 3.650682756430613e-05,
      "loss": 0.0554,
      "step": 10002
    },
    {
      "epoch": 19.04041844983357,
      "grad_norm": 0.06349057704210281,
      "learning_rate": 3.6500476341695776e-05,
      "loss": 0.0603,
      "step": 10003
    },
    {
      "epoch": 19.042320494531623,
      "grad_norm": 0.17448563873767853,
      "learning_rate": 3.649412511908543e-05,
      "loss": 0.0405,
      "step": 10004
    },
    {
      "epoch": 19.044222539229672,
      "grad_norm": 0.15310613811016083,
      "learning_rate": 3.648777389647507e-05,
      "loss": 0.0569,
      "step": 10005
    },
    {
      "epoch": 19.04612458392772,
      "grad_norm": 0.03419190272688866,
      "learning_rate": 3.648142267386472e-05,
      "loss": 0.0484,
      "step": 10006
    },
    {
      "epoch": 19.048026628625774,
      "grad_norm": 0.06719169020652771,
      "learning_rate": 3.647507145125437e-05,
      "loss": 0.0813,
      "step": 10007
    },
    {
      "epoch": 19.049928673323823,
      "grad_norm": 0.10958858579397202,
      "learning_rate": 3.6468720228644015e-05,
      "loss": 0.0634,
      "step": 10008
    },
    {
      "epoch": 19.051830718021872,
      "grad_norm": 0.12859925627708435,
      "learning_rate": 3.646236900603367e-05,
      "loss": 0.06,
      "step": 10009
    },
    {
      "epoch": 19.053732762719925,
      "grad_norm": 0.1362864226102829,
      "learning_rate": 3.645601778342331e-05,
      "loss": 0.0487,
      "step": 10010
    },
    {
      "epoch": 19.055634807417974,
      "grad_norm": 0.05309566110372543,
      "learning_rate": 3.644966656081296e-05,
      "loss": 0.0892,
      "step": 10011
    },
    {
      "epoch": 19.057536852116026,
      "grad_norm": 0.05769931152462959,
      "learning_rate": 3.64433153382026e-05,
      "loss": 0.041,
      "step": 10012
    },
    {
      "epoch": 19.059438896814076,
      "grad_norm": 0.0420185886323452,
      "learning_rate": 3.6436964115592254e-05,
      "loss": 0.0561,
      "step": 10013
    },
    {
      "epoch": 19.061340941512125,
      "grad_norm": 0.11808836460113525,
      "learning_rate": 3.6430612892981906e-05,
      "loss": 0.0507,
      "step": 10014
    },
    {
      "epoch": 19.063242986210177,
      "grad_norm": 0.10188145935535431,
      "learning_rate": 3.6424261670371544e-05,
      "loss": 0.0639,
      "step": 10015
    },
    {
      "epoch": 19.065145030908226,
      "grad_norm": 0.12096687406301498,
      "learning_rate": 3.6417910447761196e-05,
      "loss": 0.0582,
      "step": 10016
    },
    {
      "epoch": 19.067047075606276,
      "grad_norm": 0.07498153299093246,
      "learning_rate": 3.641155922515084e-05,
      "loss": 0.0539,
      "step": 10017
    },
    {
      "epoch": 19.068949120304328,
      "grad_norm": 0.15140143036842346,
      "learning_rate": 3.640520800254049e-05,
      "loss": 0.0543,
      "step": 10018
    },
    {
      "epoch": 19.070851165002377,
      "grad_norm": 0.11248353868722916,
      "learning_rate": 3.639885677993014e-05,
      "loss": 0.0522,
      "step": 10019
    },
    {
      "epoch": 19.072753209700426,
      "grad_norm": 0.13222387433052063,
      "learning_rate": 3.639250555731978e-05,
      "loss": 0.0698,
      "step": 10020
    },
    {
      "epoch": 19.07465525439848,
      "grad_norm": 0.09478526562452316,
      "learning_rate": 3.6386154334709435e-05,
      "loss": 0.041,
      "step": 10021
    },
    {
      "epoch": 19.076557299096528,
      "grad_norm": 0.049942344427108765,
      "learning_rate": 3.637980311209908e-05,
      "loss": 0.0501,
      "step": 10022
    },
    {
      "epoch": 19.07845934379458,
      "grad_norm": 0.08634361624717712,
      "learning_rate": 3.637345188948873e-05,
      "loss": 0.0555,
      "step": 10023
    },
    {
      "epoch": 19.08036138849263,
      "grad_norm": 0.14032310247421265,
      "learning_rate": 3.636710066687837e-05,
      "loss": 0.0462,
      "step": 10024
    },
    {
      "epoch": 19.08226343319068,
      "grad_norm": 0.117254339158535,
      "learning_rate": 3.636074944426802e-05,
      "loss": 0.0541,
      "step": 10025
    },
    {
      "epoch": 19.08416547788873,
      "grad_norm": 0.05373654142022133,
      "learning_rate": 3.6354398221657673e-05,
      "loss": 0.0446,
      "step": 10026
    },
    {
      "epoch": 19.08606752258678,
      "grad_norm": 0.11854445934295654,
      "learning_rate": 3.634804699904732e-05,
      "loss": 0.0596,
      "step": 10027
    },
    {
      "epoch": 19.08796956728483,
      "grad_norm": 0.18134529888629913,
      "learning_rate": 3.634169577643697e-05,
      "loss": 0.0587,
      "step": 10028
    },
    {
      "epoch": 19.089871611982883,
      "grad_norm": 0.10500608384609222,
      "learning_rate": 3.633534455382661e-05,
      "loss": 0.0683,
      "step": 10029
    },
    {
      "epoch": 19.09177365668093,
      "grad_norm": 0.10136354714632034,
      "learning_rate": 3.632899333121626e-05,
      "loss": 0.0458,
      "step": 10030
    },
    {
      "epoch": 19.09367570137898,
      "grad_norm": 0.13646340370178223,
      "learning_rate": 3.6322642108605906e-05,
      "loss": 0.0293,
      "step": 10031
    },
    {
      "epoch": 19.095577746077034,
      "grad_norm": 0.10555324703454971,
      "learning_rate": 3.631629088599556e-05,
      "loss": 0.0585,
      "step": 10032
    },
    {
      "epoch": 19.097479790775083,
      "grad_norm": 0.15480376780033112,
      "learning_rate": 3.63099396633852e-05,
      "loss": 0.0542,
      "step": 10033
    },
    {
      "epoch": 19.099381835473135,
      "grad_norm": 0.21401821076869965,
      "learning_rate": 3.630358844077485e-05,
      "loss": 0.0544,
      "step": 10034
    },
    {
      "epoch": 19.101283880171184,
      "grad_norm": 0.08610215783119202,
      "learning_rate": 3.62972372181645e-05,
      "loss": 0.0485,
      "step": 10035
    },
    {
      "epoch": 19.103185924869234,
      "grad_norm": 0.07398564368486404,
      "learning_rate": 3.6290885995554144e-05,
      "loss": 0.0466,
      "step": 10036
    },
    {
      "epoch": 19.105087969567286,
      "grad_norm": 0.05519363284111023,
      "learning_rate": 3.6284534772943796e-05,
      "loss": 0.0572,
      "step": 10037
    },
    {
      "epoch": 19.106990014265335,
      "grad_norm": 0.09300468862056732,
      "learning_rate": 3.627818355033344e-05,
      "loss": 0.0286,
      "step": 10038
    },
    {
      "epoch": 19.108892058963384,
      "grad_norm": 0.07158180326223373,
      "learning_rate": 3.6271832327723086e-05,
      "loss": 0.064,
      "step": 10039
    },
    {
      "epoch": 19.110794103661437,
      "grad_norm": 0.1243014857172966,
      "learning_rate": 3.626548110511274e-05,
      "loss": 0.0739,
      "step": 10040
    },
    {
      "epoch": 19.112696148359486,
      "grad_norm": 0.10833297669887543,
      "learning_rate": 3.625912988250238e-05,
      "loss": 0.0593,
      "step": 10041
    },
    {
      "epoch": 19.114598193057535,
      "grad_norm": 0.04479862004518509,
      "learning_rate": 3.625277865989203e-05,
      "loss": 0.0513,
      "step": 10042
    },
    {
      "epoch": 19.116500237755588,
      "grad_norm": 0.051182232797145844,
      "learning_rate": 3.624642743728168e-05,
      "loss": 0.0666,
      "step": 10043
    },
    {
      "epoch": 19.118402282453637,
      "grad_norm": 0.07713011652231216,
      "learning_rate": 3.6240076214671325e-05,
      "loss": 0.0575,
      "step": 10044
    },
    {
      "epoch": 19.12030432715169,
      "grad_norm": 0.11209198087453842,
      "learning_rate": 3.623372499206098e-05,
      "loss": 0.0653,
      "step": 10045
    },
    {
      "epoch": 19.12220637184974,
      "grad_norm": 0.1346711814403534,
      "learning_rate": 3.622737376945062e-05,
      "loss": 0.0348,
      "step": 10046
    },
    {
      "epoch": 19.124108416547788,
      "grad_norm": 0.09184443205595016,
      "learning_rate": 3.622102254684027e-05,
      "loss": 0.0624,
      "step": 10047
    },
    {
      "epoch": 19.12601046124584,
      "grad_norm": 0.08998651057481766,
      "learning_rate": 3.621467132422991e-05,
      "loss": 0.0458,
      "step": 10048
    },
    {
      "epoch": 19.12791250594389,
      "grad_norm": 0.16010458767414093,
      "learning_rate": 3.6208320101619564e-05,
      "loss": 0.0558,
      "step": 10049
    },
    {
      "epoch": 19.12981455064194,
      "grad_norm": 0.08595812320709229,
      "learning_rate": 3.6201968879009216e-05,
      "loss": 0.0514,
      "step": 10050
    },
    {
      "epoch": 19.13171659533999,
      "grad_norm": 0.09739263355731964,
      "learning_rate": 3.6195617656398854e-05,
      "loss": 0.0652,
      "step": 10051
    },
    {
      "epoch": 19.13361864003804,
      "grad_norm": 0.1122368648648262,
      "learning_rate": 3.6189266433788506e-05,
      "loss": 0.0547,
      "step": 10052
    },
    {
      "epoch": 19.13552068473609,
      "grad_norm": 0.16903916001319885,
      "learning_rate": 3.618291521117815e-05,
      "loss": 0.048,
      "step": 10053
    },
    {
      "epoch": 19.137422729434142,
      "grad_norm": 0.046694718301296234,
      "learning_rate": 3.61765639885678e-05,
      "loss": 0.0607,
      "step": 10054
    },
    {
      "epoch": 19.13932477413219,
      "grad_norm": 0.07032367587089539,
      "learning_rate": 3.617021276595745e-05,
      "loss": 0.0493,
      "step": 10055
    },
    {
      "epoch": 19.141226818830244,
      "grad_norm": 0.06717967242002487,
      "learning_rate": 3.616386154334709e-05,
      "loss": 0.0499,
      "step": 10056
    },
    {
      "epoch": 19.143128863528293,
      "grad_norm": 0.06115768849849701,
      "learning_rate": 3.6157510320736745e-05,
      "loss": 0.0614,
      "step": 10057
    },
    {
      "epoch": 19.145030908226342,
      "grad_norm": 0.0605919249355793,
      "learning_rate": 3.615115909812639e-05,
      "loss": 0.0357,
      "step": 10058
    },
    {
      "epoch": 19.146932952924395,
      "grad_norm": 0.1991409808397293,
      "learning_rate": 3.614480787551604e-05,
      "loss": 0.0538,
      "step": 10059
    },
    {
      "epoch": 19.148834997622444,
      "grad_norm": 0.19580891728401184,
      "learning_rate": 3.613845665290568e-05,
      "loss": 0.0652,
      "step": 10060
    },
    {
      "epoch": 19.150737042320493,
      "grad_norm": 0.13768815994262695,
      "learning_rate": 3.613210543029533e-05,
      "loss": 0.057,
      "step": 10061
    },
    {
      "epoch": 19.152639087018546,
      "grad_norm": 0.347734272480011,
      "learning_rate": 3.6125754207684984e-05,
      "loss": 0.0477,
      "step": 10062
    },
    {
      "epoch": 19.154541131716595,
      "grad_norm": 0.13527154922485352,
      "learning_rate": 3.611940298507463e-05,
      "loss": 0.052,
      "step": 10063
    },
    {
      "epoch": 19.156443176414644,
      "grad_norm": 0.2703625559806824,
      "learning_rate": 3.611305176246428e-05,
      "loss": 0.063,
      "step": 10064
    },
    {
      "epoch": 19.158345221112697,
      "grad_norm": 0.09101396054029465,
      "learning_rate": 3.610670053985392e-05,
      "loss": 0.0661,
      "step": 10065
    },
    {
      "epoch": 19.160247265810746,
      "grad_norm": 0.15997478365898132,
      "learning_rate": 3.610034931724357e-05,
      "loss": 0.0499,
      "step": 10066
    },
    {
      "epoch": 19.1621493105088,
      "grad_norm": 0.06663242727518082,
      "learning_rate": 3.6093998094633216e-05,
      "loss": 0.06,
      "step": 10067
    },
    {
      "epoch": 19.164051355206848,
      "grad_norm": 0.11330113559961319,
      "learning_rate": 3.608764687202287e-05,
      "loss": 0.0418,
      "step": 10068
    },
    {
      "epoch": 19.165953399904897,
      "grad_norm": 0.10328996926546097,
      "learning_rate": 3.608129564941251e-05,
      "loss": 0.0581,
      "step": 10069
    },
    {
      "epoch": 19.16785544460295,
      "grad_norm": 0.043847475200891495,
      "learning_rate": 3.607494442680216e-05,
      "loss": 0.0574,
      "step": 10070
    },
    {
      "epoch": 19.169757489301,
      "grad_norm": 0.049880996346473694,
      "learning_rate": 3.606859320419181e-05,
      "loss": 0.0479,
      "step": 10071
    },
    {
      "epoch": 19.171659533999048,
      "grad_norm": 0.2610563039779663,
      "learning_rate": 3.6062241981581455e-05,
      "loss": 0.063,
      "step": 10072
    },
    {
      "epoch": 19.1735615786971,
      "grad_norm": 0.08486849814653397,
      "learning_rate": 3.605589075897111e-05,
      "loss": 0.0641,
      "step": 10073
    },
    {
      "epoch": 19.17546362339515,
      "grad_norm": 0.0873064175248146,
      "learning_rate": 3.604953953636075e-05,
      "loss": 0.0556,
      "step": 10074
    },
    {
      "epoch": 19.1773656680932,
      "grad_norm": 0.0841386616230011,
      "learning_rate": 3.60431883137504e-05,
      "loss": 0.0572,
      "step": 10075
    },
    {
      "epoch": 19.17926771279125,
      "grad_norm": 0.1779516637325287,
      "learning_rate": 3.603683709114005e-05,
      "loss": 0.0511,
      "step": 10076
    },
    {
      "epoch": 19.1811697574893,
      "grad_norm": 0.09647682309150696,
      "learning_rate": 3.6030485868529694e-05,
      "loss": 0.0716,
      "step": 10077
    },
    {
      "epoch": 19.183071802187353,
      "grad_norm": 0.08160854876041412,
      "learning_rate": 3.602413464591934e-05,
      "loss": 0.0598,
      "step": 10078
    },
    {
      "epoch": 19.184973846885402,
      "grad_norm": 0.06882409751415253,
      "learning_rate": 3.6017783423308984e-05,
      "loss": 0.0517,
      "step": 10079
    },
    {
      "epoch": 19.18687589158345,
      "grad_norm": 0.08491438627243042,
      "learning_rate": 3.6011432200698636e-05,
      "loss": 0.0561,
      "step": 10080
    },
    {
      "epoch": 19.188777936281504,
      "grad_norm": 0.343395859003067,
      "learning_rate": 3.600508097808829e-05,
      "loss": 0.0768,
      "step": 10081
    },
    {
      "epoch": 19.190679980979553,
      "grad_norm": 0.1169079914689064,
      "learning_rate": 3.599872975547793e-05,
      "loss": 0.0423,
      "step": 10082
    },
    {
      "epoch": 19.192582025677602,
      "grad_norm": 0.24359755218029022,
      "learning_rate": 3.599237853286758e-05,
      "loss": 0.0738,
      "step": 10083
    },
    {
      "epoch": 19.194484070375655,
      "grad_norm": 0.06958276033401489,
      "learning_rate": 3.598602731025722e-05,
      "loss": 0.047,
      "step": 10084
    },
    {
      "epoch": 19.196386115073704,
      "grad_norm": 0.07158949971199036,
      "learning_rate": 3.5979676087646875e-05,
      "loss": 0.027,
      "step": 10085
    },
    {
      "epoch": 19.198288159771753,
      "grad_norm": 0.14709332585334778,
      "learning_rate": 3.597332486503652e-05,
      "loss": 0.0518,
      "step": 10086
    },
    {
      "epoch": 19.200190204469806,
      "grad_norm": 0.09249371290206909,
      "learning_rate": 3.5966973642426165e-05,
      "loss": 0.0563,
      "step": 10087
    },
    {
      "epoch": 19.202092249167855,
      "grad_norm": 0.059265051037073135,
      "learning_rate": 3.5960622419815817e-05,
      "loss": 0.0517,
      "step": 10088
    },
    {
      "epoch": 19.203994293865907,
      "grad_norm": 0.1328250616788864,
      "learning_rate": 3.595427119720546e-05,
      "loss": 0.067,
      "step": 10089
    },
    {
      "epoch": 19.205896338563957,
      "grad_norm": 0.04016127064824104,
      "learning_rate": 3.5947919974595113e-05,
      "loss": 0.0398,
      "step": 10090
    },
    {
      "epoch": 19.207798383262006,
      "grad_norm": 0.12123529613018036,
      "learning_rate": 3.594156875198476e-05,
      "loss": 0.0425,
      "step": 10091
    },
    {
      "epoch": 19.20970042796006,
      "grad_norm": 0.12374745309352875,
      "learning_rate": 3.5935217529374404e-05,
      "loss": 0.0597,
      "step": 10092
    },
    {
      "epoch": 19.211602472658107,
      "grad_norm": 0.13038672506809235,
      "learning_rate": 3.5928866306764055e-05,
      "loss": 0.06,
      "step": 10093
    },
    {
      "epoch": 19.213504517356156,
      "grad_norm": 0.10279489308595657,
      "learning_rate": 3.59225150841537e-05,
      "loss": 0.0563,
      "step": 10094
    },
    {
      "epoch": 19.21540656205421,
      "grad_norm": 0.12781091034412384,
      "learning_rate": 3.591616386154335e-05,
      "loss": 0.0603,
      "step": 10095
    },
    {
      "epoch": 19.21730860675226,
      "grad_norm": 0.09477554261684418,
      "learning_rate": 3.590981263893299e-05,
      "loss": 0.0782,
      "step": 10096
    },
    {
      "epoch": 19.219210651450307,
      "grad_norm": 0.07034879177808762,
      "learning_rate": 3.590346141632264e-05,
      "loss": 0.0505,
      "step": 10097
    },
    {
      "epoch": 19.22111269614836,
      "grad_norm": 0.06071623042225838,
      "learning_rate": 3.5897110193712294e-05,
      "loss": 0.045,
      "step": 10098
    },
    {
      "epoch": 19.22301474084641,
      "grad_norm": 0.039386387914419174,
      "learning_rate": 3.589075897110194e-05,
      "loss": 0.0487,
      "step": 10099
    },
    {
      "epoch": 19.224916785544462,
      "grad_norm": 0.1408051699399948,
      "learning_rate": 3.588440774849159e-05,
      "loss": 0.0458,
      "step": 10100
    },
    {
      "epoch": 19.22681883024251,
      "grad_norm": 0.10527674853801727,
      "learning_rate": 3.587805652588123e-05,
      "loss": 0.084,
      "step": 10101
    },
    {
      "epoch": 19.22872087494056,
      "grad_norm": 0.2394716739654541,
      "learning_rate": 3.587170530327088e-05,
      "loss": 0.0507,
      "step": 10102
    },
    {
      "epoch": 19.230622919638613,
      "grad_norm": 0.13282881677150726,
      "learning_rate": 3.5865354080660526e-05,
      "loss": 0.0416,
      "step": 10103
    },
    {
      "epoch": 19.232524964336662,
      "grad_norm": 0.13951736688613892,
      "learning_rate": 3.585900285805018e-05,
      "loss": 0.0668,
      "step": 10104
    },
    {
      "epoch": 19.23442700903471,
      "grad_norm": 0.09385409951210022,
      "learning_rate": 3.585265163543982e-05,
      "loss": 0.0596,
      "step": 10105
    },
    {
      "epoch": 19.236329053732764,
      "grad_norm": 0.11713048070669174,
      "learning_rate": 3.584630041282947e-05,
      "loss": 0.0675,
      "step": 10106
    },
    {
      "epoch": 19.238231098430813,
      "grad_norm": 0.14428676664829254,
      "learning_rate": 3.583994919021912e-05,
      "loss": 0.0602,
      "step": 10107
    },
    {
      "epoch": 19.240133143128862,
      "grad_norm": 0.04688987135887146,
      "learning_rate": 3.5833597967608765e-05,
      "loss": 0.0541,
      "step": 10108
    },
    {
      "epoch": 19.242035187826914,
      "grad_norm": 0.20087163150310516,
      "learning_rate": 3.582724674499842e-05,
      "loss": 0.0473,
      "step": 10109
    },
    {
      "epoch": 19.243937232524964,
      "grad_norm": 0.165956050157547,
      "learning_rate": 3.582089552238806e-05,
      "loss": 0.0624,
      "step": 10110
    },
    {
      "epoch": 19.245839277223016,
      "grad_norm": 0.16869060695171356,
      "learning_rate": 3.581454429977771e-05,
      "loss": 0.0597,
      "step": 10111
    },
    {
      "epoch": 19.247741321921065,
      "grad_norm": 0.10722652822732925,
      "learning_rate": 3.580819307716736e-05,
      "loss": 0.0722,
      "step": 10112
    },
    {
      "epoch": 19.249643366619114,
      "grad_norm": 0.07774709910154343,
      "learning_rate": 3.5801841854557004e-05,
      "loss": 0.0338,
      "step": 10113
    },
    {
      "epoch": 19.251545411317167,
      "grad_norm": 0.13211393356323242,
      "learning_rate": 3.579549063194665e-05,
      "loss": 0.0511,
      "step": 10114
    },
    {
      "epoch": 19.253447456015216,
      "grad_norm": 0.1211496964097023,
      "learning_rate": 3.5789139409336294e-05,
      "loss": 0.0486,
      "step": 10115
    },
    {
      "epoch": 19.255349500713265,
      "grad_norm": 0.10808859020471573,
      "learning_rate": 3.5782788186725946e-05,
      "loss": 0.0367,
      "step": 10116
    },
    {
      "epoch": 19.257251545411318,
      "grad_norm": 0.10460115969181061,
      "learning_rate": 3.57764369641156e-05,
      "loss": 0.0577,
      "step": 10117
    },
    {
      "epoch": 19.259153590109367,
      "grad_norm": 0.0897640734910965,
      "learning_rate": 3.577008574150524e-05,
      "loss": 0.0628,
      "step": 10118
    },
    {
      "epoch": 19.261055634807416,
      "grad_norm": 0.12556083500385284,
      "learning_rate": 3.576373451889489e-05,
      "loss": 0.0636,
      "step": 10119
    },
    {
      "epoch": 19.26295767950547,
      "grad_norm": 0.13245490193367004,
      "learning_rate": 3.575738329628453e-05,
      "loss": 0.0432,
      "step": 10120
    },
    {
      "epoch": 19.264859724203518,
      "grad_norm": 0.15191958844661713,
      "learning_rate": 3.5751032073674185e-05,
      "loss": 0.0657,
      "step": 10121
    },
    {
      "epoch": 19.26676176890157,
      "grad_norm": 0.04129410162568092,
      "learning_rate": 3.574468085106383e-05,
      "loss": 0.0443,
      "step": 10122
    },
    {
      "epoch": 19.26866381359962,
      "grad_norm": 0.3152913451194763,
      "learning_rate": 3.5738329628453475e-05,
      "loss": 0.0739,
      "step": 10123
    },
    {
      "epoch": 19.27056585829767,
      "grad_norm": 0.16405954957008362,
      "learning_rate": 3.573197840584313e-05,
      "loss": 0.0365,
      "step": 10124
    },
    {
      "epoch": 19.27246790299572,
      "grad_norm": 0.1917743682861328,
      "learning_rate": 3.572562718323277e-05,
      "loss": 0.0369,
      "step": 10125
    },
    {
      "epoch": 19.27436994769377,
      "grad_norm": 0.09143206477165222,
      "learning_rate": 3.5719275960622424e-05,
      "loss": 0.055,
      "step": 10126
    },
    {
      "epoch": 19.27627199239182,
      "grad_norm": 0.13609574735164642,
      "learning_rate": 3.571292473801207e-05,
      "loss": 0.0459,
      "step": 10127
    },
    {
      "epoch": 19.278174037089872,
      "grad_norm": 0.40472212433815,
      "learning_rate": 3.5706573515401714e-05,
      "loss": 0.0825,
      "step": 10128
    },
    {
      "epoch": 19.28007608178792,
      "grad_norm": 0.1614009439945221,
      "learning_rate": 3.5700222292791366e-05,
      "loss": 0.059,
      "step": 10129
    },
    {
      "epoch": 19.281978126485974,
      "grad_norm": 0.06693167984485626,
      "learning_rate": 3.569387107018101e-05,
      "loss": 0.0489,
      "step": 10130
    },
    {
      "epoch": 19.283880171184023,
      "grad_norm": 0.10688638687133789,
      "learning_rate": 3.568751984757066e-05,
      "loss": 0.0567,
      "step": 10131
    },
    {
      "epoch": 19.285782215882072,
      "grad_norm": 0.4077149033546448,
      "learning_rate": 3.56811686249603e-05,
      "loss": 0.0406,
      "step": 10132
    },
    {
      "epoch": 19.287684260580125,
      "grad_norm": 0.1552218645811081,
      "learning_rate": 3.567481740234995e-05,
      "loss": 0.0546,
      "step": 10133
    },
    {
      "epoch": 19.289586305278174,
      "grad_norm": 0.1504446119070053,
      "learning_rate": 3.56684661797396e-05,
      "loss": 0.0776,
      "step": 10134
    },
    {
      "epoch": 19.291488349976223,
      "grad_norm": 0.10397607088088989,
      "learning_rate": 3.566211495712925e-05,
      "loss": 0.0492,
      "step": 10135
    },
    {
      "epoch": 19.293390394674276,
      "grad_norm": 0.10803500562906265,
      "learning_rate": 3.56557637345189e-05,
      "loss": 0.0598,
      "step": 10136
    },
    {
      "epoch": 19.295292439372325,
      "grad_norm": 0.13279162347316742,
      "learning_rate": 3.564941251190854e-05,
      "loss": 0.0542,
      "step": 10137
    },
    {
      "epoch": 19.297194484070374,
      "grad_norm": 0.11373448371887207,
      "learning_rate": 3.564306128929819e-05,
      "loss": 0.0504,
      "step": 10138
    },
    {
      "epoch": 19.299096528768427,
      "grad_norm": 0.10581234842538834,
      "learning_rate": 3.563671006668784e-05,
      "loss": 0.0709,
      "step": 10139
    },
    {
      "epoch": 19.300998573466476,
      "grad_norm": 0.07865399122238159,
      "learning_rate": 3.563035884407749e-05,
      "loss": 0.0638,
      "step": 10140
    },
    {
      "epoch": 19.30290061816453,
      "grad_norm": 0.050489168614149094,
      "learning_rate": 3.5624007621467134e-05,
      "loss": 0.0779,
      "step": 10141
    },
    {
      "epoch": 19.304802662862578,
      "grad_norm": 0.10635179281234741,
      "learning_rate": 3.561765639885678e-05,
      "loss": 0.0646,
      "step": 10142
    },
    {
      "epoch": 19.306704707560627,
      "grad_norm": 0.12745051085948944,
      "learning_rate": 3.561130517624643e-05,
      "loss": 0.0537,
      "step": 10143
    },
    {
      "epoch": 19.30860675225868,
      "grad_norm": 0.07300863415002823,
      "learning_rate": 3.5604953953636076e-05,
      "loss": 0.0554,
      "step": 10144
    },
    {
      "epoch": 19.31050879695673,
      "grad_norm": 0.1162499487400055,
      "learning_rate": 3.559860273102573e-05,
      "loss": 0.0518,
      "step": 10145
    },
    {
      "epoch": 19.312410841654778,
      "grad_norm": 0.11278656125068665,
      "learning_rate": 3.559225150841537e-05,
      "loss": 0.0566,
      "step": 10146
    },
    {
      "epoch": 19.31431288635283,
      "grad_norm": 0.21737448871135712,
      "learning_rate": 3.558590028580502e-05,
      "loss": 0.0715,
      "step": 10147
    },
    {
      "epoch": 19.31621493105088,
      "grad_norm": 0.08031684905290604,
      "learning_rate": 3.557954906319467e-05,
      "loss": 0.0521,
      "step": 10148
    },
    {
      "epoch": 19.31811697574893,
      "grad_norm": 0.046094682067632675,
      "learning_rate": 3.5573197840584315e-05,
      "loss": 0.0427,
      "step": 10149
    },
    {
      "epoch": 19.32001902044698,
      "grad_norm": 0.10708703100681305,
      "learning_rate": 3.556684661797396e-05,
      "loss": 0.0519,
      "step": 10150
    },
    {
      "epoch": 19.32192106514503,
      "grad_norm": 0.34062817692756653,
      "learning_rate": 3.5560495395363605e-05,
      "loss": 0.07,
      "step": 10151
    },
    {
      "epoch": 19.323823109843083,
      "grad_norm": 0.05861439183354378,
      "learning_rate": 3.555414417275326e-05,
      "loss": 0.0484,
      "step": 10152
    },
    {
      "epoch": 19.325725154541132,
      "grad_norm": 0.08207020908594131,
      "learning_rate": 3.554779295014291e-05,
      "loss": 0.0504,
      "step": 10153
    },
    {
      "epoch": 19.32762719923918,
      "grad_norm": 0.10471838712692261,
      "learning_rate": 3.5541441727532554e-05,
      "loss": 0.0443,
      "step": 10154
    },
    {
      "epoch": 19.329529243937234,
      "grad_norm": 0.15948818624019623,
      "learning_rate": 3.55350905049222e-05,
      "loss": 0.0551,
      "step": 10155
    },
    {
      "epoch": 19.331431288635283,
      "grad_norm": 0.06961742043495178,
      "learning_rate": 3.5528739282311844e-05,
      "loss": 0.0376,
      "step": 10156
    },
    {
      "epoch": 19.333333333333332,
      "grad_norm": 0.04763415455818176,
      "learning_rate": 3.5522388059701495e-05,
      "loss": 0.0497,
      "step": 10157
    },
    {
      "epoch": 19.335235378031385,
      "grad_norm": 0.07544784247875214,
      "learning_rate": 3.551603683709114e-05,
      "loss": 0.0484,
      "step": 10158
    },
    {
      "epoch": 19.337137422729434,
      "grad_norm": 0.05961260944604874,
      "learning_rate": 3.5509685614480786e-05,
      "loss": 0.0528,
      "step": 10159
    },
    {
      "epoch": 19.339039467427483,
      "grad_norm": 0.03512318432331085,
      "learning_rate": 3.550333439187044e-05,
      "loss": 0.0452,
      "step": 10160
    },
    {
      "epoch": 19.340941512125536,
      "grad_norm": 0.12750087678432465,
      "learning_rate": 3.549698316926008e-05,
      "loss": 0.0486,
      "step": 10161
    },
    {
      "epoch": 19.342843556823585,
      "grad_norm": 0.1510218381881714,
      "learning_rate": 3.5490631946649734e-05,
      "loss": 0.0548,
      "step": 10162
    },
    {
      "epoch": 19.344745601521637,
      "grad_norm": 0.03664156794548035,
      "learning_rate": 3.548428072403938e-05,
      "loss": 0.0506,
      "step": 10163
    },
    {
      "epoch": 19.346647646219687,
      "grad_norm": 0.30709025263786316,
      "learning_rate": 3.5477929501429025e-05,
      "loss": 0.0758,
      "step": 10164
    },
    {
      "epoch": 19.348549690917736,
      "grad_norm": 0.1910550892353058,
      "learning_rate": 3.5471578278818676e-05,
      "loss": 0.0845,
      "step": 10165
    },
    {
      "epoch": 19.35045173561579,
      "grad_norm": 0.3427481949329376,
      "learning_rate": 3.546522705620832e-05,
      "loss": 0.0534,
      "step": 10166
    },
    {
      "epoch": 19.352353780313837,
      "grad_norm": 0.13792464137077332,
      "learning_rate": 3.545887583359797e-05,
      "loss": 0.0554,
      "step": 10167
    },
    {
      "epoch": 19.354255825011887,
      "grad_norm": 0.09045601636171341,
      "learning_rate": 3.545252461098761e-05,
      "loss": 0.0398,
      "step": 10168
    },
    {
      "epoch": 19.35615786970994,
      "grad_norm": 0.06533253937959671,
      "learning_rate": 3.544617338837726e-05,
      "loss": 0.0315,
      "step": 10169
    },
    {
      "epoch": 19.35805991440799,
      "grad_norm": 0.10264572501182556,
      "learning_rate": 3.543982216576691e-05,
      "loss": 0.0408,
      "step": 10170
    },
    {
      "epoch": 19.359961959106037,
      "grad_norm": 0.04178214818239212,
      "learning_rate": 3.543347094315656e-05,
      "loss": 0.0575,
      "step": 10171
    },
    {
      "epoch": 19.36186400380409,
      "grad_norm": 0.05626368522644043,
      "learning_rate": 3.542711972054621e-05,
      "loss": 0.0427,
      "step": 10172
    },
    {
      "epoch": 19.36376604850214,
      "grad_norm": 0.13891632854938507,
      "learning_rate": 3.542076849793585e-05,
      "loss": 0.0465,
      "step": 10173
    },
    {
      "epoch": 19.365668093200192,
      "grad_norm": 0.09107466787099838,
      "learning_rate": 3.54144172753255e-05,
      "loss": 0.0623,
      "step": 10174
    },
    {
      "epoch": 19.36757013789824,
      "grad_norm": 0.09597960114479065,
      "learning_rate": 3.540806605271515e-05,
      "loss": 0.0485,
      "step": 10175
    },
    {
      "epoch": 19.36947218259629,
      "grad_norm": 0.07946845144033432,
      "learning_rate": 3.54017148301048e-05,
      "loss": 0.0541,
      "step": 10176
    },
    {
      "epoch": 19.371374227294343,
      "grad_norm": 0.16330133378505707,
      "learning_rate": 3.5395363607494444e-05,
      "loss": 0.063,
      "step": 10177
    },
    {
      "epoch": 19.373276271992392,
      "grad_norm": 0.060695718973875046,
      "learning_rate": 3.538901238488409e-05,
      "loss": 0.0661,
      "step": 10178
    },
    {
      "epoch": 19.37517831669044,
      "grad_norm": 0.08531861007213593,
      "learning_rate": 3.538266116227374e-05,
      "loss": 0.0499,
      "step": 10179
    },
    {
      "epoch": 19.377080361388494,
      "grad_norm": 0.06099654361605644,
      "learning_rate": 3.5376309939663386e-05,
      "loss": 0.0394,
      "step": 10180
    },
    {
      "epoch": 19.378982406086543,
      "grad_norm": 0.05763452872633934,
      "learning_rate": 3.536995871705304e-05,
      "loss": 0.056,
      "step": 10181
    },
    {
      "epoch": 19.380884450784592,
      "grad_norm": 0.16771380603313446,
      "learning_rate": 3.536360749444268e-05,
      "loss": 0.0545,
      "step": 10182
    },
    {
      "epoch": 19.382786495482645,
      "grad_norm": 0.1479615420103073,
      "learning_rate": 3.535725627183233e-05,
      "loss": 0.0461,
      "step": 10183
    },
    {
      "epoch": 19.384688540180694,
      "grad_norm": 0.10853920876979828,
      "learning_rate": 3.535090504922198e-05,
      "loss": 0.0497,
      "step": 10184
    },
    {
      "epoch": 19.386590584878746,
      "grad_norm": 0.11250828206539154,
      "learning_rate": 3.5344553826611625e-05,
      "loss": 0.0582,
      "step": 10185
    },
    {
      "epoch": 19.388492629576795,
      "grad_norm": 0.08319451659917831,
      "learning_rate": 3.533820260400127e-05,
      "loss": 0.0612,
      "step": 10186
    },
    {
      "epoch": 19.390394674274845,
      "grad_norm": 0.09340374171733856,
      "learning_rate": 3.5331851381390915e-05,
      "loss": 0.046,
      "step": 10187
    },
    {
      "epoch": 19.392296718972897,
      "grad_norm": 0.12733030319213867,
      "learning_rate": 3.532550015878057e-05,
      "loss": 0.0457,
      "step": 10188
    },
    {
      "epoch": 19.394198763670946,
      "grad_norm": 0.10150488466024399,
      "learning_rate": 3.531914893617021e-05,
      "loss": 0.046,
      "step": 10189
    },
    {
      "epoch": 19.396100808368995,
      "grad_norm": 0.05435825511813164,
      "learning_rate": 3.5312797713559864e-05,
      "loss": 0.0344,
      "step": 10190
    },
    {
      "epoch": 19.398002853067048,
      "grad_norm": 0.06481046974658966,
      "learning_rate": 3.530644649094951e-05,
      "loss": 0.0543,
      "step": 10191
    },
    {
      "epoch": 19.399904897765097,
      "grad_norm": 0.1499750167131424,
      "learning_rate": 3.5300095268339154e-05,
      "loss": 0.0507,
      "step": 10192
    },
    {
      "epoch": 19.401806942463146,
      "grad_norm": 0.05150365084409714,
      "learning_rate": 3.5293744045728806e-05,
      "loss": 0.0386,
      "step": 10193
    },
    {
      "epoch": 19.4037089871612,
      "grad_norm": 0.03388383612036705,
      "learning_rate": 3.528739282311845e-05,
      "loss": 0.0366,
      "step": 10194
    },
    {
      "epoch": 19.405611031859248,
      "grad_norm": 0.19244486093521118,
      "learning_rate": 3.5281041600508096e-05,
      "loss": 0.0536,
      "step": 10195
    },
    {
      "epoch": 19.4075130765573,
      "grad_norm": 0.12983280420303345,
      "learning_rate": 3.527469037789775e-05,
      "loss": 0.037,
      "step": 10196
    },
    {
      "epoch": 19.40941512125535,
      "grad_norm": 0.08089195936918259,
      "learning_rate": 3.526833915528739e-05,
      "loss": 0.0746,
      "step": 10197
    },
    {
      "epoch": 19.4113171659534,
      "grad_norm": 0.08440740406513214,
      "learning_rate": 3.5261987932677045e-05,
      "loss": 0.0564,
      "step": 10198
    },
    {
      "epoch": 19.41321921065145,
      "grad_norm": 0.24003563821315765,
      "learning_rate": 3.525563671006669e-05,
      "loss": 0.0702,
      "step": 10199
    },
    {
      "epoch": 19.4151212553495,
      "grad_norm": 0.2483237385749817,
      "learning_rate": 3.5249285487456335e-05,
      "loss": 0.0716,
      "step": 10200
    },
    {
      "epoch": 19.41702330004755,
      "grad_norm": 0.1826210469007492,
      "learning_rate": 3.524293426484599e-05,
      "loss": 0.0598,
      "step": 10201
    },
    {
      "epoch": 19.418925344745603,
      "grad_norm": 0.09709428995847702,
      "learning_rate": 3.523658304223563e-05,
      "loss": 0.0531,
      "step": 10202
    },
    {
      "epoch": 19.42082738944365,
      "grad_norm": 0.18354997038841248,
      "learning_rate": 3.5230231819625284e-05,
      "loss": 0.0586,
      "step": 10203
    },
    {
      "epoch": 19.4227294341417,
      "grad_norm": 0.050882138311862946,
      "learning_rate": 3.522388059701492e-05,
      "loss": 0.0466,
      "step": 10204
    },
    {
      "epoch": 19.424631478839753,
      "grad_norm": 0.1463881880044937,
      "learning_rate": 3.5217529374404574e-05,
      "loss": 0.0597,
      "step": 10205
    },
    {
      "epoch": 19.426533523537802,
      "grad_norm": 0.055197589099407196,
      "learning_rate": 3.521117815179422e-05,
      "loss": 0.0411,
      "step": 10206
    },
    {
      "epoch": 19.428435568235855,
      "grad_norm": 0.13385817408561707,
      "learning_rate": 3.520482692918387e-05,
      "loss": 0.0552,
      "step": 10207
    },
    {
      "epoch": 19.430337612933904,
      "grad_norm": 0.1626228392124176,
      "learning_rate": 3.519847570657352e-05,
      "loss": 0.0351,
      "step": 10208
    },
    {
      "epoch": 19.432239657631953,
      "grad_norm": 0.11927636712789536,
      "learning_rate": 3.519212448396316e-05,
      "loss": 0.0591,
      "step": 10209
    },
    {
      "epoch": 19.434141702330006,
      "grad_norm": 0.07239444553852081,
      "learning_rate": 3.518577326135281e-05,
      "loss": 0.0627,
      "step": 10210
    },
    {
      "epoch": 19.436043747028055,
      "grad_norm": 0.07885508984327316,
      "learning_rate": 3.517942203874246e-05,
      "loss": 0.0502,
      "step": 10211
    },
    {
      "epoch": 19.437945791726104,
      "grad_norm": 0.04506470263004303,
      "learning_rate": 3.517307081613211e-05,
      "loss": 0.039,
      "step": 10212
    },
    {
      "epoch": 19.439847836424157,
      "grad_norm": 0.1872227042913437,
      "learning_rate": 3.5166719593521755e-05,
      "loss": 0.0638,
      "step": 10213
    },
    {
      "epoch": 19.441749881122206,
      "grad_norm": 0.12122814357280731,
      "learning_rate": 3.51603683709114e-05,
      "loss": 0.0661,
      "step": 10214
    },
    {
      "epoch": 19.443651925820255,
      "grad_norm": 0.06379745900630951,
      "learning_rate": 3.515401714830105e-05,
      "loss": 0.0401,
      "step": 10215
    },
    {
      "epoch": 19.445553970518308,
      "grad_norm": 0.0780811533331871,
      "learning_rate": 3.51476659256907e-05,
      "loss": 0.0553,
      "step": 10216
    },
    {
      "epoch": 19.447456015216357,
      "grad_norm": 0.07205761969089508,
      "learning_rate": 3.514131470308035e-05,
      "loss": 0.0662,
      "step": 10217
    },
    {
      "epoch": 19.44935805991441,
      "grad_norm": 0.14351381361484528,
      "learning_rate": 3.513496348046999e-05,
      "loss": 0.0624,
      "step": 10218
    },
    {
      "epoch": 19.45126010461246,
      "grad_norm": 0.042609211057424545,
      "learning_rate": 3.512861225785964e-05,
      "loss": 0.0351,
      "step": 10219
    },
    {
      "epoch": 19.453162149310508,
      "grad_norm": 0.19034677743911743,
      "learning_rate": 3.512226103524929e-05,
      "loss": 0.0662,
      "step": 10220
    },
    {
      "epoch": 19.45506419400856,
      "grad_norm": 0.16874302923679352,
      "learning_rate": 3.5115909812638936e-05,
      "loss": 0.0491,
      "step": 10221
    },
    {
      "epoch": 19.45696623870661,
      "grad_norm": 0.0731043592095375,
      "learning_rate": 3.510955859002858e-05,
      "loss": 0.0526,
      "step": 10222
    },
    {
      "epoch": 19.45886828340466,
      "grad_norm": 0.05459411069750786,
      "learning_rate": 3.5103207367418226e-05,
      "loss": 0.047,
      "step": 10223
    },
    {
      "epoch": 19.46077032810271,
      "grad_norm": 0.11637942492961884,
      "learning_rate": 3.509685614480788e-05,
      "loss": 0.0407,
      "step": 10224
    },
    {
      "epoch": 19.46267237280076,
      "grad_norm": 0.0811336413025856,
      "learning_rate": 3.509050492219752e-05,
      "loss": 0.0625,
      "step": 10225
    },
    {
      "epoch": 19.46457441749881,
      "grad_norm": 0.21679244935512543,
      "learning_rate": 3.5084153699587174e-05,
      "loss": 0.0703,
      "step": 10226
    },
    {
      "epoch": 19.466476462196862,
      "grad_norm": 0.14192962646484375,
      "learning_rate": 3.507780247697682e-05,
      "loss": 0.0709,
      "step": 10227
    },
    {
      "epoch": 19.46837850689491,
      "grad_norm": 0.10032680630683899,
      "learning_rate": 3.5071451254366465e-05,
      "loss": 0.0836,
      "step": 10228
    },
    {
      "epoch": 19.470280551592964,
      "grad_norm": 0.13514381647109985,
      "learning_rate": 3.5065100031756116e-05,
      "loss": 0.0607,
      "step": 10229
    },
    {
      "epoch": 19.472182596291013,
      "grad_norm": 0.06436719000339508,
      "learning_rate": 3.505874880914576e-05,
      "loss": 0.0532,
      "step": 10230
    },
    {
      "epoch": 19.474084640989062,
      "grad_norm": 0.09055536240339279,
      "learning_rate": 3.5052397586535407e-05,
      "loss": 0.0608,
      "step": 10231
    },
    {
      "epoch": 19.475986685687115,
      "grad_norm": 0.1204552948474884,
      "learning_rate": 3.504604636392506e-05,
      "loss": 0.0442,
      "step": 10232
    },
    {
      "epoch": 19.477888730385164,
      "grad_norm": 0.10365813970565796,
      "learning_rate": 3.5039695141314703e-05,
      "loss": 0.0474,
      "step": 10233
    },
    {
      "epoch": 19.479790775083213,
      "grad_norm": 0.07257441431283951,
      "learning_rate": 3.5033343918704355e-05,
      "loss": 0.0489,
      "step": 10234
    },
    {
      "epoch": 19.481692819781266,
      "grad_norm": 0.05598134547472,
      "learning_rate": 3.5026992696094e-05,
      "loss": 0.0486,
      "step": 10235
    },
    {
      "epoch": 19.483594864479315,
      "grad_norm": 0.07868688553571701,
      "learning_rate": 3.5020641473483645e-05,
      "loss": 0.0508,
      "step": 10236
    },
    {
      "epoch": 19.485496909177364,
      "grad_norm": 0.1124541312456131,
      "learning_rate": 3.50142902508733e-05,
      "loss": 0.0712,
      "step": 10237
    },
    {
      "epoch": 19.487398953875417,
      "grad_norm": 0.10084667801856995,
      "learning_rate": 3.500793902826294e-05,
      "loss": 0.0527,
      "step": 10238
    },
    {
      "epoch": 19.489300998573466,
      "grad_norm": 0.04100062698125839,
      "learning_rate": 3.5001587805652594e-05,
      "loss": 0.0527,
      "step": 10239
    },
    {
      "epoch": 19.49120304327152,
      "grad_norm": 0.04239039495587349,
      "learning_rate": 3.499523658304223e-05,
      "loss": 0.0576,
      "step": 10240
    },
    {
      "epoch": 19.493105087969568,
      "grad_norm": 0.119795061647892,
      "learning_rate": 3.4988885360431884e-05,
      "loss": 0.0576,
      "step": 10241
    },
    {
      "epoch": 19.495007132667617,
      "grad_norm": 0.12805697321891785,
      "learning_rate": 3.498253413782153e-05,
      "loss": 0.0574,
      "step": 10242
    },
    {
      "epoch": 19.49690917736567,
      "grad_norm": 0.1418042778968811,
      "learning_rate": 3.497618291521118e-05,
      "loss": 0.0578,
      "step": 10243
    },
    {
      "epoch": 19.49881122206372,
      "grad_norm": 0.07065989077091217,
      "learning_rate": 3.4969831692600826e-05,
      "loss": 0.0448,
      "step": 10244
    },
    {
      "epoch": 19.500713266761768,
      "grad_norm": 0.09072795510292053,
      "learning_rate": 3.496348046999047e-05,
      "loss": 0.0826,
      "step": 10245
    },
    {
      "epoch": 19.50261531145982,
      "grad_norm": 0.12471240758895874,
      "learning_rate": 3.495712924738012e-05,
      "loss": 0.0538,
      "step": 10246
    },
    {
      "epoch": 19.50451735615787,
      "grad_norm": 0.057024117559194565,
      "learning_rate": 3.495077802476977e-05,
      "loss": 0.0456,
      "step": 10247
    },
    {
      "epoch": 19.50641940085592,
      "grad_norm": 0.14013096690177917,
      "learning_rate": 3.494442680215942e-05,
      "loss": 0.061,
      "step": 10248
    },
    {
      "epoch": 19.50832144555397,
      "grad_norm": 0.18541331589221954,
      "learning_rate": 3.4938075579549065e-05,
      "loss": 0.0482,
      "step": 10249
    },
    {
      "epoch": 19.51022349025202,
      "grad_norm": 0.1770530641078949,
      "learning_rate": 3.493172435693871e-05,
      "loss": 0.083,
      "step": 10250
    },
    {
      "epoch": 19.512125534950073,
      "grad_norm": 0.15656158328056335,
      "learning_rate": 3.492537313432836e-05,
      "loss": 0.0582,
      "step": 10251
    },
    {
      "epoch": 19.514027579648122,
      "grad_norm": 0.12950193881988525,
      "learning_rate": 3.491902191171801e-05,
      "loss": 0.0575,
      "step": 10252
    },
    {
      "epoch": 19.51592962434617,
      "grad_norm": 0.07617585361003876,
      "learning_rate": 3.491267068910766e-05,
      "loss": 0.0542,
      "step": 10253
    },
    {
      "epoch": 19.517831669044224,
      "grad_norm": 0.08753058314323425,
      "learning_rate": 3.49063194664973e-05,
      "loss": 0.0687,
      "step": 10254
    },
    {
      "epoch": 19.519733713742273,
      "grad_norm": 0.1600852906703949,
      "learning_rate": 3.489996824388695e-05,
      "loss": 0.0532,
      "step": 10255
    },
    {
      "epoch": 19.521635758440322,
      "grad_norm": 0.13967394828796387,
      "learning_rate": 3.48936170212766e-05,
      "loss": 0.0824,
      "step": 10256
    },
    {
      "epoch": 19.523537803138375,
      "grad_norm": 0.10907625406980515,
      "learning_rate": 3.4887265798666246e-05,
      "loss": 0.048,
      "step": 10257
    },
    {
      "epoch": 19.525439847836424,
      "grad_norm": 0.06240085884928703,
      "learning_rate": 3.488091457605589e-05,
      "loss": 0.0527,
      "step": 10258
    },
    {
      "epoch": 19.527341892534473,
      "grad_norm": 0.044482551515102386,
      "learning_rate": 3.4874563353445536e-05,
      "loss": 0.0568,
      "step": 10259
    },
    {
      "epoch": 19.529243937232525,
      "grad_norm": 0.13326877355575562,
      "learning_rate": 3.486821213083519e-05,
      "loss": 0.0518,
      "step": 10260
    },
    {
      "epoch": 19.531145981930575,
      "grad_norm": 0.04680095613002777,
      "learning_rate": 3.486186090822483e-05,
      "loss": 0.0608,
      "step": 10261
    },
    {
      "epoch": 19.533048026628627,
      "grad_norm": 0.05832425877451897,
      "learning_rate": 3.4855509685614485e-05,
      "loss": 0.0513,
      "step": 10262
    },
    {
      "epoch": 19.534950071326676,
      "grad_norm": 0.13309139013290405,
      "learning_rate": 3.484915846300413e-05,
      "loss": 0.0508,
      "step": 10263
    },
    {
      "epoch": 19.536852116024725,
      "grad_norm": 0.22880618274211884,
      "learning_rate": 3.4842807240393775e-05,
      "loss": 0.0698,
      "step": 10264
    },
    {
      "epoch": 19.538754160722778,
      "grad_norm": 0.6374916434288025,
      "learning_rate": 3.483645601778343e-05,
      "loss": 0.0584,
      "step": 10265
    },
    {
      "epoch": 19.540656205420827,
      "grad_norm": 0.09376731514930725,
      "learning_rate": 3.483010479517307e-05,
      "loss": 0.0923,
      "step": 10266
    },
    {
      "epoch": 19.542558250118876,
      "grad_norm": 0.29286977648735046,
      "learning_rate": 3.482375357256272e-05,
      "loss": 0.0543,
      "step": 10267
    },
    {
      "epoch": 19.54446029481693,
      "grad_norm": 0.0732014924287796,
      "learning_rate": 3.481740234995237e-05,
      "loss": 0.0503,
      "step": 10268
    },
    {
      "epoch": 19.546362339514978,
      "grad_norm": 0.1219160258769989,
      "learning_rate": 3.4811051127342014e-05,
      "loss": 0.0471,
      "step": 10269
    },
    {
      "epoch": 19.548264384213027,
      "grad_norm": 0.20467039942741394,
      "learning_rate": 3.4804699904731666e-05,
      "loss": 0.0596,
      "step": 10270
    },
    {
      "epoch": 19.55016642891108,
      "grad_norm": 0.1130104586482048,
      "learning_rate": 3.479834868212131e-05,
      "loss": 0.0886,
      "step": 10271
    },
    {
      "epoch": 19.55206847360913,
      "grad_norm": 0.18084308505058289,
      "learning_rate": 3.4791997459510956e-05,
      "loss": 0.0766,
      "step": 10272
    },
    {
      "epoch": 19.55397051830718,
      "grad_norm": 0.14531627297401428,
      "learning_rate": 3.47856462369006e-05,
      "loss": 0.047,
      "step": 10273
    },
    {
      "epoch": 19.55587256300523,
      "grad_norm": 0.058127906173467636,
      "learning_rate": 3.477929501429025e-05,
      "loss": 0.0317,
      "step": 10274
    },
    {
      "epoch": 19.55777460770328,
      "grad_norm": 0.15582560002803802,
      "learning_rate": 3.4772943791679905e-05,
      "loss": 0.0742,
      "step": 10275
    },
    {
      "epoch": 19.559676652401333,
      "grad_norm": 0.3124534785747528,
      "learning_rate": 3.476659256906954e-05,
      "loss": 0.0913,
      "step": 10276
    },
    {
      "epoch": 19.56157869709938,
      "grad_norm": 0.09679864346981049,
      "learning_rate": 3.4760241346459195e-05,
      "loss": 0.0406,
      "step": 10277
    },
    {
      "epoch": 19.56348074179743,
      "grad_norm": 0.04385126382112503,
      "learning_rate": 3.475389012384884e-05,
      "loss": 0.052,
      "step": 10278
    },
    {
      "epoch": 19.565382786495483,
      "grad_norm": 0.09517911076545715,
      "learning_rate": 3.474753890123849e-05,
      "loss": 0.0772,
      "step": 10279
    },
    {
      "epoch": 19.567284831193533,
      "grad_norm": 0.11840872466564178,
      "learning_rate": 3.474118767862814e-05,
      "loss": 0.066,
      "step": 10280
    },
    {
      "epoch": 19.56918687589158,
      "grad_norm": 0.11991573125123978,
      "learning_rate": 3.473483645601778e-05,
      "loss": 0.0498,
      "step": 10281
    },
    {
      "epoch": 19.571088920589634,
      "grad_norm": 0.15331275761127472,
      "learning_rate": 3.4728485233407434e-05,
      "loss": 0.0668,
      "step": 10282
    },
    {
      "epoch": 19.572990965287683,
      "grad_norm": 0.2799055278301239,
      "learning_rate": 3.472213401079708e-05,
      "loss": 0.0768,
      "step": 10283
    },
    {
      "epoch": 19.574893009985736,
      "grad_norm": 0.1397206336259842,
      "learning_rate": 3.471578278818673e-05,
      "loss": 0.0539,
      "step": 10284
    },
    {
      "epoch": 19.576795054683785,
      "grad_norm": 0.18653438985347748,
      "learning_rate": 3.4709431565576376e-05,
      "loss": 0.0501,
      "step": 10285
    },
    {
      "epoch": 19.578697099381834,
      "grad_norm": 0.07166846096515656,
      "learning_rate": 3.470308034296602e-05,
      "loss": 0.0701,
      "step": 10286
    },
    {
      "epoch": 19.580599144079887,
      "grad_norm": 0.1638810932636261,
      "learning_rate": 3.469672912035567e-05,
      "loss": 0.0905,
      "step": 10287
    },
    {
      "epoch": 19.582501188777936,
      "grad_norm": 0.14252188801765442,
      "learning_rate": 3.469037789774532e-05,
      "loss": 0.0791,
      "step": 10288
    },
    {
      "epoch": 19.584403233475985,
      "grad_norm": 0.08229123800992966,
      "learning_rate": 3.468402667513497e-05,
      "loss": 0.0411,
      "step": 10289
    },
    {
      "epoch": 19.586305278174038,
      "grad_norm": 0.1315241903066635,
      "learning_rate": 3.467767545252461e-05,
      "loss": 0.0674,
      "step": 10290
    },
    {
      "epoch": 19.588207322872087,
      "grad_norm": 0.10541167110204697,
      "learning_rate": 3.467132422991426e-05,
      "loss": 0.0595,
      "step": 10291
    },
    {
      "epoch": 19.590109367570136,
      "grad_norm": 0.14340007305145264,
      "learning_rate": 3.466497300730391e-05,
      "loss": 0.0426,
      "step": 10292
    },
    {
      "epoch": 19.59201141226819,
      "grad_norm": 0.15541459619998932,
      "learning_rate": 3.4658621784693556e-05,
      "loss": 0.0588,
      "step": 10293
    },
    {
      "epoch": 19.593913456966238,
      "grad_norm": 0.3654743432998657,
      "learning_rate": 3.46522705620832e-05,
      "loss": 0.0756,
      "step": 10294
    },
    {
      "epoch": 19.59581550166429,
      "grad_norm": 0.10362816601991653,
      "learning_rate": 3.4645919339472847e-05,
      "loss": 0.0482,
      "step": 10295
    },
    {
      "epoch": 19.59771754636234,
      "grad_norm": 0.11532525718212128,
      "learning_rate": 3.46395681168625e-05,
      "loss": 0.053,
      "step": 10296
    },
    {
      "epoch": 19.59961959106039,
      "grad_norm": 0.20203377306461334,
      "learning_rate": 3.4633216894252143e-05,
      "loss": 0.0738,
      "step": 10297
    },
    {
      "epoch": 19.60152163575844,
      "grad_norm": 0.12903465330600739,
      "learning_rate": 3.4626865671641795e-05,
      "loss": 0.0512,
      "step": 10298
    },
    {
      "epoch": 19.60342368045649,
      "grad_norm": 0.08098254352807999,
      "learning_rate": 3.462051444903144e-05,
      "loss": 0.0439,
      "step": 10299
    },
    {
      "epoch": 19.60532572515454,
      "grad_norm": 0.0791269913315773,
      "learning_rate": 3.4614163226421085e-05,
      "loss": 0.0686,
      "step": 10300
    },
    {
      "epoch": 19.607227769852592,
      "grad_norm": 0.054660454392433167,
      "learning_rate": 3.460781200381074e-05,
      "loss": 0.0547,
      "step": 10301
    },
    {
      "epoch": 19.60912981455064,
      "grad_norm": 0.11273625493049622,
      "learning_rate": 3.460146078120038e-05,
      "loss": 0.0753,
      "step": 10302
    },
    {
      "epoch": 19.61103185924869,
      "grad_norm": 0.07658755034208298,
      "learning_rate": 3.459510955859003e-05,
      "loss": 0.0371,
      "step": 10303
    },
    {
      "epoch": 19.612933903946743,
      "grad_norm": 0.10010501742362976,
      "learning_rate": 3.458875833597968e-05,
      "loss": 0.0506,
      "step": 10304
    },
    {
      "epoch": 19.614835948644792,
      "grad_norm": 0.22586584091186523,
      "learning_rate": 3.4582407113369324e-05,
      "loss": 0.0585,
      "step": 10305
    },
    {
      "epoch": 19.616737993342845,
      "grad_norm": 0.06970422714948654,
      "learning_rate": 3.4576055890758976e-05,
      "loss": 0.0444,
      "step": 10306
    },
    {
      "epoch": 19.618640038040894,
      "grad_norm": 0.06534264981746674,
      "learning_rate": 3.456970466814862e-05,
      "loss": 0.031,
      "step": 10307
    },
    {
      "epoch": 19.620542082738943,
      "grad_norm": 0.29313337802886963,
      "learning_rate": 3.4563353445538266e-05,
      "loss": 0.0867,
      "step": 10308
    },
    {
      "epoch": 19.622444127436996,
      "grad_norm": 0.06673599034547806,
      "learning_rate": 3.455700222292791e-05,
      "loss": 0.0523,
      "step": 10309
    },
    {
      "epoch": 19.624346172135045,
      "grad_norm": 0.13317318260669708,
      "learning_rate": 3.455065100031756e-05,
      "loss": 0.0498,
      "step": 10310
    },
    {
      "epoch": 19.626248216833094,
      "grad_norm": 0.0629139170050621,
      "learning_rate": 3.4544299777707215e-05,
      "loss": 0.0643,
      "step": 10311
    },
    {
      "epoch": 19.628150261531147,
      "grad_norm": 0.1272716075181961,
      "learning_rate": 3.453794855509685e-05,
      "loss": 0.067,
      "step": 10312
    },
    {
      "epoch": 19.630052306229196,
      "grad_norm": 0.22649399936199188,
      "learning_rate": 3.4531597332486505e-05,
      "loss": 0.0505,
      "step": 10313
    },
    {
      "epoch": 19.63195435092725,
      "grad_norm": 0.09296654909849167,
      "learning_rate": 3.452524610987615e-05,
      "loss": 0.045,
      "step": 10314
    },
    {
      "epoch": 19.633856395625298,
      "grad_norm": 0.05611781030893326,
      "learning_rate": 3.45188948872658e-05,
      "loss": 0.0537,
      "step": 10315
    },
    {
      "epoch": 19.635758440323347,
      "grad_norm": 0.0745164155960083,
      "learning_rate": 3.451254366465545e-05,
      "loss": 0.0589,
      "step": 10316
    },
    {
      "epoch": 19.6376604850214,
      "grad_norm": 0.10677458345890045,
      "learning_rate": 3.450619244204509e-05,
      "loss": 0.0498,
      "step": 10317
    },
    {
      "epoch": 19.63956252971945,
      "grad_norm": 0.10812623053789139,
      "learning_rate": 3.4499841219434744e-05,
      "loss": 0.0477,
      "step": 10318
    },
    {
      "epoch": 19.641464574417498,
      "grad_norm": 0.13432307541370392,
      "learning_rate": 3.449348999682439e-05,
      "loss": 0.0711,
      "step": 10319
    },
    {
      "epoch": 19.64336661911555,
      "grad_norm": 0.143142431974411,
      "learning_rate": 3.448713877421404e-05,
      "loss": 0.0575,
      "step": 10320
    },
    {
      "epoch": 19.6452686638136,
      "grad_norm": 0.1534140706062317,
      "learning_rate": 3.4480787551603686e-05,
      "loss": 0.0714,
      "step": 10321
    },
    {
      "epoch": 19.64717070851165,
      "grad_norm": 0.08591202646493912,
      "learning_rate": 3.447443632899333e-05,
      "loss": 0.0554,
      "step": 10322
    },
    {
      "epoch": 19.6490727532097,
      "grad_norm": 0.19998444616794586,
      "learning_rate": 3.446808510638298e-05,
      "loss": 0.0403,
      "step": 10323
    },
    {
      "epoch": 19.65097479790775,
      "grad_norm": 0.14870277047157288,
      "learning_rate": 3.446173388377263e-05,
      "loss": 0.061,
      "step": 10324
    },
    {
      "epoch": 19.652876842605803,
      "grad_norm": 0.2212769240140915,
      "learning_rate": 3.445538266116228e-05,
      "loss": 0.0625,
      "step": 10325
    },
    {
      "epoch": 19.654778887303852,
      "grad_norm": 0.188511461019516,
      "learning_rate": 3.444903143855192e-05,
      "loss": 0.0668,
      "step": 10326
    },
    {
      "epoch": 19.6566809320019,
      "grad_norm": 0.13568802177906036,
      "learning_rate": 3.444268021594157e-05,
      "loss": 0.0565,
      "step": 10327
    },
    {
      "epoch": 19.658582976699954,
      "grad_norm": 0.141657292842865,
      "learning_rate": 3.4436328993331215e-05,
      "loss": 0.0692,
      "step": 10328
    },
    {
      "epoch": 19.660485021398003,
      "grad_norm": 0.0863770842552185,
      "learning_rate": 3.442997777072087e-05,
      "loss": 0.0594,
      "step": 10329
    },
    {
      "epoch": 19.662387066096052,
      "grad_norm": 0.13790996372699738,
      "learning_rate": 3.442362654811051e-05,
      "loss": 0.0615,
      "step": 10330
    },
    {
      "epoch": 19.664289110794105,
      "grad_norm": 0.14185914397239685,
      "learning_rate": 3.441727532550016e-05,
      "loss": 0.0661,
      "step": 10331
    },
    {
      "epoch": 19.666191155492154,
      "grad_norm": 0.18941640853881836,
      "learning_rate": 3.441092410288981e-05,
      "loss": 0.0657,
      "step": 10332
    },
    {
      "epoch": 19.668093200190203,
      "grad_norm": 0.13069617748260498,
      "learning_rate": 3.4404572880279454e-05,
      "loss": 0.0482,
      "step": 10333
    },
    {
      "epoch": 19.669995244888256,
      "grad_norm": 0.12583699822425842,
      "learning_rate": 3.4398221657669106e-05,
      "loss": 0.0487,
      "step": 10334
    },
    {
      "epoch": 19.671897289586305,
      "grad_norm": 0.10469093173742294,
      "learning_rate": 3.439187043505875e-05,
      "loss": 0.0415,
      "step": 10335
    },
    {
      "epoch": 19.673799334284357,
      "grad_norm": 0.05414561554789543,
      "learning_rate": 3.4385519212448396e-05,
      "loss": 0.0554,
      "step": 10336
    },
    {
      "epoch": 19.675701378982406,
      "grad_norm": 0.22049076855182648,
      "learning_rate": 3.437916798983805e-05,
      "loss": 0.0663,
      "step": 10337
    },
    {
      "epoch": 19.677603423680456,
      "grad_norm": 0.10063643753528595,
      "learning_rate": 3.437281676722769e-05,
      "loss": 0.0678,
      "step": 10338
    },
    {
      "epoch": 19.67950546837851,
      "grad_norm": 0.05266137793660164,
      "learning_rate": 3.436646554461734e-05,
      "loss": 0.0629,
      "step": 10339
    },
    {
      "epoch": 19.681407513076557,
      "grad_norm": 0.1943947821855545,
      "learning_rate": 3.436011432200699e-05,
      "loss": 0.0345,
      "step": 10340
    },
    {
      "epoch": 19.683309557774606,
      "grad_norm": 0.10251480340957642,
      "learning_rate": 3.4353763099396635e-05,
      "loss": 0.0664,
      "step": 10341
    },
    {
      "epoch": 19.68521160247266,
      "grad_norm": 0.058005135506391525,
      "learning_rate": 3.4347411876786287e-05,
      "loss": 0.0534,
      "step": 10342
    },
    {
      "epoch": 19.687113647170708,
      "grad_norm": 0.10651689022779465,
      "learning_rate": 3.434106065417593e-05,
      "loss": 0.06,
      "step": 10343
    },
    {
      "epoch": 19.689015691868757,
      "grad_norm": 0.17154528200626373,
      "learning_rate": 3.433470943156558e-05,
      "loss": 0.0594,
      "step": 10344
    },
    {
      "epoch": 19.69091773656681,
      "grad_norm": 0.055961765348911285,
      "learning_rate": 3.432835820895522e-05,
      "loss": 0.0647,
      "step": 10345
    },
    {
      "epoch": 19.69281978126486,
      "grad_norm": 0.3030036389827728,
      "learning_rate": 3.4322006986344874e-05,
      "loss": 0.0497,
      "step": 10346
    },
    {
      "epoch": 19.69472182596291,
      "grad_norm": 0.04463787376880646,
      "learning_rate": 3.4315655763734525e-05,
      "loss": 0.0558,
      "step": 10347
    },
    {
      "epoch": 19.69662387066096,
      "grad_norm": 0.13033027946949005,
      "learning_rate": 3.4309304541124164e-05,
      "loss": 0.0599,
      "step": 10348
    },
    {
      "epoch": 19.69852591535901,
      "grad_norm": 0.056211188435554504,
      "learning_rate": 3.4302953318513816e-05,
      "loss": 0.0574,
      "step": 10349
    },
    {
      "epoch": 19.700427960057063,
      "grad_norm": 0.14974373579025269,
      "learning_rate": 3.429660209590346e-05,
      "loss": 0.0634,
      "step": 10350
    },
    {
      "epoch": 19.70233000475511,
      "grad_norm": 0.20371678471565247,
      "learning_rate": 3.429025087329311e-05,
      "loss": 0.0621,
      "step": 10351
    },
    {
      "epoch": 19.70423204945316,
      "grad_norm": 0.1817709505558014,
      "learning_rate": 3.428389965068276e-05,
      "loss": 0.0649,
      "step": 10352
    },
    {
      "epoch": 19.706134094151214,
      "grad_norm": 0.04039102420210838,
      "learning_rate": 3.42775484280724e-05,
      "loss": 0.0461,
      "step": 10353
    },
    {
      "epoch": 19.708036138849263,
      "grad_norm": 0.24125699698925018,
      "learning_rate": 3.4271197205462054e-05,
      "loss": 0.0501,
      "step": 10354
    },
    {
      "epoch": 19.70993818354731,
      "grad_norm": 0.0726756826043129,
      "learning_rate": 3.42648459828517e-05,
      "loss": 0.0522,
      "step": 10355
    },
    {
      "epoch": 19.711840228245364,
      "grad_norm": 0.12377038598060608,
      "learning_rate": 3.425849476024135e-05,
      "loss": 0.0573,
      "step": 10356
    },
    {
      "epoch": 19.713742272943414,
      "grad_norm": 0.11465056985616684,
      "learning_rate": 3.425214353763099e-05,
      "loss": 0.0681,
      "step": 10357
    },
    {
      "epoch": 19.715644317641466,
      "grad_norm": 0.04917832091450691,
      "learning_rate": 3.424579231502064e-05,
      "loss": 0.045,
      "step": 10358
    },
    {
      "epoch": 19.717546362339515,
      "grad_norm": 0.059877946972846985,
      "learning_rate": 3.423944109241029e-05,
      "loss": 0.047,
      "step": 10359
    },
    {
      "epoch": 19.719448407037564,
      "grad_norm": 0.10934069007635117,
      "learning_rate": 3.423308986979994e-05,
      "loss": 0.0543,
      "step": 10360
    },
    {
      "epoch": 19.721350451735617,
      "grad_norm": 0.18297024071216583,
      "learning_rate": 3.422673864718959e-05,
      "loss": 0.0492,
      "step": 10361
    },
    {
      "epoch": 19.723252496433666,
      "grad_norm": 0.11962675303220749,
      "learning_rate": 3.422038742457923e-05,
      "loss": 0.0739,
      "step": 10362
    },
    {
      "epoch": 19.725154541131715,
      "grad_norm": 0.18719187378883362,
      "learning_rate": 3.421403620196888e-05,
      "loss": 0.0809,
      "step": 10363
    },
    {
      "epoch": 19.727056585829768,
      "grad_norm": 0.0692734643816948,
      "learning_rate": 3.4207684979358525e-05,
      "loss": 0.0345,
      "step": 10364
    },
    {
      "epoch": 19.728958630527817,
      "grad_norm": 0.056538041681051254,
      "learning_rate": 3.420133375674818e-05,
      "loss": 0.0407,
      "step": 10365
    },
    {
      "epoch": 19.730860675225866,
      "grad_norm": 0.04328737035393715,
      "learning_rate": 3.419498253413782e-05,
      "loss": 0.0443,
      "step": 10366
    },
    {
      "epoch": 19.73276271992392,
      "grad_norm": 0.0875651016831398,
      "learning_rate": 3.418863131152747e-05,
      "loss": 0.073,
      "step": 10367
    },
    {
      "epoch": 19.734664764621968,
      "grad_norm": 0.10402372479438782,
      "learning_rate": 3.418228008891712e-05,
      "loss": 0.0613,
      "step": 10368
    },
    {
      "epoch": 19.73656680932002,
      "grad_norm": 0.09834878146648407,
      "learning_rate": 3.4175928866306764e-05,
      "loss": 0.058,
      "step": 10369
    },
    {
      "epoch": 19.73846885401807,
      "grad_norm": 0.11881900578737259,
      "learning_rate": 3.4169577643696416e-05,
      "loss": 0.1091,
      "step": 10370
    },
    {
      "epoch": 19.74037089871612,
      "grad_norm": 0.21092288196086884,
      "learning_rate": 3.416322642108606e-05,
      "loss": 0.064,
      "step": 10371
    },
    {
      "epoch": 19.74227294341417,
      "grad_norm": 0.06970597803592682,
      "learning_rate": 3.4156875198475706e-05,
      "loss": 0.0847,
      "step": 10372
    },
    {
      "epoch": 19.74417498811222,
      "grad_norm": 0.219869002699852,
      "learning_rate": 3.415052397586536e-05,
      "loss": 0.0689,
      "step": 10373
    },
    {
      "epoch": 19.74607703281027,
      "grad_norm": 0.16127139329910278,
      "learning_rate": 3.4144172753255e-05,
      "loss": 0.0505,
      "step": 10374
    },
    {
      "epoch": 19.747979077508322,
      "grad_norm": 0.24652771651744843,
      "learning_rate": 3.413782153064465e-05,
      "loss": 0.0548,
      "step": 10375
    },
    {
      "epoch": 19.74988112220637,
      "grad_norm": 0.07329868525266647,
      "learning_rate": 3.41314703080343e-05,
      "loss": 0.0465,
      "step": 10376
    },
    {
      "epoch": 19.75178316690442,
      "grad_norm": 0.1390024572610855,
      "learning_rate": 3.4125119085423945e-05,
      "loss": 0.0475,
      "step": 10377
    },
    {
      "epoch": 19.753685211602473,
      "grad_norm": 0.2154022455215454,
      "learning_rate": 3.41187678628136e-05,
      "loss": 0.0493,
      "step": 10378
    },
    {
      "epoch": 19.755587256300522,
      "grad_norm": 0.13576474785804749,
      "learning_rate": 3.411241664020324e-05,
      "loss": 0.0483,
      "step": 10379
    },
    {
      "epoch": 19.757489300998575,
      "grad_norm": 0.05765054374933243,
      "learning_rate": 3.410606541759289e-05,
      "loss": 0.0463,
      "step": 10380
    },
    {
      "epoch": 19.759391345696624,
      "grad_norm": 0.04597964510321617,
      "learning_rate": 3.409971419498253e-05,
      "loss": 0.0471,
      "step": 10381
    },
    {
      "epoch": 19.761293390394673,
      "grad_norm": 0.10041475296020508,
      "learning_rate": 3.4093362972372184e-05,
      "loss": 0.0485,
      "step": 10382
    },
    {
      "epoch": 19.763195435092726,
      "grad_norm": 0.1566227674484253,
      "learning_rate": 3.408701174976183e-05,
      "loss": 0.0803,
      "step": 10383
    },
    {
      "epoch": 19.765097479790775,
      "grad_norm": 0.13510426878929138,
      "learning_rate": 3.4080660527151474e-05,
      "loss": 0.056,
      "step": 10384
    },
    {
      "epoch": 19.766999524488824,
      "grad_norm": 0.11006447672843933,
      "learning_rate": 3.4074309304541126e-05,
      "loss": 0.0502,
      "step": 10385
    },
    {
      "epoch": 19.768901569186877,
      "grad_norm": 0.0916387289762497,
      "learning_rate": 3.406795808193077e-05,
      "loss": 0.0463,
      "step": 10386
    },
    {
      "epoch": 19.770803613884926,
      "grad_norm": 0.04085199907422066,
      "learning_rate": 3.406160685932042e-05,
      "loss": 0.0612,
      "step": 10387
    },
    {
      "epoch": 19.772705658582975,
      "grad_norm": 0.05123887583613396,
      "learning_rate": 3.405525563671007e-05,
      "loss": 0.0495,
      "step": 10388
    },
    {
      "epoch": 19.774607703281028,
      "grad_norm": 0.12037782371044159,
      "learning_rate": 3.404890441409971e-05,
      "loss": 0.0837,
      "step": 10389
    },
    {
      "epoch": 19.776509747979077,
      "grad_norm": 0.11001364886760712,
      "learning_rate": 3.4042553191489365e-05,
      "loss": 0.0525,
      "step": 10390
    },
    {
      "epoch": 19.77841179267713,
      "grad_norm": 0.25168079137802124,
      "learning_rate": 3.403620196887901e-05,
      "loss": 0.0759,
      "step": 10391
    },
    {
      "epoch": 19.78031383737518,
      "grad_norm": 0.07635489851236343,
      "learning_rate": 3.402985074626866e-05,
      "loss": 0.051,
      "step": 10392
    },
    {
      "epoch": 19.782215882073228,
      "grad_norm": 0.06394420564174652,
      "learning_rate": 3.40234995236583e-05,
      "loss": 0.0479,
      "step": 10393
    },
    {
      "epoch": 19.78411792677128,
      "grad_norm": 0.15731367468833923,
      "learning_rate": 3.401714830104795e-05,
      "loss": 0.0593,
      "step": 10394
    },
    {
      "epoch": 19.78601997146933,
      "grad_norm": 0.05368688702583313,
      "learning_rate": 3.4010797078437604e-05,
      "loss": 0.0519,
      "step": 10395
    },
    {
      "epoch": 19.78792201616738,
      "grad_norm": 0.07790486514568329,
      "learning_rate": 3.400444585582725e-05,
      "loss": 0.0569,
      "step": 10396
    },
    {
      "epoch": 19.78982406086543,
      "grad_norm": 0.11774320900440216,
      "learning_rate": 3.39980946332169e-05,
      "loss": 0.0529,
      "step": 10397
    },
    {
      "epoch": 19.79172610556348,
      "grad_norm": 0.06621931493282318,
      "learning_rate": 3.399174341060654e-05,
      "loss": 0.0541,
      "step": 10398
    },
    {
      "epoch": 19.79362815026153,
      "grad_norm": 0.2989416718482971,
      "learning_rate": 3.398539218799619e-05,
      "loss": 0.0608,
      "step": 10399
    },
    {
      "epoch": 19.795530194959582,
      "grad_norm": 0.18191273510456085,
      "learning_rate": 3.3979040965385836e-05,
      "loss": 0.05,
      "step": 10400
    },
    {
      "epoch": 19.79743223965763,
      "grad_norm": 0.13151490688323975,
      "learning_rate": 3.397268974277549e-05,
      "loss": 0.0614,
      "step": 10401
    },
    {
      "epoch": 19.799334284355684,
      "grad_norm": 0.16322055459022522,
      "learning_rate": 3.396633852016513e-05,
      "loss": 0.0675,
      "step": 10402
    },
    {
      "epoch": 19.801236329053733,
      "grad_norm": 0.17609618604183197,
      "learning_rate": 3.395998729755478e-05,
      "loss": 0.048,
      "step": 10403
    },
    {
      "epoch": 19.803138373751782,
      "grad_norm": 0.17454537749290466,
      "learning_rate": 3.395363607494443e-05,
      "loss": 0.0587,
      "step": 10404
    },
    {
      "epoch": 19.805040418449835,
      "grad_norm": 0.06503864377737045,
      "learning_rate": 3.3947284852334075e-05,
      "loss": 0.0449,
      "step": 10405
    },
    {
      "epoch": 19.806942463147884,
      "grad_norm": 0.09223943948745728,
      "learning_rate": 3.394093362972373e-05,
      "loss": 0.0659,
      "step": 10406
    },
    {
      "epoch": 19.808844507845933,
      "grad_norm": 0.06822570413351059,
      "learning_rate": 3.393458240711337e-05,
      "loss": 0.0858,
      "step": 10407
    },
    {
      "epoch": 19.810746552543986,
      "grad_norm": 0.2909882068634033,
      "learning_rate": 3.392823118450302e-05,
      "loss": 0.0472,
      "step": 10408
    },
    {
      "epoch": 19.812648597242035,
      "grad_norm": 0.05018160492181778,
      "learning_rate": 3.392187996189267e-05,
      "loss": 0.0711,
      "step": 10409
    },
    {
      "epoch": 19.814550641940087,
      "grad_norm": 0.21213145554065704,
      "learning_rate": 3.3915528739282314e-05,
      "loss": 0.0512,
      "step": 10410
    },
    {
      "epoch": 19.816452686638137,
      "grad_norm": 0.10144142806529999,
      "learning_rate": 3.390917751667196e-05,
      "loss": 0.0535,
      "step": 10411
    },
    {
      "epoch": 19.818354731336186,
      "grad_norm": 0.14137786626815796,
      "learning_rate": 3.3902826294061604e-05,
      "loss": 0.0491,
      "step": 10412
    },
    {
      "epoch": 19.82025677603424,
      "grad_norm": 0.09976770728826523,
      "learning_rate": 3.3896475071451256e-05,
      "loss": 0.0621,
      "step": 10413
    },
    {
      "epoch": 19.822158820732287,
      "grad_norm": 0.07440201938152313,
      "learning_rate": 3.389012384884091e-05,
      "loss": 0.059,
      "step": 10414
    },
    {
      "epoch": 19.824060865430337,
      "grad_norm": 0.028187036514282227,
      "learning_rate": 3.388377262623055e-05,
      "loss": 0.0376,
      "step": 10415
    },
    {
      "epoch": 19.82596291012839,
      "grad_norm": 0.24573412537574768,
      "learning_rate": 3.38774214036202e-05,
      "loss": 0.1015,
      "step": 10416
    },
    {
      "epoch": 19.82786495482644,
      "grad_norm": 0.12045437842607498,
      "learning_rate": 3.387107018100984e-05,
      "loss": 0.0506,
      "step": 10417
    },
    {
      "epoch": 19.829766999524487,
      "grad_norm": 0.3481733500957489,
      "learning_rate": 3.3864718958399495e-05,
      "loss": 0.0791,
      "step": 10418
    },
    {
      "epoch": 19.83166904422254,
      "grad_norm": 0.1278395652770996,
      "learning_rate": 3.385836773578914e-05,
      "loss": 0.0351,
      "step": 10419
    },
    {
      "epoch": 19.83357108892059,
      "grad_norm": 0.04318676143884659,
      "learning_rate": 3.3852016513178785e-05,
      "loss": 0.0456,
      "step": 10420
    },
    {
      "epoch": 19.835473133618642,
      "grad_norm": 0.15962186455726624,
      "learning_rate": 3.3845665290568436e-05,
      "loss": 0.072,
      "step": 10421
    },
    {
      "epoch": 19.83737517831669,
      "grad_norm": 0.2105236053466797,
      "learning_rate": 3.383931406795808e-05,
      "loss": 0.0533,
      "step": 10422
    },
    {
      "epoch": 19.83927722301474,
      "grad_norm": 0.17966818809509277,
      "learning_rate": 3.3832962845347733e-05,
      "loss": 0.0576,
      "step": 10423
    },
    {
      "epoch": 19.841179267712793,
      "grad_norm": 0.10667653381824493,
      "learning_rate": 3.382661162273738e-05,
      "loss": 0.0466,
      "step": 10424
    },
    {
      "epoch": 19.843081312410842,
      "grad_norm": 0.13803623616695404,
      "learning_rate": 3.3820260400127024e-05,
      "loss": 0.0625,
      "step": 10425
    },
    {
      "epoch": 19.84498335710889,
      "grad_norm": 0.1208210214972496,
      "learning_rate": 3.3813909177516675e-05,
      "loss": 0.0544,
      "step": 10426
    },
    {
      "epoch": 19.846885401806944,
      "grad_norm": 0.09105329215526581,
      "learning_rate": 3.380755795490632e-05,
      "loss": 0.0758,
      "step": 10427
    },
    {
      "epoch": 19.848787446504993,
      "grad_norm": 0.07196736335754395,
      "learning_rate": 3.380120673229597e-05,
      "loss": 0.0597,
      "step": 10428
    },
    {
      "epoch": 19.850689491203042,
      "grad_norm": 0.21177725493907928,
      "learning_rate": 3.379485550968561e-05,
      "loss": 0.0538,
      "step": 10429
    },
    {
      "epoch": 19.852591535901094,
      "grad_norm": 0.15468494594097137,
      "learning_rate": 3.378850428707526e-05,
      "loss": 0.0612,
      "step": 10430
    },
    {
      "epoch": 19.854493580599144,
      "grad_norm": 0.15754562616348267,
      "learning_rate": 3.3782153064464914e-05,
      "loss": 0.0537,
      "step": 10431
    },
    {
      "epoch": 19.856395625297196,
      "grad_norm": 0.21864154934883118,
      "learning_rate": 3.377580184185456e-05,
      "loss": 0.057,
      "step": 10432
    },
    {
      "epoch": 19.858297669995245,
      "grad_norm": 0.19252237677574158,
      "learning_rate": 3.376945061924421e-05,
      "loss": 0.0537,
      "step": 10433
    },
    {
      "epoch": 19.860199714693294,
      "grad_norm": 0.1940947026014328,
      "learning_rate": 3.376309939663385e-05,
      "loss": 0.0606,
      "step": 10434
    },
    {
      "epoch": 19.862101759391347,
      "grad_norm": 0.14336170256137848,
      "learning_rate": 3.37567481740235e-05,
      "loss": 0.0673,
      "step": 10435
    },
    {
      "epoch": 19.864003804089396,
      "grad_norm": 0.09128237515687943,
      "learning_rate": 3.3750396951413146e-05,
      "loss": 0.0648,
      "step": 10436
    },
    {
      "epoch": 19.865905848787445,
      "grad_norm": 0.10571899265050888,
      "learning_rate": 3.37440457288028e-05,
      "loss": 0.0566,
      "step": 10437
    },
    {
      "epoch": 19.867807893485498,
      "grad_norm": 0.0520542673766613,
      "learning_rate": 3.373769450619244e-05,
      "loss": 0.083,
      "step": 10438
    },
    {
      "epoch": 19.869709938183547,
      "grad_norm": 0.0973014086484909,
      "learning_rate": 3.373134328358209e-05,
      "loss": 0.0561,
      "step": 10439
    },
    {
      "epoch": 19.871611982881596,
      "grad_norm": 0.1153886690735817,
      "learning_rate": 3.372499206097174e-05,
      "loss": 0.0594,
      "step": 10440
    },
    {
      "epoch": 19.87351402757965,
      "grad_norm": 0.0537719763815403,
      "learning_rate": 3.3718640838361385e-05,
      "loss": 0.0385,
      "step": 10441
    },
    {
      "epoch": 19.875416072277698,
      "grad_norm": 0.0892142504453659,
      "learning_rate": 3.371228961575104e-05,
      "loss": 0.0516,
      "step": 10442
    },
    {
      "epoch": 19.87731811697575,
      "grad_norm": 0.11739884316921234,
      "learning_rate": 3.370593839314068e-05,
      "loss": 0.0526,
      "step": 10443
    },
    {
      "epoch": 19.8792201616738,
      "grad_norm": 0.08801928907632828,
      "learning_rate": 3.369958717053033e-05,
      "loss": 0.0581,
      "step": 10444
    },
    {
      "epoch": 19.88112220637185,
      "grad_norm": 0.12499243021011353,
      "learning_rate": 3.369323594791998e-05,
      "loss": 0.0594,
      "step": 10445
    },
    {
      "epoch": 19.8830242510699,
      "grad_norm": 0.14099840819835663,
      "learning_rate": 3.3686884725309624e-05,
      "loss": 0.0634,
      "step": 10446
    },
    {
      "epoch": 19.88492629576795,
      "grad_norm": 0.12861575186252594,
      "learning_rate": 3.368053350269927e-05,
      "loss": 0.0371,
      "step": 10447
    },
    {
      "epoch": 19.886828340466,
      "grad_norm": 0.16485625505447388,
      "learning_rate": 3.3674182280088914e-05,
      "loss": 0.0504,
      "step": 10448
    },
    {
      "epoch": 19.888730385164052,
      "grad_norm": 0.07762710005044937,
      "learning_rate": 3.3667831057478566e-05,
      "loss": 0.0523,
      "step": 10449
    },
    {
      "epoch": 19.8906324298621,
      "grad_norm": 0.07982052117586136,
      "learning_rate": 3.366147983486822e-05,
      "loss": 0.0543,
      "step": 10450
    },
    {
      "epoch": 19.89253447456015,
      "grad_norm": 0.05542469024658203,
      "learning_rate": 3.365512861225786e-05,
      "loss": 0.0465,
      "step": 10451
    },
    {
      "epoch": 19.894436519258203,
      "grad_norm": 0.1404300183057785,
      "learning_rate": 3.364877738964751e-05,
      "loss": 0.0462,
      "step": 10452
    },
    {
      "epoch": 19.896338563956252,
      "grad_norm": 0.14150568842887878,
      "learning_rate": 3.364242616703715e-05,
      "loss": 0.0528,
      "step": 10453
    },
    {
      "epoch": 19.898240608654305,
      "grad_norm": 0.3389735817909241,
      "learning_rate": 3.3636074944426805e-05,
      "loss": 0.0537,
      "step": 10454
    },
    {
      "epoch": 19.900142653352354,
      "grad_norm": 0.0653960183262825,
      "learning_rate": 3.362972372181645e-05,
      "loss": 0.0503,
      "step": 10455
    },
    {
      "epoch": 19.902044698050403,
      "grad_norm": 0.128154456615448,
      "learning_rate": 3.3623372499206095e-05,
      "loss": 0.0627,
      "step": 10456
    },
    {
      "epoch": 19.903946742748456,
      "grad_norm": 0.17091745138168335,
      "learning_rate": 3.361702127659575e-05,
      "loss": 0.0774,
      "step": 10457
    },
    {
      "epoch": 19.905848787446505,
      "grad_norm": 0.28092002868652344,
      "learning_rate": 3.361067005398539e-05,
      "loss": 0.0446,
      "step": 10458
    },
    {
      "epoch": 19.907750832144554,
      "grad_norm": 0.10894166678190231,
      "learning_rate": 3.3604318831375044e-05,
      "loss": 0.0457,
      "step": 10459
    },
    {
      "epoch": 19.909652876842607,
      "grad_norm": 0.13073298335075378,
      "learning_rate": 3.359796760876469e-05,
      "loss": 0.0335,
      "step": 10460
    },
    {
      "epoch": 19.911554921540656,
      "grad_norm": 0.1777719408273697,
      "learning_rate": 3.3591616386154334e-05,
      "loss": 0.0654,
      "step": 10461
    },
    {
      "epoch": 19.913456966238705,
      "grad_norm": 0.24284358322620392,
      "learning_rate": 3.3585265163543986e-05,
      "loss": 0.0787,
      "step": 10462
    },
    {
      "epoch": 19.915359010936758,
      "grad_norm": 0.0868089497089386,
      "learning_rate": 3.357891394093363e-05,
      "loss": 0.0881,
      "step": 10463
    },
    {
      "epoch": 19.917261055634807,
      "grad_norm": 0.19045250117778778,
      "learning_rate": 3.357256271832328e-05,
      "loss": 0.0762,
      "step": 10464
    },
    {
      "epoch": 19.91916310033286,
      "grad_norm": 0.10188870877027512,
      "learning_rate": 3.356621149571292e-05,
      "loss": 0.0672,
      "step": 10465
    },
    {
      "epoch": 19.92106514503091,
      "grad_norm": 0.06768245249986649,
      "learning_rate": 3.355986027310257e-05,
      "loss": 0.0564,
      "step": 10466
    },
    {
      "epoch": 19.922967189728958,
      "grad_norm": 0.20635931193828583,
      "learning_rate": 3.355350905049222e-05,
      "loss": 0.0504,
      "step": 10467
    },
    {
      "epoch": 19.92486923442701,
      "grad_norm": 0.126325324177742,
      "learning_rate": 3.354715782788187e-05,
      "loss": 0.0491,
      "step": 10468
    },
    {
      "epoch": 19.92677127912506,
      "grad_norm": 0.118792824447155,
      "learning_rate": 3.354080660527152e-05,
      "loss": 0.0579,
      "step": 10469
    },
    {
      "epoch": 19.92867332382311,
      "grad_norm": 0.12369075417518616,
      "learning_rate": 3.353445538266116e-05,
      "loss": 0.0572,
      "step": 10470
    },
    {
      "epoch": 19.93057536852116,
      "grad_norm": 0.12031448632478714,
      "learning_rate": 3.352810416005081e-05,
      "loss": 0.0564,
      "step": 10471
    },
    {
      "epoch": 19.93247741321921,
      "grad_norm": 0.12101674824953079,
      "learning_rate": 3.352175293744046e-05,
      "loss": 0.0612,
      "step": 10472
    },
    {
      "epoch": 19.93437945791726,
      "grad_norm": 0.05656763166189194,
      "learning_rate": 3.351540171483011e-05,
      "loss": 0.0558,
      "step": 10473
    },
    {
      "epoch": 19.936281502615312,
      "grad_norm": 0.1553266942501068,
      "learning_rate": 3.3509050492219754e-05,
      "loss": 0.0457,
      "step": 10474
    },
    {
      "epoch": 19.93818354731336,
      "grad_norm": 0.21062347292900085,
      "learning_rate": 3.35026992696094e-05,
      "loss": 0.0558,
      "step": 10475
    },
    {
      "epoch": 19.940085592011414,
      "grad_norm": 0.1241660863161087,
      "learning_rate": 3.349634804699905e-05,
      "loss": 0.0851,
      "step": 10476
    },
    {
      "epoch": 19.941987636709463,
      "grad_norm": 0.10217336565256119,
      "learning_rate": 3.3489996824388696e-05,
      "loss": 0.0683,
      "step": 10477
    },
    {
      "epoch": 19.943889681407512,
      "grad_norm": 0.2679067850112915,
      "learning_rate": 3.348364560177835e-05,
      "loss": 0.0585,
      "step": 10478
    },
    {
      "epoch": 19.945791726105565,
      "grad_norm": 0.07839953899383545,
      "learning_rate": 3.347729437916799e-05,
      "loss": 0.0568,
      "step": 10479
    },
    {
      "epoch": 19.947693770803614,
      "grad_norm": 0.04855075106024742,
      "learning_rate": 3.347094315655764e-05,
      "loss": 0.063,
      "step": 10480
    },
    {
      "epoch": 19.949595815501663,
      "grad_norm": 0.10905822366476059,
      "learning_rate": 3.346459193394729e-05,
      "loss": 0.0612,
      "step": 10481
    },
    {
      "epoch": 19.951497860199716,
      "grad_norm": 0.055349789559841156,
      "learning_rate": 3.3458240711336935e-05,
      "loss": 0.0628,
      "step": 10482
    },
    {
      "epoch": 19.953399904897765,
      "grad_norm": 0.05278193950653076,
      "learning_rate": 3.345188948872658e-05,
      "loss": 0.0554,
      "step": 10483
    },
    {
      "epoch": 19.955301949595814,
      "grad_norm": 0.10295037180185318,
      "learning_rate": 3.3445538266116225e-05,
      "loss": 0.0515,
      "step": 10484
    },
    {
      "epoch": 19.957203994293867,
      "grad_norm": 0.123154416680336,
      "learning_rate": 3.3439187043505877e-05,
      "loss": 0.047,
      "step": 10485
    },
    {
      "epoch": 19.959106038991916,
      "grad_norm": 0.0768277496099472,
      "learning_rate": 3.343283582089553e-05,
      "loss": 0.0667,
      "step": 10486
    },
    {
      "epoch": 19.96100808368997,
      "grad_norm": 0.15484827756881714,
      "learning_rate": 3.3426484598285173e-05,
      "loss": 0.0769,
      "step": 10487
    },
    {
      "epoch": 19.962910128388017,
      "grad_norm": 0.06407146155834198,
      "learning_rate": 3.342013337567482e-05,
      "loss": 0.0435,
      "step": 10488
    },
    {
      "epoch": 19.964812173086067,
      "grad_norm": 0.04490051791071892,
      "learning_rate": 3.3413782153064464e-05,
      "loss": 0.0437,
      "step": 10489
    },
    {
      "epoch": 19.96671421778412,
      "grad_norm": 0.1047554686665535,
      "learning_rate": 3.3407430930454115e-05,
      "loss": 0.034,
      "step": 10490
    },
    {
      "epoch": 19.96861626248217,
      "grad_norm": 0.14887818694114685,
      "learning_rate": 3.340107970784376e-05,
      "loss": 0.0666,
      "step": 10491
    },
    {
      "epoch": 19.970518307180217,
      "grad_norm": 0.16302020847797394,
      "learning_rate": 3.3394728485233406e-05,
      "loss": 0.054,
      "step": 10492
    },
    {
      "epoch": 19.97242035187827,
      "grad_norm": 0.10449177771806717,
      "learning_rate": 3.338837726262306e-05,
      "loss": 0.0347,
      "step": 10493
    },
    {
      "epoch": 19.97432239657632,
      "grad_norm": 0.05189422518014908,
      "learning_rate": 3.33820260400127e-05,
      "loss": 0.0609,
      "step": 10494
    },
    {
      "epoch": 19.97622444127437,
      "grad_norm": 0.10788179188966751,
      "learning_rate": 3.3375674817402354e-05,
      "loss": 0.0685,
      "step": 10495
    },
    {
      "epoch": 19.97812648597242,
      "grad_norm": 0.20024745166301727,
      "learning_rate": 3.3369323594792e-05,
      "loss": 0.0687,
      "step": 10496
    },
    {
      "epoch": 19.98002853067047,
      "grad_norm": 0.14526651799678802,
      "learning_rate": 3.3362972372181644e-05,
      "loss": 0.0574,
      "step": 10497
    },
    {
      "epoch": 19.981930575368523,
      "grad_norm": 0.06175341457128525,
      "learning_rate": 3.3356621149571296e-05,
      "loss": 0.0566,
      "step": 10498
    },
    {
      "epoch": 19.983832620066572,
      "grad_norm": 0.08872699737548828,
      "learning_rate": 3.335026992696094e-05,
      "loss": 0.0389,
      "step": 10499
    },
    {
      "epoch": 19.98573466476462,
      "grad_norm": 0.10846307873725891,
      "learning_rate": 3.334391870435059e-05,
      "loss": 0.0648,
      "step": 10500
    },
    {
      "epoch": 19.987636709462674,
      "grad_norm": 0.13726340234279633,
      "learning_rate": 3.333756748174023e-05,
      "loss": 0.0515,
      "step": 10501
    },
    {
      "epoch": 19.989538754160723,
      "grad_norm": 0.08092505484819412,
      "learning_rate": 3.333121625912988e-05,
      "loss": 0.0444,
      "step": 10502
    },
    {
      "epoch": 19.991440798858772,
      "grad_norm": 0.1583898663520813,
      "learning_rate": 3.332486503651953e-05,
      "loss": 0.044,
      "step": 10503
    },
    {
      "epoch": 19.993342843556825,
      "grad_norm": 0.3377038538455963,
      "learning_rate": 3.331851381390918e-05,
      "loss": 0.0705,
      "step": 10504
    },
    {
      "epoch": 19.995244888254874,
      "grad_norm": 0.14866070449352264,
      "learning_rate": 3.331216259129883e-05,
      "loss": 0.0688,
      "step": 10505
    },
    {
      "epoch": 19.997146932952923,
      "grad_norm": 0.08868202567100525,
      "learning_rate": 3.330581136868847e-05,
      "loss": 0.0554,
      "step": 10506
    },
    {
      "epoch": 19.999048977650975,
      "grad_norm": 0.058857131749391556,
      "learning_rate": 3.329946014607812e-05,
      "loss": 0.0576,
      "step": 10507
    },
    {
      "epoch": 20.000951022349025,
      "grad_norm": 0.24435585737228394,
      "learning_rate": 3.329310892346777e-05,
      "loss": 0.1003,
      "step": 10508
    },
    {
      "epoch": 20.002853067047077,
      "grad_norm": 0.1501217633485794,
      "learning_rate": 3.328675770085742e-05,
      "loss": 0.0406,
      "step": 10509
    },
    {
      "epoch": 20.004755111745126,
      "grad_norm": 0.04581620916724205,
      "learning_rate": 3.3280406478247064e-05,
      "loss": 0.0585,
      "step": 10510
    },
    {
      "epoch": 20.006657156443175,
      "grad_norm": 0.034946098923683167,
      "learning_rate": 3.327405525563671e-05,
      "loss": 0.0249,
      "step": 10511
    },
    {
      "epoch": 20.008559201141228,
      "grad_norm": 0.057416193187236786,
      "learning_rate": 3.326770403302636e-05,
      "loss": 0.0498,
      "step": 10512
    },
    {
      "epoch": 20.010461245839277,
      "grad_norm": 0.04225662723183632,
      "learning_rate": 3.3261352810416006e-05,
      "loss": 0.0572,
      "step": 10513
    },
    {
      "epoch": 20.012363290537326,
      "grad_norm": 0.05202086642384529,
      "learning_rate": 3.325500158780566e-05,
      "loss": 0.0347,
      "step": 10514
    },
    {
      "epoch": 20.01426533523538,
      "grad_norm": 0.03269398584961891,
      "learning_rate": 3.32486503651953e-05,
      "loss": 0.0497,
      "step": 10515
    },
    {
      "epoch": 20.016167379933428,
      "grad_norm": 0.0800165981054306,
      "learning_rate": 3.324229914258495e-05,
      "loss": 0.0473,
      "step": 10516
    },
    {
      "epoch": 20.018069424631477,
      "grad_norm": 0.033391691744327545,
      "learning_rate": 3.32359479199746e-05,
      "loss": 0.0513,
      "step": 10517
    },
    {
      "epoch": 20.01997146932953,
      "grad_norm": 0.04623610898852348,
      "learning_rate": 3.3229596697364245e-05,
      "loss": 0.0658,
      "step": 10518
    },
    {
      "epoch": 20.02187351402758,
      "grad_norm": 0.10310270637273788,
      "learning_rate": 3.322324547475389e-05,
      "loss": 0.0581,
      "step": 10519
    },
    {
      "epoch": 20.02377555872563,
      "grad_norm": 0.0826338678598404,
      "learning_rate": 3.3216894252143535e-05,
      "loss": 0.0469,
      "step": 10520
    },
    {
      "epoch": 20.02567760342368,
      "grad_norm": 0.12246903777122498,
      "learning_rate": 3.321054302953319e-05,
      "loss": 0.0397,
      "step": 10521
    },
    {
      "epoch": 20.02757964812173,
      "grad_norm": 0.1892233043909073,
      "learning_rate": 3.320419180692283e-05,
      "loss": 0.0516,
      "step": 10522
    },
    {
      "epoch": 20.029481692819783,
      "grad_norm": 0.355392187833786,
      "learning_rate": 3.3197840584312484e-05,
      "loss": 0.0651,
      "step": 10523
    },
    {
      "epoch": 20.03138373751783,
      "grad_norm": 0.11444549262523651,
      "learning_rate": 3.319148936170213e-05,
      "loss": 0.0398,
      "step": 10524
    },
    {
      "epoch": 20.03328578221588,
      "grad_norm": 0.041265226900577545,
      "learning_rate": 3.3185138139091774e-05,
      "loss": 0.0417,
      "step": 10525
    },
    {
      "epoch": 20.035187826913933,
      "grad_norm": 0.16662882268428802,
      "learning_rate": 3.3178786916481426e-05,
      "loss": 0.0507,
      "step": 10526
    },
    {
      "epoch": 20.037089871611983,
      "grad_norm": 0.040374815464019775,
      "learning_rate": 3.317243569387107e-05,
      "loss": 0.0491,
      "step": 10527
    },
    {
      "epoch": 20.03899191631003,
      "grad_norm": 0.09320390224456787,
      "learning_rate": 3.3166084471260716e-05,
      "loss": 0.0601,
      "step": 10528
    },
    {
      "epoch": 20.040893961008084,
      "grad_norm": 0.09097753465175629,
      "learning_rate": 3.315973324865037e-05,
      "loss": 0.0487,
      "step": 10529
    },
    {
      "epoch": 20.042796005706133,
      "grad_norm": 0.06951908022165298,
      "learning_rate": 3.315338202604001e-05,
      "loss": 0.0507,
      "step": 10530
    },
    {
      "epoch": 20.044698050404186,
      "grad_norm": 0.11662323772907257,
      "learning_rate": 3.3147030803429665e-05,
      "loss": 0.0621,
      "step": 10531
    },
    {
      "epoch": 20.046600095102235,
      "grad_norm": 0.04502786695957184,
      "learning_rate": 3.314067958081931e-05,
      "loss": 0.0721,
      "step": 10532
    },
    {
      "epoch": 20.048502139800284,
      "grad_norm": 0.11359043419361115,
      "learning_rate": 3.3134328358208955e-05,
      "loss": 0.0604,
      "step": 10533
    },
    {
      "epoch": 20.050404184498337,
      "grad_norm": 0.06094936281442642,
      "learning_rate": 3.312797713559861e-05,
      "loss": 0.0504,
      "step": 10534
    },
    {
      "epoch": 20.052306229196386,
      "grad_norm": 0.07370217144489288,
      "learning_rate": 3.312162591298825e-05,
      "loss": 0.0448,
      "step": 10535
    },
    {
      "epoch": 20.054208273894435,
      "grad_norm": 0.0950629934668541,
      "learning_rate": 3.3115274690377904e-05,
      "loss": 0.0699,
      "step": 10536
    },
    {
      "epoch": 20.056110318592488,
      "grad_norm": 0.39011773467063904,
      "learning_rate": 3.310892346776754e-05,
      "loss": 0.0969,
      "step": 10537
    },
    {
      "epoch": 20.058012363290537,
      "grad_norm": 0.10561854392290115,
      "learning_rate": 3.3102572245157194e-05,
      "loss": 0.0477,
      "step": 10538
    },
    {
      "epoch": 20.059914407988586,
      "grad_norm": 0.041889406740665436,
      "learning_rate": 3.309622102254684e-05,
      "loss": 0.0513,
      "step": 10539
    },
    {
      "epoch": 20.06181645268664,
      "grad_norm": 0.07895664125680923,
      "learning_rate": 3.308986979993649e-05,
      "loss": 0.0527,
      "step": 10540
    },
    {
      "epoch": 20.063718497384688,
      "grad_norm": 0.10135366022586823,
      "learning_rate": 3.308351857732614e-05,
      "loss": 0.0405,
      "step": 10541
    },
    {
      "epoch": 20.06562054208274,
      "grad_norm": 0.05142286792397499,
      "learning_rate": 3.307716735471578e-05,
      "loss": 0.0487,
      "step": 10542
    },
    {
      "epoch": 20.06752258678079,
      "grad_norm": 0.08144795894622803,
      "learning_rate": 3.307081613210543e-05,
      "loss": 0.0546,
      "step": 10543
    },
    {
      "epoch": 20.06942463147884,
      "grad_norm": 0.04765758290886879,
      "learning_rate": 3.306446490949508e-05,
      "loss": 0.0493,
      "step": 10544
    },
    {
      "epoch": 20.07132667617689,
      "grad_norm": 0.13438722491264343,
      "learning_rate": 3.305811368688473e-05,
      "loss": 0.0552,
      "step": 10545
    },
    {
      "epoch": 20.07322872087494,
      "grad_norm": 0.07511651515960693,
      "learning_rate": 3.3051762464274375e-05,
      "loss": 0.05,
      "step": 10546
    },
    {
      "epoch": 20.07513076557299,
      "grad_norm": 0.18269193172454834,
      "learning_rate": 3.304541124166402e-05,
      "loss": 0.0574,
      "step": 10547
    },
    {
      "epoch": 20.077032810271042,
      "grad_norm": 0.1806280016899109,
      "learning_rate": 3.303906001905367e-05,
      "loss": 0.0662,
      "step": 10548
    },
    {
      "epoch": 20.07893485496909,
      "grad_norm": 0.030926980078220367,
      "learning_rate": 3.3032708796443317e-05,
      "loss": 0.0418,
      "step": 10549
    },
    {
      "epoch": 20.08083689966714,
      "grad_norm": 0.14629711210727692,
      "learning_rate": 3.302635757383297e-05,
      "loss": 0.0574,
      "step": 10550
    },
    {
      "epoch": 20.082738944365193,
      "grad_norm": 0.17619511485099792,
      "learning_rate": 3.302000635122261e-05,
      "loss": 0.0517,
      "step": 10551
    },
    {
      "epoch": 20.084640989063242,
      "grad_norm": 0.05990555137395859,
      "learning_rate": 3.301365512861226e-05,
      "loss": 0.0466,
      "step": 10552
    },
    {
      "epoch": 20.086543033761295,
      "grad_norm": 0.0348682776093483,
      "learning_rate": 3.300730390600191e-05,
      "loss": 0.0341,
      "step": 10553
    },
    {
      "epoch": 20.088445078459344,
      "grad_norm": 0.13899552822113037,
      "learning_rate": 3.3000952683391555e-05,
      "loss": 0.0481,
      "step": 10554
    },
    {
      "epoch": 20.090347123157393,
      "grad_norm": 0.07599950581789017,
      "learning_rate": 3.29946014607812e-05,
      "loss": 0.0579,
      "step": 10555
    },
    {
      "epoch": 20.092249167855446,
      "grad_norm": 0.117378830909729,
      "learning_rate": 3.2988250238170846e-05,
      "loss": 0.0689,
      "step": 10556
    },
    {
      "epoch": 20.094151212553495,
      "grad_norm": 0.30050474405288696,
      "learning_rate": 3.29818990155605e-05,
      "loss": 0.1331,
      "step": 10557
    },
    {
      "epoch": 20.096053257251544,
      "grad_norm": 0.040973104536533356,
      "learning_rate": 3.297554779295014e-05,
      "loss": 0.0588,
      "step": 10558
    },
    {
      "epoch": 20.097955301949597,
      "grad_norm": 0.05349220335483551,
      "learning_rate": 3.2969196570339794e-05,
      "loss": 0.0563,
      "step": 10559
    },
    {
      "epoch": 20.099857346647646,
      "grad_norm": 0.14277498424053192,
      "learning_rate": 3.296284534772944e-05,
      "loss": 0.0468,
      "step": 10560
    },
    {
      "epoch": 20.101759391345695,
      "grad_norm": 0.11612348258495331,
      "learning_rate": 3.2956494125119084e-05,
      "loss": 0.0646,
      "step": 10561
    },
    {
      "epoch": 20.103661436043748,
      "grad_norm": 0.08324389159679413,
      "learning_rate": 3.2950142902508736e-05,
      "loss": 0.0504,
      "step": 10562
    },
    {
      "epoch": 20.105563480741797,
      "grad_norm": 0.12529146671295166,
      "learning_rate": 3.294379167989838e-05,
      "loss": 0.0433,
      "step": 10563
    },
    {
      "epoch": 20.10746552543985,
      "grad_norm": 0.20667465031147003,
      "learning_rate": 3.2937440457288026e-05,
      "loss": 0.0534,
      "step": 10564
    },
    {
      "epoch": 20.1093675701379,
      "grad_norm": 0.06124941259622574,
      "learning_rate": 3.293108923467768e-05,
      "loss": 0.0474,
      "step": 10565
    },
    {
      "epoch": 20.111269614835948,
      "grad_norm": 0.05799207463860512,
      "learning_rate": 3.292473801206732e-05,
      "loss": 0.0506,
      "step": 10566
    },
    {
      "epoch": 20.113171659534,
      "grad_norm": 0.07804218679666519,
      "learning_rate": 3.2918386789456975e-05,
      "loss": 0.0523,
      "step": 10567
    },
    {
      "epoch": 20.11507370423205,
      "grad_norm": 0.044461559504270554,
      "learning_rate": 3.291203556684662e-05,
      "loss": 0.063,
      "step": 10568
    },
    {
      "epoch": 20.1169757489301,
      "grad_norm": 0.07837837934494019,
      "learning_rate": 3.2905684344236265e-05,
      "loss": 0.0563,
      "step": 10569
    },
    {
      "epoch": 20.11887779362815,
      "grad_norm": 0.04336393624544144,
      "learning_rate": 3.289933312162591e-05,
      "loss": 0.0471,
      "step": 10570
    },
    {
      "epoch": 20.1207798383262,
      "grad_norm": 0.10369131714105606,
      "learning_rate": 3.289298189901556e-05,
      "loss": 0.0577,
      "step": 10571
    },
    {
      "epoch": 20.12268188302425,
      "grad_norm": 0.16776293516159058,
      "learning_rate": 3.2886630676405214e-05,
      "loss": 0.0368,
      "step": 10572
    },
    {
      "epoch": 20.124583927722302,
      "grad_norm": 0.20040836930274963,
      "learning_rate": 3.288027945379485e-05,
      "loss": 0.0735,
      "step": 10573
    },
    {
      "epoch": 20.12648597242035,
      "grad_norm": 0.11139001697301865,
      "learning_rate": 3.2873928231184504e-05,
      "loss": 0.0689,
      "step": 10574
    },
    {
      "epoch": 20.128388017118404,
      "grad_norm": 0.08649170398712158,
      "learning_rate": 3.286757700857415e-05,
      "loss": 0.057,
      "step": 10575
    },
    {
      "epoch": 20.130290061816453,
      "grad_norm": 0.09043058007955551,
      "learning_rate": 3.28612257859638e-05,
      "loss": 0.0806,
      "step": 10576
    },
    {
      "epoch": 20.132192106514502,
      "grad_norm": 0.03538661077618599,
      "learning_rate": 3.2854874563353446e-05,
      "loss": 0.0419,
      "step": 10577
    },
    {
      "epoch": 20.134094151212555,
      "grad_norm": 0.2169920653104782,
      "learning_rate": 3.284852334074309e-05,
      "loss": 0.0542,
      "step": 10578
    },
    {
      "epoch": 20.135996195910604,
      "grad_norm": 0.1910613477230072,
      "learning_rate": 3.284217211813274e-05,
      "loss": 0.0606,
      "step": 10579
    },
    {
      "epoch": 20.137898240608653,
      "grad_norm": 0.0717383474111557,
      "learning_rate": 3.283582089552239e-05,
      "loss": 0.058,
      "step": 10580
    },
    {
      "epoch": 20.139800285306706,
      "grad_norm": 0.15653537213802338,
      "learning_rate": 3.282946967291204e-05,
      "loss": 0.0464,
      "step": 10581
    },
    {
      "epoch": 20.141702330004755,
      "grad_norm": 0.07739397138357162,
      "learning_rate": 3.2823118450301685e-05,
      "loss": 0.06,
      "step": 10582
    },
    {
      "epoch": 20.143604374702804,
      "grad_norm": 0.10936830937862396,
      "learning_rate": 3.281676722769133e-05,
      "loss": 0.0569,
      "step": 10583
    },
    {
      "epoch": 20.145506419400856,
      "grad_norm": 0.1430310606956482,
      "learning_rate": 3.281041600508098e-05,
      "loss": 0.0563,
      "step": 10584
    },
    {
      "epoch": 20.147408464098906,
      "grad_norm": 0.0776374563574791,
      "learning_rate": 3.280406478247063e-05,
      "loss": 0.0461,
      "step": 10585
    },
    {
      "epoch": 20.149310508796958,
      "grad_norm": 0.11177509278059006,
      "learning_rate": 3.279771355986028e-05,
      "loss": 0.047,
      "step": 10586
    },
    {
      "epoch": 20.151212553495007,
      "grad_norm": 0.06614487618207932,
      "learning_rate": 3.279136233724992e-05,
      "loss": 0.0497,
      "step": 10587
    },
    {
      "epoch": 20.153114598193056,
      "grad_norm": 0.1806342899799347,
      "learning_rate": 3.278501111463957e-05,
      "loss": 0.0645,
      "step": 10588
    },
    {
      "epoch": 20.15501664289111,
      "grad_norm": 0.11989549547433853,
      "learning_rate": 3.277865989202922e-05,
      "loss": 0.0521,
      "step": 10589
    },
    {
      "epoch": 20.156918687589158,
      "grad_norm": 0.04524906724691391,
      "learning_rate": 3.2772308669418866e-05,
      "loss": 0.0651,
      "step": 10590
    },
    {
      "epoch": 20.158820732287207,
      "grad_norm": 0.07851403206586838,
      "learning_rate": 3.276595744680851e-05,
      "loss": 0.0574,
      "step": 10591
    },
    {
      "epoch": 20.16072277698526,
      "grad_norm": 0.13270311057567596,
      "learning_rate": 3.2759606224198156e-05,
      "loss": 0.0432,
      "step": 10592
    },
    {
      "epoch": 20.16262482168331,
      "grad_norm": 0.0337836928665638,
      "learning_rate": 3.275325500158781e-05,
      "loss": 0.0444,
      "step": 10593
    },
    {
      "epoch": 20.16452686638136,
      "grad_norm": 0.12775632739067078,
      "learning_rate": 3.274690377897745e-05,
      "loss": 0.0475,
      "step": 10594
    },
    {
      "epoch": 20.16642891107941,
      "grad_norm": 0.05581698939204216,
      "learning_rate": 3.2740552556367105e-05,
      "loss": 0.0341,
      "step": 10595
    },
    {
      "epoch": 20.16833095577746,
      "grad_norm": 0.06510213762521744,
      "learning_rate": 3.273420133375675e-05,
      "loss": 0.0529,
      "step": 10596
    },
    {
      "epoch": 20.170233000475513,
      "grad_norm": 0.09131671488285065,
      "learning_rate": 3.2727850111146395e-05,
      "loss": 0.0457,
      "step": 10597
    },
    {
      "epoch": 20.17213504517356,
      "grad_norm": 0.21071843802928925,
      "learning_rate": 3.272149888853605e-05,
      "loss": 0.0515,
      "step": 10598
    },
    {
      "epoch": 20.17403708987161,
      "grad_norm": 0.06667834520339966,
      "learning_rate": 3.271514766592569e-05,
      "loss": 0.0865,
      "step": 10599
    },
    {
      "epoch": 20.175939134569663,
      "grad_norm": 0.11008747667074203,
      "learning_rate": 3.270879644331534e-05,
      "loss": 0.0742,
      "step": 10600
    },
    {
      "epoch": 20.177841179267713,
      "grad_norm": 0.2698976397514343,
      "learning_rate": 3.270244522070499e-05,
      "loss": 0.0775,
      "step": 10601
    },
    {
      "epoch": 20.17974322396576,
      "grad_norm": 0.10627546161413193,
      "learning_rate": 3.2696093998094634e-05,
      "loss": 0.0513,
      "step": 10602
    },
    {
      "epoch": 20.181645268663814,
      "grad_norm": 0.041233550757169724,
      "learning_rate": 3.2689742775484286e-05,
      "loss": 0.0397,
      "step": 10603
    },
    {
      "epoch": 20.183547313361863,
      "grad_norm": 0.07916425168514252,
      "learning_rate": 3.268339155287393e-05,
      "loss": 0.0485,
      "step": 10604
    },
    {
      "epoch": 20.185449358059916,
      "grad_norm": 0.060306694358587265,
      "learning_rate": 3.2677040330263576e-05,
      "loss": 0.0489,
      "step": 10605
    },
    {
      "epoch": 20.187351402757965,
      "grad_norm": 0.08756151050329208,
      "learning_rate": 3.267068910765322e-05,
      "loss": 0.0771,
      "step": 10606
    },
    {
      "epoch": 20.189253447456014,
      "grad_norm": 0.26062679290771484,
      "learning_rate": 3.266433788504287e-05,
      "loss": 0.0882,
      "step": 10607
    },
    {
      "epoch": 20.191155492154067,
      "grad_norm": 0.03976987302303314,
      "learning_rate": 3.2657986662432524e-05,
      "loss": 0.0528,
      "step": 10608
    },
    {
      "epoch": 20.193057536852116,
      "grad_norm": 0.056227684020996094,
      "learning_rate": 3.265163543982216e-05,
      "loss": 0.0552,
      "step": 10609
    },
    {
      "epoch": 20.194959581550165,
      "grad_norm": 0.13015170395374298,
      "learning_rate": 3.2645284217211815e-05,
      "loss": 0.0598,
      "step": 10610
    },
    {
      "epoch": 20.196861626248218,
      "grad_norm": 0.05390501394867897,
      "learning_rate": 3.263893299460146e-05,
      "loss": 0.0467,
      "step": 10611
    },
    {
      "epoch": 20.198763670946267,
      "grad_norm": 0.08726727217435837,
      "learning_rate": 3.263258177199111e-05,
      "loss": 0.0368,
      "step": 10612
    },
    {
      "epoch": 20.200665715644316,
      "grad_norm": 0.11441841721534729,
      "learning_rate": 3.2626230549380757e-05,
      "loss": 0.0651,
      "step": 10613
    },
    {
      "epoch": 20.20256776034237,
      "grad_norm": 0.1922193467617035,
      "learning_rate": 3.26198793267704e-05,
      "loss": 0.0534,
      "step": 10614
    },
    {
      "epoch": 20.204469805040418,
      "grad_norm": 0.03666595742106438,
      "learning_rate": 3.2613528104160053e-05,
      "loss": 0.0385,
      "step": 10615
    },
    {
      "epoch": 20.20637184973847,
      "grad_norm": 0.04062128812074661,
      "learning_rate": 3.26071768815497e-05,
      "loss": 0.0422,
      "step": 10616
    },
    {
      "epoch": 20.20827389443652,
      "grad_norm": 0.1665174514055252,
      "learning_rate": 3.260082565893935e-05,
      "loss": 0.0449,
      "step": 10617
    },
    {
      "epoch": 20.21017593913457,
      "grad_norm": 0.16695557534694672,
      "learning_rate": 3.2594474436328995e-05,
      "loss": 0.0513,
      "step": 10618
    },
    {
      "epoch": 20.21207798383262,
      "grad_norm": 0.14871883392333984,
      "learning_rate": 3.258812321371864e-05,
      "loss": 0.0531,
      "step": 10619
    },
    {
      "epoch": 20.21398002853067,
      "grad_norm": 0.04111943021416664,
      "learning_rate": 3.258177199110829e-05,
      "loss": 0.0468,
      "step": 10620
    },
    {
      "epoch": 20.21588207322872,
      "grad_norm": 0.07345671206712723,
      "learning_rate": 3.257542076849794e-05,
      "loss": 0.0476,
      "step": 10621
    },
    {
      "epoch": 20.217784117926772,
      "grad_norm": 0.09153008460998535,
      "learning_rate": 3.256906954588759e-05,
      "loss": 0.0476,
      "step": 10622
    },
    {
      "epoch": 20.21968616262482,
      "grad_norm": 0.07967954874038696,
      "learning_rate": 3.256271832327723e-05,
      "loss": 0.0705,
      "step": 10623
    },
    {
      "epoch": 20.22158820732287,
      "grad_norm": 0.08356297761201859,
      "learning_rate": 3.255636710066688e-05,
      "loss": 0.0411,
      "step": 10624
    },
    {
      "epoch": 20.223490252020923,
      "grad_norm": 0.0991276279091835,
      "learning_rate": 3.2550015878056524e-05,
      "loss": 0.0682,
      "step": 10625
    },
    {
      "epoch": 20.225392296718972,
      "grad_norm": 0.05717969685792923,
      "learning_rate": 3.2543664655446176e-05,
      "loss": 0.0605,
      "step": 10626
    },
    {
      "epoch": 20.227294341417025,
      "grad_norm": 0.14118197560310364,
      "learning_rate": 3.253731343283582e-05,
      "loss": 0.0532,
      "step": 10627
    },
    {
      "epoch": 20.229196386115074,
      "grad_norm": 0.09270215779542923,
      "learning_rate": 3.2530962210225466e-05,
      "loss": 0.0666,
      "step": 10628
    },
    {
      "epoch": 20.231098430813123,
      "grad_norm": 0.09237004071474075,
      "learning_rate": 3.252461098761512e-05,
      "loss": 0.0587,
      "step": 10629
    },
    {
      "epoch": 20.233000475511176,
      "grad_norm": 0.16020096838474274,
      "learning_rate": 3.251825976500476e-05,
      "loss": 0.0743,
      "step": 10630
    },
    {
      "epoch": 20.234902520209225,
      "grad_norm": 0.11568637937307358,
      "learning_rate": 3.2511908542394415e-05,
      "loss": 0.0558,
      "step": 10631
    },
    {
      "epoch": 20.236804564907274,
      "grad_norm": 0.049830514937639236,
      "learning_rate": 3.250555731978406e-05,
      "loss": 0.0593,
      "step": 10632
    },
    {
      "epoch": 20.238706609605327,
      "grad_norm": 0.1097976490855217,
      "learning_rate": 3.2499206097173705e-05,
      "loss": 0.0614,
      "step": 10633
    },
    {
      "epoch": 20.240608654303376,
      "grad_norm": 0.07119275629520416,
      "learning_rate": 3.249285487456336e-05,
      "loss": 0.064,
      "step": 10634
    },
    {
      "epoch": 20.242510699001425,
      "grad_norm": 0.11860954761505127,
      "learning_rate": 3.2486503651953e-05,
      "loss": 0.0656,
      "step": 10635
    },
    {
      "epoch": 20.244412743699478,
      "grad_norm": 0.0620465911924839,
      "learning_rate": 3.248015242934265e-05,
      "loss": 0.0463,
      "step": 10636
    },
    {
      "epoch": 20.246314788397527,
      "grad_norm": 0.09755115956068039,
      "learning_rate": 3.24738012067323e-05,
      "loss": 0.0398,
      "step": 10637
    },
    {
      "epoch": 20.24821683309558,
      "grad_norm": 0.21140997111797333,
      "learning_rate": 3.2467449984121944e-05,
      "loss": 0.0716,
      "step": 10638
    },
    {
      "epoch": 20.25011887779363,
      "grad_norm": 0.38212889432907104,
      "learning_rate": 3.2461098761511596e-05,
      "loss": 0.07,
      "step": 10639
    },
    {
      "epoch": 20.252020922491678,
      "grad_norm": 0.11554834991693497,
      "learning_rate": 3.245474753890124e-05,
      "loss": 0.0524,
      "step": 10640
    },
    {
      "epoch": 20.25392296718973,
      "grad_norm": 0.14220361411571503,
      "learning_rate": 3.2448396316290886e-05,
      "loss": 0.0562,
      "step": 10641
    },
    {
      "epoch": 20.25582501188778,
      "grad_norm": 0.09919548779726028,
      "learning_rate": 3.244204509368053e-05,
      "loss": 0.0471,
      "step": 10642
    },
    {
      "epoch": 20.25772705658583,
      "grad_norm": 0.04594764858484268,
      "learning_rate": 3.243569387107018e-05,
      "loss": 0.0479,
      "step": 10643
    },
    {
      "epoch": 20.25962910128388,
      "grad_norm": 0.1690056025981903,
      "learning_rate": 3.2429342648459835e-05,
      "loss": 0.0657,
      "step": 10644
    },
    {
      "epoch": 20.26153114598193,
      "grad_norm": 0.07409583777189255,
      "learning_rate": 3.242299142584947e-05,
      "loss": 0.0647,
      "step": 10645
    },
    {
      "epoch": 20.26343319067998,
      "grad_norm": 0.049228012561798096,
      "learning_rate": 3.2416640203239125e-05,
      "loss": 0.0732,
      "step": 10646
    },
    {
      "epoch": 20.265335235378032,
      "grad_norm": 0.15843163430690765,
      "learning_rate": 3.241028898062877e-05,
      "loss": 0.0509,
      "step": 10647
    },
    {
      "epoch": 20.26723728007608,
      "grad_norm": 0.15096253156661987,
      "learning_rate": 3.240393775801842e-05,
      "loss": 0.0542,
      "step": 10648
    },
    {
      "epoch": 20.269139324774134,
      "grad_norm": 0.18820707499980927,
      "learning_rate": 3.239758653540807e-05,
      "loss": 0.0638,
      "step": 10649
    },
    {
      "epoch": 20.271041369472183,
      "grad_norm": 0.09935000538825989,
      "learning_rate": 3.239123531279771e-05,
      "loss": 0.0653,
      "step": 10650
    },
    {
      "epoch": 20.272943414170232,
      "grad_norm": 0.05760449171066284,
      "learning_rate": 3.2384884090187364e-05,
      "loss": 0.0413,
      "step": 10651
    },
    {
      "epoch": 20.274845458868285,
      "grad_norm": 0.08053761720657349,
      "learning_rate": 3.237853286757701e-05,
      "loss": 0.033,
      "step": 10652
    },
    {
      "epoch": 20.276747503566334,
      "grad_norm": 0.04330631345510483,
      "learning_rate": 3.237218164496666e-05,
      "loss": 0.0395,
      "step": 10653
    },
    {
      "epoch": 20.278649548264383,
      "grad_norm": 0.11636463552713394,
      "learning_rate": 3.23658304223563e-05,
      "loss": 0.0588,
      "step": 10654
    },
    {
      "epoch": 20.280551592962436,
      "grad_norm": 0.16587413847446442,
      "learning_rate": 3.235947919974595e-05,
      "loss": 0.0788,
      "step": 10655
    },
    {
      "epoch": 20.282453637660485,
      "grad_norm": 0.10477234423160553,
      "learning_rate": 3.23531279771356e-05,
      "loss": 0.0715,
      "step": 10656
    },
    {
      "epoch": 20.284355682358534,
      "grad_norm": 0.05622519552707672,
      "learning_rate": 3.234677675452525e-05,
      "loss": 0.0691,
      "step": 10657
    },
    {
      "epoch": 20.286257727056586,
      "grad_norm": 0.0460372194647789,
      "learning_rate": 3.23404255319149e-05,
      "loss": 0.0574,
      "step": 10658
    },
    {
      "epoch": 20.288159771754636,
      "grad_norm": 0.07962246239185333,
      "learning_rate": 3.233407430930454e-05,
      "loss": 0.0682,
      "step": 10659
    },
    {
      "epoch": 20.29006181645269,
      "grad_norm": 0.10750008374452591,
      "learning_rate": 3.232772308669419e-05,
      "loss": 0.0617,
      "step": 10660
    },
    {
      "epoch": 20.291963861150737,
      "grad_norm": 0.2016543298959732,
      "learning_rate": 3.2321371864083835e-05,
      "loss": 0.0619,
      "step": 10661
    },
    {
      "epoch": 20.293865905848786,
      "grad_norm": 0.05132821947336197,
      "learning_rate": 3.231502064147349e-05,
      "loss": 0.0388,
      "step": 10662
    },
    {
      "epoch": 20.29576795054684,
      "grad_norm": 0.1965903788805008,
      "learning_rate": 3.230866941886313e-05,
      "loss": 0.0657,
      "step": 10663
    },
    {
      "epoch": 20.29766999524489,
      "grad_norm": 0.22762592136859894,
      "learning_rate": 3.230231819625278e-05,
      "loss": 0.0711,
      "step": 10664
    },
    {
      "epoch": 20.299572039942937,
      "grad_norm": 0.06518333405256271,
      "learning_rate": 3.229596697364243e-05,
      "loss": 0.0841,
      "step": 10665
    },
    {
      "epoch": 20.30147408464099,
      "grad_norm": 0.048307519406080246,
      "learning_rate": 3.2289615751032074e-05,
      "loss": 0.0357,
      "step": 10666
    },
    {
      "epoch": 20.30337612933904,
      "grad_norm": 0.04469428211450577,
      "learning_rate": 3.2283264528421726e-05,
      "loss": 0.062,
      "step": 10667
    },
    {
      "epoch": 20.30527817403709,
      "grad_norm": 0.1484098583459854,
      "learning_rate": 3.227691330581137e-05,
      "loss": 0.0607,
      "step": 10668
    },
    {
      "epoch": 20.30718021873514,
      "grad_norm": 0.0734637901186943,
      "learning_rate": 3.2270562083201016e-05,
      "loss": 0.0993,
      "step": 10669
    },
    {
      "epoch": 20.30908226343319,
      "grad_norm": 0.03457577899098396,
      "learning_rate": 3.226421086059067e-05,
      "loss": 0.0516,
      "step": 10670
    },
    {
      "epoch": 20.310984308131243,
      "grad_norm": 0.05358421057462692,
      "learning_rate": 3.225785963798031e-05,
      "loss": 0.066,
      "step": 10671
    },
    {
      "epoch": 20.312886352829292,
      "grad_norm": 0.033646367490291595,
      "learning_rate": 3.225150841536996e-05,
      "loss": 0.0477,
      "step": 10672
    },
    {
      "epoch": 20.31478839752734,
      "grad_norm": 0.10537423938512802,
      "learning_rate": 3.224515719275961e-05,
      "loss": 0.0797,
      "step": 10673
    },
    {
      "epoch": 20.316690442225394,
      "grad_norm": 0.16200460493564606,
      "learning_rate": 3.2238805970149255e-05,
      "loss": 0.0626,
      "step": 10674
    },
    {
      "epoch": 20.318592486923443,
      "grad_norm": 0.07985933870077133,
      "learning_rate": 3.2232454747538906e-05,
      "loss": 0.0667,
      "step": 10675
    },
    {
      "epoch": 20.32049453162149,
      "grad_norm": 0.047756507992744446,
      "learning_rate": 3.222610352492855e-05,
      "loss": 0.0641,
      "step": 10676
    },
    {
      "epoch": 20.322396576319544,
      "grad_norm": 0.04638848453760147,
      "learning_rate": 3.22197523023182e-05,
      "loss": 0.0489,
      "step": 10677
    },
    {
      "epoch": 20.324298621017594,
      "grad_norm": 0.07590161263942719,
      "learning_rate": 3.221340107970784e-05,
      "loss": 0.0556,
      "step": 10678
    },
    {
      "epoch": 20.326200665715643,
      "grad_norm": 0.04426351562142372,
      "learning_rate": 3.2207049857097494e-05,
      "loss": 0.0592,
      "step": 10679
    },
    {
      "epoch": 20.328102710413695,
      "grad_norm": 0.03708796575665474,
      "learning_rate": 3.220069863448714e-05,
      "loss": 0.0463,
      "step": 10680
    },
    {
      "epoch": 20.330004755111744,
      "grad_norm": 0.053088005632162094,
      "learning_rate": 3.2194347411876784e-05,
      "loss": 0.063,
      "step": 10681
    },
    {
      "epoch": 20.331906799809797,
      "grad_norm": 0.12782998383045197,
      "learning_rate": 3.2187996189266436e-05,
      "loss": 0.0604,
      "step": 10682
    },
    {
      "epoch": 20.333808844507846,
      "grad_norm": 0.06219214200973511,
      "learning_rate": 3.218164496665608e-05,
      "loss": 0.0762,
      "step": 10683
    },
    {
      "epoch": 20.335710889205895,
      "grad_norm": 0.15822537243366241,
      "learning_rate": 3.217529374404573e-05,
      "loss": 0.0525,
      "step": 10684
    },
    {
      "epoch": 20.337612933903948,
      "grad_norm": 0.0818241685628891,
      "learning_rate": 3.216894252143538e-05,
      "loss": 0.0487,
      "step": 10685
    },
    {
      "epoch": 20.339514978601997,
      "grad_norm": 0.1592274010181427,
      "learning_rate": 3.216259129882502e-05,
      "loss": 0.0469,
      "step": 10686
    },
    {
      "epoch": 20.341417023300046,
      "grad_norm": 0.10056457668542862,
      "learning_rate": 3.2156240076214674e-05,
      "loss": 0.0555,
      "step": 10687
    },
    {
      "epoch": 20.3433190679981,
      "grad_norm": 0.04942634701728821,
      "learning_rate": 3.214988885360432e-05,
      "loss": 0.0548,
      "step": 10688
    },
    {
      "epoch": 20.345221112696148,
      "grad_norm": 0.13030976057052612,
      "learning_rate": 3.214353763099397e-05,
      "loss": 0.0523,
      "step": 10689
    },
    {
      "epoch": 20.347123157394197,
      "grad_norm": 0.07076043635606766,
      "learning_rate": 3.213718640838361e-05,
      "loss": 0.0471,
      "step": 10690
    },
    {
      "epoch": 20.34902520209225,
      "grad_norm": 0.06825718283653259,
      "learning_rate": 3.213083518577326e-05,
      "loss": 0.0574,
      "step": 10691
    },
    {
      "epoch": 20.3509272467903,
      "grad_norm": 0.17072850465774536,
      "learning_rate": 3.212448396316291e-05,
      "loss": 0.0439,
      "step": 10692
    },
    {
      "epoch": 20.35282929148835,
      "grad_norm": 0.05790894106030464,
      "learning_rate": 3.211813274055256e-05,
      "loss": 0.037,
      "step": 10693
    },
    {
      "epoch": 20.3547313361864,
      "grad_norm": 0.1315700113773346,
      "learning_rate": 3.211178151794221e-05,
      "loss": 0.0632,
      "step": 10694
    },
    {
      "epoch": 20.35663338088445,
      "grad_norm": 0.16016748547554016,
      "learning_rate": 3.210543029533185e-05,
      "loss": 0.0655,
      "step": 10695
    },
    {
      "epoch": 20.358535425582502,
      "grad_norm": 0.07636621594429016,
      "learning_rate": 3.20990790727215e-05,
      "loss": 0.0591,
      "step": 10696
    },
    {
      "epoch": 20.36043747028055,
      "grad_norm": 0.18454785645008087,
      "learning_rate": 3.2092727850111145e-05,
      "loss": 0.0617,
      "step": 10697
    },
    {
      "epoch": 20.3623395149786,
      "grad_norm": 0.1629074215888977,
      "learning_rate": 3.20863766275008e-05,
      "loss": 0.0581,
      "step": 10698
    },
    {
      "epoch": 20.364241559676653,
      "grad_norm": 0.1100032702088356,
      "learning_rate": 3.208002540489044e-05,
      "loss": 0.0496,
      "step": 10699
    },
    {
      "epoch": 20.366143604374702,
      "grad_norm": 0.20545372366905212,
      "learning_rate": 3.207367418228009e-05,
      "loss": 0.0573,
      "step": 10700
    },
    {
      "epoch": 20.36804564907275,
      "grad_norm": 0.07017530500888824,
      "learning_rate": 3.206732295966974e-05,
      "loss": 0.0367,
      "step": 10701
    },
    {
      "epoch": 20.369947693770804,
      "grad_norm": 0.16110143065452576,
      "learning_rate": 3.2060971737059384e-05,
      "loss": 0.0636,
      "step": 10702
    },
    {
      "epoch": 20.371849738468853,
      "grad_norm": 0.20994989573955536,
      "learning_rate": 3.2054620514449036e-05,
      "loss": 0.056,
      "step": 10703
    },
    {
      "epoch": 20.373751783166906,
      "grad_norm": 0.10679108649492264,
      "learning_rate": 3.204826929183868e-05,
      "loss": 0.0571,
      "step": 10704
    },
    {
      "epoch": 20.375653827864955,
      "grad_norm": 0.0616210512816906,
      "learning_rate": 3.2041918069228326e-05,
      "loss": 0.0654,
      "step": 10705
    },
    {
      "epoch": 20.377555872563004,
      "grad_norm": 0.04933302849531174,
      "learning_rate": 3.203556684661798e-05,
      "loss": 0.0616,
      "step": 10706
    },
    {
      "epoch": 20.379457917261057,
      "grad_norm": 0.029315801337361336,
      "learning_rate": 3.202921562400762e-05,
      "loss": 0.0435,
      "step": 10707
    },
    {
      "epoch": 20.381359961959106,
      "grad_norm": 0.0827157124876976,
      "learning_rate": 3.202286440139727e-05,
      "loss": 0.0386,
      "step": 10708
    },
    {
      "epoch": 20.383262006657155,
      "grad_norm": 0.12883013486862183,
      "learning_rate": 3.201651317878691e-05,
      "loss": 0.0499,
      "step": 10709
    },
    {
      "epoch": 20.385164051355208,
      "grad_norm": 0.12958528101444244,
      "learning_rate": 3.2010161956176565e-05,
      "loss": 0.0619,
      "step": 10710
    },
    {
      "epoch": 20.387066096053257,
      "grad_norm": 0.057290948927402496,
      "learning_rate": 3.200381073356622e-05,
      "loss": 0.0534,
      "step": 10711
    },
    {
      "epoch": 20.388968140751306,
      "grad_norm": 0.06070676073431969,
      "learning_rate": 3.199745951095586e-05,
      "loss": 0.0652,
      "step": 10712
    },
    {
      "epoch": 20.39087018544936,
      "grad_norm": 0.10806555300951004,
      "learning_rate": 3.199110828834551e-05,
      "loss": 0.0411,
      "step": 10713
    },
    {
      "epoch": 20.392772230147408,
      "grad_norm": 0.07100493460893631,
      "learning_rate": 3.198475706573515e-05,
      "loss": 0.0441,
      "step": 10714
    },
    {
      "epoch": 20.39467427484546,
      "grad_norm": 0.1170794889330864,
      "learning_rate": 3.1978405843124804e-05,
      "loss": 0.0545,
      "step": 10715
    },
    {
      "epoch": 20.39657631954351,
      "grad_norm": 0.07554541528224945,
      "learning_rate": 3.197205462051445e-05,
      "loss": 0.0424,
      "step": 10716
    },
    {
      "epoch": 20.39847836424156,
      "grad_norm": 0.17473971843719482,
      "learning_rate": 3.1965703397904094e-05,
      "loss": 0.0444,
      "step": 10717
    },
    {
      "epoch": 20.40038040893961,
      "grad_norm": 0.11114981025457382,
      "learning_rate": 3.1959352175293746e-05,
      "loss": 0.0478,
      "step": 10718
    },
    {
      "epoch": 20.40228245363766,
      "grad_norm": 0.15614411234855652,
      "learning_rate": 3.195300095268339e-05,
      "loss": 0.0689,
      "step": 10719
    },
    {
      "epoch": 20.40418449833571,
      "grad_norm": 0.07727070897817612,
      "learning_rate": 3.194664973007304e-05,
      "loss": 0.0489,
      "step": 10720
    },
    {
      "epoch": 20.406086543033762,
      "grad_norm": 0.11544694751501083,
      "learning_rate": 3.194029850746269e-05,
      "loss": 0.0627,
      "step": 10721
    },
    {
      "epoch": 20.40798858773181,
      "grad_norm": 0.07351727038621902,
      "learning_rate": 3.193394728485233e-05,
      "loss": 0.07,
      "step": 10722
    },
    {
      "epoch": 20.409890632429864,
      "grad_norm": 0.04631667584180832,
      "learning_rate": 3.1927596062241985e-05,
      "loss": 0.0702,
      "step": 10723
    },
    {
      "epoch": 20.411792677127913,
      "grad_norm": 0.06886739283800125,
      "learning_rate": 3.192124483963163e-05,
      "loss": 0.0393,
      "step": 10724
    },
    {
      "epoch": 20.413694721825962,
      "grad_norm": 0.05337328091263771,
      "learning_rate": 3.191489361702128e-05,
      "loss": 0.0609,
      "step": 10725
    },
    {
      "epoch": 20.415596766524015,
      "grad_norm": 0.11911852657794952,
      "learning_rate": 3.190854239441092e-05,
      "loss": 0.0726,
      "step": 10726
    },
    {
      "epoch": 20.417498811222064,
      "grad_norm": 0.16817975044250488,
      "learning_rate": 3.190219117180057e-05,
      "loss": 0.0719,
      "step": 10727
    },
    {
      "epoch": 20.419400855920113,
      "grad_norm": 0.04718000814318657,
      "learning_rate": 3.1895839949190224e-05,
      "loss": 0.0559,
      "step": 10728
    },
    {
      "epoch": 20.421302900618166,
      "grad_norm": 0.09537462890148163,
      "learning_rate": 3.188948872657987e-05,
      "loss": 0.0762,
      "step": 10729
    },
    {
      "epoch": 20.423204945316215,
      "grad_norm": 0.18325896561145782,
      "learning_rate": 3.188313750396952e-05,
      "loss": 0.058,
      "step": 10730
    },
    {
      "epoch": 20.425106990014264,
      "grad_norm": 0.28283315896987915,
      "learning_rate": 3.187678628135916e-05,
      "loss": 0.0598,
      "step": 10731
    },
    {
      "epoch": 20.427009034712317,
      "grad_norm": 0.15887336432933807,
      "learning_rate": 3.187043505874881e-05,
      "loss": 0.0394,
      "step": 10732
    },
    {
      "epoch": 20.428911079410366,
      "grad_norm": 0.08057612180709839,
      "learning_rate": 3.1864083836138456e-05,
      "loss": 0.0644,
      "step": 10733
    },
    {
      "epoch": 20.43081312410842,
      "grad_norm": 0.05225507542490959,
      "learning_rate": 3.185773261352811e-05,
      "loss": 0.044,
      "step": 10734
    },
    {
      "epoch": 20.432715168806467,
      "grad_norm": 0.07600190490484238,
      "learning_rate": 3.185138139091775e-05,
      "loss": 0.0595,
      "step": 10735
    },
    {
      "epoch": 20.434617213504517,
      "grad_norm": 0.3140774965286255,
      "learning_rate": 3.18450301683074e-05,
      "loss": 0.0707,
      "step": 10736
    },
    {
      "epoch": 20.43651925820257,
      "grad_norm": 0.09039638191461563,
      "learning_rate": 3.183867894569705e-05,
      "loss": 0.0476,
      "step": 10737
    },
    {
      "epoch": 20.43842130290062,
      "grad_norm": 0.0431450791656971,
      "learning_rate": 3.1832327723086695e-05,
      "loss": 0.0644,
      "step": 10738
    },
    {
      "epoch": 20.440323347598667,
      "grad_norm": 0.11939884722232819,
      "learning_rate": 3.1825976500476347e-05,
      "loss": 0.0681,
      "step": 10739
    },
    {
      "epoch": 20.44222539229672,
      "grad_norm": 0.06911275535821915,
      "learning_rate": 3.181962527786599e-05,
      "loss": 0.0501,
      "step": 10740
    },
    {
      "epoch": 20.44412743699477,
      "grad_norm": 0.18807461857795715,
      "learning_rate": 3.181327405525564e-05,
      "loss": 0.0616,
      "step": 10741
    },
    {
      "epoch": 20.44602948169282,
      "grad_norm": 0.11017563194036484,
      "learning_rate": 3.180692283264529e-05,
      "loss": 0.0593,
      "step": 10742
    },
    {
      "epoch": 20.44793152639087,
      "grad_norm": 0.09336445480585098,
      "learning_rate": 3.1800571610034934e-05,
      "loss": 0.0429,
      "step": 10743
    },
    {
      "epoch": 20.44983357108892,
      "grad_norm": 0.14734186232089996,
      "learning_rate": 3.179422038742458e-05,
      "loss": 0.0414,
      "step": 10744
    },
    {
      "epoch": 20.451735615786973,
      "grad_norm": 0.06388304382562637,
      "learning_rate": 3.1787869164814224e-05,
      "loss": 0.0492,
      "step": 10745
    },
    {
      "epoch": 20.453637660485022,
      "grad_norm": 0.08828426897525787,
      "learning_rate": 3.1781517942203876e-05,
      "loss": 0.0416,
      "step": 10746
    },
    {
      "epoch": 20.45553970518307,
      "grad_norm": 0.10653724521398544,
      "learning_rate": 3.177516671959353e-05,
      "loss": 0.0627,
      "step": 10747
    },
    {
      "epoch": 20.457441749881124,
      "grad_norm": 0.058555085211992264,
      "learning_rate": 3.176881549698317e-05,
      "loss": 0.0529,
      "step": 10748
    },
    {
      "epoch": 20.459343794579173,
      "grad_norm": 0.08683603256940842,
      "learning_rate": 3.176246427437282e-05,
      "loss": 0.0595,
      "step": 10749
    },
    {
      "epoch": 20.461245839277222,
      "grad_norm": 0.1622881293296814,
      "learning_rate": 3.175611305176246e-05,
      "loss": 0.0635,
      "step": 10750
    },
    {
      "epoch": 20.463147883975275,
      "grad_norm": 0.09806717187166214,
      "learning_rate": 3.1749761829152114e-05,
      "loss": 0.0549,
      "step": 10751
    },
    {
      "epoch": 20.465049928673324,
      "grad_norm": 0.17263513803482056,
      "learning_rate": 3.174341060654176e-05,
      "loss": 0.0649,
      "step": 10752
    },
    {
      "epoch": 20.466951973371373,
      "grad_norm": 0.061647363007068634,
      "learning_rate": 3.1737059383931405e-05,
      "loss": 0.065,
      "step": 10753
    },
    {
      "epoch": 20.468854018069425,
      "grad_norm": 0.08864830434322357,
      "learning_rate": 3.1730708161321056e-05,
      "loss": 0.0564,
      "step": 10754
    },
    {
      "epoch": 20.470756062767475,
      "grad_norm": 0.06646085530519485,
      "learning_rate": 3.17243569387107e-05,
      "loss": 0.0717,
      "step": 10755
    },
    {
      "epoch": 20.472658107465527,
      "grad_norm": 0.04390653595328331,
      "learning_rate": 3.171800571610035e-05,
      "loss": 0.0627,
      "step": 10756
    },
    {
      "epoch": 20.474560152163576,
      "grad_norm": 0.0896117091178894,
      "learning_rate": 3.171165449349e-05,
      "loss": 0.0781,
      "step": 10757
    },
    {
      "epoch": 20.476462196861625,
      "grad_norm": 0.07033764570951462,
      "learning_rate": 3.1705303270879643e-05,
      "loss": 0.0548,
      "step": 10758
    },
    {
      "epoch": 20.478364241559678,
      "grad_norm": 0.056597303599119186,
      "learning_rate": 3.1698952048269295e-05,
      "loss": 0.0499,
      "step": 10759
    },
    {
      "epoch": 20.480266286257727,
      "grad_norm": 0.04619000479578972,
      "learning_rate": 3.169260082565894e-05,
      "loss": 0.0613,
      "step": 10760
    },
    {
      "epoch": 20.482168330955776,
      "grad_norm": 0.15724436938762665,
      "learning_rate": 3.168624960304859e-05,
      "loss": 0.0471,
      "step": 10761
    },
    {
      "epoch": 20.48407037565383,
      "grad_norm": 0.06588472425937653,
      "learning_rate": 3.167989838043823e-05,
      "loss": 0.0569,
      "step": 10762
    },
    {
      "epoch": 20.485972420351878,
      "grad_norm": 0.038958970457315445,
      "learning_rate": 3.167354715782788e-05,
      "loss": 0.052,
      "step": 10763
    },
    {
      "epoch": 20.487874465049927,
      "grad_norm": 0.13805849850177765,
      "learning_rate": 3.166719593521753e-05,
      "loss": 0.0502,
      "step": 10764
    },
    {
      "epoch": 20.48977650974798,
      "grad_norm": 0.1267121285200119,
      "learning_rate": 3.166084471260718e-05,
      "loss": 0.0529,
      "step": 10765
    },
    {
      "epoch": 20.49167855444603,
      "grad_norm": 0.04090994969010353,
      "learning_rate": 3.165449348999683e-05,
      "loss": 0.0403,
      "step": 10766
    },
    {
      "epoch": 20.49358059914408,
      "grad_norm": 0.0730077475309372,
      "learning_rate": 3.164814226738647e-05,
      "loss": 0.073,
      "step": 10767
    },
    {
      "epoch": 20.49548264384213,
      "grad_norm": 0.050069231539964676,
      "learning_rate": 3.164179104477612e-05,
      "loss": 0.0624,
      "step": 10768
    },
    {
      "epoch": 20.49738468854018,
      "grad_norm": 0.04610844701528549,
      "learning_rate": 3.1635439822165766e-05,
      "loss": 0.04,
      "step": 10769
    },
    {
      "epoch": 20.499286733238232,
      "grad_norm": 0.19118037819862366,
      "learning_rate": 3.162908859955542e-05,
      "loss": 0.0596,
      "step": 10770
    },
    {
      "epoch": 20.50118877793628,
      "grad_norm": 0.09393315762281418,
      "learning_rate": 3.162273737694506e-05,
      "loss": 0.0526,
      "step": 10771
    },
    {
      "epoch": 20.50309082263433,
      "grad_norm": 0.08483687788248062,
      "learning_rate": 3.161638615433471e-05,
      "loss": 0.0662,
      "step": 10772
    },
    {
      "epoch": 20.504992867332383,
      "grad_norm": 0.1932431012392044,
      "learning_rate": 3.161003493172436e-05,
      "loss": 0.0451,
      "step": 10773
    },
    {
      "epoch": 20.506894912030432,
      "grad_norm": 0.10873065888881683,
      "learning_rate": 3.1603683709114005e-05,
      "loss": 0.0485,
      "step": 10774
    },
    {
      "epoch": 20.50879695672848,
      "grad_norm": 0.08355213701725006,
      "learning_rate": 3.159733248650366e-05,
      "loss": 0.0419,
      "step": 10775
    },
    {
      "epoch": 20.510699001426534,
      "grad_norm": 0.0491744689643383,
      "learning_rate": 3.15909812638933e-05,
      "loss": 0.0466,
      "step": 10776
    },
    {
      "epoch": 20.512601046124583,
      "grad_norm": 0.09892838448286057,
      "learning_rate": 3.158463004128295e-05,
      "loss": 0.0563,
      "step": 10777
    },
    {
      "epoch": 20.514503090822636,
      "grad_norm": 0.09677373617887497,
      "learning_rate": 3.15782788186726e-05,
      "loss": 0.0774,
      "step": 10778
    },
    {
      "epoch": 20.516405135520685,
      "grad_norm": 0.11103257536888123,
      "learning_rate": 3.1571927596062244e-05,
      "loss": 0.0595,
      "step": 10779
    },
    {
      "epoch": 20.518307180218734,
      "grad_norm": 0.19784103333950043,
      "learning_rate": 3.156557637345189e-05,
      "loss": 0.0512,
      "step": 10780
    },
    {
      "epoch": 20.520209224916787,
      "grad_norm": 0.04738070070743561,
      "learning_rate": 3.1559225150841534e-05,
      "loss": 0.0466,
      "step": 10781
    },
    {
      "epoch": 20.522111269614836,
      "grad_norm": 0.11906538903713226,
      "learning_rate": 3.1552873928231186e-05,
      "loss": 0.0806,
      "step": 10782
    },
    {
      "epoch": 20.524013314312885,
      "grad_norm": 0.038742128759622574,
      "learning_rate": 3.154652270562084e-05,
      "loss": 0.0581,
      "step": 10783
    },
    {
      "epoch": 20.525915359010938,
      "grad_norm": 0.1178169995546341,
      "learning_rate": 3.154017148301048e-05,
      "loss": 0.0491,
      "step": 10784
    },
    {
      "epoch": 20.527817403708987,
      "grad_norm": 0.03254316374659538,
      "learning_rate": 3.153382026040013e-05,
      "loss": 0.0409,
      "step": 10785
    },
    {
      "epoch": 20.529719448407036,
      "grad_norm": 0.04235231876373291,
      "learning_rate": 3.152746903778977e-05,
      "loss": 0.0597,
      "step": 10786
    },
    {
      "epoch": 20.53162149310509,
      "grad_norm": 0.16329194605350494,
      "learning_rate": 3.1521117815179425e-05,
      "loss": 0.0644,
      "step": 10787
    },
    {
      "epoch": 20.533523537803138,
      "grad_norm": 0.04553667828440666,
      "learning_rate": 3.151476659256907e-05,
      "loss": 0.0485,
      "step": 10788
    },
    {
      "epoch": 20.53542558250119,
      "grad_norm": 0.13178446888923645,
      "learning_rate": 3.1508415369958715e-05,
      "loss": 0.0569,
      "step": 10789
    },
    {
      "epoch": 20.53732762719924,
      "grad_norm": 0.04360739141702652,
      "learning_rate": 3.150206414734837e-05,
      "loss": 0.0662,
      "step": 10790
    },
    {
      "epoch": 20.53922967189729,
      "grad_norm": 0.06077009066939354,
      "learning_rate": 3.149571292473801e-05,
      "loss": 0.0437,
      "step": 10791
    },
    {
      "epoch": 20.54113171659534,
      "grad_norm": 0.15785424411296844,
      "learning_rate": 3.1489361702127664e-05,
      "loss": 0.0656,
      "step": 10792
    },
    {
      "epoch": 20.54303376129339,
      "grad_norm": 0.05192839354276657,
      "learning_rate": 3.148301047951731e-05,
      "loss": 0.0518,
      "step": 10793
    },
    {
      "epoch": 20.54493580599144,
      "grad_norm": 0.05656667798757553,
      "learning_rate": 3.1476659256906954e-05,
      "loss": 0.0665,
      "step": 10794
    },
    {
      "epoch": 20.546837850689492,
      "grad_norm": 0.044597215950489044,
      "learning_rate": 3.1470308034296606e-05,
      "loss": 0.0619,
      "step": 10795
    },
    {
      "epoch": 20.54873989538754,
      "grad_norm": 0.22907453775405884,
      "learning_rate": 3.146395681168625e-05,
      "loss": 0.0696,
      "step": 10796
    },
    {
      "epoch": 20.55064194008559,
      "grad_norm": 0.05552733689546585,
      "learning_rate": 3.14576055890759e-05,
      "loss": 0.0772,
      "step": 10797
    },
    {
      "epoch": 20.552543984783643,
      "grad_norm": 0.026332750916481018,
      "learning_rate": 3.145125436646554e-05,
      "loss": 0.0393,
      "step": 10798
    },
    {
      "epoch": 20.554446029481692,
      "grad_norm": 0.13045305013656616,
      "learning_rate": 3.144490314385519e-05,
      "loss": 0.0458,
      "step": 10799
    },
    {
      "epoch": 20.556348074179745,
      "grad_norm": 0.051635902374982834,
      "learning_rate": 3.143855192124484e-05,
      "loss": 0.0564,
      "step": 10800
    },
    {
      "epoch": 20.558250118877794,
      "grad_norm": 0.056866101920604706,
      "learning_rate": 3.143220069863449e-05,
      "loss": 0.0663,
      "step": 10801
    },
    {
      "epoch": 20.560152163575843,
      "grad_norm": 0.05314189940690994,
      "learning_rate": 3.142584947602414e-05,
      "loss": 0.0668,
      "step": 10802
    },
    {
      "epoch": 20.562054208273896,
      "grad_norm": 0.1597507894039154,
      "learning_rate": 3.141949825341378e-05,
      "loss": 0.0569,
      "step": 10803
    },
    {
      "epoch": 20.563956252971945,
      "grad_norm": 0.06841409206390381,
      "learning_rate": 3.141314703080343e-05,
      "loss": 0.0699,
      "step": 10804
    },
    {
      "epoch": 20.565858297669994,
      "grad_norm": 0.06581047922372818,
      "learning_rate": 3.140679580819308e-05,
      "loss": 0.0535,
      "step": 10805
    },
    {
      "epoch": 20.567760342368047,
      "grad_norm": 0.11147249490022659,
      "learning_rate": 3.140044458558273e-05,
      "loss": 0.0545,
      "step": 10806
    },
    {
      "epoch": 20.569662387066096,
      "grad_norm": 0.1470136195421219,
      "learning_rate": 3.1394093362972374e-05,
      "loss": 0.0508,
      "step": 10807
    },
    {
      "epoch": 20.571564431764145,
      "grad_norm": 0.12944050133228302,
      "learning_rate": 3.138774214036202e-05,
      "loss": 0.0582,
      "step": 10808
    },
    {
      "epoch": 20.573466476462198,
      "grad_norm": 0.042375169694423676,
      "learning_rate": 3.138139091775167e-05,
      "loss": 0.0541,
      "step": 10809
    },
    {
      "epoch": 20.575368521160247,
      "grad_norm": 0.03796051815152168,
      "learning_rate": 3.1375039695141316e-05,
      "loss": 0.0475,
      "step": 10810
    },
    {
      "epoch": 20.5772705658583,
      "grad_norm": 0.03297824040055275,
      "learning_rate": 3.136868847253097e-05,
      "loss": 0.0434,
      "step": 10811
    },
    {
      "epoch": 20.57917261055635,
      "grad_norm": 0.0863594189286232,
      "learning_rate": 3.136233724992061e-05,
      "loss": 0.0382,
      "step": 10812
    },
    {
      "epoch": 20.581074655254397,
      "grad_norm": 0.13227379322052002,
      "learning_rate": 3.135598602731026e-05,
      "loss": 0.0576,
      "step": 10813
    },
    {
      "epoch": 20.58297669995245,
      "grad_norm": 0.07175970822572708,
      "learning_rate": 3.134963480469991e-05,
      "loss": 0.0525,
      "step": 10814
    },
    {
      "epoch": 20.5848787446505,
      "grad_norm": 0.0745711550116539,
      "learning_rate": 3.1343283582089554e-05,
      "loss": 0.0528,
      "step": 10815
    },
    {
      "epoch": 20.58678078934855,
      "grad_norm": 0.10481584072113037,
      "learning_rate": 3.13369323594792e-05,
      "loss": 0.0495,
      "step": 10816
    },
    {
      "epoch": 20.5886828340466,
      "grad_norm": 0.19122248888015747,
      "learning_rate": 3.1330581136868845e-05,
      "loss": 0.0593,
      "step": 10817
    },
    {
      "epoch": 20.59058487874465,
      "grad_norm": 0.12699630856513977,
      "learning_rate": 3.1324229914258496e-05,
      "loss": 0.0621,
      "step": 10818
    },
    {
      "epoch": 20.5924869234427,
      "grad_norm": 0.11133062094449997,
      "learning_rate": 3.131787869164814e-05,
      "loss": 0.0439,
      "step": 10819
    },
    {
      "epoch": 20.594388968140752,
      "grad_norm": 0.21265095472335815,
      "learning_rate": 3.131152746903779e-05,
      "loss": 0.032,
      "step": 10820
    },
    {
      "epoch": 20.5962910128388,
      "grad_norm": 0.11187831312417984,
      "learning_rate": 3.130517624642744e-05,
      "loss": 0.0674,
      "step": 10821
    },
    {
      "epoch": 20.598193057536854,
      "grad_norm": 0.09113247692584991,
      "learning_rate": 3.1298825023817083e-05,
      "loss": 0.0431,
      "step": 10822
    },
    {
      "epoch": 20.600095102234903,
      "grad_norm": 0.20912741124629974,
      "learning_rate": 3.1292473801206735e-05,
      "loss": 0.0459,
      "step": 10823
    },
    {
      "epoch": 20.601997146932952,
      "grad_norm": 0.3858474791049957,
      "learning_rate": 3.128612257859638e-05,
      "loss": 0.0699,
      "step": 10824
    },
    {
      "epoch": 20.603899191631005,
      "grad_norm": 0.13949212431907654,
      "learning_rate": 3.1279771355986025e-05,
      "loss": 0.0672,
      "step": 10825
    },
    {
      "epoch": 20.605801236329054,
      "grad_norm": 0.05572618916630745,
      "learning_rate": 3.127342013337568e-05,
      "loss": 0.0607,
      "step": 10826
    },
    {
      "epoch": 20.607703281027103,
      "grad_norm": 0.24470728635787964,
      "learning_rate": 3.126706891076532e-05,
      "loss": 0.0976,
      "step": 10827
    },
    {
      "epoch": 20.609605325725155,
      "grad_norm": 0.10430791974067688,
      "learning_rate": 3.1260717688154974e-05,
      "loss": 0.0522,
      "step": 10828
    },
    {
      "epoch": 20.611507370423205,
      "grad_norm": 0.05254757031798363,
      "learning_rate": 3.125436646554462e-05,
      "loss": 0.0684,
      "step": 10829
    },
    {
      "epoch": 20.613409415121254,
      "grad_norm": 0.10193728655576706,
      "learning_rate": 3.1248015242934264e-05,
      "loss": 0.0571,
      "step": 10830
    },
    {
      "epoch": 20.615311459819306,
      "grad_norm": 0.0633193850517273,
      "learning_rate": 3.1241664020323916e-05,
      "loss": 0.0625,
      "step": 10831
    },
    {
      "epoch": 20.617213504517355,
      "grad_norm": 0.15912435948848724,
      "learning_rate": 3.123531279771356e-05,
      "loss": 0.077,
      "step": 10832
    },
    {
      "epoch": 20.619115549215408,
      "grad_norm": 0.1428174376487732,
      "learning_rate": 3.122896157510321e-05,
      "loss": 0.057,
      "step": 10833
    },
    {
      "epoch": 20.621017593913457,
      "grad_norm": 0.18162624537944794,
      "learning_rate": 3.122261035249285e-05,
      "loss": 0.061,
      "step": 10834
    },
    {
      "epoch": 20.622919638611506,
      "grad_norm": 0.34952130913734436,
      "learning_rate": 3.12162591298825e-05,
      "loss": 0.1014,
      "step": 10835
    },
    {
      "epoch": 20.62482168330956,
      "grad_norm": 0.06106233596801758,
      "learning_rate": 3.120990790727215e-05,
      "loss": 0.0434,
      "step": 10836
    },
    {
      "epoch": 20.626723728007608,
      "grad_norm": 0.08583278954029083,
      "learning_rate": 3.12035566846618e-05,
      "loss": 0.0609,
      "step": 10837
    },
    {
      "epoch": 20.628625772705657,
      "grad_norm": 0.3161568343639374,
      "learning_rate": 3.119720546205145e-05,
      "loss": 0.0511,
      "step": 10838
    },
    {
      "epoch": 20.63052781740371,
      "grad_norm": 0.04114923253655434,
      "learning_rate": 3.119085423944109e-05,
      "loss": 0.0592,
      "step": 10839
    },
    {
      "epoch": 20.63242986210176,
      "grad_norm": 0.33630475401878357,
      "learning_rate": 3.118450301683074e-05,
      "loss": 0.0786,
      "step": 10840
    },
    {
      "epoch": 20.634331906799808,
      "grad_norm": 0.19183726608753204,
      "learning_rate": 3.117815179422039e-05,
      "loss": 0.0512,
      "step": 10841
    },
    {
      "epoch": 20.63623395149786,
      "grad_norm": 0.0717271938920021,
      "learning_rate": 3.117180057161004e-05,
      "loss": 0.0655,
      "step": 10842
    },
    {
      "epoch": 20.63813599619591,
      "grad_norm": 0.0996527448296547,
      "learning_rate": 3.1165449348999684e-05,
      "loss": 0.0629,
      "step": 10843
    },
    {
      "epoch": 20.640038040893963,
      "grad_norm": 0.18561841547489166,
      "learning_rate": 3.115909812638933e-05,
      "loss": 0.046,
      "step": 10844
    },
    {
      "epoch": 20.64194008559201,
      "grad_norm": 0.08247508108615875,
      "learning_rate": 3.115274690377898e-05,
      "loss": 0.0628,
      "step": 10845
    },
    {
      "epoch": 20.64384213029006,
      "grad_norm": 0.056294526904821396,
      "learning_rate": 3.1146395681168626e-05,
      "loss": 0.0415,
      "step": 10846
    },
    {
      "epoch": 20.645744174988113,
      "grad_norm": 0.14750158786773682,
      "learning_rate": 3.114004445855828e-05,
      "loss": 0.0636,
      "step": 10847
    },
    {
      "epoch": 20.647646219686163,
      "grad_norm": 0.12316614389419556,
      "learning_rate": 3.1133693235947916e-05,
      "loss": 0.0507,
      "step": 10848
    },
    {
      "epoch": 20.64954826438421,
      "grad_norm": 0.12978743016719818,
      "learning_rate": 3.112734201333757e-05,
      "loss": 0.0622,
      "step": 10849
    },
    {
      "epoch": 20.651450309082264,
      "grad_norm": 0.08613703399896622,
      "learning_rate": 3.112099079072722e-05,
      "loss": 0.0529,
      "step": 10850
    },
    {
      "epoch": 20.653352353780313,
      "grad_norm": 0.1138317883014679,
      "learning_rate": 3.1114639568116865e-05,
      "loss": 0.0408,
      "step": 10851
    },
    {
      "epoch": 20.655254398478363,
      "grad_norm": 0.14546433091163635,
      "learning_rate": 3.110828834550651e-05,
      "loss": 0.0593,
      "step": 10852
    },
    {
      "epoch": 20.657156443176415,
      "grad_norm": 0.11413826048374176,
      "learning_rate": 3.1101937122896155e-05,
      "loss": 0.0464,
      "step": 10853
    },
    {
      "epoch": 20.659058487874464,
      "grad_norm": 0.11910077184438705,
      "learning_rate": 3.109558590028581e-05,
      "loss": 0.0645,
      "step": 10854
    },
    {
      "epoch": 20.660960532572517,
      "grad_norm": 0.058692291378974915,
      "learning_rate": 3.108923467767545e-05,
      "loss": 0.0482,
      "step": 10855
    },
    {
      "epoch": 20.662862577270566,
      "grad_norm": 0.13482342660427094,
      "learning_rate": 3.1082883455065104e-05,
      "loss": 0.0626,
      "step": 10856
    },
    {
      "epoch": 20.664764621968615,
      "grad_norm": 0.07138006389141083,
      "learning_rate": 3.107653223245475e-05,
      "loss": 0.0412,
      "step": 10857
    },
    {
      "epoch": 20.666666666666668,
      "grad_norm": 0.06352374702692032,
      "learning_rate": 3.1070181009844394e-05,
      "loss": 0.0512,
      "step": 10858
    },
    {
      "epoch": 20.668568711364717,
      "grad_norm": 0.13332059979438782,
      "learning_rate": 3.1063829787234046e-05,
      "loss": 0.0546,
      "step": 10859
    },
    {
      "epoch": 20.670470756062766,
      "grad_norm": 0.14220471680164337,
      "learning_rate": 3.105747856462369e-05,
      "loss": 0.0702,
      "step": 10860
    },
    {
      "epoch": 20.67237280076082,
      "grad_norm": 0.2089843600988388,
      "learning_rate": 3.1051127342013336e-05,
      "loss": 0.0652,
      "step": 10861
    },
    {
      "epoch": 20.674274845458868,
      "grad_norm": 0.03339073434472084,
      "learning_rate": 3.104477611940299e-05,
      "loss": 0.0281,
      "step": 10862
    },
    {
      "epoch": 20.676176890156917,
      "grad_norm": 0.04318920150399208,
      "learning_rate": 3.103842489679263e-05,
      "loss": 0.0523,
      "step": 10863
    },
    {
      "epoch": 20.67807893485497,
      "grad_norm": 0.21177279949188232,
      "learning_rate": 3.1032073674182285e-05,
      "loss": 0.0564,
      "step": 10864
    },
    {
      "epoch": 20.67998097955302,
      "grad_norm": 0.08779855817556381,
      "learning_rate": 3.102572245157193e-05,
      "loss": 0.0443,
      "step": 10865
    },
    {
      "epoch": 20.68188302425107,
      "grad_norm": 0.18260325491428375,
      "learning_rate": 3.1019371228961575e-05,
      "loss": 0.0574,
      "step": 10866
    },
    {
      "epoch": 20.68378506894912,
      "grad_norm": 0.18878425657749176,
      "learning_rate": 3.1013020006351227e-05,
      "loss": 0.067,
      "step": 10867
    },
    {
      "epoch": 20.68568711364717,
      "grad_norm": 0.27964478731155396,
      "learning_rate": 3.100666878374087e-05,
      "loss": 0.0713,
      "step": 10868
    },
    {
      "epoch": 20.687589158345222,
      "grad_norm": 0.08817431330680847,
      "learning_rate": 3.1000317561130524e-05,
      "loss": 0.0418,
      "step": 10869
    },
    {
      "epoch": 20.68949120304327,
      "grad_norm": 0.40673351287841797,
      "learning_rate": 3.099396633852016e-05,
      "loss": 0.0635,
      "step": 10870
    },
    {
      "epoch": 20.69139324774132,
      "grad_norm": 0.09810397773981094,
      "learning_rate": 3.0987615115909814e-05,
      "loss": 0.055,
      "step": 10871
    },
    {
      "epoch": 20.693295292439373,
      "grad_norm": 0.1463918834924698,
      "learning_rate": 3.098126389329946e-05,
      "loss": 0.0435,
      "step": 10872
    },
    {
      "epoch": 20.695197337137422,
      "grad_norm": 0.18395887315273285,
      "learning_rate": 3.097491267068911e-05,
      "loss": 0.0578,
      "step": 10873
    },
    {
      "epoch": 20.69709938183547,
      "grad_norm": 0.10693424940109253,
      "learning_rate": 3.0968561448078756e-05,
      "loss": 0.0653,
      "step": 10874
    },
    {
      "epoch": 20.699001426533524,
      "grad_norm": 0.1919068992137909,
      "learning_rate": 3.09622102254684e-05,
      "loss": 0.068,
      "step": 10875
    },
    {
      "epoch": 20.700903471231573,
      "grad_norm": 0.06700928509235382,
      "learning_rate": 3.095585900285805e-05,
      "loss": 0.0469,
      "step": 10876
    },
    {
      "epoch": 20.702805515929626,
      "grad_norm": 0.10424041748046875,
      "learning_rate": 3.09495077802477e-05,
      "loss": 0.0499,
      "step": 10877
    },
    {
      "epoch": 20.704707560627675,
      "grad_norm": 0.09712348133325577,
      "learning_rate": 3.094315655763735e-05,
      "loss": 0.0631,
      "step": 10878
    },
    {
      "epoch": 20.706609605325724,
      "grad_norm": 0.06249113380908966,
      "learning_rate": 3.0936805335026994e-05,
      "loss": 0.0542,
      "step": 10879
    },
    {
      "epoch": 20.708511650023777,
      "grad_norm": 0.1309369057416916,
      "learning_rate": 3.093045411241664e-05,
      "loss": 0.0502,
      "step": 10880
    },
    {
      "epoch": 20.710413694721826,
      "grad_norm": 0.09550520032644272,
      "learning_rate": 3.092410288980629e-05,
      "loss": 0.0572,
      "step": 10881
    },
    {
      "epoch": 20.712315739419875,
      "grad_norm": 0.055158138275146484,
      "learning_rate": 3.0917751667195936e-05,
      "loss": 0.0622,
      "step": 10882
    },
    {
      "epoch": 20.714217784117928,
      "grad_norm": 0.20897230505943298,
      "learning_rate": 3.091140044458559e-05,
      "loss": 0.0584,
      "step": 10883
    },
    {
      "epoch": 20.716119828815977,
      "grad_norm": 0.06125312298536301,
      "learning_rate": 3.0905049221975227e-05,
      "loss": 0.0517,
      "step": 10884
    },
    {
      "epoch": 20.718021873514026,
      "grad_norm": 0.09136564284563065,
      "learning_rate": 3.089869799936488e-05,
      "loss": 0.0502,
      "step": 10885
    },
    {
      "epoch": 20.71992391821208,
      "grad_norm": 0.06094088777899742,
      "learning_rate": 3.089234677675453e-05,
      "loss": 0.0575,
      "step": 10886
    },
    {
      "epoch": 20.721825962910128,
      "grad_norm": 0.12295346707105637,
      "learning_rate": 3.0885995554144175e-05,
      "loss": 0.0672,
      "step": 10887
    },
    {
      "epoch": 20.72372800760818,
      "grad_norm": 0.061890989542007446,
      "learning_rate": 3.087964433153382e-05,
      "loss": 0.0631,
      "step": 10888
    },
    {
      "epoch": 20.72563005230623,
      "grad_norm": 0.08851854503154755,
      "learning_rate": 3.0873293108923465e-05,
      "loss": 0.0553,
      "step": 10889
    },
    {
      "epoch": 20.72753209700428,
      "grad_norm": 0.13970603048801422,
      "learning_rate": 3.086694188631312e-05,
      "loss": 0.0537,
      "step": 10890
    },
    {
      "epoch": 20.72943414170233,
      "grad_norm": 0.13283051550388336,
      "learning_rate": 3.086059066370276e-05,
      "loss": 0.0567,
      "step": 10891
    },
    {
      "epoch": 20.73133618640038,
      "grad_norm": 0.12537896633148193,
      "learning_rate": 3.0854239441092414e-05,
      "loss": 0.0542,
      "step": 10892
    },
    {
      "epoch": 20.73323823109843,
      "grad_norm": 0.1433955729007721,
      "learning_rate": 3.084788821848206e-05,
      "loss": 0.0602,
      "step": 10893
    },
    {
      "epoch": 20.735140275796482,
      "grad_norm": 0.04828595370054245,
      "learning_rate": 3.0841536995871704e-05,
      "loss": 0.0453,
      "step": 10894
    },
    {
      "epoch": 20.73704232049453,
      "grad_norm": 0.08964021503925323,
      "learning_rate": 3.0835185773261356e-05,
      "loss": 0.04,
      "step": 10895
    },
    {
      "epoch": 20.73894436519258,
      "grad_norm": 0.06883792579174042,
      "learning_rate": 3.0828834550651e-05,
      "loss": 0.0591,
      "step": 10896
    },
    {
      "epoch": 20.740846409890633,
      "grad_norm": 0.06266488879919052,
      "learning_rate": 3.0822483328040646e-05,
      "loss": 0.0572,
      "step": 10897
    },
    {
      "epoch": 20.742748454588682,
      "grad_norm": 0.12018706649541855,
      "learning_rate": 3.08161321054303e-05,
      "loss": 0.0529,
      "step": 10898
    },
    {
      "epoch": 20.744650499286735,
      "grad_norm": 0.11263476312160492,
      "learning_rate": 3.080978088281994e-05,
      "loss": 0.0564,
      "step": 10899
    },
    {
      "epoch": 20.746552543984784,
      "grad_norm": 0.05064086988568306,
      "learning_rate": 3.0803429660209595e-05,
      "loss": 0.051,
      "step": 10900
    },
    {
      "epoch": 20.748454588682833,
      "grad_norm": 0.05158840864896774,
      "learning_rate": 3.079707843759924e-05,
      "loss": 0.0469,
      "step": 10901
    },
    {
      "epoch": 20.750356633380886,
      "grad_norm": 0.1185307428240776,
      "learning_rate": 3.0790727214988885e-05,
      "loss": 0.059,
      "step": 10902
    },
    {
      "epoch": 20.752258678078935,
      "grad_norm": 0.09362038969993591,
      "learning_rate": 3.078437599237853e-05,
      "loss": 0.0561,
      "step": 10903
    },
    {
      "epoch": 20.754160722776984,
      "grad_norm": 0.11035136878490448,
      "learning_rate": 3.077802476976818e-05,
      "loss": 0.0342,
      "step": 10904
    },
    {
      "epoch": 20.756062767475036,
      "grad_norm": 0.1329907476902008,
      "learning_rate": 3.0771673547157834e-05,
      "loss": 0.0597,
      "step": 10905
    },
    {
      "epoch": 20.757964812173086,
      "grad_norm": 0.17069882154464722,
      "learning_rate": 3.076532232454747e-05,
      "loss": 0.038,
      "step": 10906
    },
    {
      "epoch": 20.759866856871135,
      "grad_norm": 0.0572471059858799,
      "learning_rate": 3.0758971101937124e-05,
      "loss": 0.0602,
      "step": 10907
    },
    {
      "epoch": 20.761768901569187,
      "grad_norm": 0.0768604725599289,
      "learning_rate": 3.075261987932677e-05,
      "loss": 0.0556,
      "step": 10908
    },
    {
      "epoch": 20.763670946267236,
      "grad_norm": 0.09952004253864288,
      "learning_rate": 3.074626865671642e-05,
      "loss": 0.0541,
      "step": 10909
    },
    {
      "epoch": 20.76557299096529,
      "grad_norm": 0.07463855296373367,
      "learning_rate": 3.0739917434106066e-05,
      "loss": 0.0424,
      "step": 10910
    },
    {
      "epoch": 20.767475035663338,
      "grad_norm": 0.04839564859867096,
      "learning_rate": 3.073356621149571e-05,
      "loss": 0.0537,
      "step": 10911
    },
    {
      "epoch": 20.769377080361387,
      "grad_norm": 0.13054612278938293,
      "learning_rate": 3.072721498888536e-05,
      "loss": 0.0589,
      "step": 10912
    },
    {
      "epoch": 20.77127912505944,
      "grad_norm": 0.05364089086651802,
      "learning_rate": 3.072086376627501e-05,
      "loss": 0.0448,
      "step": 10913
    },
    {
      "epoch": 20.77318116975749,
      "grad_norm": 0.05852840840816498,
      "learning_rate": 3.071451254366466e-05,
      "loss": 0.0526,
      "step": 10914
    },
    {
      "epoch": 20.775083214455538,
      "grad_norm": 0.08128932118415833,
      "learning_rate": 3.0708161321054305e-05,
      "loss": 0.0709,
      "step": 10915
    },
    {
      "epoch": 20.77698525915359,
      "grad_norm": 0.057719286531209946,
      "learning_rate": 3.070181009844395e-05,
      "loss": 0.0422,
      "step": 10916
    },
    {
      "epoch": 20.77888730385164,
      "grad_norm": 0.16810031235218048,
      "learning_rate": 3.06954588758336e-05,
      "loss": 0.0791,
      "step": 10917
    },
    {
      "epoch": 20.780789348549693,
      "grad_norm": 0.11165507137775421,
      "learning_rate": 3.068910765322325e-05,
      "loss": 0.0465,
      "step": 10918
    },
    {
      "epoch": 20.78269139324774,
      "grad_norm": 0.09537408500909805,
      "learning_rate": 3.06827564306129e-05,
      "loss": 0.0683,
      "step": 10919
    },
    {
      "epoch": 20.78459343794579,
      "grad_norm": 0.039117906242609024,
      "learning_rate": 3.067640520800254e-05,
      "loss": 0.0376,
      "step": 10920
    },
    {
      "epoch": 20.786495482643844,
      "grad_norm": 0.05061280354857445,
      "learning_rate": 3.067005398539219e-05,
      "loss": 0.0524,
      "step": 10921
    },
    {
      "epoch": 20.788397527341893,
      "grad_norm": 0.19125020503997803,
      "learning_rate": 3.066370276278184e-05,
      "loss": 0.0582,
      "step": 10922
    },
    {
      "epoch": 20.79029957203994,
      "grad_norm": 0.07217196375131607,
      "learning_rate": 3.0657351540171486e-05,
      "loss": 0.0569,
      "step": 10923
    },
    {
      "epoch": 20.792201616737994,
      "grad_norm": 0.04245905950665474,
      "learning_rate": 3.065100031756113e-05,
      "loss": 0.0615,
      "step": 10924
    },
    {
      "epoch": 20.794103661436043,
      "grad_norm": 0.12768147885799408,
      "learning_rate": 3.0644649094950776e-05,
      "loss": 0.0597,
      "step": 10925
    },
    {
      "epoch": 20.796005706134093,
      "grad_norm": 0.21361514925956726,
      "learning_rate": 3.063829787234043e-05,
      "loss": 0.0492,
      "step": 10926
    },
    {
      "epoch": 20.797907750832145,
      "grad_norm": 0.07605758309364319,
      "learning_rate": 3.063194664973007e-05,
      "loss": 0.0569,
      "step": 10927
    },
    {
      "epoch": 20.799809795530194,
      "grad_norm": 0.2108287811279297,
      "learning_rate": 3.0625595427119725e-05,
      "loss": 0.0536,
      "step": 10928
    },
    {
      "epoch": 20.801711840228247,
      "grad_norm": 0.08548277616500854,
      "learning_rate": 3.061924420450937e-05,
      "loss": 0.047,
      "step": 10929
    },
    {
      "epoch": 20.803613884926296,
      "grad_norm": 0.14243417978286743,
      "learning_rate": 3.0612892981899015e-05,
      "loss": 0.0505,
      "step": 10930
    },
    {
      "epoch": 20.805515929624345,
      "grad_norm": 0.12442498654127121,
      "learning_rate": 3.060654175928867e-05,
      "loss": 0.0621,
      "step": 10931
    },
    {
      "epoch": 20.807417974322398,
      "grad_norm": 0.11285270750522614,
      "learning_rate": 3.060019053667831e-05,
      "loss": 0.0724,
      "step": 10932
    },
    {
      "epoch": 20.809320019020447,
      "grad_norm": 0.1174018383026123,
      "learning_rate": 3.059383931406796e-05,
      "loss": 0.0671,
      "step": 10933
    },
    {
      "epoch": 20.811222063718496,
      "grad_norm": 0.048765238374471664,
      "learning_rate": 3.058748809145761e-05,
      "loss": 0.0535,
      "step": 10934
    },
    {
      "epoch": 20.81312410841655,
      "grad_norm": 0.11892572790384293,
      "learning_rate": 3.0581136868847254e-05,
      "loss": 0.0509,
      "step": 10935
    },
    {
      "epoch": 20.815026153114598,
      "grad_norm": 0.11556268483400345,
      "learning_rate": 3.0574785646236906e-05,
      "loss": 0.056,
      "step": 10936
    },
    {
      "epoch": 20.816928197812647,
      "grad_norm": 0.12910594046115875,
      "learning_rate": 3.056843442362655e-05,
      "loss": 0.0657,
      "step": 10937
    },
    {
      "epoch": 20.8188302425107,
      "grad_norm": 0.17326128482818604,
      "learning_rate": 3.0562083201016196e-05,
      "loss": 0.0569,
      "step": 10938
    },
    {
      "epoch": 20.82073228720875,
      "grad_norm": 0.16716662049293518,
      "learning_rate": 3.055573197840584e-05,
      "loss": 0.0531,
      "step": 10939
    },
    {
      "epoch": 20.8226343319068,
      "grad_norm": 0.06517747044563293,
      "learning_rate": 3.054938075579549e-05,
      "loss": 0.0406,
      "step": 10940
    },
    {
      "epoch": 20.82453637660485,
      "grad_norm": 0.07059972733259201,
      "learning_rate": 3.0543029533185144e-05,
      "loss": 0.0558,
      "step": 10941
    },
    {
      "epoch": 20.8264384213029,
      "grad_norm": 0.05612773075699806,
      "learning_rate": 3.053667831057478e-05,
      "loss": 0.0493,
      "step": 10942
    },
    {
      "epoch": 20.828340466000952,
      "grad_norm": 0.169120654463768,
      "learning_rate": 3.0530327087964435e-05,
      "loss": 0.056,
      "step": 10943
    },
    {
      "epoch": 20.830242510699,
      "grad_norm": 0.07485458999872208,
      "learning_rate": 3.052397586535408e-05,
      "loss": 0.0534,
      "step": 10944
    },
    {
      "epoch": 20.83214455539705,
      "grad_norm": 0.08926668763160706,
      "learning_rate": 3.051762464274373e-05,
      "loss": 0.049,
      "step": 10945
    },
    {
      "epoch": 20.834046600095103,
      "grad_norm": 0.12054421752691269,
      "learning_rate": 3.051127342013338e-05,
      "loss": 0.0455,
      "step": 10946
    },
    {
      "epoch": 20.835948644793152,
      "grad_norm": 0.04459091275930405,
      "learning_rate": 3.050492219752302e-05,
      "loss": 0.0416,
      "step": 10947
    },
    {
      "epoch": 20.8378506894912,
      "grad_norm": 0.057703930884599686,
      "learning_rate": 3.049857097491267e-05,
      "loss": 0.07,
      "step": 10948
    },
    {
      "epoch": 20.839752734189254,
      "grad_norm": 0.13715192675590515,
      "learning_rate": 3.049221975230232e-05,
      "loss": 0.0488,
      "step": 10949
    },
    {
      "epoch": 20.841654778887303,
      "grad_norm": 0.0581369549036026,
      "learning_rate": 3.048586852969197e-05,
      "loss": 0.0609,
      "step": 10950
    },
    {
      "epoch": 20.843556823585356,
      "grad_norm": 0.2517063319683075,
      "learning_rate": 3.0479517307081612e-05,
      "loss": 0.0628,
      "step": 10951
    },
    {
      "epoch": 20.845458868283405,
      "grad_norm": 0.05997535586357117,
      "learning_rate": 3.047316608447126e-05,
      "loss": 0.0546,
      "step": 10952
    },
    {
      "epoch": 20.847360912981454,
      "grad_norm": 0.09997560828924179,
      "learning_rate": 3.046681486186091e-05,
      "loss": 0.0532,
      "step": 10953
    },
    {
      "epoch": 20.849262957679507,
      "grad_norm": 0.15614329278469086,
      "learning_rate": 3.0460463639250557e-05,
      "loss": 0.0592,
      "step": 10954
    },
    {
      "epoch": 20.851165002377556,
      "grad_norm": 0.09338696300983429,
      "learning_rate": 3.0454112416640206e-05,
      "loss": 0.0468,
      "step": 10955
    },
    {
      "epoch": 20.853067047075605,
      "grad_norm": 0.04714600369334221,
      "learning_rate": 3.044776119402985e-05,
      "loss": 0.0514,
      "step": 10956
    },
    {
      "epoch": 20.854969091773658,
      "grad_norm": 0.12118776142597198,
      "learning_rate": 3.04414099714195e-05,
      "loss": 0.06,
      "step": 10957
    },
    {
      "epoch": 20.856871136471707,
      "grad_norm": 0.053253013640642166,
      "learning_rate": 3.0435058748809148e-05,
      "loss": 0.0443,
      "step": 10958
    },
    {
      "epoch": 20.858773181169756,
      "grad_norm": 0.08597280830144882,
      "learning_rate": 3.0428707526198796e-05,
      "loss": 0.049,
      "step": 10959
    },
    {
      "epoch": 20.86067522586781,
      "grad_norm": 0.045080091804265976,
      "learning_rate": 3.0422356303588438e-05,
      "loss": 0.0498,
      "step": 10960
    },
    {
      "epoch": 20.862577270565858,
      "grad_norm": 0.09501554816961288,
      "learning_rate": 3.041600508097809e-05,
      "loss": 0.0977,
      "step": 10961
    },
    {
      "epoch": 20.86447931526391,
      "grad_norm": 0.13913887739181519,
      "learning_rate": 3.0409653858367738e-05,
      "loss": 0.0466,
      "step": 10962
    },
    {
      "epoch": 20.86638135996196,
      "grad_norm": 0.1571308970451355,
      "learning_rate": 3.0403302635757387e-05,
      "loss": 0.0637,
      "step": 10963
    },
    {
      "epoch": 20.86828340466001,
      "grad_norm": 0.10953457653522491,
      "learning_rate": 3.0396951413147035e-05,
      "loss": 0.0535,
      "step": 10964
    },
    {
      "epoch": 20.87018544935806,
      "grad_norm": 0.08388474583625793,
      "learning_rate": 3.0390600190536677e-05,
      "loss": 0.0448,
      "step": 10965
    },
    {
      "epoch": 20.87208749405611,
      "grad_norm": 0.07518678158521652,
      "learning_rate": 3.0384248967926325e-05,
      "loss": 0.0478,
      "step": 10966
    },
    {
      "epoch": 20.87398953875416,
      "grad_norm": 0.1467273086309433,
      "learning_rate": 3.0377897745315974e-05,
      "loss": 0.0452,
      "step": 10967
    },
    {
      "epoch": 20.875891583452212,
      "grad_norm": 0.05250808969140053,
      "learning_rate": 3.0371546522705626e-05,
      "loss": 0.0577,
      "step": 10968
    },
    {
      "epoch": 20.87779362815026,
      "grad_norm": 0.07057197391986847,
      "learning_rate": 3.0365195300095267e-05,
      "loss": 0.049,
      "step": 10969
    },
    {
      "epoch": 20.87969567284831,
      "grad_norm": 0.0900682657957077,
      "learning_rate": 3.0358844077484916e-05,
      "loss": 0.0478,
      "step": 10970
    },
    {
      "epoch": 20.881597717546363,
      "grad_norm": 0.16264529526233673,
      "learning_rate": 3.0352492854874564e-05,
      "loss": 0.062,
      "step": 10971
    },
    {
      "epoch": 20.883499762244412,
      "grad_norm": 0.10332495719194412,
      "learning_rate": 3.0346141632264213e-05,
      "loss": 0.065,
      "step": 10972
    },
    {
      "epoch": 20.885401806942465,
      "grad_norm": 0.04506974667310715,
      "learning_rate": 3.033979040965386e-05,
      "loss": 0.046,
      "step": 10973
    },
    {
      "epoch": 20.887303851640514,
      "grad_norm": 0.1779114454984665,
      "learning_rate": 3.0333439187043506e-05,
      "loss": 0.0544,
      "step": 10974
    },
    {
      "epoch": 20.889205896338563,
      "grad_norm": 0.07354097068309784,
      "learning_rate": 3.0327087964433155e-05,
      "loss": 0.061,
      "step": 10975
    },
    {
      "epoch": 20.891107941036616,
      "grad_norm": 0.0837244987487793,
      "learning_rate": 3.0320736741822803e-05,
      "loss": 0.0682,
      "step": 10976
    },
    {
      "epoch": 20.893009985734665,
      "grad_norm": 0.03713522478938103,
      "learning_rate": 3.031438551921245e-05,
      "loss": 0.0564,
      "step": 10977
    },
    {
      "epoch": 20.894912030432714,
      "grad_norm": 0.09520062804222107,
      "learning_rate": 3.0308034296602093e-05,
      "loss": 0.0504,
      "step": 10978
    },
    {
      "epoch": 20.896814075130766,
      "grad_norm": 0.10911743342876434,
      "learning_rate": 3.0301683073991745e-05,
      "loss": 0.0537,
      "step": 10979
    },
    {
      "epoch": 20.898716119828816,
      "grad_norm": 0.07108257710933685,
      "learning_rate": 3.0295331851381393e-05,
      "loss": 0.0442,
      "step": 10980
    },
    {
      "epoch": 20.900618164526865,
      "grad_norm": 0.12098158895969391,
      "learning_rate": 3.0288980628771042e-05,
      "loss": 0.0633,
      "step": 10981
    },
    {
      "epoch": 20.902520209224917,
      "grad_norm": 0.06423678994178772,
      "learning_rate": 3.028262940616069e-05,
      "loss": 0.0583,
      "step": 10982
    },
    {
      "epoch": 20.904422253922966,
      "grad_norm": 0.20237389206886292,
      "learning_rate": 3.0276278183550332e-05,
      "loss": 0.0751,
      "step": 10983
    },
    {
      "epoch": 20.90632429862102,
      "grad_norm": 0.24858830869197845,
      "learning_rate": 3.026992696093998e-05,
      "loss": 0.0766,
      "step": 10984
    },
    {
      "epoch": 20.90822634331907,
      "grad_norm": 0.03466055914759636,
      "learning_rate": 3.026357573832963e-05,
      "loss": 0.0486,
      "step": 10985
    },
    {
      "epoch": 20.910128388017117,
      "grad_norm": 0.12383092939853668,
      "learning_rate": 3.025722451571928e-05,
      "loss": 0.046,
      "step": 10986
    },
    {
      "epoch": 20.91203043271517,
      "grad_norm": 0.06242097169160843,
      "learning_rate": 3.0250873293108922e-05,
      "loss": 0.0532,
      "step": 10987
    },
    {
      "epoch": 20.91393247741322,
      "grad_norm": 0.09692638367414474,
      "learning_rate": 3.024452207049857e-05,
      "loss": 0.0585,
      "step": 10988
    },
    {
      "epoch": 20.91583452211127,
      "grad_norm": 0.10633906722068787,
      "learning_rate": 3.023817084788822e-05,
      "loss": 0.0508,
      "step": 10989
    },
    {
      "epoch": 20.91773656680932,
      "grad_norm": 0.08840779960155487,
      "learning_rate": 3.0231819625277868e-05,
      "loss": 0.0547,
      "step": 10990
    },
    {
      "epoch": 20.91963861150737,
      "grad_norm": 0.08700352907180786,
      "learning_rate": 3.0225468402667516e-05,
      "loss": 0.0617,
      "step": 10991
    },
    {
      "epoch": 20.92154065620542,
      "grad_norm": 0.08046238124370575,
      "learning_rate": 3.021911718005716e-05,
      "loss": 0.0522,
      "step": 10992
    },
    {
      "epoch": 20.923442700903472,
      "grad_norm": 0.17197772860527039,
      "learning_rate": 3.021276595744681e-05,
      "loss": 0.0668,
      "step": 10993
    },
    {
      "epoch": 20.92534474560152,
      "grad_norm": 0.15205800533294678,
      "learning_rate": 3.0206414734836458e-05,
      "loss": 0.0748,
      "step": 10994
    },
    {
      "epoch": 20.927246790299574,
      "grad_norm": 0.11835450679063797,
      "learning_rate": 3.0200063512226107e-05,
      "loss": 0.0563,
      "step": 10995
    },
    {
      "epoch": 20.929148834997623,
      "grad_norm": 0.09660196304321289,
      "learning_rate": 3.019371228961575e-05,
      "loss": 0.0703,
      "step": 10996
    },
    {
      "epoch": 20.931050879695672,
      "grad_norm": 0.04566892236471176,
      "learning_rate": 3.01873610670054e-05,
      "loss": 0.0536,
      "step": 10997
    },
    {
      "epoch": 20.932952924393724,
      "grad_norm": 0.11894775927066803,
      "learning_rate": 3.018100984439505e-05,
      "loss": 0.0624,
      "step": 10998
    },
    {
      "epoch": 20.934854969091774,
      "grad_norm": 0.05920516699552536,
      "learning_rate": 3.0174658621784697e-05,
      "loss": 0.0709,
      "step": 10999
    },
    {
      "epoch": 20.936757013789823,
      "grad_norm": 0.037830401211977005,
      "learning_rate": 3.0168307399174346e-05,
      "loss": 0.0576,
      "step": 11000
    },
    {
      "epoch": 20.938659058487875,
      "grad_norm": 0.29501423239707947,
      "learning_rate": 3.0161956176563987e-05,
      "loss": 0.0718,
      "step": 11001
    },
    {
      "epoch": 20.940561103185924,
      "grad_norm": 0.17233391106128693,
      "learning_rate": 3.0155604953953636e-05,
      "loss": 0.0622,
      "step": 11002
    },
    {
      "epoch": 20.942463147883974,
      "grad_norm": 0.10794918239116669,
      "learning_rate": 3.0149253731343284e-05,
      "loss": 0.0566,
      "step": 11003
    },
    {
      "epoch": 20.944365192582026,
      "grad_norm": 0.08928713202476501,
      "learning_rate": 3.0142902508732933e-05,
      "loss": 0.0667,
      "step": 11004
    },
    {
      "epoch": 20.946267237280075,
      "grad_norm": 0.11571650952100754,
      "learning_rate": 3.0136551286122578e-05,
      "loss": 0.048,
      "step": 11005
    },
    {
      "epoch": 20.948169281978128,
      "grad_norm": 0.09128861874341965,
      "learning_rate": 3.0130200063512226e-05,
      "loss": 0.0675,
      "step": 11006
    },
    {
      "epoch": 20.950071326676177,
      "grad_norm": 0.06981503218412399,
      "learning_rate": 3.0123848840901875e-05,
      "loss": 0.0471,
      "step": 11007
    },
    {
      "epoch": 20.951973371374226,
      "grad_norm": 0.0702807605266571,
      "learning_rate": 3.0117497618291523e-05,
      "loss": 0.0706,
      "step": 11008
    },
    {
      "epoch": 20.95387541607228,
      "grad_norm": 0.07049190998077393,
      "learning_rate": 3.011114639568117e-05,
      "loss": 0.0763,
      "step": 11009
    },
    {
      "epoch": 20.955777460770328,
      "grad_norm": 0.07275652885437012,
      "learning_rate": 3.0104795173070817e-05,
      "loss": 0.0522,
      "step": 11010
    },
    {
      "epoch": 20.957679505468377,
      "grad_norm": 0.10747750848531723,
      "learning_rate": 3.0098443950460465e-05,
      "loss": 0.0555,
      "step": 11011
    },
    {
      "epoch": 20.95958155016643,
      "grad_norm": 0.20730341970920563,
      "learning_rate": 3.0092092727850113e-05,
      "loss": 0.0609,
      "step": 11012
    },
    {
      "epoch": 20.96148359486448,
      "grad_norm": 0.1200687512755394,
      "learning_rate": 3.0085741505239762e-05,
      "loss": 0.0621,
      "step": 11013
    },
    {
      "epoch": 20.96338563956253,
      "grad_norm": 0.04054923355579376,
      "learning_rate": 3.0079390282629404e-05,
      "loss": 0.0502,
      "step": 11014
    },
    {
      "epoch": 20.96528768426058,
      "grad_norm": 0.08221112191677094,
      "learning_rate": 3.0073039060019052e-05,
      "loss": 0.0436,
      "step": 11015
    },
    {
      "epoch": 20.96718972895863,
      "grad_norm": 0.134440079331398,
      "learning_rate": 3.0066687837408704e-05,
      "loss": 0.0751,
      "step": 11016
    },
    {
      "epoch": 20.969091773656682,
      "grad_norm": 0.07171544432640076,
      "learning_rate": 3.0060336614798352e-05,
      "loss": 0.0511,
      "step": 11017
    },
    {
      "epoch": 20.97099381835473,
      "grad_norm": 0.07600289583206177,
      "learning_rate": 3.0053985392188e-05,
      "loss": 0.0471,
      "step": 11018
    },
    {
      "epoch": 20.97289586305278,
      "grad_norm": 0.06850898265838623,
      "learning_rate": 3.0047634169577642e-05,
      "loss": 0.0529,
      "step": 11019
    },
    {
      "epoch": 20.974797907750833,
      "grad_norm": 0.19148507714271545,
      "learning_rate": 3.004128294696729e-05,
      "loss": 0.0604,
      "step": 11020
    },
    {
      "epoch": 20.976699952448882,
      "grad_norm": 0.06374479085206985,
      "learning_rate": 3.003493172435694e-05,
      "loss": 0.0543,
      "step": 11021
    },
    {
      "epoch": 20.97860199714693,
      "grad_norm": 0.16483956575393677,
      "learning_rate": 3.0028580501746588e-05,
      "loss": 0.0646,
      "step": 11022
    },
    {
      "epoch": 20.980504041844984,
      "grad_norm": 0.2773119807243347,
      "learning_rate": 3.0022229279136233e-05,
      "loss": 0.0552,
      "step": 11023
    },
    {
      "epoch": 20.982406086543033,
      "grad_norm": 0.03720984607934952,
      "learning_rate": 3.001587805652588e-05,
      "loss": 0.0346,
      "step": 11024
    },
    {
      "epoch": 20.984308131241086,
      "grad_norm": 0.05585749074816704,
      "learning_rate": 3.000952683391553e-05,
      "loss": 0.0461,
      "step": 11025
    }
  ],
  "logging_steps": 1,
  "max_steps": 15750,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 30,
  "save_steps": 525,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.8240761081897124e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}